2006-12-31 01:02:27 +01:00
|
|
|
/* Copyright (C) 2000-2006 MySQL AB
|
2001-12-06 13:10:51 +01:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
This program is free software; you can redistribute it and/or modify
|
|
|
|
it under the terms of the GNU General Public License as published by
|
2006-12-23 20:17:15 +01:00
|
|
|
the Free Software Foundation; version 2 of the License.
|
2001-12-06 13:10:51 +01:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
This program is distributed in the hope that it will be useful,
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
GNU General Public License for more details.
|
2001-12-06 13:10:51 +01:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
You should have received a copy of the GNU General Public License
|
|
|
|
along with this program; if not, write to the Free Software
|
|
|
|
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */
|
|
|
|
|
|
|
|
|
|
|
|
/* A lexical scanner on a temporary buffer with a yacc interface */
|
|
|
|
|
2006-03-09 19:09:52 +01:00
|
|
|
#define MYSQL_LEX 1
|
2000-07-31 21:29:14 +02:00
|
|
|
#include "mysql_priv.h"
|
|
|
|
#include "item_create.h"
|
|
|
|
#include <m_ctype.h>
|
|
|
|
#include <hash.h>
|
2003-02-26 19:22:29 +01:00
|
|
|
#include "sp.h"
|
|
|
|
#include "sp_head.h"
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2004-09-07 14:29:46 +02:00
|
|
|
/*
|
|
|
|
We are using pointer to this variable for distinguishing between assignment
|
|
|
|
to NEW row field (when parsing trigger definition) and structured variable.
|
|
|
|
*/
|
2005-11-06 01:36:40 +01:00
|
|
|
|
|
|
|
sys_var *trg_new_row_fake_var= (sys_var*) 0x01;
|
2004-08-10 10:42:31 +02:00
|
|
|
|
2003-11-03 13:01:59 +01:00
|
|
|
/* Longest standard keyword name */
|
2000-07-31 21:29:14 +02:00
|
|
|
#define TOCK_NAME_LENGTH 24
|
|
|
|
|
2003-11-03 13:01:59 +01:00
|
|
|
/*
|
|
|
|
The following data is based on the latin1 character set, and is only
|
2000-07-31 21:29:14 +02:00
|
|
|
used when comparing keywords
|
|
|
|
*/
|
|
|
|
|
2005-05-18 18:00:21 +02:00
|
|
|
static uchar to_upper_lex[]=
|
|
|
|
{
|
2000-07-31 21:29:14 +02:00
|
|
|
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
|
|
|
|
16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31,
|
|
|
|
32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47,
|
|
|
|
48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63,
|
|
|
|
64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79,
|
|
|
|
80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95,
|
|
|
|
96, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79,
|
|
|
|
80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90,123,124,125,126,127,
|
|
|
|
128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,
|
|
|
|
144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,
|
|
|
|
160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,
|
|
|
|
176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,
|
|
|
|
192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,
|
|
|
|
208,209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,
|
|
|
|
192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,
|
|
|
|
208,209,210,211,212,213,214,247,216,217,218,219,220,221,222,255
|
|
|
|
};
|
|
|
|
|
2007-03-05 18:08:41 +01:00
|
|
|
/*
|
|
|
|
Names of the index hints (for error messages). Keep in sync with
|
|
|
|
index_hint_type
|
|
|
|
*/
|
|
|
|
|
|
|
|
const char * index_hint_type_name[] =
|
|
|
|
{
|
|
|
|
"IGNORE INDEX",
|
|
|
|
"USE INDEX",
|
|
|
|
"FORCE INDEX"
|
|
|
|
};
|
2003-11-03 13:01:59 +01:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
inline int lex_casecmp(const char *s, const char *t, uint len)
|
|
|
|
{
|
|
|
|
while (len-- != 0 &&
|
|
|
|
to_upper_lex[(uchar) *s++] == to_upper_lex[(uchar) *t++]) ;
|
|
|
|
return (int) len+1;
|
|
|
|
}
|
|
|
|
|
2006-04-13 09:57:51 +02:00
|
|
|
#include <lex_hash.h>
|
2000-07-31 21:29:14 +02:00
|
|
|
|
|
|
|
|
|
|
|
void lex_init(void)
|
|
|
|
{
|
|
|
|
uint i;
|
|
|
|
DBUG_ENTER("lex_init");
|
|
|
|
for (i=0 ; i < array_elements(symbols) ; i++)
|
|
|
|
symbols[i].length=(uchar) strlen(symbols[i].name);
|
|
|
|
for (i=0 ; i < array_elements(sql_functions) ; i++)
|
|
|
|
sql_functions[i].length=(uchar) strlen(sql_functions[i].name);
|
|
|
|
|
|
|
|
DBUG_VOID_RETURN;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void lex_free(void)
|
|
|
|
{ // Call this when daemon ends
|
|
|
|
DBUG_ENTER("lex_free");
|
|
|
|
DBUG_VOID_RETURN;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2007-02-04 14:49:24 +01:00
|
|
|
void
|
|
|
|
st_parsing_options::reset()
|
|
|
|
{
|
|
|
|
allows_variable= TRUE;
|
|
|
|
allows_select_into= TRUE;
|
|
|
|
allows_select_procedure= TRUE;
|
|
|
|
allows_derived= TRUE;
|
|
|
|
}
|
|
|
|
|
2007-04-26 05:38:12 +02:00
|
|
|
Lex_input_stream::Lex_input_stream(THD *thd,
|
|
|
|
const char* buffer,
|
|
|
|
unsigned int length)
|
|
|
|
: m_thd(thd),
|
|
|
|
yylineno(1),
|
|
|
|
yytoklen(0),
|
|
|
|
yylval(NULL),
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
m_ptr(buffer),
|
|
|
|
m_tok_start(NULL),
|
|
|
|
m_tok_end(NULL),
|
|
|
|
m_end_of_query(buffer + length),
|
|
|
|
m_tok_start_prev(NULL),
|
|
|
|
m_buf(buffer),
|
|
|
|
m_echo(true),
|
|
|
|
m_cpp_tok_start(NULL),
|
|
|
|
m_cpp_tok_start_prev(NULL),
|
|
|
|
m_cpp_tok_end(NULL),
|
2007-04-26 05:38:12 +02:00
|
|
|
next_state(MY_LEX_START),
|
|
|
|
found_semicolon(NULL),
|
2007-05-11 15:26:12 +02:00
|
|
|
ignore_space(test(thd->variables.sql_mode & MODE_IGNORE_SPACE)),
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
stmt_prepare_mode(FALSE),
|
|
|
|
in_comment(NO_COMMENT)
|
2007-04-26 05:38:12 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
m_cpp_buf= (char*) thd->alloc(length + 1);
|
|
|
|
m_cpp_ptr= m_cpp_buf;
|
2007-04-26 05:38:12 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
Lex_input_stream::~Lex_input_stream()
|
|
|
|
{}
|
|
|
|
|
2007-02-04 14:49:24 +01:00
|
|
|
|
2003-02-12 20:55:37 +01:00
|
|
|
/*
|
|
|
|
This is called before every query that is to be parsed.
|
|
|
|
Because of this, it's critical to not do too much things here.
|
|
|
|
(We already do too much here)
|
|
|
|
*/
|
|
|
|
|
2007-04-26 05:38:12 +02:00
|
|
|
void lex_start(THD *thd)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2003-05-05 20:54:37 +02:00
|
|
|
LEX *lex= thd->lex;
|
2005-02-05 15:05:46 +01:00
|
|
|
DBUG_ENTER("lex_start");
|
|
|
|
|
|
|
|
lex->thd= lex->unit.thd= thd;
|
|
|
|
|
2005-08-12 16:57:19 +02:00
|
|
|
lex->context_stack.empty();
|
2004-10-14 00:53:59 +02:00
|
|
|
lex->unit.init_query();
|
|
|
|
lex->unit.init_select();
|
2005-08-12 16:57:19 +02:00
|
|
|
/* 'parent_lex' is used in init_query() so it must be before it. */
|
|
|
|
lex->select_lex.parent_lex= lex;
|
2004-10-14 00:53:59 +02:00
|
|
|
lex->select_lex.init_query();
|
|
|
|
lex->value_list.empty();
|
2004-12-13 13:26:28 +01:00
|
|
|
lex->update_list.empty();
|
2004-10-14 00:53:59 +02:00
|
|
|
lex->param_list.empty();
|
2004-10-29 18:26:52 +02:00
|
|
|
lex->view_list.empty();
|
Implement WL#2661 "Prepared Statements: Dynamic SQL in Stored Procedures".
The idea of the patch is to separate statement processing logic,
such as parsing, validation of the parsed tree, execution and cleanup,
from global query processing logic, such as logging, resetting
priorities of a thread, resetting stored procedure cache, resetting
thread count of errors and warnings.
This makes PREPARE and EXECUTE behave similarly to the rest of SQL
statements and allows their use in stored procedures.
This patch contains a change in behaviour:
until recently for each SQL prepared statement command, 2 queries
were written to the general log, e.g.
[Query] prepare stmt from @stmt_text;
[Prepare] select * from t1 <-- contents of @stmt_text
The chagne was necessary to prevent [Prepare] commands from being written
to the general log when executing a stored procedure with Dynamic SQL.
We should consider whether the old behavior is preferrable and probably
restore it.
This patch refixes Bug#7115, Bug#10975 (partially), Bug#10605 (various bugs
in Dynamic SQL reported before it was disabled).
2005-09-03 01:13:18 +02:00
|
|
|
lex->prepared_stmt_params.empty();
|
2006-07-11 21:39:51 +02:00
|
|
|
lex->auxiliary_table_list.empty();
|
2004-10-14 00:53:59 +02:00
|
|
|
lex->unit.next= lex->unit.master=
|
|
|
|
lex->unit.link_next= lex->unit.return_to= 0;
|
|
|
|
lex->unit.prev= lex->unit.link_prev= 0;
|
|
|
|
lex->unit.slave= lex->unit.global_parameters= lex->current_select=
|
|
|
|
lex->all_selects_list= &lex->select_lex;
|
|
|
|
lex->select_lex.master= &lex->unit;
|
|
|
|
lex->select_lex.prev= &lex->unit.slave;
|
|
|
|
lex->select_lex.link_next= lex->select_lex.slave= lex->select_lex.next= 0;
|
|
|
|
lex->select_lex.link_prev= (st_select_lex_node**)&(lex->all_selects_list);
|
|
|
|
lex->select_lex.options= 0;
|
2006-06-27 19:28:32 +02:00
|
|
|
lex->select_lex.sql_cache= SELECT_LEX::SQL_CACHE_UNSPECIFIED;
|
2004-10-29 18:26:52 +02:00
|
|
|
lex->select_lex.init_order();
|
|
|
|
lex->select_lex.group_list.empty();
|
2004-10-14 00:53:59 +02:00
|
|
|
lex->describe= 0;
|
2004-12-06 16:15:54 +01:00
|
|
|
lex->subqueries= FALSE;
|
2004-10-29 18:26:52 +02:00
|
|
|
lex->view_prepare_mode= FALSE;
|
2004-12-06 16:15:54 +01:00
|
|
|
lex->derived_tables= 0;
|
2004-10-14 00:53:59 +02:00
|
|
|
lex->lock_option= TL_READ;
|
|
|
|
lex->safe_to_cache_query= 1;
|
2006-05-30 07:45:23 +02:00
|
|
|
lex->leaf_tables_insert= 0;
|
2007-02-04 14:49:24 +01:00
|
|
|
lex->parsing_options.reset();
|
2004-10-29 18:26:52 +02:00
|
|
|
lex->empty_field_list_on_rset= 0;
|
2004-10-14 00:53:59 +02:00
|
|
|
lex->select_lex.select_number= 1;
|
2000-07-31 21:29:14 +02:00
|
|
|
lex->length=0;
|
2005-07-18 13:31:02 +02:00
|
|
|
lex->part_info= 0;
|
2002-10-03 15:35:08 +02:00
|
|
|
lex->select_lex.in_sum_expr=0;
|
|
|
|
lex->select_lex.expr_list.empty();
|
2002-10-04 13:15:59 +02:00
|
|
|
lex->select_lex.ftfunc_list_alloc.empty();
|
2002-10-30 12:18:52 +01:00
|
|
|
lex->select_lex.ftfunc_list= &lex->select_lex.ftfunc_list_alloc;
|
2004-01-21 12:47:47 +01:00
|
|
|
lex->select_lex.group_list.empty();
|
|
|
|
lex->select_lex.order_list.empty();
|
2006-06-20 12:20:32 +02:00
|
|
|
lex->sql_command= SQLCOM_END;
|
2003-06-25 00:19:09 +02:00
|
|
|
lex->duplicates= DUP_ERROR;
|
2004-12-31 11:04:35 +01:00
|
|
|
lex->ignore= 0;
|
2006-07-13 10:59:58 +02:00
|
|
|
lex->spname= NULL;
|
Simplistic, experimental framework for Stored Procedures (SPs).
Implements creation and dropping of PROCEDUREs, IN, OUT, and INOUT parameters,
single-statement procedures, rudimentary multi-statement (begin-end) prodedures
(when the client can handle it), and local variables.
Missing most of the embedded SQL language, all attributes, FUNCTIONs, error handling,
reparses procedures at each call (no caching), etc, etc.
Certainly buggy too, but procedures can actually be created and called....
2002-12-08 19:59:22 +01:00
|
|
|
lex->sphead= NULL;
|
|
|
|
lex->spcont= NULL;
|
2004-11-05 16:29:47 +01:00
|
|
|
lex->proc_list.first= 0;
|
2006-06-29 00:42:25 +02:00
|
|
|
lex->escape_used= FALSE;
|
2007-03-02 17:43:45 +01:00
|
|
|
lex->query_tables= 0;
|
2006-05-30 07:45:23 +02:00
|
|
|
lex->reset_query_tables_list(FALSE);
|
2006-08-23 16:53:04 +02:00
|
|
|
lex->expr_allows_subselect= TRUE;
|
2005-12-02 13:07:02 +01:00
|
|
|
|
2006-10-16 18:57:33 +02:00
|
|
|
lex->name.str= 0;
|
|
|
|
lex->name.length= 0;
|
2006-06-29 00:42:25 +02:00
|
|
|
lex->event_parse_data= NULL;
|
2003-07-07 14:55:10 +02:00
|
|
|
|
2005-10-15 23:32:37 +02:00
|
|
|
lex->nest_level=0 ;
|
|
|
|
lex->allow_sum_func= 0;
|
|
|
|
lex->in_sum_func= NULL;
|
2006-12-02 01:47:45 +01:00
|
|
|
/*
|
|
|
|
ok, there must be a better solution for this, long-term
|
|
|
|
I tried "bzero" in the sql_yacc.yy code, but that for
|
|
|
|
some reason made the values zero, even if they were set
|
|
|
|
*/
|
|
|
|
lex->server_options.server_name= 0;
|
|
|
|
lex->server_options.server_name_length= 0;
|
|
|
|
lex->server_options.host= 0;
|
|
|
|
lex->server_options.db= 0;
|
|
|
|
lex->server_options.username= 0;
|
|
|
|
lex->server_options.password= 0;
|
|
|
|
lex->server_options.scheme= 0;
|
|
|
|
lex->server_options.socket= 0;
|
|
|
|
lex->server_options.owner= 0;
|
|
|
|
lex->server_options.port= -1;
|
2005-02-05 15:05:46 +01:00
|
|
|
DBUG_VOID_RETURN;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
void lex_end(LEX *lex)
|
|
|
|
{
|
2006-05-04 18:39:47 +02:00
|
|
|
DBUG_ENTER("lex_end");
|
2006-05-04 21:19:31 +02:00
|
|
|
DBUG_PRINT("enter", ("lex: 0x%lx", (long) lex));
|
2006-05-04 18:39:47 +02:00
|
|
|
if (lex->yacc_yyss)
|
|
|
|
{
|
|
|
|
my_free(lex->yacc_yyss, MYF(0));
|
|
|
|
my_free(lex->yacc_yyvs, MYF(0));
|
|
|
|
lex->yacc_yyss= 0;
|
|
|
|
lex->yacc_yyvs= 0;
|
|
|
|
}
|
2007-03-02 17:43:45 +01:00
|
|
|
|
|
|
|
/* release used plugins */
|
|
|
|
plugin_unlock_list(0, (plugin_ref*)lex->plugins.buffer,
|
|
|
|
lex->plugins.elements);
|
|
|
|
reset_dynamic(&lex->plugins);
|
|
|
|
|
2006-05-04 18:39:47 +02:00
|
|
|
DBUG_VOID_RETURN;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2007-04-26 05:38:12 +02:00
|
|
|
static int find_keyword(Lex_input_stream *lip, uint len, bool function)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
const char *tok= lip->get_tok_start();
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2007-03-27 19:09:56 +02:00
|
|
|
SYMBOL *symbol= get_hash_symbol(tok, len, function);
|
2000-07-31 21:29:14 +02:00
|
|
|
if (symbol)
|
|
|
|
{
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->yylval->symbol.symbol=symbol;
|
|
|
|
lip->yylval->symbol.str= (char*) tok;
|
|
|
|
lip->yylval->symbol.length=len;
|
|
|
|
|
2004-11-17 16:49:10 +01:00
|
|
|
if ((symbol->tok == NOT_SYM) &&
|
2007-04-26 05:38:12 +02:00
|
|
|
(lip->m_thd->variables.sql_mode & MODE_HIGH_NOT_PRECEDENCE))
|
2004-11-17 16:49:10 +01:00
|
|
|
return NOT2_SYM;
|
|
|
|
if ((symbol->tok == OR_OR_SYM) &&
|
2007-04-26 05:38:12 +02:00
|
|
|
!(lip->m_thd->variables.sql_mode & MODE_PIPES_AS_CONCAT))
|
2004-11-17 16:49:10 +01:00
|
|
|
return OR2_SYM;
|
2007-04-26 05:38:12 +02:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
return symbol->tok;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2004-02-06 21:57:22 +01:00
|
|
|
/*
|
|
|
|
Check if name is a keyword
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
is_keyword()
|
2006-03-09 01:10:39 +01:00
|
|
|
name checked name (must not be empty)
|
2004-02-06 21:57:22 +01:00
|
|
|
len length of checked name
|
|
|
|
|
|
|
|
RETURN VALUES
|
|
|
|
0 name is a keyword
|
|
|
|
1 name isn't a keyword
|
|
|
|
*/
|
|
|
|
|
|
|
|
bool is_keyword(const char *name, uint len)
|
|
|
|
{
|
2006-03-09 01:10:39 +01:00
|
|
|
DBUG_ASSERT(len != 0);
|
2004-02-06 21:57:22 +01:00
|
|
|
return get_hash_symbol(name,len,0)!=0;
|
|
|
|
}
|
2000-07-31 21:29:14 +02:00
|
|
|
|
Bug#18239 (Possible to overload internal functions with stored functions)
Bug#21025 (misleading error message when creating functions named 'x', or 'y')
Bug#22619 (Spaces considered harmful)
This change contains a fix to report warnings or errors, and multiple tests
cases.
Before this fix, name collisions between:
- Native functions
- User Defined Functions
- Stored Functions
were not systematically reported, leading to confusing behavior.
I) Native / User Defined Function
Before this fix, is was possible to create a UDF named "foo", with the same
name as a native function "foo", but it was impossible to invoke the UDF,
since the syntax "foo()" always refer to the native function.
After this fix, creating a UDF fails with an error if there is a name
collision with a native function.
II) Native / Stored Function
Before this fix, is was possible to create a SF named "db.foo", with the same
name as a native function "foo", but this was confusing since the syntax
"foo()" would refer to the native function. To refer to the Stored Function,
the user had to use the "db.foo()" syntax.
After this fix, creating a Stored Function reports a warning if there is a
name collision with a native function.
III) User Defined Function / Stored Function
Before this fix, creating a User Defined Function "foo" and a Stored Function
"db.foo" are mutually exclusive operations. Whenever the second function is
created, an error is reported. However, the test suite did not cover this
behavior.
After this fix, the behavior is unchanged, and is now covered by test cases.
Note that the code change in this patch depends on the fix for Bug 21114.
2006-11-15 03:34:16 +01:00
|
|
|
bool is_lex_native_function(const LEX_STRING *name)
|
|
|
|
{
|
|
|
|
DBUG_ASSERT(name != NULL);
|
|
|
|
return (get_hash_symbol(name->str, name->length, 1) != 0);
|
|
|
|
}
|
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
/* make a copy of token before ptr and set yytoklen */
|
|
|
|
|
Bug#21513 (SP having body starting with quoted label rendered unusable)
Before this fix, the parser would sometime change where a token starts by
altering Lex_input_string::tok_start, which later confused the code in
sql_yacc.yy that needs to capture the source code of a SQL statement,
like to represent the body of a stored procedure.
This line of code in sql_lex.cc :
case MY_LEX_USER_VARIABLE_DELIMITER:
lip->tok_start= lip->ptr; // Skip first `
would <skip the first back quote> ... and cause the bug reported.
In general, the responsibility of sql_lex.cc is to *find* where token are
in the SQL text, but is *not* to make up fake or incomplete tokens.
With a quoted label like `my_label`, the token starts on the first quote.
Extracting the token value should not change that (it did).
With this fix, the lexical analysis has been cleaned up to not change
lip->tok_start (in the case found for this bug).
The functions get_token() and get_quoted_token() now have an extra
parameters, used when some characters from the beginning of the token need
to be skipped when extracting a token value, like when extracting 'AB' from
'0xAB', for example, for a HEX_NUM token.
This exposed a bad assumption in Item_hex_string and Item_bin_string,
which has been fixed:
The assumption was that the string given, 'AB', was in fact preceded in
memory by '0x', which might be false (it can be preceded by "x'" and
followed by "'" -- or not be preceded by valid memory at all)
If a name is needed for Item_hex_string or Item_bin_string, the name is
taken from the original and true source code ('0xAB'), and assigned in
the select_item rule, instead of relying on assumptions related to how
memory is used.
2007-04-28 01:14:25 +02:00
|
|
|
static LEX_STRING get_token(Lex_input_stream *lip, uint skip, uint length)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
|
|
|
LEX_STRING tmp;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yyUnget(); // ptr points now after last token char
|
2007-04-26 05:38:12 +02:00
|
|
|
tmp.length=lip->yytoklen=length;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
tmp.str= lip->m_thd->strmake(lip->get_tok_start() + skip, tmp.length);
|
2000-07-31 21:29:14 +02:00
|
|
|
return tmp;
|
|
|
|
}
|
|
|
|
|
2004-02-10 21:47:18 +01:00
|
|
|
/*
|
|
|
|
todo:
|
|
|
|
There are no dangerous charsets in mysql for function
|
|
|
|
get_quoted_token yet. But it should be fixed in the
|
|
|
|
future to operate multichar strings (like ucs2)
|
|
|
|
*/
|
|
|
|
|
2007-04-26 05:38:12 +02:00
|
|
|
static LEX_STRING get_quoted_token(Lex_input_stream *lip,
|
Bug#21513 (SP having body starting with quoted label rendered unusable)
Before this fix, the parser would sometime change where a token starts by
altering Lex_input_string::tok_start, which later confused the code in
sql_yacc.yy that needs to capture the source code of a SQL statement,
like to represent the body of a stored procedure.
This line of code in sql_lex.cc :
case MY_LEX_USER_VARIABLE_DELIMITER:
lip->tok_start= lip->ptr; // Skip first `
would <skip the first back quote> ... and cause the bug reported.
In general, the responsibility of sql_lex.cc is to *find* where token are
in the SQL text, but is *not* to make up fake or incomplete tokens.
With a quoted label like `my_label`, the token starts on the first quote.
Extracting the token value should not change that (it did).
With this fix, the lexical analysis has been cleaned up to not change
lip->tok_start (in the case found for this bug).
The functions get_token() and get_quoted_token() now have an extra
parameters, used when some characters from the beginning of the token need
to be skipped when extracting a token value, like when extracting 'AB' from
'0xAB', for example, for a HEX_NUM token.
This exposed a bad assumption in Item_hex_string and Item_bin_string,
which has been fixed:
The assumption was that the string given, 'AB', was in fact preceded in
memory by '0x', which might be false (it can be preceded by "x'" and
followed by "'" -- or not be preceded by valid memory at all)
If a name is needed for Item_hex_string or Item_bin_string, the name is
taken from the original and true source code ('0xAB'), and assigned in
the select_item rule, instead of relying on assumptions related to how
memory is used.
2007-04-28 01:14:25 +02:00
|
|
|
uint skip,
|
2007-04-26 05:38:12 +02:00
|
|
|
uint length, char quote)
|
2002-12-01 13:59:06 +01:00
|
|
|
{
|
|
|
|
LEX_STRING tmp;
|
2007-03-27 19:09:56 +02:00
|
|
|
const char *from, *end;
|
|
|
|
char *to;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yyUnget(); // ptr points now after last token char
|
2007-04-26 05:38:12 +02:00
|
|
|
tmp.length= lip->yytoklen=length;
|
|
|
|
tmp.str=(char*) lip->m_thd->alloc(tmp.length+1);
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
from= lip->get_tok_start() + skip;
|
2007-04-30 19:41:51 +02:00
|
|
|
to= tmp.str;
|
Bug#21513 (SP having body starting with quoted label rendered unusable)
Before this fix, the parser would sometime change where a token starts by
altering Lex_input_string::tok_start, which later confused the code in
sql_yacc.yy that needs to capture the source code of a SQL statement,
like to represent the body of a stored procedure.
This line of code in sql_lex.cc :
case MY_LEX_USER_VARIABLE_DELIMITER:
lip->tok_start= lip->ptr; // Skip first `
would <skip the first back quote> ... and cause the bug reported.
In general, the responsibility of sql_lex.cc is to *find* where token are
in the SQL text, but is *not* to make up fake or incomplete tokens.
With a quoted label like `my_label`, the token starts on the first quote.
Extracting the token value should not change that (it did).
With this fix, the lexical analysis has been cleaned up to not change
lip->tok_start (in the case found for this bug).
The functions get_token() and get_quoted_token() now have an extra
parameters, used when some characters from the beginning of the token need
to be skipped when extracting a token value, like when extracting 'AB' from
'0xAB', for example, for a HEX_NUM token.
This exposed a bad assumption in Item_hex_string and Item_bin_string,
which has been fixed:
The assumption was that the string given, 'AB', was in fact preceded in
memory by '0x', which might be false (it can be preceded by "x'" and
followed by "'" -- or not be preceded by valid memory at all)
If a name is needed for Item_hex_string or Item_bin_string, the name is
taken from the original and true source code ('0xAB'), and assigned in
the select_item rule, instead of relying on assumptions related to how
memory is used.
2007-04-28 01:14:25 +02:00
|
|
|
end= to+length;
|
|
|
|
for ( ; to != end; )
|
2002-12-01 13:59:06 +01:00
|
|
|
{
|
2007-03-27 19:09:56 +02:00
|
|
|
if ((*to++= *from++) == quote)
|
2002-12-01 13:59:06 +01:00
|
|
|
from++; // Skip double quotes
|
|
|
|
}
|
|
|
|
*to= 0; // End null for safety
|
|
|
|
return tmp;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
Return an unescaped text literal without quotes
|
|
|
|
Fix sometimes to do only one scan of the string
|
|
|
|
*/
|
2000-07-31 21:29:14 +02:00
|
|
|
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
static char *get_text(Lex_input_stream *lip, int pre_skip, int post_skip)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
|
|
|
reg1 uchar c,sep;
|
|
|
|
uint found_escape=0;
|
2007-04-26 05:38:12 +02:00
|
|
|
CHARSET_INFO *cs= lip->m_thd->charset();
|
2000-07-31 21:29:14 +02:00
|
|
|
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
sep= lip->yyGetLast(); // String should end with this
|
|
|
|
while (! lip->eof())
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
c= lip->yyGet();
|
2000-07-31 21:29:14 +02:00
|
|
|
#ifdef USE_MB
|
2006-12-14 23:51:37 +01:00
|
|
|
{
|
|
|
|
int l;
|
|
|
|
if (use_mb(cs) &&
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
(l = my_ismbchar(cs,
|
|
|
|
lip->get_ptr() -1,
|
|
|
|
lip->get_end_of_query()))) {
|
|
|
|
lip->skip_binary(l-1);
|
|
|
|
continue;
|
2006-12-14 23:51:37 +01:00
|
|
|
}
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
#endif
|
2004-07-07 14:26:43 +02:00
|
|
|
if (c == '\\' &&
|
2007-04-26 05:38:12 +02:00
|
|
|
!(lip->m_thd->variables.sql_mode & MODE_NO_BACKSLASH_ESCAPES))
|
2000-07-31 21:29:14 +02:00
|
|
|
{ // Escaped character
|
|
|
|
found_escape=1;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->eof())
|
2000-07-31 21:29:14 +02:00
|
|
|
return 0;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip();
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
else if (c == sep)
|
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (c == lip->yyGet()) // Check if two separators in a row
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
found_escape=1; // duplicate. Remember for delete
|
2000-07-31 21:29:14 +02:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
else
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yyUnget();
|
2000-07-31 21:29:14 +02:00
|
|
|
|
|
|
|
/* Found end. Unescape and return string */
|
2007-03-27 19:09:56 +02:00
|
|
|
const char *str, *end;
|
|
|
|
char *start;
|
2000-07-31 21:29:14 +02:00
|
|
|
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
str= lip->get_tok_start();
|
|
|
|
end= lip->get_ptr();
|
|
|
|
/* Extract the text from the token */
|
|
|
|
str += pre_skip;
|
|
|
|
end -= post_skip;
|
|
|
|
DBUG_ASSERT(end >= str);
|
|
|
|
|
2007-04-26 05:38:12 +02:00
|
|
|
if (!(start= (char*) lip->m_thd->alloc((uint) (end-str)+1)))
|
2001-03-11 20:20:15 +01:00
|
|
|
return (char*) ""; // Sql_alloc has set error flag
|
2000-07-31 21:29:14 +02:00
|
|
|
if (!found_escape)
|
|
|
|
{
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->yytoklen=(uint) (end-str);
|
|
|
|
memcpy(start,str,lip->yytoklen);
|
|
|
|
start[lip->yytoklen]=0;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2007-03-27 19:09:56 +02:00
|
|
|
char *to;
|
2005-02-04 01:07:32 +01:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
for (to=start ; str != end ; str++)
|
|
|
|
{
|
|
|
|
#ifdef USE_MB
|
|
|
|
int l;
|
2003-03-15 12:19:06 +01:00
|
|
|
if (use_mb(cs) &&
|
2007-03-27 19:09:56 +02:00
|
|
|
(l = my_ismbchar(cs, str, end))) {
|
2000-07-31 21:29:14 +02:00
|
|
|
while (l--)
|
|
|
|
*to++ = *str++;
|
|
|
|
str--;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
#endif
|
2007-04-26 05:38:12 +02:00
|
|
|
if (!(lip->m_thd->variables.sql_mode & MODE_NO_BACKSLASH_ESCAPES) &&
|
2005-02-04 01:14:02 +01:00
|
|
|
*str == '\\' && str+1 != end)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
|
|
|
switch(*++str) {
|
|
|
|
case 'n':
|
|
|
|
*to++='\n';
|
|
|
|
break;
|
|
|
|
case 't':
|
|
|
|
*to++= '\t';
|
|
|
|
break;
|
|
|
|
case 'r':
|
|
|
|
*to++ = '\r';
|
|
|
|
break;
|
|
|
|
case 'b':
|
|
|
|
*to++ = '\b';
|
|
|
|
break;
|
|
|
|
case '0':
|
|
|
|
*to++= 0; // Ascii null
|
|
|
|
break;
|
|
|
|
case 'Z': // ^Z must be escaped on Win32
|
|
|
|
*to++='\032';
|
|
|
|
break;
|
|
|
|
case '_':
|
|
|
|
case '%':
|
|
|
|
*to++= '\\'; // remember prefix for wildcard
|
|
|
|
/* Fall through */
|
|
|
|
default:
|
2006-05-24 00:55:53 +02:00
|
|
|
*to++= *str;
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2006-05-24 00:55:53 +02:00
|
|
|
else if (*str == sep)
|
|
|
|
*to++= *str++; // Two ' or "
|
2000-07-31 21:29:14 +02:00
|
|
|
else
|
|
|
|
*to++ = *str;
|
|
|
|
}
|
|
|
|
*to=0;
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->yytoklen=(uint) (to-start);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
2007-03-27 19:09:56 +02:00
|
|
|
return start;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
return 0; // unexpected end of query
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
** Calc type of integer; long integer, longlong integer or real.
|
|
|
|
** Returns smallest type that match the string.
|
|
|
|
** When using unsigned long long values the result is converted to a real
|
|
|
|
** because else they will be unexpected sign changes because all calculation
|
|
|
|
** is done with longlong or double.
|
|
|
|
*/
|
|
|
|
|
|
|
|
static const char *long_str="2147483647";
|
|
|
|
static const uint long_len=10;
|
|
|
|
static const char *signed_long_str="-2147483648";
|
|
|
|
static const char *longlong_str="9223372036854775807";
|
|
|
|
static const uint longlong_len=19;
|
|
|
|
static const char *signed_longlong_str="-9223372036854775808";
|
|
|
|
static const uint signed_longlong_len=19;
|
2001-09-14 01:54:33 +02:00
|
|
|
static const char *unsigned_longlong_str="18446744073709551615";
|
|
|
|
static const uint unsigned_longlong_len=20;
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2005-05-25 11:56:47 +02:00
|
|
|
static inline uint int_token(const char *str,uint length)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
|
|
|
if (length < long_len) // quick normal case
|
|
|
|
return NUM;
|
|
|
|
bool neg=0;
|
|
|
|
|
|
|
|
if (*str == '+') // Remove sign and pre-zeros
|
|
|
|
{
|
|
|
|
str++; length--;
|
|
|
|
}
|
|
|
|
else if (*str == '-')
|
|
|
|
{
|
|
|
|
str++; length--;
|
|
|
|
neg=1;
|
|
|
|
}
|
|
|
|
while (*str == '0' && length)
|
|
|
|
{
|
|
|
|
str++; length --;
|
|
|
|
}
|
|
|
|
if (length < long_len)
|
|
|
|
return NUM;
|
|
|
|
|
|
|
|
uint smaller,bigger;
|
|
|
|
const char *cmp;
|
|
|
|
if (neg)
|
|
|
|
{
|
|
|
|
if (length == long_len)
|
|
|
|
{
|
|
|
|
cmp= signed_long_str+1;
|
|
|
|
smaller=NUM; // If <= signed_long_str
|
|
|
|
bigger=LONG_NUM; // If >= signed_long_str
|
|
|
|
}
|
|
|
|
else if (length < signed_longlong_len)
|
|
|
|
return LONG_NUM;
|
|
|
|
else if (length > signed_longlong_len)
|
2005-02-08 23:50:45 +01:00
|
|
|
return DECIMAL_NUM;
|
2000-07-31 21:29:14 +02:00
|
|
|
else
|
|
|
|
{
|
|
|
|
cmp=signed_longlong_str+1;
|
|
|
|
smaller=LONG_NUM; // If <= signed_longlong_str
|
2005-02-08 23:50:45 +01:00
|
|
|
bigger=DECIMAL_NUM;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
if (length == long_len)
|
|
|
|
{
|
|
|
|
cmp= long_str;
|
|
|
|
smaller=NUM;
|
|
|
|
bigger=LONG_NUM;
|
|
|
|
}
|
|
|
|
else if (length < longlong_len)
|
|
|
|
return LONG_NUM;
|
|
|
|
else if (length > longlong_len)
|
2001-09-14 01:54:33 +02:00
|
|
|
{
|
|
|
|
if (length > unsigned_longlong_len)
|
2005-02-08 23:50:45 +01:00
|
|
|
return DECIMAL_NUM;
|
2001-09-14 01:54:33 +02:00
|
|
|
cmp=unsigned_longlong_str;
|
|
|
|
smaller=ULONGLONG_NUM;
|
2005-02-08 23:50:45 +01:00
|
|
|
bigger=DECIMAL_NUM;
|
2001-09-14 01:54:33 +02:00
|
|
|
}
|
2000-07-31 21:29:14 +02:00
|
|
|
else
|
|
|
|
{
|
|
|
|
cmp=longlong_str;
|
|
|
|
smaller=LONG_NUM;
|
2003-02-27 01:10:19 +01:00
|
|
|
bigger= ULONGLONG_NUM;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
while (*cmp && *cmp++ == *str++) ;
|
|
|
|
return ((uchar) str[-1] <= (uchar) cmp[-1]) ? smaller : bigger;
|
|
|
|
}
|
|
|
|
|
2003-11-03 13:01:59 +01:00
|
|
|
/*
|
2006-03-10 01:44:08 +01:00
|
|
|
MYSQLlex remember the following states from the following MYSQLlex()
|
2003-11-03 13:01:59 +01:00
|
|
|
|
|
|
|
- MY_LEX_EOQ Found end of query
|
|
|
|
- MY_LEX_OPERATOR_OR_IDENT Last state was an ident, text or number
|
|
|
|
(which can't be followed by a signed number)
|
|
|
|
*/
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2006-03-10 01:44:08 +01:00
|
|
|
int MYSQLlex(void *arg, void *yythd)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
|
|
|
reg1 uchar c;
|
2003-11-03 13:01:59 +01:00
|
|
|
int tokval, result_state;
|
2000-07-31 21:29:14 +02:00
|
|
|
uint length;
|
2003-11-20 21:06:25 +01:00
|
|
|
enum my_lex_states state;
|
2007-04-26 05:38:12 +02:00
|
|
|
THD *thd= (THD *)yythd;
|
|
|
|
Lex_input_stream *lip= thd->m_lip;
|
|
|
|
LEX *lex= thd->lex;
|
2000-07-31 21:29:14 +02:00
|
|
|
YYSTYPE *yylval=(YYSTYPE*) arg;
|
2007-04-26 05:38:12 +02:00
|
|
|
CHARSET_INFO *cs= thd->charset();
|
2003-03-14 15:08:12 +01:00
|
|
|
uchar *state_map= cs->state_map;
|
|
|
|
uchar *ident_map= cs->ident_map;
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->yylval=yylval; // The global state
|
2005-08-25 15:34:34 +02:00
|
|
|
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->start_token();
|
2007-04-26 05:38:12 +02:00
|
|
|
state=lip->next_state;
|
|
|
|
lip->next_state=MY_LEX_OPERATOR_OR_IDENT;
|
2000-07-31 21:29:14 +02:00
|
|
|
LINT_INIT(c);
|
|
|
|
for (;;)
|
|
|
|
{
|
2002-12-02 16:52:22 +01:00
|
|
|
switch (state) {
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_OPERATOR_OR_IDENT: // Next is operator or keyword
|
|
|
|
case MY_LEX_START: // Start of token
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
// Skip starting whitespace
|
|
|
|
while(state_map[c= lip->yyPeek()] == MY_LEX_SKIP)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
|
|
|
if (c == '\n')
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->yylineno++;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
|
|
|
|
lip->yySkip();
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
|
|
|
|
/* Start of real token */
|
|
|
|
lip->restart_token();
|
|
|
|
c= lip->yyGet();
|
2003-03-14 15:08:12 +01:00
|
|
|
state= (enum my_lex_states) state_map[c];
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_ESCAPE:
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->yyGet() == 'N')
|
2000-07-31 21:29:14 +02:00
|
|
|
{ // Allow \N as shortcut for NULL
|
|
|
|
yylval->lex_str.str=(char*) "\\N";
|
|
|
|
yylval->lex_str.length=2;
|
|
|
|
return NULL_SYM;
|
|
|
|
}
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_CHAR: // Unknown or single char token
|
|
|
|
case MY_LEX_SKIP: // This should not happen
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (c == '-' && lip->yyPeek() == '-' &&
|
|
|
|
(my_isspace(cs,lip->yyPeekn(1)) ||
|
|
|
|
my_iscntrl(cs,lip->yyPeekn(1))))
|
2003-07-20 12:26:18 +02:00
|
|
|
{
|
|
|
|
state=MY_LEX_COMMENT;
|
|
|
|
break;
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
if (c != ')')
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state= MY_LEX_START; // Allow signed numbers
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
if (c == ',')
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
{
|
|
|
|
/*
|
|
|
|
Warning:
|
|
|
|
This is a work around, to make the "remember_name" rule in
|
|
|
|
sql/sql_yacc.yy work properly.
|
|
|
|
The problem is that, when parsing "select expr1, expr2",
|
|
|
|
the code generated by bison executes the *pre* action
|
|
|
|
remember_name (see select_item) *before* actually parsing the
|
|
|
|
first token of expr2.
|
|
|
|
*/
|
|
|
|
lip->restart_token();
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
Check for a placeholder: it should not precede a possible identifier
|
|
|
|
because of binlogging: when a placeholder is replaced with
|
|
|
|
its value in a query for the binlog, the query must stay
|
|
|
|
grammatically correct.
|
|
|
|
*/
|
|
|
|
if (c == '?' && lip->stmt_prepare_mode && !ident_map[lip->yyPeek()])
|
2005-07-14 22:01:49 +02:00
|
|
|
return(PARAM_MARKER);
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
}
|
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
return((int) c);
|
|
|
|
|
2003-03-20 19:01:03 +01:00
|
|
|
case MY_LEX_IDENT_OR_NCHAR:
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->yyPeek() != '\'')
|
|
|
|
{
|
2003-03-20 19:01:03 +01:00
|
|
|
state= MY_LEX_IDENT;
|
|
|
|
break;
|
|
|
|
}
|
2006-07-31 09:47:01 +02:00
|
|
|
/* Found N'string' */
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip(); // Skip '
|
|
|
|
if (!(yylval->lex_str.str = get_text(lip, 2, 1)))
|
2003-03-20 19:01:03 +01:00
|
|
|
{
|
2006-07-31 09:47:01 +02:00
|
|
|
state= MY_LEX_CHAR; // Read char by char
|
|
|
|
break;
|
2003-03-20 19:01:03 +01:00
|
|
|
}
|
2007-04-26 05:38:12 +02:00
|
|
|
yylval->lex_str.length= lip->yytoklen;
|
2006-07-31 09:47:01 +02:00
|
|
|
return(NCHAR_STRING);
|
2003-03-20 19:01:03 +01:00
|
|
|
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_IDENT_OR_HEX:
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->yyPeek() == '\'')
|
2001-07-04 08:39:58 +02:00
|
|
|
{ // Found x'hex-number'
|
2003-03-14 15:08:12 +01:00
|
|
|
state= MY_LEX_HEX_NUMBER;
|
2001-07-04 08:39:58 +02:00
|
|
|
break;
|
|
|
|
}
|
2004-12-17 15:06:05 +01:00
|
|
|
case MY_LEX_IDENT_OR_BIN:
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->yyPeek() == '\'')
|
2004-12-17 15:06:05 +01:00
|
|
|
{ // Found b'bin-number'
|
|
|
|
state= MY_LEX_BIN_NUMBER;
|
|
|
|
break;
|
|
|
|
}
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_IDENT:
|
2007-03-27 19:09:56 +02:00
|
|
|
const char *start;
|
2000-07-31 21:29:14 +02:00
|
|
|
#if defined(USE_MB) && defined(USE_MB_IDENT)
|
2003-03-14 15:08:12 +01:00
|
|
|
if (use_mb(cs))
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2003-11-03 13:01:59 +01:00
|
|
|
result_state= IDENT_QUOTED;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (my_mbcharlen(cs, lip->yyGetLast()) > 1)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
int l = my_ismbchar(cs,
|
|
|
|
lip->get_ptr() -1,
|
|
|
|
lip->get_end_of_query());
|
2000-07-31 21:29:14 +02:00
|
|
|
if (l == 0) {
|
2003-03-14 15:08:12 +01:00
|
|
|
state = MY_LEX_CHAR;
|
2000-07-31 21:29:14 +02:00
|
|
|
continue;
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->skip_binary(l - 1);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
while (ident_map[c=lip->yyGet()])
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2003-04-01 12:52:09 +02:00
|
|
|
if (my_mbcharlen(cs, c) > 1)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
|
|
|
int l;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if ((l = my_ismbchar(cs,
|
|
|
|
lip->get_ptr() -1,
|
|
|
|
lip->get_end_of_query())) == 0)
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->skip_binary(l-1);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
#endif
|
2003-11-03 13:01:59 +01:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
for (result_state= c; ident_map[c= lip->yyGet()]; result_state|= c);
|
2004-06-10 16:10:21 +02:00
|
|
|
/* If there were non-ASCII characters, mark that we must convert */
|
|
|
|
result_state= result_state & 0x80 ? IDENT_QUOTED : IDENT;
|
2003-11-03 13:01:59 +01:00
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
length= lip->yyLength();
|
|
|
|
start= lip->get_ptr();
|
2007-04-26 05:38:12 +02:00
|
|
|
if (lip->ignore_space)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2004-10-14 17:03:46 +02:00
|
|
|
/*
|
|
|
|
If we find a space then this can't be an identifier. We notice this
|
|
|
|
below by checking start != lex->ptr.
|
|
|
|
*/
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
for (; state_map[c] == MY_LEX_SKIP ; c= lip->yyGet());
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (start == lip->get_ptr() && c == '.' && ident_map[lip->yyPeek()])
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state=MY_LEX_IDENT_SEP;
|
2000-07-31 21:29:14 +02:00
|
|
|
else
|
|
|
|
{ // '(' must follow directly if function
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yyUnget();
|
|
|
|
if ((tokval = find_keyword(lip, length, c == '(')))
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state= MY_LEX_START; // Allow signed numbers
|
2000-07-31 21:29:14 +02:00
|
|
|
return(tokval); // Was keyword
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip(); // next state does a unget
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
Bug#21513 (SP having body starting with quoted label rendered unusable)
Before this fix, the parser would sometime change where a token starts by
altering Lex_input_string::tok_start, which later confused the code in
sql_yacc.yy that needs to capture the source code of a SQL statement,
like to represent the body of a stored procedure.
This line of code in sql_lex.cc :
case MY_LEX_USER_VARIABLE_DELIMITER:
lip->tok_start= lip->ptr; // Skip first `
would <skip the first back quote> ... and cause the bug reported.
In general, the responsibility of sql_lex.cc is to *find* where token are
in the SQL text, but is *not* to make up fake or incomplete tokens.
With a quoted label like `my_label`, the token starts on the first quote.
Extracting the token value should not change that (it did).
With this fix, the lexical analysis has been cleaned up to not change
lip->tok_start (in the case found for this bug).
The functions get_token() and get_quoted_token() now have an extra
parameters, used when some characters from the beginning of the token need
to be skipped when extracting a token value, like when extracting 'AB' from
'0xAB', for example, for a HEX_NUM token.
This exposed a bad assumption in Item_hex_string and Item_bin_string,
which has been fixed:
The assumption was that the string given, 'AB', was in fact preceded in
memory by '0x', which might be false (it can be preceded by "x'" and
followed by "'" -- or not be preceded by valid memory at all)
If a name is needed for Item_hex_string or Item_bin_string, the name is
taken from the original and true source code ('0xAB'), and assigned in
the select_item rule, instead of relying on assumptions related to how
memory is used.
2007-04-28 01:14:25 +02:00
|
|
|
yylval->lex_str=get_token(lip, 0, length);
|
2002-06-20 15:47:55 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
Note: "SELECT _bla AS 'alias'"
|
|
|
|
_bla should be considered as a IDENT if charset haven't been found.
|
2003-07-29 14:12:14 +02:00
|
|
|
So we don't use MYF(MY_WME) with get_charset_by_csname to avoid
|
2002-06-20 15:47:55 +02:00
|
|
|
producing an error.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if ((yylval->lex_str.str[0]=='_') &&
|
2006-08-15 12:24:07 +02:00
|
|
|
(lex->underscore_charset=
|
|
|
|
get_charset_by_csname(yylval->lex_str.str + 1,
|
|
|
|
MY_CS_PRIMARY,MYF(0))))
|
2002-06-20 15:47:55 +02:00
|
|
|
return(UNDERSCORE_CHARSET);
|
2003-11-03 13:01:59 +01:00
|
|
|
return(result_state); // IDENT or IDENT_QUOTED
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_IDENT_SEP: // Found ident and now '.'
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
yylval->lex_str.str= (char*) lip->get_ptr();
|
|
|
|
yylval->lex_str.length= 1;
|
|
|
|
c= lip->yyGet(); // should be '.'
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state= MY_LEX_IDENT_START;// Next is an ident (not a keyword)
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (!ident_map[lip->yyPeek()]) // Probably ` or "
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state= MY_LEX_START;
|
2000-07-31 21:29:14 +02:00
|
|
|
return((int) c);
|
|
|
|
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_NUMBER_IDENT: // number or ident which num-start
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->yyGetLast() == '0')
|
|
|
|
{
|
|
|
|
c= lip->yyGet();
|
|
|
|
if (c == 'x')
|
|
|
|
{
|
|
|
|
while (my_isxdigit(cs,(c = lip->yyGet()))) ;
|
|
|
|
if ((lip->yyLength() >= 3) && !ident_map[c])
|
|
|
|
{
|
|
|
|
/* skip '0x' */
|
|
|
|
yylval->lex_str=get_token(lip, 2, lip->yyLength()-2);
|
|
|
|
return (HEX_NUM);
|
|
|
|
}
|
|
|
|
lip->yyUnget();
|
|
|
|
state= MY_LEX_IDENT_START;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
else if (c == 'b')
|
|
|
|
{
|
|
|
|
while ((c= lip->yyGet()) == '0' || c == '1');
|
|
|
|
if ((lip->yyLength() >= 3) && !ident_map[c])
|
|
|
|
{
|
|
|
|
/* Skip '0b' */
|
|
|
|
yylval->lex_str= get_token(lip, 2, lip->yyLength()-2);
|
|
|
|
return (BIN_NUM);
|
|
|
|
}
|
|
|
|
lip->yyUnget();
|
|
|
|
state= MY_LEX_IDENT_START;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
lip->yyUnget();
|
|
|
|
}
|
|
|
|
|
|
|
|
while (my_isdigit(cs, (c = lip->yyGet()))) ;
|
2002-11-25 11:11:16 +01:00
|
|
|
if (!ident_map[c])
|
2000-07-31 21:29:14 +02:00
|
|
|
{ // Can't be identifier
|
2003-03-14 15:08:12 +01:00
|
|
|
state=MY_LEX_INT_OR_REAL;
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (c == 'e' || c == 'E')
|
|
|
|
{
|
2000-10-10 23:48:03 +02:00
|
|
|
// The following test is written this way to allow numbers of type 1e1
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (my_isdigit(cs,lip->yyPeek()) ||
|
|
|
|
(c=(lip->yyGet())) == '+' || c == '-')
|
2000-07-31 21:29:14 +02:00
|
|
|
{ // Allow 1E+10
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (my_isdigit(cs,lip->yyPeek())) // Number must have digit after sign
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip();
|
|
|
|
while (my_isdigit(cs,lip->yyGet())) ;
|
|
|
|
yylval->lex_str=get_token(lip, 0, lip->yyLength());
|
2000-07-31 21:29:14 +02:00
|
|
|
return(FLOAT_NUM);
|
|
|
|
}
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yyUnget();
|
2004-12-17 15:06:05 +01:00
|
|
|
}
|
2000-07-31 21:29:14 +02:00
|
|
|
// fall through
|
2003-07-06 18:09:57 +02:00
|
|
|
case MY_LEX_IDENT_START: // We come here after '.'
|
2003-11-03 13:01:59 +01:00
|
|
|
result_state= IDENT;
|
2000-07-31 21:29:14 +02:00
|
|
|
#if defined(USE_MB) && defined(USE_MB_IDENT)
|
2003-03-14 15:08:12 +01:00
|
|
|
if (use_mb(cs))
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2003-11-03 13:01:59 +01:00
|
|
|
result_state= IDENT_QUOTED;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
while (ident_map[c=lip->yyGet()])
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2003-04-01 12:52:09 +02:00
|
|
|
if (my_mbcharlen(cs, c) > 1)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
|
|
|
int l;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if ((l = my_ismbchar(cs,
|
|
|
|
lip->get_ptr() -1,
|
|
|
|
lip->get_end_of_query())) == 0)
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->skip_binary(l-1);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
#endif
|
2004-06-10 16:10:21 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
for (result_state=0; ident_map[c= lip->yyGet()]; result_state|= c);
|
2004-06-10 16:10:21 +02:00
|
|
|
/* If there were non-ASCII characters, mark that we must convert */
|
|
|
|
result_state= result_state & 0x80 ? IDENT_QUOTED : IDENT;
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (c == '.' && ident_map[lip->yyPeek()])
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state=MY_LEX_IDENT_SEP;// Next is '.'
|
2000-07-31 21:29:14 +02:00
|
|
|
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
yylval->lex_str= get_token(lip, 0, lip->yyLength());
|
2003-11-03 13:01:59 +01:00
|
|
|
return(result_state);
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2004-02-17 00:35:17 +01:00
|
|
|
case MY_LEX_USER_VARIABLE_DELIMITER: // Found quote char
|
2003-01-17 15:33:54 +01:00
|
|
|
{
|
2004-02-06 23:22:12 +01:00
|
|
|
uint double_quotes= 0;
|
|
|
|
char quote_char= c; // Used char
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
while ((c=lip->yyGet()))
|
2004-02-13 23:26:27 +01:00
|
|
|
{
|
2006-12-14 23:51:37 +01:00
|
|
|
int var_length;
|
|
|
|
if ((var_length= my_mbcharlen(cs, c)) == 1)
|
2002-12-01 13:59:06 +01:00
|
|
|
{
|
|
|
|
if (c == quote_char)
|
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->yyPeek() != quote_char)
|
2002-12-01 13:59:06 +01:00
|
|
|
break;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
c=lip->yyGet();
|
2002-12-01 13:59:06 +01:00
|
|
|
double_quotes++;
|
|
|
|
continue;
|
|
|
|
}
|
2004-02-10 21:47:18 +01:00
|
|
|
}
|
|
|
|
#ifdef USE_MB
|
2006-12-14 23:51:37 +01:00
|
|
|
else if (var_length < 1)
|
2004-02-17 00:35:17 +01:00
|
|
|
break; // Error
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->skip_binary(var_length-1);
|
2004-02-10 21:47:18 +01:00
|
|
|
#endif
|
2001-10-30 15:31:35 +01:00
|
|
|
}
|
2004-02-06 23:22:12 +01:00
|
|
|
if (double_quotes)
|
Bug#21513 (SP having body starting with quoted label rendered unusable)
Before this fix, the parser would sometime change where a token starts by
altering Lex_input_string::tok_start, which later confused the code in
sql_yacc.yy that needs to capture the source code of a SQL statement,
like to represent the body of a stored procedure.
This line of code in sql_lex.cc :
case MY_LEX_USER_VARIABLE_DELIMITER:
lip->tok_start= lip->ptr; // Skip first `
would <skip the first back quote> ... and cause the bug reported.
In general, the responsibility of sql_lex.cc is to *find* where token are
in the SQL text, but is *not* to make up fake or incomplete tokens.
With a quoted label like `my_label`, the token starts on the first quote.
Extracting the token value should not change that (it did).
With this fix, the lexical analysis has been cleaned up to not change
lip->tok_start (in the case found for this bug).
The functions get_token() and get_quoted_token() now have an extra
parameters, used when some characters from the beginning of the token need
to be skipped when extracting a token value, like when extracting 'AB' from
'0xAB', for example, for a HEX_NUM token.
This exposed a bad assumption in Item_hex_string and Item_bin_string,
which has been fixed:
The assumption was that the string given, 'AB', was in fact preceded in
memory by '0x', which might be false (it can be preceded by "x'" and
followed by "'" -- or not be preceded by valid memory at all)
If a name is needed for Item_hex_string or Item_bin_string, the name is
taken from the original and true source code ('0xAB'), and assigned in
the select_item rule, instead of relying on assumptions related to how
memory is used.
2007-04-28 01:14:25 +02:00
|
|
|
yylval->lex_str=get_quoted_token(lip, 1,
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yyLength() - double_quotes -1,
|
2004-02-06 23:22:12 +01:00
|
|
|
quote_char);
|
|
|
|
else
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
yylval->lex_str=get_token(lip, 1, lip->yyLength() -1);
|
2004-02-06 23:22:12 +01:00
|
|
|
if (c == quote_char)
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip(); // Skip end `
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state= MY_LEX_START;
|
2003-11-03 13:01:59 +01:00
|
|
|
return(IDENT_QUOTED);
|
2003-01-17 15:33:54 +01:00
|
|
|
}
|
Bug#21513 (SP having body starting with quoted label rendered unusable)
Before this fix, the parser would sometime change where a token starts by
altering Lex_input_string::tok_start, which later confused the code in
sql_yacc.yy that needs to capture the source code of a SQL statement,
like to represent the body of a stored procedure.
This line of code in sql_lex.cc :
case MY_LEX_USER_VARIABLE_DELIMITER:
lip->tok_start= lip->ptr; // Skip first `
would <skip the first back quote> ... and cause the bug reported.
In general, the responsibility of sql_lex.cc is to *find* where token are
in the SQL text, but is *not* to make up fake or incomplete tokens.
With a quoted label like `my_label`, the token starts on the first quote.
Extracting the token value should not change that (it did).
With this fix, the lexical analysis has been cleaned up to not change
lip->tok_start (in the case found for this bug).
The functions get_token() and get_quoted_token() now have an extra
parameters, used when some characters from the beginning of the token need
to be skipped when extracting a token value, like when extracting 'AB' from
'0xAB', for example, for a HEX_NUM token.
This exposed a bad assumption in Item_hex_string and Item_bin_string,
which has been fixed:
The assumption was that the string given, 'AB', was in fact preceded in
memory by '0x', which might be false (it can be preceded by "x'" and
followed by "'" -- or not be preceded by valid memory at all)
If a name is needed for Item_hex_string or Item_bin_string, the name is
taken from the original and true source code ('0xAB'), and assigned in
the select_item rule, instead of relying on assumptions related to how
memory is used.
2007-04-28 01:14:25 +02:00
|
|
|
case MY_LEX_INT_OR_REAL: // Complete int or incomplete real
|
2000-07-31 21:29:14 +02:00
|
|
|
if (c != '.')
|
|
|
|
{ // Found complete integer number.
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
yylval->lex_str=get_token(lip, 0, lip->yyLength());
|
2000-07-31 21:29:14 +02:00
|
|
|
return int_token(yylval->lex_str.str,yylval->lex_str.length);
|
|
|
|
}
|
|
|
|
// fall through
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_REAL: // Incomplete real number
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
while (my_isdigit(cs,c = lip->yyGet())) ;
|
2000-07-31 21:29:14 +02:00
|
|
|
|
|
|
|
if (c == 'e' || c == 'E')
|
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
c = lip->yyGet();
|
2001-06-28 14:24:28 +02:00
|
|
|
if (c == '-' || c == '+')
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
c = lip->yyGet(); // Skip sign
|
2003-03-14 15:08:12 +01:00
|
|
|
if (!my_isdigit(cs,c))
|
2000-07-31 21:29:14 +02:00
|
|
|
{ // No digit after sign
|
2003-03-14 15:08:12 +01:00
|
|
|
state= MY_LEX_CHAR;
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
while (my_isdigit(cs,lip->yyGet())) ;
|
|
|
|
yylval->lex_str=get_token(lip, 0, lip->yyLength());
|
2000-07-31 21:29:14 +02:00
|
|
|
return(FLOAT_NUM);
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
yylval->lex_str=get_token(lip, 0, lip->yyLength());
|
2005-02-08 23:50:45 +01:00
|
|
|
return(DECIMAL_NUM);
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_HEX_NUMBER: // Found x'hexstring'
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip(); // Accept opening '
|
|
|
|
while (my_isxdigit(cs, (c= lip->yyGet()))) ;
|
|
|
|
if (c != '\'')
|
|
|
|
return(ABORT_SYM); // Illegal hex constant
|
|
|
|
lip->yySkip(); // Accept closing '
|
|
|
|
length= lip->yyLength(); // Length of hexnum+3
|
|
|
|
if ((length % 2) == 0)
|
|
|
|
return(ABORT_SYM); // odd number of hex digits
|
Bug#21513 (SP having body starting with quoted label rendered unusable)
Before this fix, the parser would sometime change where a token starts by
altering Lex_input_string::tok_start, which later confused the code in
sql_yacc.yy that needs to capture the source code of a SQL statement,
like to represent the body of a stored procedure.
This line of code in sql_lex.cc :
case MY_LEX_USER_VARIABLE_DELIMITER:
lip->tok_start= lip->ptr; // Skip first `
would <skip the first back quote> ... and cause the bug reported.
In general, the responsibility of sql_lex.cc is to *find* where token are
in the SQL text, but is *not* to make up fake or incomplete tokens.
With a quoted label like `my_label`, the token starts on the first quote.
Extracting the token value should not change that (it did).
With this fix, the lexical analysis has been cleaned up to not change
lip->tok_start (in the case found for this bug).
The functions get_token() and get_quoted_token() now have an extra
parameters, used when some characters from the beginning of the token need
to be skipped when extracting a token value, like when extracting 'AB' from
'0xAB', for example, for a HEX_NUM token.
This exposed a bad assumption in Item_hex_string and Item_bin_string,
which has been fixed:
The assumption was that the string given, 'AB', was in fact preceded in
memory by '0x', which might be false (it can be preceded by "x'" and
followed by "'" -- or not be preceded by valid memory at all)
If a name is needed for Item_hex_string or Item_bin_string, the name is
taken from the original and true source code ('0xAB'), and assigned in
the select_item rule, instead of relying on assumptions related to how
memory is used.
2007-04-28 01:14:25 +02:00
|
|
|
yylval->lex_str=get_token(lip,
|
|
|
|
2, // skip x'
|
|
|
|
length-3); // don't count x' and last '
|
2001-07-04 08:39:58 +02:00
|
|
|
return (HEX_NUM);
|
|
|
|
|
2004-12-17 15:06:05 +01:00
|
|
|
case MY_LEX_BIN_NUMBER: // Found b'bin-string'
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip(); // Accept opening '
|
|
|
|
while ((c= lip->yyGet()) == '0' || c == '1');
|
2004-12-17 15:06:05 +01:00
|
|
|
if (c != '\'')
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
return(ABORT_SYM); // Illegal hex constant
|
|
|
|
lip->yySkip(); // Accept closing '
|
|
|
|
length= lip->yyLength(); // Length of bin-num + 3
|
Bug#21513 (SP having body starting with quoted label rendered unusable)
Before this fix, the parser would sometime change where a token starts by
altering Lex_input_string::tok_start, which later confused the code in
sql_yacc.yy that needs to capture the source code of a SQL statement,
like to represent the body of a stored procedure.
This line of code in sql_lex.cc :
case MY_LEX_USER_VARIABLE_DELIMITER:
lip->tok_start= lip->ptr; // Skip first `
would <skip the first back quote> ... and cause the bug reported.
In general, the responsibility of sql_lex.cc is to *find* where token are
in the SQL text, but is *not* to make up fake or incomplete tokens.
With a quoted label like `my_label`, the token starts on the first quote.
Extracting the token value should not change that (it did).
With this fix, the lexical analysis has been cleaned up to not change
lip->tok_start (in the case found for this bug).
The functions get_token() and get_quoted_token() now have an extra
parameters, used when some characters from the beginning of the token need
to be skipped when extracting a token value, like when extracting 'AB' from
'0xAB', for example, for a HEX_NUM token.
This exposed a bad assumption in Item_hex_string and Item_bin_string,
which has been fixed:
The assumption was that the string given, 'AB', was in fact preceded in
memory by '0x', which might be false (it can be preceded by "x'" and
followed by "'" -- or not be preceded by valid memory at all)
If a name is needed for Item_hex_string or Item_bin_string, the name is
taken from the original and true source code ('0xAB'), and assigned in
the select_item rule, instead of relying on assumptions related to how
memory is used.
2007-04-28 01:14:25 +02:00
|
|
|
yylval->lex_str= get_token(lip,
|
|
|
|
2, // skip b'
|
|
|
|
length-3); // don't count b' and last '
|
|
|
|
return (BIN_NUM);
|
2004-12-17 15:06:05 +01:00
|
|
|
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_CMP_OP: // Incomplete comparison operator
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (state_map[lip->yyPeek()] == MY_LEX_CMP_OP ||
|
|
|
|
state_map[lip->yyPeek()] == MY_LEX_LONG_CMP_OP)
|
|
|
|
lip->yySkip();
|
|
|
|
if ((tokval = find_keyword(lip, lip->yyLength() + 1, 0)))
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state= MY_LEX_START; // Allow signed numbers
|
2000-07-31 21:29:14 +02:00
|
|
|
return(tokval);
|
|
|
|
}
|
2003-03-14 15:08:12 +01:00
|
|
|
state = MY_LEX_CHAR; // Something fishy found
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
|
|
|
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_LONG_CMP_OP: // Incomplete comparison operator
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (state_map[lip->yyPeek()] == MY_LEX_CMP_OP ||
|
|
|
|
state_map[lip->yyPeek()] == MY_LEX_LONG_CMP_OP)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip();
|
|
|
|
if (state_map[lip->yyPeek()] == MY_LEX_CMP_OP)
|
|
|
|
lip->yySkip();
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if ((tokval = find_keyword(lip, lip->yyLength() + 1, 0)))
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state= MY_LEX_START; // Found long op
|
2000-07-31 21:29:14 +02:00
|
|
|
return(tokval);
|
|
|
|
}
|
2003-03-14 15:08:12 +01:00
|
|
|
state = MY_LEX_CHAR; // Something fishy found
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
|
|
|
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_BOOL:
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (c != lip->yyPeek())
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2003-03-14 15:08:12 +01:00
|
|
|
state=MY_LEX_CHAR;
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip();
|
2007-04-26 05:38:12 +02:00
|
|
|
tokval = find_keyword(lip,2,0); // Is a bool operator
|
|
|
|
lip->next_state= MY_LEX_START; // Allow signed numbers
|
2000-07-31 21:29:14 +02:00
|
|
|
return(tokval);
|
|
|
|
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_STRING_OR_DELIMITER:
|
2007-04-26 05:38:12 +02:00
|
|
|
if (thd->variables.sql_mode & MODE_ANSI_QUOTES)
|
2003-01-17 15:33:54 +01:00
|
|
|
{
|
2003-03-14 15:08:12 +01:00
|
|
|
state= MY_LEX_USER_VARIABLE_DELIMITER;
|
2003-01-17 15:33:54 +01:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
/* " used for strings */
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_STRING: // Incomplete text string
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (!(yylval->lex_str.str = get_text(lip, 1, 1)))
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2003-03-14 15:08:12 +01:00
|
|
|
state= MY_LEX_CHAR; // Read char by char
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
|
|
|
}
|
2007-04-26 05:38:12 +02:00
|
|
|
yylval->lex_str.length=lip->yytoklen;
|
2000-07-31 21:29:14 +02:00
|
|
|
return(TEXT_STRING);
|
|
|
|
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_COMMENT: // Comment
|
2002-02-15 01:49:02 +01:00
|
|
|
lex->select_lex.options|= OPTION_FOUND_COMMENT;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
while ((c = lip->yyGet()) != '\n' && c) ;
|
|
|
|
lip->yyUnget(); // Safety against eof
|
2003-03-14 15:08:12 +01:00
|
|
|
state = MY_LEX_START; // Try again
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_LONG_COMMENT: /* Long C comment? */
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->yyPeek() != '*')
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2003-03-14 15:08:12 +01:00
|
|
|
state=MY_LEX_CHAR; // Probable division
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
|
|
|
}
|
2002-02-15 01:49:02 +01:00
|
|
|
lex->select_lex.options|= OPTION_FOUND_COMMENT;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
/* Reject '/' '*', since we might need to turn off the echo */
|
|
|
|
lip->yyUnget();
|
|
|
|
|
|
|
|
if (lip->yyPeekn(2) == '!')
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->in_comment= DISCARD_COMMENT;
|
|
|
|
/* Accept '/' '*' '!', but do not keep this marker. */
|
|
|
|
lip->set_echo(false);
|
|
|
|
lip->yySkip();
|
|
|
|
lip->yySkip();
|
|
|
|
lip->yySkip();
|
|
|
|
|
|
|
|
/*
|
|
|
|
The special comment format is very strict:
|
|
|
|
'/' '*' '!', followed by exactly
|
|
|
|
2 digits (major), then 3 digits (minor).
|
|
|
|
*/
|
|
|
|
char version_str[6];
|
|
|
|
version_str[0]= lip->yyPeekn(0);
|
|
|
|
version_str[1]= lip->yyPeekn(1);
|
|
|
|
version_str[2]= lip->yyPeekn(2);
|
|
|
|
version_str[3]= lip->yyPeekn(3);
|
|
|
|
version_str[4]= lip->yyPeekn(4);
|
|
|
|
version_str[5]= 0;
|
|
|
|
if ( my_isdigit(cs, version_str[0])
|
|
|
|
&& my_isdigit(cs, version_str[1])
|
|
|
|
&& my_isdigit(cs, version_str[2])
|
|
|
|
&& my_isdigit(cs, version_str[3])
|
|
|
|
&& my_isdigit(cs, version_str[4])
|
|
|
|
)
|
|
|
|
{
|
|
|
|
ulong version;
|
|
|
|
version=strtol(version_str, NULL, 10);
|
|
|
|
|
|
|
|
/* Accept 'M' 'M' 'm' 'm' 'm' */
|
|
|
|
lip->yySkipn(5);
|
|
|
|
|
|
|
|
if (version <= MYSQL_VERSION_ID)
|
|
|
|
{
|
|
|
|
/* Expand the content of the special comment as real code */
|
|
|
|
lip->set_echo(true);
|
|
|
|
state=MY_LEX_START;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
state=MY_LEX_START;
|
|
|
|
lip->set_echo(true);
|
|
|
|
break;
|
|
|
|
}
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
else
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->in_comment= PRESERVE_COMMENT;
|
|
|
|
lip->yySkip(); // Accept /
|
|
|
|
lip->yySkip(); // Accept *
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
|
|
|
|
while (! lip->eof() &&
|
|
|
|
((c=lip->yyGet()) != '*' || lip->yyPeek() != '/'))
|
|
|
|
{
|
|
|
|
if (c == '\n')
|
|
|
|
lip->yylineno++;
|
|
|
|
}
|
|
|
|
if (! lip->eof())
|
|
|
|
lip->yySkip(); // remove last '/'
|
|
|
|
state = MY_LEX_START; // Try again
|
|
|
|
lip->set_echo(true);
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_END_LONG_COMMENT:
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if ((lip->in_comment != NO_COMMENT) && lip->yyPeek() == '/')
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
/* Reject '*' '/' */
|
|
|
|
lip->yyUnget();
|
|
|
|
/* Accept '*' '/', with the proper echo */
|
|
|
|
lip->set_echo(lip->in_comment == PRESERVE_COMMENT);
|
|
|
|
lip->yySkipn(2);
|
|
|
|
/* And start recording the tokens again */
|
|
|
|
lip->set_echo(true);
|
|
|
|
lip->in_comment=NO_COMMENT;
|
|
|
|
state=MY_LEX_START;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
else
|
2003-03-14 15:08:12 +01:00
|
|
|
state=MY_LEX_CHAR; // Return '*'
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
2003-07-06 18:09:57 +02:00
|
|
|
case MY_LEX_SET_VAR: // Check if ':='
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->yyPeek() != '=')
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2003-03-14 15:08:12 +01:00
|
|
|
state=MY_LEX_CHAR; // Return ':'
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip();
|
2000-07-31 21:29:14 +02:00
|
|
|
return (SET_VAR);
|
2004-04-27 23:49:05 +02:00
|
|
|
case MY_LEX_SEMICOLON: // optional line terminator
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->yyPeek())
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2004-04-27 01:44:41 +02:00
|
|
|
if ((thd->client_capabilities & CLIENT_MULTI_STATEMENTS) &&
|
2007-05-11 15:26:12 +02:00
|
|
|
!lip->stmt_prepare_mode)
|
2003-01-18 20:53:38 +01:00
|
|
|
{
|
2005-01-18 03:03:26 +01:00
|
|
|
lex->safe_to_cache_query= 0;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->found_semicolon= lip->get_ptr();
|
2004-12-30 23:44:00 +01:00
|
|
|
thd->server_status|= SERVER_MORE_RESULTS_EXISTS;
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state= MY_LEX_END;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->set_echo(true);
|
2004-12-30 23:44:00 +01:00
|
|
|
return (END_OF_INPUT);
|
2003-01-18 20:53:38 +01:00
|
|
|
}
|
2004-12-30 23:44:00 +01:00
|
|
|
state= MY_LEX_CHAR; // Return ';'
|
2000-07-31 21:29:14 +02:00
|
|
|
break;
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->next_state=MY_LEX_END; // Mark for next loop
|
|
|
|
return(END_OF_INPUT);
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_EOL:
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (lip->eof())
|
2004-02-12 13:01:27 +01:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yyUnget(); // Reject the last '\0'
|
|
|
|
lip->set_echo(false);
|
|
|
|
lip->yySkip();
|
|
|
|
lip->set_echo(true);
|
|
|
|
lip->next_state=MY_LEX_END; // Mark for next loop
|
|
|
|
return(END_OF_INPUT);
|
2004-02-12 13:01:27 +01:00
|
|
|
}
|
|
|
|
state=MY_LEX_CHAR;
|
|
|
|
break;
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_END:
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state=MY_LEX_END;
|
2000-07-31 21:29:14 +02:00
|
|
|
return(0); // We found end of input last time
|
2002-11-21 01:07:14 +01:00
|
|
|
|
|
|
|
/* Actually real shouldn't start with . but allow them anyhow */
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_REAL_OR_POINT:
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
if (my_isdigit(cs,lip->yyPeek()))
|
2003-03-14 15:08:12 +01:00
|
|
|
state = MY_LEX_REAL; // Real
|
2000-07-31 21:29:14 +02:00
|
|
|
else
|
|
|
|
{
|
2003-07-06 18:09:57 +02:00
|
|
|
state= MY_LEX_IDENT_SEP; // return '.'
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yyUnget(); // Put back '.'
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
break;
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_USER_END: // end '@' of user@hostname
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
switch (state_map[lip->yyPeek()]) {
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_STRING:
|
|
|
|
case MY_LEX_USER_VARIABLE_DELIMITER:
|
|
|
|
case MY_LEX_STRING_OR_DELIMITER:
|
2001-09-11 01:30:29 +02:00
|
|
|
break;
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_USER_END:
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state=MY_LEX_SYSTEM_VAR;
|
2001-09-11 01:30:29 +02:00
|
|
|
break;
|
|
|
|
default:
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state=MY_LEX_HOSTNAME;
|
2001-09-11 01:30:29 +02:00
|
|
|
break;
|
|
|
|
}
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
yylval->lex_str.str=(char*) lip->get_ptr();
|
2000-07-31 21:29:14 +02:00
|
|
|
yylval->lex_str.length=1;
|
|
|
|
return((int) '@');
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_HOSTNAME: // end '@' of user@hostname
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
for (c=lip->yyGet() ;
|
2003-03-14 15:08:12 +01:00
|
|
|
my_isalnum(cs,c) || c == '.' || c == '_' || c == '$';
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
c= lip->yyGet()) ;
|
|
|
|
yylval->lex_str=get_token(lip, 0, lip->yyLength());
|
2000-07-31 21:29:14 +02:00
|
|
|
return(LEX_HOSTNAME);
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_SYSTEM_VAR:
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
yylval->lex_str.str=(char*) lip->get_ptr();
|
2002-07-23 17:31:22 +02:00
|
|
|
yylval->lex_str.length=1;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yySkip(); // Skip '@'
|
|
|
|
lip->next_state= (state_map[lip->yyPeek()] ==
|
2003-07-06 18:09:57 +02:00
|
|
|
MY_LEX_USER_VARIABLE_DELIMITER ?
|
|
|
|
MY_LEX_OPERATOR_OR_IDENT :
|
|
|
|
MY_LEX_IDENT_OR_KEYWORD);
|
2002-07-23 17:31:22 +02:00
|
|
|
return((int) '@');
|
2003-03-14 15:08:12 +01:00
|
|
|
case MY_LEX_IDENT_OR_KEYWORD:
|
2002-07-23 17:31:22 +02:00
|
|
|
/*
|
|
|
|
We come here when we have found two '@' in a row.
|
|
|
|
We should now be able to handle:
|
|
|
|
[(global | local | session) .]variable_name
|
|
|
|
*/
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
|
|
|
|
for (result_state= 0; ident_map[c= lip->yyGet()]; result_state|= c);
|
2004-06-10 16:10:21 +02:00
|
|
|
/* If there were non-ASCII characters, mark that we must convert */
|
|
|
|
result_state= result_state & 0x80 ? IDENT_QUOTED : IDENT;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
|
2002-07-23 17:31:22 +02:00
|
|
|
if (c == '.')
|
2007-04-26 05:38:12 +02:00
|
|
|
lip->next_state=MY_LEX_IDENT_SEP;
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
length= lip->yyLength();
|
|
|
|
if (length == 0)
|
2006-08-15 18:41:21 +02:00
|
|
|
return(ABORT_SYM); // Names must be nonempty.
|
2007-04-26 05:38:12 +02:00
|
|
|
if ((tokval= find_keyword(lip, length,0)))
|
2002-07-23 17:31:22 +02:00
|
|
|
{
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
lip->yyUnget(); // Put back 'c'
|
2002-07-23 17:31:22 +02:00
|
|
|
return(tokval); // Was keyword
|
|
|
|
}
|
Bug#21513 (SP having body starting with quoted label rendered unusable)
Before this fix, the parser would sometime change where a token starts by
altering Lex_input_string::tok_start, which later confused the code in
sql_yacc.yy that needs to capture the source code of a SQL statement,
like to represent the body of a stored procedure.
This line of code in sql_lex.cc :
case MY_LEX_USER_VARIABLE_DELIMITER:
lip->tok_start= lip->ptr; // Skip first `
would <skip the first back quote> ... and cause the bug reported.
In general, the responsibility of sql_lex.cc is to *find* where token are
in the SQL text, but is *not* to make up fake or incomplete tokens.
With a quoted label like `my_label`, the token starts on the first quote.
Extracting the token value should not change that (it did).
With this fix, the lexical analysis has been cleaned up to not change
lip->tok_start (in the case found for this bug).
The functions get_token() and get_quoted_token() now have an extra
parameters, used when some characters from the beginning of the token need
to be skipped when extracting a token value, like when extracting 'AB' from
'0xAB', for example, for a HEX_NUM token.
This exposed a bad assumption in Item_hex_string and Item_bin_string,
which has been fixed:
The assumption was that the string given, 'AB', was in fact preceded in
memory by '0x', which might be false (it can be preceded by "x'" and
followed by "'" -- or not be preceded by valid memory at all)
If a name is needed for Item_hex_string or Item_bin_string, the name is
taken from the original and true source code ('0xAB'), and assigned in
the select_item rule, instead of relying on assumptions related to how
memory is used.
2007-04-28 01:14:25 +02:00
|
|
|
yylval->lex_str=get_token(lip, 0, length);
|
2003-11-03 13:01:59 +01:00
|
|
|
return(result_state);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2002-05-06 23:04:16 +02:00
|
|
|
|
2006-07-28 00:49:18 +02:00
|
|
|
|
5.1 version of a fix and test cases for bugs:
Bug#4968 ""Stored procedure crash if cursor opened on altered table"
Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing"
Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from
stored procedure."
Bug#19733 "Repeated alter, or repeated create/drop, fails"
Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server"
Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a
growing key length" (this bug is not fixed in 5.0)
Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE
statements in stored routines or as prepared statements caused
incorrect results (and crashes in versions prior to 5.0.25).
In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE
SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options).
The problem of bugs 4968, 19733, 19282 and 6895 was that functions
mysql_prepare_table, mysql_create_table and mysql_alter_table are not
re-execution friendly: during their operation they modify contents
of LEX (members create_info, alter_info, key_list, create_list),
thus making the LEX unusable for the next execution.
In particular, these functions removed processed columns and keys from
create_list, key_list and drop_list. Search the code in sql_table.cc
for drop_it.remove() and similar patterns to find evidence.
The fix is to supply to these functions a usable copy of each of the
above structures at every re-execution of an SQL statement.
To simplify memory management, LEX::key_list and LEX::create_list
were added to LEX::alter_info, a fresh copy of which is created for
every execution.
The problem of crashing bug 22060 stemmed from the fact that the above
metnioned functions were not only modifying HA_CREATE_INFO structure
in LEX, but also were changing it to point to areas in volatile memory
of the execution memory root.
The patch solves this problem by creating and using an on-stack
copy of HA_CREATE_INFO in mysql_execute_command.
Additionally, this patch splits the part of mysql_alter_table
that analizes and rewrites information from the parser into
a separate function - mysql_prepare_alter_table, in analogy with
mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
|
|
|
Alter_info::Alter_info(const Alter_info &rhs, MEM_ROOT *mem_root)
|
|
|
|
:drop_list(rhs.drop_list, mem_root),
|
|
|
|
alter_list(rhs.alter_list, mem_root),
|
|
|
|
key_list(rhs.key_list, mem_root),
|
|
|
|
create_list(rhs.create_list, mem_root),
|
|
|
|
flags(rhs.flags),
|
|
|
|
keys_onoff(rhs.keys_onoff),
|
|
|
|
tablespace_op(rhs.tablespace_op),
|
|
|
|
partition_names(rhs.partition_names, mem_root),
|
2007-06-04 12:03:15 +02:00
|
|
|
no_parts(rhs.no_parts),
|
|
|
|
change_level(rhs.change_level),
|
|
|
|
datetime_field(rhs.datetime_field),
|
|
|
|
error_if_not_empty(rhs.error_if_not_empty)
|
5.1 version of a fix and test cases for bugs:
Bug#4968 ""Stored procedure crash if cursor opened on altered table"
Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing"
Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from
stored procedure."
Bug#19733 "Repeated alter, or repeated create/drop, fails"
Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server"
Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a
growing key length" (this bug is not fixed in 5.0)
Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE
statements in stored routines or as prepared statements caused
incorrect results (and crashes in versions prior to 5.0.25).
In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE
SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options).
The problem of bugs 4968, 19733, 19282 and 6895 was that functions
mysql_prepare_table, mysql_create_table and mysql_alter_table are not
re-execution friendly: during their operation they modify contents
of LEX (members create_info, alter_info, key_list, create_list),
thus making the LEX unusable for the next execution.
In particular, these functions removed processed columns and keys from
create_list, key_list and drop_list. Search the code in sql_table.cc
for drop_it.remove() and similar patterns to find evidence.
The fix is to supply to these functions a usable copy of each of the
above structures at every re-execution of an SQL statement.
To simplify memory management, LEX::key_list and LEX::create_list
were added to LEX::alter_info, a fresh copy of which is created for
every execution.
The problem of crashing bug 22060 stemmed from the fact that the above
metnioned functions were not only modifying HA_CREATE_INFO structure
in LEX, but also were changing it to point to areas in volatile memory
of the execution memory root.
The patch solves this problem by creating and using an on-stack
copy of HA_CREATE_INFO in mysql_execute_command.
Additionally, this patch splits the part of mysql_alter_table
that analizes and rewrites information from the parser into
a separate function - mysql_prepare_alter_table, in analogy with
mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
|
|
|
{
|
|
|
|
/*
|
|
|
|
Make deep copies of used objects.
|
|
|
|
This is not a fully deep copy - clone() implementations
|
2007-06-10 12:43:57 +02:00
|
|
|
of Alter_drop, Alter_column, Key, foreign_key, Key_part_spec
|
5.1 version of a fix and test cases for bugs:
Bug#4968 ""Stored procedure crash if cursor opened on altered table"
Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing"
Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from
stored procedure."
Bug#19733 "Repeated alter, or repeated create/drop, fails"
Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server"
Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a
growing key length" (this bug is not fixed in 5.0)
Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE
statements in stored routines or as prepared statements caused
incorrect results (and crashes in versions prior to 5.0.25).
In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE
SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options).
The problem of bugs 4968, 19733, 19282 and 6895 was that functions
mysql_prepare_table, mysql_create_table and mysql_alter_table are not
re-execution friendly: during their operation they modify contents
of LEX (members create_info, alter_info, key_list, create_list),
thus making the LEX unusable for the next execution.
In particular, these functions removed processed columns and keys from
create_list, key_list and drop_list. Search the code in sql_table.cc
for drop_it.remove() and similar patterns to find evidence.
The fix is to supply to these functions a usable copy of each of the
above structures at every re-execution of an SQL statement.
To simplify memory management, LEX::key_list and LEX::create_list
were added to LEX::alter_info, a fresh copy of which is created for
every execution.
The problem of crashing bug 22060 stemmed from the fact that the above
metnioned functions were not only modifying HA_CREATE_INFO structure
in LEX, but also were changing it to point to areas in volatile memory
of the execution memory root.
The patch solves this problem by creating and using an on-stack
copy of HA_CREATE_INFO in mysql_execute_command.
Additionally, this patch splits the part of mysql_alter_table
that analizes and rewrites information from the parser into
a separate function - mysql_prepare_alter_table, in analogy with
mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
|
|
|
do not copy string constants. At the same length the only
|
|
|
|
reason we make a copy currently is that ALTER/CREATE TABLE
|
|
|
|
code changes input Alter_info definitions, but string
|
|
|
|
constants never change.
|
|
|
|
*/
|
|
|
|
list_copy_and_replace_each_value(drop_list, mem_root);
|
|
|
|
list_copy_and_replace_each_value(alter_list, mem_root);
|
|
|
|
list_copy_and_replace_each_value(key_list, mem_root);
|
|
|
|
list_copy_and_replace_each_value(create_list, mem_root);
|
|
|
|
/* partition_names are not deeply copied currently */
|
|
|
|
}
|
|
|
|
|
|
|
|
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
void trim_whitespace(CHARSET_INFO *cs, LEX_STRING *str)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
TODO:
|
|
|
|
This code assumes that there are no multi-bytes characters
|
|
|
|
that can be considered white-space.
|
|
|
|
*/
|
2006-07-28 00:49:18 +02:00
|
|
|
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
while ((str->length > 0) && (my_isspace(cs, str->str[0])))
|
|
|
|
{
|
|
|
|
str->length --;
|
|
|
|
str->str ++;
|
|
|
|
}
|
2006-07-28 00:49:18 +02:00
|
|
|
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
/*
|
|
|
|
FIXME:
|
|
|
|
Also, parsing backward is not safe with multi bytes characters
|
|
|
|
*/
|
|
|
|
while ((str->length > 0) && (my_isspace(cs, str->str[str->length-1])))
|
|
|
|
{
|
|
|
|
str->length --;
|
|
|
|
}
|
2006-07-28 00:49:18 +02:00
|
|
|
}
|
|
|
|
|
Bug#25411 (trigger code truncated), PART II
Bug 28127 (Some valid identifiers names are not parsed correctly)
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
This patch is the second part of a major cleanup, required to fix
Bug 25411 (trigger code truncated).
The root cause of the issue stems from the function skip_rear_comments,
which was a work around to remove "extra" "*/" characters from the query
text, when parsing a query and reusing the text fragments to represent a
view, trigger, function or stored procedure.
The reason for this work around is that "special comments",
like /*!50002 XXX */, were not parsed properly, so that a query like:
AAA /*!50002 BBB */ CCC
would be seen by the parser as "AAA BBB */ CCC" when the current version
is greater or equal to 5.0.2
The root cause of this stems from how special comments are parsed.
Special comments are really out-of-bound text that appear inside a query,
that affects how the parser behave.
In nature, /*!50002 XXX */ in MySQL is similar to the C concept
of preprocessing :
#if VERSION >= 50002
XXX
#endif
Depending on the current VERSION of the server, either the special comment
should be expanded or it should be ignored, but in all cases the "text" of
the query should be re-written to strip the "/*!50002" and "*/" markers,
which does not belong to the SQL language itself.
Prior to this fix, these markers would leak into :
- the storage format for VIEW,
- the storage format for FUNCTION,
- the storage format for FUNCTION parameters, in mysql.proc (param_list),
- the storage format for PROCEDURE,
- the storage format for PROCEDURE parameters, in mysql.proc (param_list),
- the storage format for TRIGGER,
- the binary log used for replication.
In all cases, not only this cause format corruption, but also provide a vector
for dormant security issues, by allowing to tunnel code that will be activated
after an upgrade.
The proper solution is to deal with special comments strictly during parsing,
when accepting a query from the outside world.
Once a query is parsed and an object is created with a persistant
representation, this object should not arbitrarily mutate after an upgrade.
In short, special comments are a useful but limited feature for MYSQLdump,
when used at an *interface* level to facilitate import/export,
but bloating the server *internal* storage format is *not* the proper way
to deal with configuration management of the user logic.
With this fix:
- the Lex_input_stream class now acts as a comment pre-processor,
and either expands or ignore special comments on the fly.
- MYSQLlex and sql_yacc.yy have been cleaned up to strictly use the
public interface of Lex_input_stream. In particular, how the input stream
accepts or rejects a character is private to Lex_input_stream, and the
internal buffer pointers of that class are strictly private, and should not
be tempered with during parsing.
This caused many changes mostly in sql_lex.cc.
During the code cleanup in case MY_LEX_NUMBER_IDENT,
Bug 28127 (Some valid identifiers names are not parsed correctly)
was found and fixed.
By parsing special comments properly, and removing the function
'skip_rear_comments' [sic],
Bug 26302 (MySQL server cuts off trailing "*/" from comments in SP/func)
has been fixed as well.
2007-06-12 23:23:58 +02:00
|
|
|
|
2002-05-06 23:04:16 +02:00
|
|
|
/*
|
|
|
|
st_select_lex structures initialisations
|
|
|
|
*/
|
|
|
|
|
|
|
|
void st_select_lex_node::init_query()
|
|
|
|
{
|
2003-03-05 22:32:59 +01:00
|
|
|
options= 0;
|
2006-06-27 19:28:32 +02:00
|
|
|
sql_cache= SQL_CACHE_UNSPECIFIED;
|
2003-03-05 22:32:59 +01:00
|
|
|
linkage= UNSPECIFIED_TYPE;
|
2003-11-17 19:53:40 +01:00
|
|
|
no_error= no_table_names_allowed= 0;
|
|
|
|
uncacheable= 0;
|
2002-05-06 23:04:16 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
void st_select_lex_node::init_select()
|
|
|
|
{
|
|
|
|
}
|
|
|
|
|
|
|
|
void st_select_lex_unit::init_query()
|
|
|
|
{
|
|
|
|
st_select_lex_node::init_query();
|
2003-03-05 22:32:59 +01:00
|
|
|
linkage= GLOBAL_OPTIONS_TYPE;
|
2003-07-03 01:30:52 +02:00
|
|
|
global_parameters= first_select();
|
2002-05-08 22:14:40 +02:00
|
|
|
select_limit_cnt= HA_POS_ERROR;
|
|
|
|
offset_limit_cnt= 0;
|
2004-03-23 14:43:24 +01:00
|
|
|
union_distinct= 0;
|
2002-10-27 20:29:40 +01:00
|
|
|
prepared= optimized= executed= 0;
|
2002-09-03 08:50:36 +02:00
|
|
|
item= 0;
|
2003-01-25 01:25:52 +01:00
|
|
|
union_result= 0;
|
|
|
|
table= 0;
|
2003-07-03 01:30:52 +02:00
|
|
|
fake_select_lex= 0;
|
2004-02-10 01:18:22 +01:00
|
|
|
cleaned= 0;
|
2004-03-23 13:26:54 +01:00
|
|
|
item_list.empty();
|
2004-05-06 19:40:21 +02:00
|
|
|
describe= 0;
|
2003-11-21 20:19:56 +01:00
|
|
|
found_rows_for_union= 0;
|
2002-05-06 23:04:16 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
void st_select_lex::init_query()
|
|
|
|
{
|
|
|
|
st_select_lex_node::init_query();
|
2003-03-05 22:32:59 +01:00
|
|
|
table_list.empty();
|
2004-06-11 07:27:21 +02:00
|
|
|
top_join_list.empty();
|
|
|
|
join_list= &top_join_list;
|
2004-09-14 18:28:29 +02:00
|
|
|
embedding= leaf_tables= 0;
|
2002-05-06 23:04:16 +02:00
|
|
|
item_list.empty();
|
2002-10-15 23:42:59 +02:00
|
|
|
join= 0;
|
2006-02-03 06:23:36 +01:00
|
|
|
having= prep_having= where= prep_where= 0;
|
2002-11-09 14:40:46 +01:00
|
|
|
olap= UNSPECIFIED_OLAP_TYPE;
|
2003-07-29 15:59:46 +02:00
|
|
|
having_fix_field= 0;
|
2005-07-01 06:05:42 +02:00
|
|
|
context.select_lex= this;
|
|
|
|
context.init();
|
2005-08-12 16:57:19 +02:00
|
|
|
/*
|
|
|
|
Add the name resolution context of the current (sub)query to the
|
|
|
|
stack of contexts for the whole query.
|
2005-11-28 20:57:50 +01:00
|
|
|
TODO:
|
|
|
|
push_context may return an error if there is no memory for a new
|
|
|
|
element in the stack, however this method has no return value,
|
|
|
|
thus push_context should be moved to a place where query
|
|
|
|
initialization is checked for failure.
|
2005-08-12 16:57:19 +02:00
|
|
|
*/
|
|
|
|
parent_lex->push_context(&context);
|
2006-10-16 23:25:28 +02:00
|
|
|
cond_count= between_count= with_wild= 0;
|
2004-05-20 01:02:49 +02:00
|
|
|
conds_processed_with_permanent_arena= 0;
|
2003-07-03 01:30:52 +02:00
|
|
|
ref_pointer_array= 0;
|
2007-02-24 21:04:15 +01:00
|
|
|
select_n_where_fields= 0;
|
2003-08-16 12:26:48 +02:00
|
|
|
select_n_having_items= 0;
|
2004-06-09 22:32:20 +02:00
|
|
|
subquery_in_having= explicit_limit= 0;
|
2005-10-12 22:58:59 +02:00
|
|
|
is_item_list_lookup= 0;
|
2004-05-20 01:02:49 +02:00
|
|
|
first_execution= 1;
|
2004-06-30 14:54:32 +02:00
|
|
|
first_cond_optimization= 1;
|
2004-09-09 05:59:26 +02:00
|
|
|
parsing_place= NO_MATTER;
|
2005-03-28 14:13:31 +02:00
|
|
|
exclude_from_table_unique_test= no_wrap_view_item= FALSE;
|
2005-10-15 23:32:37 +02:00
|
|
|
nest_level= 0;
|
2004-11-13 11:56:39 +01:00
|
|
|
link_next= 0;
|
2002-05-06 23:04:16 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
void st_select_lex::init_select()
|
|
|
|
{
|
|
|
|
st_select_lex_node::init_select();
|
2003-03-05 22:32:59 +01:00
|
|
|
group_list.empty();
|
2005-08-12 16:57:19 +02:00
|
|
|
type= db= 0;
|
2003-03-05 22:32:59 +01:00
|
|
|
having= 0;
|
|
|
|
table_join_options= 0;
|
|
|
|
in_sum_expr= with_wild= 0;
|
2002-05-06 23:04:16 +02:00
|
|
|
options= 0;
|
2006-06-27 19:28:32 +02:00
|
|
|
sql_cache= SQL_CACHE_UNSPECIFIED;
|
2003-03-05 22:32:59 +01:00
|
|
|
braces= 0;
|
2002-05-06 23:04:16 +02:00
|
|
|
expr_list.empty();
|
2004-07-21 21:44:12 +02:00
|
|
|
interval_list.empty();
|
2002-09-03 08:50:36 +02:00
|
|
|
ftfunc_list_alloc.empty();
|
2005-10-15 23:32:37 +02:00
|
|
|
inner_sum_func_list= 0;
|
2002-09-03 08:50:36 +02:00
|
|
|
ftfunc_list= &ftfunc_list_alloc;
|
2003-02-10 16:59:16 +01:00
|
|
|
linkage= UNSPECIFIED_TYPE;
|
2003-07-03 01:30:52 +02:00
|
|
|
order_list.elements= 0;
|
|
|
|
order_list.first= 0;
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
|
|
|
order_list.next= (uchar**) &order_list.first;
|
2005-06-07 12:11:36 +02:00
|
|
|
/* Set limit and offset to default values */
|
|
|
|
select_limit= 0; /* denotes the default limit = HA_POS_ERROR */
|
|
|
|
offset_limit= 0; /* denotes the default offset = 0 */
|
2003-07-03 01:30:52 +02:00
|
|
|
with_sum_func= 0;
|
2006-10-31 18:51:09 +01:00
|
|
|
is_correlated= 0;
|
2007-01-11 21:18:01 +01:00
|
|
|
cur_pos_in_select_list= UNDEF_POS;
|
|
|
|
non_agg_fields.empty();
|
2007-03-07 19:44:58 +01:00
|
|
|
cond_value= having_value= Item::COND_UNDEF;
|
2007-02-21 21:00:32 +01:00
|
|
|
inner_refs_list.empty();
|
2002-05-06 23:04:16 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
st_select_lex structures linking
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* include on level down */
|
|
|
|
void st_select_lex_node::include_down(st_select_lex_node *upper)
|
|
|
|
{
|
|
|
|
if ((next= upper->slave))
|
|
|
|
next->prev= &next;
|
|
|
|
prev= &upper->slave;
|
|
|
|
upper->slave= this;
|
|
|
|
master= upper;
|
2002-12-15 21:01:09 +01:00
|
|
|
slave= 0;
|
2002-05-06 23:04:16 +02:00
|
|
|
}
|
|
|
|
|
2003-07-03 01:30:52 +02:00
|
|
|
/*
|
|
|
|
include on level down (but do not link)
|
2004-07-21 21:44:12 +02:00
|
|
|
|
2003-07-03 01:30:52 +02:00
|
|
|
SYNOPSYS
|
|
|
|
st_select_lex_node::include_standalone()
|
|
|
|
upper - reference on node underr which this node should be included
|
|
|
|
ref - references on reference on this node
|
|
|
|
*/
|
|
|
|
void st_select_lex_node::include_standalone(st_select_lex_node *upper,
|
|
|
|
st_select_lex_node **ref)
|
|
|
|
{
|
|
|
|
next= 0;
|
|
|
|
prev= ref;
|
|
|
|
master= upper;
|
|
|
|
slave= 0;
|
|
|
|
}
|
|
|
|
|
2002-05-06 23:04:16 +02:00
|
|
|
/* include neighbour (on same level) */
|
|
|
|
void st_select_lex_node::include_neighbour(st_select_lex_node *before)
|
|
|
|
{
|
|
|
|
if ((next= before->next))
|
|
|
|
next->prev= &next;
|
|
|
|
prev= &before->next;
|
|
|
|
before->next= this;
|
|
|
|
master= before->master;
|
2002-12-15 21:01:09 +01:00
|
|
|
slave= 0;
|
2002-05-06 23:04:16 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/* including in global SELECT_LEX list */
|
|
|
|
void st_select_lex_node::include_global(st_select_lex_node **plink)
|
|
|
|
{
|
|
|
|
if ((link_next= *plink))
|
|
|
|
link_next->link_prev= &link_next;
|
|
|
|
link_prev= plink;
|
|
|
|
*plink= this;
|
|
|
|
}
|
|
|
|
|
|
|
|
//excluding from global list (internal function)
|
|
|
|
void st_select_lex_node::fast_exclude()
|
|
|
|
{
|
2002-12-28 00:01:05 +01:00
|
|
|
if (link_prev)
|
2002-05-06 23:04:16 +02:00
|
|
|
{
|
|
|
|
if ((*link_prev= link_next))
|
|
|
|
link_next->link_prev= link_prev;
|
|
|
|
}
|
2003-02-16 19:37:51 +01:00
|
|
|
// Remove slave structure
|
|
|
|
for (; slave; slave= slave->next)
|
|
|
|
slave->fast_exclude();
|
|
|
|
|
2002-05-06 23:04:16 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
excluding select_lex structure (except first (first select can't be
|
|
|
|
deleted, because it is most upper select))
|
|
|
|
*/
|
|
|
|
void st_select_lex_node::exclude()
|
|
|
|
{
|
|
|
|
//exclude from global list
|
|
|
|
fast_exclude();
|
|
|
|
//exclude from other structures
|
|
|
|
if ((*prev= next))
|
|
|
|
next->prev= prev;
|
|
|
|
/*
|
|
|
|
We do not need following statements, because prev pointer of first
|
|
|
|
list element point to master->slave
|
|
|
|
if (master->slave == this)
|
|
|
|
master->slave= next;
|
|
|
|
*/
|
|
|
|
}
|
2002-05-09 14:23:57 +02:00
|
|
|
|
2003-10-17 14:18:57 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
Exclude level of current unit from tree of SELECTs
|
|
|
|
|
|
|
|
SYNOPSYS
|
|
|
|
st_select_lex_unit::exclude_level()
|
|
|
|
|
|
|
|
NOTE: units which belong to current will be brought up on level of
|
|
|
|
currernt unit
|
|
|
|
*/
|
2002-11-28 18:29:26 +01:00
|
|
|
void st_select_lex_unit::exclude_level()
|
|
|
|
{
|
|
|
|
SELECT_LEX_UNIT *units= 0, **units_last= &units;
|
2002-12-28 00:01:05 +01:00
|
|
|
for (SELECT_LEX *sl= first_select(); sl; sl= sl->next_select())
|
2002-11-28 18:29:26 +01:00
|
|
|
{
|
2003-10-25 15:29:35 +02:00
|
|
|
// unlink current level from global SELECTs list
|
2002-11-28 18:29:26 +01:00
|
|
|
if (sl->link_prev && (*sl->link_prev= sl->link_next))
|
|
|
|
sl->link_next->link_prev= sl->link_prev;
|
2003-10-25 15:29:35 +02:00
|
|
|
|
|
|
|
// bring up underlay levels
|
2002-11-28 18:29:26 +01:00
|
|
|
SELECT_LEX_UNIT **last= 0;
|
|
|
|
for (SELECT_LEX_UNIT *u= sl->first_inner_unit(); u; u= u->next_unit())
|
2002-11-29 09:44:30 +01:00
|
|
|
{
|
|
|
|
u->master= master;
|
2002-11-28 18:29:26 +01:00
|
|
|
last= (SELECT_LEX_UNIT**)&(u->next);
|
2002-11-29 09:44:30 +01:00
|
|
|
}
|
2002-11-28 18:29:26 +01:00
|
|
|
if (last)
|
|
|
|
{
|
|
|
|
(*units_last)= sl->first_inner_unit();
|
|
|
|
units_last= last;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (units)
|
|
|
|
{
|
2003-10-25 15:29:35 +02:00
|
|
|
// include brought up levels in place of current
|
2002-11-28 18:29:26 +01:00
|
|
|
(*prev)= units;
|
|
|
|
(*units_last)= (SELECT_LEX_UNIT*)next;
|
2003-10-25 15:29:35 +02:00
|
|
|
if (next)
|
|
|
|
next->prev= (SELECT_LEX_NODE**)units_last;
|
|
|
|
units->prev= prev;
|
2002-11-28 18:29:26 +01:00
|
|
|
}
|
|
|
|
else
|
2003-10-25 15:29:35 +02:00
|
|
|
{
|
|
|
|
// exclude currect unit from list of nodes
|
2002-11-28 18:29:26 +01:00
|
|
|
(*prev)= next;
|
2003-10-25 15:29:35 +02:00
|
|
|
if (next)
|
|
|
|
next->prev= prev;
|
|
|
|
}
|
2002-11-28 18:29:26 +01:00
|
|
|
}
|
|
|
|
|
2003-10-17 14:18:57 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
Exclude subtree of current unit from tree of SELECTs
|
|
|
|
|
|
|
|
SYNOPSYS
|
|
|
|
st_select_lex_unit::exclude_tree()
|
|
|
|
*/
|
|
|
|
void st_select_lex_unit::exclude_tree()
|
|
|
|
{
|
|
|
|
for (SELECT_LEX *sl= first_select(); sl; sl= sl->next_select())
|
|
|
|
{
|
2003-10-25 15:29:35 +02:00
|
|
|
// unlink current level from global SELECTs list
|
2003-10-17 14:18:57 +02:00
|
|
|
if (sl->link_prev && (*sl->link_prev= sl->link_next))
|
|
|
|
sl->link_next->link_prev= sl->link_prev;
|
|
|
|
|
2003-10-25 15:29:35 +02:00
|
|
|
// unlink underlay levels
|
2003-10-17 14:18:57 +02:00
|
|
|
for (SELECT_LEX_UNIT *u= sl->first_inner_unit(); u; u= u->next_unit())
|
|
|
|
{
|
|
|
|
u->exclude_level();
|
|
|
|
}
|
|
|
|
}
|
2003-10-25 15:29:35 +02:00
|
|
|
// exclude currect unit from list of nodes
|
2003-10-17 14:18:57 +02:00
|
|
|
(*prev)= next;
|
2003-10-25 15:29:35 +02:00
|
|
|
if (next)
|
|
|
|
next->prev= prev;
|
2003-10-17 14:18:57 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2002-11-04 21:12:45 +01:00
|
|
|
/*
|
|
|
|
st_select_lex_node::mark_as_dependent mark all st_select_lex struct from
|
|
|
|
this to 'last' as dependent
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
last - pointer to last st_select_lex struct, before wich all
|
|
|
|
st_select_lex have to be marked as dependent
|
|
|
|
|
|
|
|
NOTE
|
|
|
|
'last' should be reachable from this st_select_lex_node
|
|
|
|
*/
|
|
|
|
|
2003-07-03 01:30:52 +02:00
|
|
|
void st_select_lex::mark_as_dependent(SELECT_LEX *last)
|
2002-11-04 21:12:45 +01:00
|
|
|
{
|
|
|
|
/*
|
|
|
|
Mark all selects from resolved to 1 before select where was
|
|
|
|
found table as depended (of select where was found table)
|
|
|
|
*/
|
2003-07-03 01:30:52 +02:00
|
|
|
for (SELECT_LEX *s= this;
|
2003-11-02 15:31:22 +01:00
|
|
|
s && s != last;
|
2002-11-04 21:12:45 +01:00
|
|
|
s= s->outer_select())
|
2003-11-17 19:53:40 +01:00
|
|
|
if (!(s->uncacheable & UNCACHEABLE_DEPENDENT))
|
2002-11-04 21:12:45 +01:00
|
|
|
{
|
|
|
|
// Select is dependent of outer select
|
2007-01-19 09:17:28 +01:00
|
|
|
s->uncacheable= (s->uncacheable & ~UNCACHEABLE_UNITED) |
|
|
|
|
UNCACHEABLE_DEPENDENT;
|
2003-11-02 15:31:22 +01:00
|
|
|
SELECT_LEX_UNIT *munit= s->master_unit();
|
2007-01-19 09:17:28 +01:00
|
|
|
munit->uncacheable= (munit->uncacheable & ~UNCACHEABLE_UNITED) |
|
|
|
|
UNCACHEABLE_DEPENDENT;
|
|
|
|
for (SELECT_LEX *sl= munit->first_select(); sl ; sl= sl->next_select())
|
|
|
|
{
|
|
|
|
if (sl != s &&
|
|
|
|
!(sl->uncacheable & (UNCACHEABLE_DEPENDENT | UNCACHEABLE_UNITED)))
|
|
|
|
sl->uncacheable|= UNCACHEABLE_UNITED;
|
|
|
|
}
|
2002-11-04 21:12:45 +01:00
|
|
|
}
|
2006-10-31 18:51:09 +01:00
|
|
|
is_correlated= TRUE;
|
|
|
|
this->master_unit()->item->is_correlated= TRUE;
|
2002-11-04 21:12:45 +01:00
|
|
|
}
|
|
|
|
|
2002-10-30 12:18:52 +01:00
|
|
|
bool st_select_lex_node::set_braces(bool value) { return 1; }
|
|
|
|
bool st_select_lex_node::inc_in_sum_expr() { return 1; }
|
|
|
|
uint st_select_lex_node::get_in_sum_expr() { return 0; }
|
|
|
|
TABLE_LIST* st_select_lex_node::get_table_list() { return 0; }
|
|
|
|
List<Item>* st_select_lex_node::get_item_list() { return 0; }
|
2007-03-05 18:08:41 +01:00
|
|
|
TABLE_LIST *st_select_lex_node::add_table_to_list (THD *thd, Table_ident *table,
|
2002-10-30 12:18:52 +01:00
|
|
|
LEX_STRING *alias,
|
2003-01-09 21:42:31 +01:00
|
|
|
ulong table_join_options,
|
2002-10-30 12:18:52 +01:00
|
|
|
thr_lock_type flags,
|
2007-03-05 18:08:41 +01:00
|
|
|
List<index_hint> *hints,
|
2003-07-16 21:30:49 +02:00
|
|
|
LEX_STRING *option)
|
2003-08-27 00:14:13 +02:00
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
2003-07-02 00:45:22 +02:00
|
|
|
ulong st_select_lex_node::get_table_join_options()
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
prohibit using LIMIT clause
|
|
|
|
*/
|
2003-07-03 01:30:52 +02:00
|
|
|
bool st_select_lex::test_limit()
|
2003-07-02 00:45:22 +02:00
|
|
|
{
|
2005-06-07 12:11:36 +02:00
|
|
|
if (select_limit != 0)
|
2003-07-02 00:45:22 +02:00
|
|
|
{
|
|
|
|
my_error(ER_NOT_SUPPORTED_YET, MYF(0),
|
2004-11-12 13:34:00 +01:00
|
|
|
"LIMIT & IN/ALL/ANY/SOME subquery");
|
2003-07-02 00:45:22 +02:00
|
|
|
return(1);
|
|
|
|
}
|
|
|
|
// no sense in ORDER BY without LIMIT
|
|
|
|
order_list.empty();
|
|
|
|
return(0);
|
|
|
|
}
|
2002-05-09 14:23:57 +02:00
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-10-30 12:18:52 +01:00
|
|
|
st_select_lex_unit* st_select_lex_unit::master_unit()
|
|
|
|
{
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-10-30 12:18:52 +01:00
|
|
|
st_select_lex* st_select_lex_unit::outer_select()
|
|
|
|
{
|
|
|
|
return (st_select_lex*) master;
|
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2003-07-03 01:30:52 +02:00
|
|
|
bool st_select_lex::add_order_to_list(THD *thd, Item *item, bool asc)
|
2002-11-04 21:12:45 +01:00
|
|
|
{
|
2003-07-03 01:30:52 +02:00
|
|
|
return add_to_list(thd, order_list, item, asc);
|
2002-11-04 21:12:45 +01:00
|
|
|
}
|
2002-10-30 12:18:52 +01:00
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-12-06 20:11:27 +01:00
|
|
|
bool st_select_lex::add_item_to_list(THD *thd, Item *item)
|
2002-10-30 12:18:52 +01:00
|
|
|
{
|
2004-05-20 01:02:49 +02:00
|
|
|
DBUG_ENTER("st_select_lex::add_item_to_list");
|
2006-11-27 00:47:38 +01:00
|
|
|
DBUG_PRINT("info", ("Item: 0x%lx", (long) item));
|
2004-05-20 01:02:49 +02:00
|
|
|
DBUG_RETURN(item_list.push_back(item));
|
2002-10-30 12:18:52 +01:00
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-12-06 20:11:27 +01:00
|
|
|
bool st_select_lex::add_group_to_list(THD *thd, Item *item, bool asc)
|
2002-10-30 12:18:52 +01:00
|
|
|
{
|
2002-12-06 20:11:27 +01:00
|
|
|
return add_to_list(thd, group_list, item, asc);
|
2002-10-30 12:18:52 +01:00
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-10-30 12:18:52 +01:00
|
|
|
bool st_select_lex::add_ftfunc_to_list(Item_func_match *func)
|
|
|
|
{
|
|
|
|
return !func || ftfunc_list->push_back(func); // end of memory?
|
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-10-30 12:18:52 +01:00
|
|
|
st_select_lex_unit* st_select_lex::master_unit()
|
|
|
|
{
|
|
|
|
return (st_select_lex_unit*) master;
|
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-10-30 12:18:52 +01:00
|
|
|
st_select_lex* st_select_lex::outer_select()
|
|
|
|
{
|
|
|
|
return (st_select_lex*) master->get_master();
|
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-10-30 12:18:52 +01:00
|
|
|
bool st_select_lex::set_braces(bool value)
|
|
|
|
{
|
|
|
|
braces= value;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-10-30 12:18:52 +01:00
|
|
|
bool st_select_lex::inc_in_sum_expr()
|
|
|
|
{
|
|
|
|
in_sum_expr++;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-10-30 12:18:52 +01:00
|
|
|
uint st_select_lex::get_in_sum_expr()
|
|
|
|
{
|
|
|
|
return in_sum_expr;
|
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2002-10-30 12:18:52 +01:00
|
|
|
TABLE_LIST* st_select_lex::get_table_list()
|
|
|
|
{
|
|
|
|
return (TABLE_LIST*) table_list.first;
|
|
|
|
}
|
|
|
|
|
|
|
|
List<Item>* st_select_lex::get_item_list()
|
|
|
|
{
|
|
|
|
return &item_list;
|
|
|
|
}
|
|
|
|
|
2003-01-09 21:42:31 +01:00
|
|
|
ulong st_select_lex::get_table_join_options()
|
|
|
|
{
|
|
|
|
return table_join_options;
|
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2003-08-16 12:26:48 +02:00
|
|
|
bool st_select_lex::setup_ref_array(THD *thd, uint order_group_num)
|
|
|
|
{
|
|
|
|
if (ref_pointer_array)
|
|
|
|
return 0;
|
2004-02-12 02:10:26 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
We have to create array in prepared statement memory if it is
|
|
|
|
prepared statement
|
|
|
|
*/
|
2005-09-02 15:21:19 +02:00
|
|
|
Query_arena *arena= thd->stmt_arena;
|
2005-06-15 19:58:35 +02:00
|
|
|
return (ref_pointer_array=
|
2006-09-01 11:23:43 +02:00
|
|
|
(Item **)arena->alloc(sizeof(Item*) * (n_child_sum_items +
|
|
|
|
item_list.elements +
|
|
|
|
select_n_having_items +
|
2007-02-24 21:04:15 +01:00
|
|
|
select_n_where_fields +
|
2006-09-01 11:23:43 +02:00
|
|
|
order_group_num)*5)) == 0;
|
2003-08-16 12:26:48 +02:00
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2003-10-16 14:54:47 +02:00
|
|
|
void st_select_lex_unit::print(String *str)
|
|
|
|
{
|
2005-08-15 21:05:05 +02:00
|
|
|
bool union_all= !union_distinct;
|
2003-10-16 14:54:47 +02:00
|
|
|
for (SELECT_LEX *sl= first_select(); sl; sl= sl->next_select())
|
|
|
|
{
|
|
|
|
if (sl != first_select())
|
|
|
|
{
|
2005-11-20 19:47:07 +01:00
|
|
|
str->append(STRING_WITH_LEN(" union "));
|
2005-08-15 21:05:05 +02:00
|
|
|
if (union_all)
|
2005-11-20 19:47:07 +01:00
|
|
|
str->append(STRING_WITH_LEN("all "));
|
2005-08-15 21:05:05 +02:00
|
|
|
else if (union_distinct == sl)
|
2005-08-22 00:13:37 +02:00
|
|
|
union_all= TRUE;
|
2003-10-16 14:54:47 +02:00
|
|
|
}
|
|
|
|
if (sl->braces)
|
|
|
|
str->append('(');
|
|
|
|
sl->print(thd, str);
|
|
|
|
if (sl->braces)
|
|
|
|
str->append(')');
|
|
|
|
}
|
|
|
|
if (fake_select_lex == global_parameters)
|
|
|
|
{
|
|
|
|
if (fake_select_lex->order_list.elements)
|
|
|
|
{
|
2005-11-20 19:47:07 +01:00
|
|
|
str->append(STRING_WITH_LEN(" order by "));
|
2003-10-16 14:54:47 +02:00
|
|
|
fake_select_lex->print_order(str,
|
|
|
|
(ORDER *) fake_select_lex->
|
|
|
|
order_list.first);
|
|
|
|
}
|
|
|
|
fake_select_lex->print_limit(thd, str);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void st_select_lex::print_order(String *str, ORDER *order)
|
|
|
|
{
|
|
|
|
for (; order; order= order->next)
|
|
|
|
{
|
2004-08-31 10:58:45 +02:00
|
|
|
if (order->counter_used)
|
|
|
|
{
|
|
|
|
char buffer[20];
|
2005-05-18 18:00:21 +02:00
|
|
|
uint length= my_snprintf(buffer, 20, "%d", order->counter);
|
|
|
|
str->append(buffer, length);
|
2004-08-31 10:58:45 +02:00
|
|
|
}
|
|
|
|
else
|
|
|
|
(*order->item)->print(str);
|
2003-10-16 14:54:47 +02:00
|
|
|
if (!order->asc)
|
2005-11-20 19:47:07 +01:00
|
|
|
str->append(STRING_WITH_LEN(" desc"));
|
2003-10-16 14:54:47 +02:00
|
|
|
if (order->next)
|
|
|
|
str->append(',');
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2004-02-01 14:30:32 +01:00
|
|
|
|
2003-10-16 14:54:47 +02:00
|
|
|
void st_select_lex::print_limit(THD *thd, String *str)
|
|
|
|
{
|
2004-08-23 12:19:59 +02:00
|
|
|
SELECT_LEX_UNIT *unit= master_unit();
|
|
|
|
Item_subselect *item= unit->item;
|
|
|
|
if (item && unit->global_parameters == this &&
|
2004-07-16 00:15:55 +02:00
|
|
|
(item->substype() == Item_subselect::EXISTS_SUBS ||
|
|
|
|
item->substype() == Item_subselect::IN_SUBS ||
|
|
|
|
item->substype() == Item_subselect::ALL_SUBS))
|
|
|
|
{
|
2005-06-07 12:11:36 +02:00
|
|
|
DBUG_ASSERT(!item->fixed ||
|
|
|
|
select_limit->val_int() == LL(1) && offset_limit == 0);
|
2004-07-16 00:15:55 +02:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2004-05-05 20:21:41 +02:00
|
|
|
if (explicit_limit)
|
2003-10-16 14:54:47 +02:00
|
|
|
{
|
2005-11-20 19:47:07 +01:00
|
|
|
str->append(STRING_WITH_LEN(" limit "));
|
2003-10-16 14:54:47 +02:00
|
|
|
if (offset_limit)
|
|
|
|
{
|
2005-06-07 12:11:36 +02:00
|
|
|
offset_limit->print(str);
|
2003-10-16 14:54:47 +02:00
|
|
|
str->append(',');
|
|
|
|
}
|
2005-06-07 12:11:36 +02:00
|
|
|
select_limit->print(str);
|
2003-10-16 14:54:47 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review
fixes).
The legend: on a replication slave, in case a trigger creation
was filtered out because of application of replicate-do-table/
replicate-ignore-table rule, the parsed definition of a trigger was not
cleaned up properly. LEX::sphead member was left around and leaked
memory. Until the actual implementation of support of
replicate-ignore-table rules for triggers by the patch for Bug 24478 it
was never the case that "case SQLCOM_CREATE_TRIGGER"
was not executed once a trigger was parsed,
so the deletion of lex->sphead there worked and the memory did not leak.
The fix:
The real cause of the bug is that there is no 1 or 2 places where
we can clean up the main LEX after parse. And the reason we
can not have just one or two places where we clean up the LEX is
asymmetric behaviour of MYSQLparse in case of success or error.
One of the root causes of this behaviour is the code in Item::Item()
constructor. There, a newly created item adds itself to THD::free_list
- a single-linked list of Items used in a statement. Yuck. This code
is unaware that we may have more than one statement active at a time,
and always assumes that the free_list of the current statement is
located in THD::free_list. One day we need to be able to explicitly
allocate an item in a given Query_arena.
Thus, when parsing a definition of a stored procedure, like
CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END;
we actually need to reset THD::mem_root, THD::free_list and THD::lex
to parse the nested procedure statement (SELECT *).
The actual reset and restore is implemented in semantic actions
attached to sp_proc_stmt grammar rule.
The problem is that in case of a parsing error inside a nested statement
Bison generated parser would abort immediately, without executing the
restore part of the semantic action. This would leave THD in an
in-the-middle-of-parsing state.
This is why we couldn't have had a single place where we clean up the LEX
after MYSQLparse - in case of an error we needed to do a clean up
immediately, in case of success a clean up could have been delayed.
This left the door open for a memory leak.
One of the following possibilities were considered when working on a fix:
- patch the replication logic to do the clean up. Rejected
as breaks module borders, replication code should not need to know the
gory details of clean up procedure after CREATE TRIGGER.
- wrap MYSQLparse with a function that would do a clean up.
Rejected as ideally we should fix the problem when it happens, not
adjust for it outside of the problematic code.
- make sure MYSQLparse cleans up after itself by invoking the clean up
functionality in the appropriate places before return. Implemented in
this patch.
- use %destructor rule for sp_proc_stmt to restore THD - cleaner
than the prevoius approach, but rejected
because needs a careful analysis of the side effects, and this patch is
for 5.0, and long term we need to use the next alternative anyway
- make sure that sp_proc_stmt doesn't juggle with THD - this is a
large work that will affect many modules.
Cleanup: move main_lex and main_mem_root from Statement to its
only two descendants Prepared_statement and THD. This ensures that
when a Statement instance was created for purposes of statement backup,
we do not involve LEX constructor/destructor, which is fairly expensive.
In order to track that the transformation produces equivalent
functionality please check the respective constructors and destructors
of Statement, Prepared_statement and THD - these members were
used only there.
This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
|
|
|
/**
|
|
|
|
@brief Restore the LEX and THD in case of a parse error.
|
|
|
|
|
|
|
|
This is a clean up call that is invoked by the Bison generated
|
|
|
|
parser before returning an error from MYSQLparse. If your
|
|
|
|
semantic actions manipulate with the global thread state (which
|
|
|
|
is a very bad practice and should not normally be employed) and
|
|
|
|
need a clean-up in case of error, and you can not use %destructor
|
|
|
|
rule in the grammar file itself, this function should be used
|
|
|
|
to implement the clean up.
|
|
|
|
*/
|
|
|
|
|
|
|
|
void st_lex::cleanup_lex_after_parse_error(THD *thd)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
Delete sphead for the side effect of restoring of the original
|
|
|
|
LEX state, thd->lex, thd->mem_root and thd->free_list if they
|
|
|
|
were replaced when parsing stored procedure statements. We
|
|
|
|
will never use sphead object after a parse error, so it's okay
|
|
|
|
to delete it only for the sake of the side effect.
|
|
|
|
TODO: make this functionality explicit in sp_head class.
|
|
|
|
Sic: we must nullify the member of the main lex, not the
|
|
|
|
current one that will be thrown away
|
|
|
|
*/
|
2007-03-07 12:03:44 +01:00
|
|
|
if (thd->lex->sphead)
|
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review
fixes).
The legend: on a replication slave, in case a trigger creation
was filtered out because of application of replicate-do-table/
replicate-ignore-table rule, the parsed definition of a trigger was not
cleaned up properly. LEX::sphead member was left around and leaked
memory. Until the actual implementation of support of
replicate-ignore-table rules for triggers by the patch for Bug 24478 it
was never the case that "case SQLCOM_CREATE_TRIGGER"
was not executed once a trigger was parsed,
so the deletion of lex->sphead there worked and the memory did not leak.
The fix:
The real cause of the bug is that there is no 1 or 2 places where
we can clean up the main LEX after parse. And the reason we
can not have just one or two places where we clean up the LEX is
asymmetric behaviour of MYSQLparse in case of success or error.
One of the root causes of this behaviour is the code in Item::Item()
constructor. There, a newly created item adds itself to THD::free_list
- a single-linked list of Items used in a statement. Yuck. This code
is unaware that we may have more than one statement active at a time,
and always assumes that the free_list of the current statement is
located in THD::free_list. One day we need to be able to explicitly
allocate an item in a given Query_arena.
Thus, when parsing a definition of a stored procedure, like
CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END;
we actually need to reset THD::mem_root, THD::free_list and THD::lex
to parse the nested procedure statement (SELECT *).
The actual reset and restore is implemented in semantic actions
attached to sp_proc_stmt grammar rule.
The problem is that in case of a parsing error inside a nested statement
Bison generated parser would abort immediately, without executing the
restore part of the semantic action. This would leave THD in an
in-the-middle-of-parsing state.
This is why we couldn't have had a single place where we clean up the LEX
after MYSQLparse - in case of an error we needed to do a clean up
immediately, in case of success a clean up could have been delayed.
This left the door open for a memory leak.
One of the following possibilities were considered when working on a fix:
- patch the replication logic to do the clean up. Rejected
as breaks module borders, replication code should not need to know the
gory details of clean up procedure after CREATE TRIGGER.
- wrap MYSQLparse with a function that would do a clean up.
Rejected as ideally we should fix the problem when it happens, not
adjust for it outside of the problematic code.
- make sure MYSQLparse cleans up after itself by invoking the clean up
functionality in the appropriate places before return. Implemented in
this patch.
- use %destructor rule for sp_proc_stmt to restore THD - cleaner
than the prevoius approach, but rejected
because needs a careful analysis of the side effects, and this patch is
for 5.0, and long term we need to use the next alternative anyway
- make sure that sp_proc_stmt doesn't juggle with THD - this is a
large work that will affect many modules.
Cleanup: move main_lex and main_mem_root from Statement to its
only two descendants Prepared_statement and THD. This ensures that
when a Statement instance was created for purposes of statement backup,
we do not involve LEX constructor/destructor, which is fairly expensive.
In order to track that the transformation produces equivalent
functionality please check the respective constructors and destructors
of Statement, Prepared_statement and THD - these members were
used only there.
This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
|
|
|
{
|
|
|
|
delete thd->lex->sphead;
|
|
|
|
thd->lex->sphead= NULL;
|
|
|
|
}
|
|
|
|
}
|
2004-07-16 00:15:55 +02:00
|
|
|
|
2006-05-30 07:45:23 +02:00
|
|
|
/*
|
|
|
|
Initialize (or reset) Query_tables_list object.
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
reset_query_tables_list()
|
|
|
|
init TRUE - we should perform full initialization of object with
|
|
|
|
allocating needed memory
|
|
|
|
FALSE - object is already initialized so we should only reset
|
|
|
|
its state so it can be used for parsing/processing
|
|
|
|
of new statement
|
|
|
|
|
|
|
|
DESCRIPTION
|
|
|
|
This method initializes Query_tables_list so it can be used as part
|
|
|
|
of LEX object for parsing/processing of statement. One can also use
|
|
|
|
this method to reset state of already initialized Query_tables_list
|
|
|
|
so it can be used for processing of new statement.
|
|
|
|
*/
|
|
|
|
|
|
|
|
void Query_tables_list::reset_query_tables_list(bool init)
|
|
|
|
{
|
2007-03-02 17:43:45 +01:00
|
|
|
if (!init && query_tables)
|
|
|
|
{
|
|
|
|
TABLE_LIST *table= query_tables;
|
|
|
|
for (;;)
|
|
|
|
{
|
|
|
|
delete table->view;
|
|
|
|
if (query_tables_last == &table->next_global ||
|
|
|
|
!(table= table->next_global))
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2006-05-30 07:45:23 +02:00
|
|
|
query_tables= 0;
|
|
|
|
query_tables_last= &query_tables;
|
|
|
|
query_tables_own_last= 0;
|
|
|
|
if (init)
|
2006-11-01 13:41:48 +01:00
|
|
|
{
|
|
|
|
/*
|
|
|
|
We delay real initialization of hash (and therefore related
|
|
|
|
memory allocation) until first insertion into this hash.
|
|
|
|
*/
|
|
|
|
hash_clear(&sroutines);
|
|
|
|
}
|
2006-05-30 07:45:23 +02:00
|
|
|
else if (sroutines.records)
|
2006-11-01 13:41:48 +01:00
|
|
|
{
|
|
|
|
/* Non-zero sroutines.records means that hash was initialized. */
|
2006-05-30 07:45:23 +02:00
|
|
|
my_hash_reset(&sroutines);
|
2006-11-01 13:41:48 +01:00
|
|
|
}
|
2006-05-30 07:45:23 +02:00
|
|
|
sroutines_list.empty();
|
|
|
|
sroutines_list_own_last= sroutines_list.next;
|
|
|
|
sroutines_list_own_elements= 0;
|
2007-05-14 14:45:38 +02:00
|
|
|
binlog_stmt_flags= 0;
|
2006-05-30 07:45:23 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
Destroy Query_tables_list object with freeing all resources used by it.
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
destroy_query_tables_list()
|
|
|
|
*/
|
|
|
|
|
|
|
|
void Query_tables_list::destroy_query_tables_list()
|
|
|
|
{
|
|
|
|
hash_free(&sroutines);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2005-07-09 19:51:59 +02:00
|
|
|
/*
|
|
|
|
Initialize LEX object.
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
st_lex::st_lex()
|
|
|
|
|
|
|
|
NOTE
|
|
|
|
LEX object initialized with this constructor can be used as part of
|
|
|
|
THD object for which one can safely call open_tables(), lock_tables()
|
|
|
|
and close_thread_tables() functions. But it is not yet ready for
|
|
|
|
statement parsing. On should use lex_start() function to prepare LEX
|
|
|
|
for this.
|
|
|
|
*/
|
|
|
|
|
|
|
|
st_lex::st_lex()
|
2006-05-04 18:39:47 +02:00
|
|
|
:result(0), yacc_yyss(0), yacc_yyvs(0),
|
2007-05-17 10:27:39 +02:00
|
|
|
sql_command(SQLCOM_END), option_type(OPT_DEFAULT)
|
2005-07-09 19:51:59 +02:00
|
|
|
{
|
2007-03-23 18:14:46 +01:00
|
|
|
|
2007-03-02 17:43:45 +01:00
|
|
|
my_init_dynamic_array2(&plugins, sizeof(plugin_ref),
|
|
|
|
plugins_static_buffer,
|
|
|
|
INITIAL_LEX_PLUGIN_LIST_SIZE,
|
|
|
|
INITIAL_LEX_PLUGIN_LIST_SIZE);
|
2006-05-30 07:45:23 +02:00
|
|
|
reset_query_tables_list(TRUE);
|
2005-07-09 19:51:59 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2004-07-16 00:15:55 +02:00
|
|
|
/*
|
2004-10-07 00:45:06 +02:00
|
|
|
Check whether the merging algorithm can be used on this VIEW
|
2004-07-16 00:15:55 +02:00
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
st_lex::can_be_merged()
|
|
|
|
|
2004-10-07 00:45:06 +02:00
|
|
|
DESCRIPTION
|
2004-10-07 21:54:31 +02:00
|
|
|
We can apply merge algorithm if it is single SELECT view with
|
|
|
|
subqueries only in WHERE clause (we do not count SELECTs of underlying
|
|
|
|
views, and second level subqueries) and we have not grpouping, ordering,
|
2004-10-07 00:45:06 +02:00
|
|
|
HAVING clause, aggregate functions, DISTINCT clause, LIMIT clause and
|
|
|
|
several underlying tables.
|
|
|
|
|
2004-07-16 00:15:55 +02:00
|
|
|
RETURN
|
|
|
|
FALSE - only temporary table algorithm can be used
|
|
|
|
TRUE - merge algorithm can be used
|
|
|
|
*/
|
|
|
|
|
|
|
|
bool st_lex::can_be_merged()
|
|
|
|
{
|
|
|
|
// TODO: do not forget implement case when select_lex.table_list.elements==0
|
|
|
|
|
|
|
|
/* find non VIEW subqueries/unions */
|
2004-10-07 21:54:31 +02:00
|
|
|
bool selects_allow_merge= select_lex.next_select() == 0;
|
|
|
|
if (selects_allow_merge)
|
|
|
|
{
|
2006-12-14 23:51:37 +01:00
|
|
|
for (SELECT_LEX_UNIT *tmp_unit= select_lex.first_inner_unit();
|
|
|
|
tmp_unit;
|
|
|
|
tmp_unit= tmp_unit->next_unit())
|
2004-10-07 21:54:31 +02:00
|
|
|
{
|
2006-12-14 23:51:37 +01:00
|
|
|
if (tmp_unit->first_select()->parent_lex == this &&
|
|
|
|
(tmp_unit->item == 0 ||
|
|
|
|
(tmp_unit->item->place() != IN_WHERE &&
|
|
|
|
tmp_unit->item->place() != IN_ON)))
|
2004-10-07 21:54:31 +02:00
|
|
|
{
|
|
|
|
selects_allow_merge= 0;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return (selects_allow_merge &&
|
2004-07-16 00:15:55 +02:00
|
|
|
select_lex.group_list.elements == 0 &&
|
|
|
|
select_lex.having == 0 &&
|
2004-08-26 13:34:56 +02:00
|
|
|
select_lex.with_sum_func == 0 &&
|
2004-09-14 18:28:29 +02:00
|
|
|
select_lex.table_list.elements >= 1 &&
|
2004-07-16 00:15:55 +02:00
|
|
|
!(select_lex.options & SELECT_DISTINCT) &&
|
2005-06-07 12:11:36 +02:00
|
|
|
select_lex.select_limit == 0);
|
2004-07-16 00:15:55 +02:00
|
|
|
}
|
|
|
|
|
2004-10-07 00:45:06 +02:00
|
|
|
|
2004-07-16 00:15:55 +02:00
|
|
|
/*
|
2004-08-24 21:51:23 +02:00
|
|
|
check if command can use VIEW with MERGE algorithm (for top VIEWs)
|
2004-07-16 00:15:55 +02:00
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
st_lex::can_use_merged()
|
|
|
|
|
2004-10-07 00:45:06 +02:00
|
|
|
DESCRIPTION
|
|
|
|
Only listed here commands can use merge algorithm in top level
|
|
|
|
SELECT_LEX (for subqueries will be used merge algorithm if
|
|
|
|
st_lex::can_not_use_merged() is not TRUE).
|
|
|
|
|
2004-07-16 00:15:55 +02:00
|
|
|
RETURN
|
|
|
|
FALSE - command can't use merged VIEWs
|
|
|
|
TRUE - VIEWs with MERGE algorithms can be used
|
|
|
|
*/
|
|
|
|
|
|
|
|
bool st_lex::can_use_merged()
|
|
|
|
{
|
|
|
|
switch (sql_command)
|
|
|
|
{
|
|
|
|
case SQLCOM_SELECT:
|
|
|
|
case SQLCOM_CREATE_TABLE:
|
|
|
|
case SQLCOM_UPDATE:
|
|
|
|
case SQLCOM_UPDATE_MULTI:
|
|
|
|
case SQLCOM_DELETE:
|
|
|
|
case SQLCOM_DELETE_MULTI:
|
|
|
|
case SQLCOM_INSERT:
|
|
|
|
case SQLCOM_INSERT_SELECT:
|
|
|
|
case SQLCOM_REPLACE:
|
|
|
|
case SQLCOM_REPLACE_SELECT:
|
2004-10-21 20:53:27 +02:00
|
|
|
case SQLCOM_LOAD:
|
2004-07-16 00:15:55 +02:00
|
|
|
return TRUE;
|
|
|
|
default:
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2004-08-24 21:51:23 +02:00
|
|
|
/*
|
2004-10-07 00:45:06 +02:00
|
|
|
Check if command can't use merged views in any part of command
|
2004-08-24 21:51:23 +02:00
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
st_lex::can_not_use_merged()
|
|
|
|
|
2004-10-07 00:45:06 +02:00
|
|
|
DESCRIPTION
|
|
|
|
Temporary table algorithm will be used on all SELECT levels for queries
|
|
|
|
listed here (see also st_lex::can_use_merged()).
|
|
|
|
|
2004-08-24 21:51:23 +02:00
|
|
|
RETURN
|
|
|
|
FALSE - command can't use merged VIEWs
|
|
|
|
TRUE - VIEWs with MERGE algorithms can be used
|
|
|
|
*/
|
|
|
|
|
|
|
|
bool st_lex::can_not_use_merged()
|
|
|
|
{
|
|
|
|
switch (sql_command)
|
|
|
|
{
|
|
|
|
case SQLCOM_CREATE_VIEW:
|
|
|
|
case SQLCOM_SHOW_CREATE:
|
2005-02-16 11:00:03 +01:00
|
|
|
/*
|
|
|
|
SQLCOM_SHOW_FIELDS is necessary to make
|
|
|
|
information schema tables working correctly with views.
|
|
|
|
see get_schema_tables_result function
|
|
|
|
*/
|
|
|
|
case SQLCOM_SHOW_FIELDS:
|
2004-08-24 21:51:23 +02:00
|
|
|
return TRUE;
|
|
|
|
default:
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2004-07-16 00:15:55 +02:00
|
|
|
/*
|
|
|
|
Detect that we need only table structure of derived table/view
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
only_view_structure()
|
|
|
|
|
|
|
|
RETURN
|
|
|
|
TRUE yes, we need only structure
|
|
|
|
FALSE no, we need data
|
|
|
|
*/
|
2004-11-24 18:48:30 +01:00
|
|
|
|
2004-07-16 00:15:55 +02:00
|
|
|
bool st_lex::only_view_structure()
|
|
|
|
{
|
2005-07-01 06:05:42 +02:00
|
|
|
switch (sql_command) {
|
2004-07-16 00:15:55 +02:00
|
|
|
case SQLCOM_SHOW_CREATE:
|
|
|
|
case SQLCOM_SHOW_TABLES:
|
|
|
|
case SQLCOM_SHOW_FIELDS:
|
|
|
|
case SQLCOM_REVOKE_ALL:
|
|
|
|
case SQLCOM_REVOKE:
|
|
|
|
case SQLCOM_GRANT:
|
|
|
|
case SQLCOM_CREATE_VIEW:
|
|
|
|
return TRUE;
|
|
|
|
default:
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2004-11-24 18:48:30 +01:00
|
|
|
/*
|
|
|
|
Should Items_ident be printed correctly
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
need_correct_ident()
|
|
|
|
|
|
|
|
RETURN
|
|
|
|
TRUE yes, we need only structure
|
|
|
|
FALSE no, we need data
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
|
|
bool st_lex::need_correct_ident()
|
|
|
|
{
|
|
|
|
switch(sql_command)
|
|
|
|
{
|
|
|
|
case SQLCOM_SHOW_CREATE:
|
|
|
|
case SQLCOM_SHOW_TABLES:
|
|
|
|
case SQLCOM_CREATE_VIEW:
|
|
|
|
return TRUE;
|
|
|
|
default:
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-07-01 06:05:42 +02:00
|
|
|
/*
|
|
|
|
Get effective type of CHECK OPTION for given view
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
get_effective_with_check()
|
|
|
|
view given view
|
|
|
|
|
|
|
|
NOTE
|
|
|
|
It have not sense to set CHECK OPTION for SELECT satement or subqueries,
|
|
|
|
so we do not.
|
|
|
|
|
|
|
|
RETURN
|
|
|
|
VIEW_CHECK_NONE no need CHECK OPTION
|
|
|
|
VIEW_CHECK_LOCAL CHECK OPTION LOCAL
|
|
|
|
VIEW_CHECK_CASCADED CHECK OPTION CASCADED
|
|
|
|
*/
|
|
|
|
|
|
|
|
uint8 st_lex::get_effective_with_check(st_table_list *view)
|
|
|
|
{
|
|
|
|
if (view->select_lex->master_unit() == &unit &&
|
|
|
|
which_check_option_applicable())
|
|
|
|
return (uint8)view->with_check;
|
|
|
|
return VIEW_CHECK_NONE;
|
|
|
|
}
|
|
|
|
|
2004-11-24 18:48:30 +01:00
|
|
|
|
2004-03-29 21:40:49 +02:00
|
|
|
/*
|
|
|
|
initialize limit counters
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
st_select_lex_unit::set_limit()
|
|
|
|
values - SELECT_LEX with initial values for counters
|
|
|
|
*/
|
2004-07-16 00:15:55 +02:00
|
|
|
|
2005-06-07 12:11:36 +02:00
|
|
|
void st_select_lex_unit::set_limit(SELECT_LEX *sl)
|
2003-11-21 20:19:56 +01:00
|
|
|
{
|
2005-06-18 01:55:42 +02:00
|
|
|
ha_rows select_limit_val;
|
2005-06-07 12:11:36 +02:00
|
|
|
|
2005-09-02 15:21:19 +02:00
|
|
|
DBUG_ASSERT(! thd->stmt_arena->is_stmt_prepare());
|
2005-06-18 01:55:42 +02:00
|
|
|
select_limit_val= (ha_rows)(sl->select_limit ? sl->select_limit->val_uint() :
|
|
|
|
HA_POS_ERROR);
|
|
|
|
offset_limit_cnt= (ha_rows)(sl->offset_limit ? sl->offset_limit->val_uint() :
|
|
|
|
ULL(0));
|
2005-06-07 12:11:36 +02:00
|
|
|
select_limit_cnt= select_limit_val + offset_limit_cnt;
|
|
|
|
if (select_limit_cnt < select_limit_val)
|
2003-11-21 20:19:56 +01:00
|
|
|
select_limit_cnt= HA_POS_ERROR; // no limit
|
|
|
|
}
|
|
|
|
|
2004-04-10 00:14:32 +02:00
|
|
|
|
2004-04-07 23:16:17 +02:00
|
|
|
/*
|
2004-10-07 00:45:06 +02:00
|
|
|
Unlink the first table from the global table list and the first table from
|
|
|
|
outer select (lex->select_lex) local list
|
2004-04-07 23:16:17 +02:00
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
unlink_first_table()
|
2004-10-07 00:45:06 +02:00
|
|
|
link_to_local Set to 1 if caller should link this table to local list
|
2004-04-12 02:26:32 +02:00
|
|
|
|
2004-07-16 00:15:55 +02:00
|
|
|
NOTES
|
2004-10-07 00:45:06 +02:00
|
|
|
We assume that first tables in both lists is the same table or the local
|
|
|
|
list is empty.
|
2004-04-07 23:16:17 +02:00
|
|
|
|
|
|
|
RETURN
|
2004-09-03 20:43:04 +02:00
|
|
|
0 If 'query_tables' == 0
|
2004-07-16 00:15:55 +02:00
|
|
|
unlinked table
|
2004-09-03 20:43:04 +02:00
|
|
|
In this case link_to_local is set.
|
2004-04-12 02:26:32 +02:00
|
|
|
|
2004-04-07 23:16:17 +02:00
|
|
|
*/
|
2004-07-16 00:15:55 +02:00
|
|
|
TABLE_LIST *st_lex::unlink_first_table(bool *link_to_local)
|
2004-04-07 23:16:17 +02:00
|
|
|
{
|
2004-07-16 00:15:55 +02:00
|
|
|
TABLE_LIST *first;
|
|
|
|
if ((first= query_tables))
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
Exclude from global table list
|
|
|
|
*/
|
|
|
|
if ((query_tables= query_tables->next_global))
|
|
|
|
query_tables->prev_global= &query_tables;
|
2005-03-04 14:35:28 +01:00
|
|
|
else
|
|
|
|
query_tables_last= &query_tables;
|
2004-07-16 00:15:55 +02:00
|
|
|
first->next_global= 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
and from local list if it is not empty
|
|
|
|
*/
|
|
|
|
if ((*link_to_local= test(select_lex.table_list.first)))
|
|
|
|
{
|
2005-08-18 02:12:42 +02:00
|
|
|
select_lex.context.table_list=
|
|
|
|
select_lex.context.first_name_resolution_table= first->next_local;
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
|
|
|
select_lex.table_list.first= (uchar*) (first->next_local);
|
2004-07-16 00:15:55 +02:00
|
|
|
select_lex.table_list.elements--; //safety
|
|
|
|
first->next_local= 0;
|
|
|
|
/*
|
2004-10-07 00:45:06 +02:00
|
|
|
Ensure that the global list has the same first table as the local
|
|
|
|
list.
|
2004-07-16 00:15:55 +02:00
|
|
|
*/
|
|
|
|
first_lists_tables_same();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return first;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
Bring first local table of first most outer select to first place in global
|
|
|
|
table list
|
|
|
|
|
|
|
|
SYNOPSYS
|
|
|
|
st_lex::first_lists_tables_same()
|
|
|
|
|
|
|
|
NOTES
|
2004-10-07 00:45:06 +02:00
|
|
|
In many cases (for example, usual INSERT/DELETE/...) the first table of
|
|
|
|
main SELECT_LEX have special meaning => check that it is the first table
|
|
|
|
in global list and re-link to be first in the global list if it is
|
|
|
|
necessary. We need such re-linking only for queries with sub-queries in
|
|
|
|
the select list, as only in this case tables of sub-queries will go to
|
|
|
|
the global list first.
|
2004-07-16 00:15:55 +02:00
|
|
|
*/
|
|
|
|
|
|
|
|
void st_lex::first_lists_tables_same()
|
|
|
|
{
|
|
|
|
TABLE_LIST *first_table= (TABLE_LIST*) select_lex.table_list.first;
|
|
|
|
if (query_tables != first_table && first_table != 0)
|
|
|
|
{
|
2004-09-03 20:43:04 +02:00
|
|
|
TABLE_LIST *next;
|
2004-07-16 00:15:55 +02:00
|
|
|
if (query_tables_last == &first_table->next_global)
|
|
|
|
query_tables_last= first_table->prev_global;
|
2004-10-07 00:45:06 +02:00
|
|
|
|
2004-09-03 20:43:04 +02:00
|
|
|
if ((next= *first_table->prev_global= first_table->next_global))
|
2004-07-16 00:15:55 +02:00
|
|
|
next->prev_global= first_table->prev_global;
|
|
|
|
/* include in new place */
|
|
|
|
first_table->next_global= query_tables;
|
|
|
|
/*
|
2004-10-07 00:45:06 +02:00
|
|
|
We are sure that query_tables is not 0, because first_table was not
|
|
|
|
first table in the global list => we can use
|
|
|
|
query_tables->prev_global without check of query_tables
|
2004-07-16 00:15:55 +02:00
|
|
|
*/
|
|
|
|
query_tables->prev_global= &first_table->next_global;
|
|
|
|
first_table->prev_global= &query_tables;
|
|
|
|
query_tables= first_table;
|
|
|
|
}
|
2004-04-07 23:16:17 +02:00
|
|
|
}
|
|
|
|
|
2004-04-10 00:14:32 +02:00
|
|
|
|
2004-04-07 23:16:17 +02:00
|
|
|
/*
|
2004-04-10 00:14:32 +02:00
|
|
|
Link table back that was unlinked with unlink_first_table()
|
2004-04-07 23:16:17 +02:00
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
link_first_table_back()
|
2004-07-16 00:15:55 +02:00
|
|
|
link_to_local do we need link this table to local
|
2004-04-07 23:16:17 +02:00
|
|
|
|
|
|
|
RETURN
|
|
|
|
global list
|
|
|
|
*/
|
2004-07-16 00:15:55 +02:00
|
|
|
|
|
|
|
void st_lex::link_first_table_back(TABLE_LIST *first,
|
|
|
|
bool link_to_local)
|
2004-04-07 23:16:17 +02:00
|
|
|
{
|
2004-07-16 00:15:55 +02:00
|
|
|
if (first)
|
2004-04-07 23:16:17 +02:00
|
|
|
{
|
2004-07-16 00:15:55 +02:00
|
|
|
if ((first->next_global= query_tables))
|
|
|
|
query_tables->prev_global= &first->next_global;
|
2005-03-04 14:35:28 +01:00
|
|
|
else
|
|
|
|
query_tables_last= &first->next_global;
|
2004-07-16 00:15:55 +02:00
|
|
|
query_tables= first;
|
|
|
|
|
|
|
|
if (link_to_local)
|
|
|
|
{
|
|
|
|
first->next_local= (TABLE_LIST*) select_lex.table_list.first;
|
2005-08-12 16:57:19 +02:00
|
|
|
select_lex.context.table_list= first;
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
|
|
|
select_lex.table_list.first= (uchar*) first;
|
2004-07-16 00:15:55 +02:00
|
|
|
select_lex.table_list.elements++; //safety
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2005-07-05 12:37:02 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
cleanup lex for case when we open table by table for processing
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
st_lex::cleanup_after_one_table_open()
|
2006-05-30 07:45:23 +02:00
|
|
|
|
|
|
|
NOTE
|
|
|
|
This method is mostly responsible for cleaning up of selects lists and
|
|
|
|
derived tables state. To rollback changes in Query_tables_list one has
|
|
|
|
to call Query_tables_list::reset_query_tables_list(FALSE).
|
2005-07-05 12:37:02 +02:00
|
|
|
*/
|
|
|
|
|
|
|
|
void st_lex::cleanup_after_one_table_open()
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
thd->lex->derived_tables & additional units may be set if we open
|
|
|
|
a view. It is necessary to clear thd->lex->derived_tables flag
|
|
|
|
to prevent processing of derived tables during next open_and_lock_tables
|
|
|
|
if next table is a real table and cleanup & remove underlying units
|
|
|
|
NOTE: all units will be connected to thd->lex->select_lex, because we
|
|
|
|
have not UNION on most upper level.
|
|
|
|
*/
|
|
|
|
if (all_selects_list != &select_lex)
|
|
|
|
{
|
|
|
|
derived_tables= 0;
|
|
|
|
/* cleunup underlying units (units of VIEW) */
|
|
|
|
for (SELECT_LEX_UNIT *un= select_lex.first_inner_unit();
|
|
|
|
un;
|
|
|
|
un= un->next_unit())
|
|
|
|
un->cleanup();
|
|
|
|
/* reduce all selects list to default state */
|
|
|
|
all_selects_list= &select_lex;
|
|
|
|
/* remove underlying units (units of VIEW) subtree */
|
|
|
|
select_lex.cut_subtree();
|
|
|
|
}
|
2006-05-30 07:45:23 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
Save current state of Query_tables_list for this LEX, and prepare it
|
|
|
|
for processing of new statemnt.
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
reset_n_backup_query_tables_list()
|
|
|
|
backup Pointer to Query_tables_list instance to be used for backup
|
|
|
|
*/
|
|
|
|
|
|
|
|
void st_lex::reset_n_backup_query_tables_list(Query_tables_list *backup)
|
|
|
|
{
|
|
|
|
backup->set_query_tables_list(this);
|
|
|
|
/*
|
|
|
|
We have to perform full initialization here since otherwise we
|
|
|
|
will damage backed up state.
|
|
|
|
*/
|
|
|
|
this->reset_query_tables_list(TRUE);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
Restore state of Query_tables_list for this LEX from backup.
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
restore_backup_query_tables_list()
|
|
|
|
backup Pointer to Query_tables_list instance used for backup
|
|
|
|
*/
|
|
|
|
|
|
|
|
void st_lex::restore_backup_query_tables_list(Query_tables_list *backup)
|
|
|
|
{
|
|
|
|
this->destroy_query_tables_list();
|
|
|
|
this->set_query_tables_list(backup);
|
2005-07-05 12:37:02 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2006-11-02 13:51:43 +01:00
|
|
|
/*
|
|
|
|
Checks for usage of routines and/or tables in a parsed statement
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
st_lex:table_or_sp_used()
|
|
|
|
|
|
|
|
RETURN
|
|
|
|
FALSE No routines and tables used
|
|
|
|
TRUE Either or both routines and tables are used.
|
|
|
|
*/
|
|
|
|
|
|
|
|
bool st_lex::table_or_sp_used()
|
|
|
|
{
|
|
|
|
DBUG_ENTER("table_or_sp_used");
|
|
|
|
|
|
|
|
if (sroutines.records || query_tables)
|
|
|
|
DBUG_RETURN(TRUE);
|
|
|
|
|
|
|
|
DBUG_RETURN(FALSE);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2005-11-02 05:05:19 +01:00
|
|
|
/*
|
|
|
|
Do end-of-prepare fixup for list of tables and their merge-VIEWed tables
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
fix_prepare_info_in_table_list()
|
|
|
|
thd Thread handle
|
|
|
|
tbl List of tables to process
|
|
|
|
|
|
|
|
DESCRIPTION
|
|
|
|
Perform end-end-of prepare fixup for list of tables, if any of the tables
|
|
|
|
is a merge-algorithm VIEW, recursively fix up its underlying tables as
|
|
|
|
well.
|
|
|
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
static void fix_prepare_info_in_table_list(THD *thd, TABLE_LIST *tbl)
|
|
|
|
{
|
|
|
|
for (; tbl; tbl= tbl->next_local)
|
|
|
|
{
|
|
|
|
if (tbl->on_expr)
|
|
|
|
{
|
|
|
|
tbl->prep_on_expr= tbl->on_expr;
|
|
|
|
tbl->on_expr= tbl->on_expr->copy_andor_structure(thd);
|
|
|
|
}
|
|
|
|
fix_prepare_info_in_table_list(thd, tbl->merge_underlying_list);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2004-07-16 00:15:55 +02:00
|
|
|
/*
|
2006-09-16 18:50:48 +02:00
|
|
|
Save WHERE/HAVING/ON clauses and replace them with disposable copies
|
2004-07-16 00:15:55 +02:00
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
st_select_lex::fix_prepare_information
|
2006-09-16 18:50:48 +02:00
|
|
|
thd thread handler
|
|
|
|
conds in/out pointer to WHERE condition to be met at execution
|
|
|
|
having_conds in/out pointer to HAVING condition to be met at execution
|
|
|
|
|
|
|
|
DESCRIPTION
|
|
|
|
The passed WHERE and HAVING are to be saved for the future executions.
|
|
|
|
This function saves it, and returns a copy which can be thrashed during
|
|
|
|
this execution of the statement. By saving/thrashing here we mean only
|
|
|
|
AND/OR trees.
|
|
|
|
The function also calls fix_prepare_info_in_table_list that saves all
|
|
|
|
ON expressions.
|
2004-07-16 00:15:55 +02:00
|
|
|
*/
|
|
|
|
|
2006-09-16 18:50:48 +02:00
|
|
|
void st_select_lex::fix_prepare_information(THD *thd, Item **conds,
|
|
|
|
Item **having_conds)
|
2004-07-16 00:15:55 +02:00
|
|
|
{
|
2005-09-02 15:21:19 +02:00
|
|
|
if (!thd->stmt_arena->is_conventional() && first_execution)
|
2004-07-16 00:15:55 +02:00
|
|
|
{
|
|
|
|
first_execution= 0;
|
2005-07-01 06:05:42 +02:00
|
|
|
if (*conds)
|
|
|
|
{
|
|
|
|
prep_where= *conds;
|
|
|
|
*conds= where= prep_where->copy_andor_structure(thd);
|
|
|
|
}
|
2006-09-16 18:50:48 +02:00
|
|
|
if (*having_conds)
|
|
|
|
{
|
|
|
|
prep_having= *having_conds;
|
|
|
|
*having_conds= having= prep_having->copy_andor_structure(thd);
|
|
|
|
}
|
2005-11-02 05:05:19 +01:00
|
|
|
fix_prepare_info_in_table_list(thd, (TABLE_LIST *)table_list.first);
|
2004-04-07 23:16:17 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-11-02 13:51:43 +01:00
|
|
|
|
2002-11-23 17:54:15 +01:00
|
|
|
/*
|
2004-07-16 00:15:55 +02:00
|
|
|
There are st_select_lex::add_table_to_list &
|
2004-02-08 19:14:13 +01:00
|
|
|
st_select_lex::set_lock_for_tables are in sql_parse.cc
|
2003-10-16 14:54:47 +02:00
|
|
|
|
2004-07-16 00:15:55 +02:00
|
|
|
st_select_lex::print is in sql_select.cc
|
2004-02-08 19:14:13 +01:00
|
|
|
|
|
|
|
st_select_lex_unit::prepare, st_select_lex_unit::exec,
|
2004-05-07 22:06:11 +02:00
|
|
|
st_select_lex_unit::cleanup, st_select_lex_unit::reinit_exec_mechanism,
|
|
|
|
st_select_lex_unit::change_result
|
2004-02-08 19:14:13 +01:00
|
|
|
are in sql_union.cc
|
2002-11-23 17:54:15 +01:00
|
|
|
*/
|
2005-07-01 06:05:42 +02:00
|
|
|
|
2007-03-05 18:08:41 +01:00
|
|
|
/*
|
|
|
|
Sets the kind of hints to be added by the calls to add_index_hint().
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
set_index_hint_type()
|
|
|
|
type the kind of hints to be added from now on.
|
|
|
|
clause the clause to use for hints to be added from now on.
|
|
|
|
|
|
|
|
DESCRIPTION
|
|
|
|
Used in filling up the tagged hints list.
|
|
|
|
This list is filled by first setting the kind of the hint as a
|
|
|
|
context variable and then adding hints of the current kind.
|
|
|
|
Then the context variable index_hint_type can be reset to the
|
|
|
|
next hint type.
|
|
|
|
*/
|
|
|
|
void st_select_lex::set_index_hint_type(enum index_hint_type type,
|
|
|
|
index_clause_map clause)
|
|
|
|
{
|
|
|
|
current_index_hint_type= type;
|
|
|
|
current_index_hint_clause= clause;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
Makes an array to store index usage hints (ADD/FORCE/IGNORE INDEX).
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
alloc_index_hints()
|
|
|
|
thd current thread.
|
|
|
|
*/
|
|
|
|
|
|
|
|
void st_select_lex::alloc_index_hints (THD *thd)
|
|
|
|
{
|
|
|
|
index_hints= new (thd->mem_root) List<index_hint>();
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
adds an element to the array storing index usage hints
|
|
|
|
(ADD/FORCE/IGNORE INDEX).
|
|
|
|
|
|
|
|
SYNOPSIS
|
|
|
|
add_index_hint()
|
|
|
|
thd current thread.
|
|
|
|
str name of the index.
|
|
|
|
length number of characters in str.
|
|
|
|
|
|
|
|
RETURN VALUE
|
|
|
|
0 on success, non-zero otherwise
|
|
|
|
*/
|
|
|
|
bool st_select_lex::add_index_hint (THD *thd, char *str, uint length)
|
|
|
|
{
|
|
|
|
return index_hints->push_front (new (thd->mem_root)
|
|
|
|
index_hint(current_index_hint_type,
|
|
|
|
current_index_hint_clause,
|
|
|
|
str, length));
|
|
|
|
}
|
5.1 version of a fix and test cases for bugs:
Bug#4968 ""Stored procedure crash if cursor opened on altered table"
Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing"
Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from
stored procedure."
Bug#19733 "Repeated alter, or repeated create/drop, fails"
Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server"
Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a
growing key length" (this bug is not fixed in 5.0)
Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE
statements in stored routines or as prepared statements caused
incorrect results (and crashes in versions prior to 5.0.25).
In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE
SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options).
The problem of bugs 4968, 19733, 19282 and 6895 was that functions
mysql_prepare_table, mysql_create_table and mysql_alter_table are not
re-execution friendly: during their operation they modify contents
of LEX (members create_info, alter_info, key_list, create_list),
thus making the LEX unusable for the next execution.
In particular, these functions removed processed columns and keys from
create_list, key_list and drop_list. Search the code in sql_table.cc
for drop_it.remove() and similar patterns to find evidence.
The fix is to supply to these functions a usable copy of each of the
above structures at every re-execution of an SQL statement.
To simplify memory management, LEX::key_list and LEX::create_list
were added to LEX::alter_info, a fresh copy of which is created for
every execution.
The problem of crashing bug 22060 stemmed from the fact that the above
metnioned functions were not only modifying HA_CREATE_INFO structure
in LEX, but also were changing it to point to areas in volatile memory
of the execution memory root.
The patch solves this problem by creating and using an on-stack
copy of HA_CREATE_INFO in mysql_execute_command.
Additionally, this patch splits the part of mysql_alter_table
that analizes and rewrites information from the parser into
a separate function - mysql_prepare_alter_table, in analogy with
mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
A routine used by the parser to decide whether we are specifying a full
|
|
|
|
partitioning or if only partitions to add or to split.
|
|
|
|
|
|
|
|
@note This needs to be outside of WITH_PARTITION_STORAGE_ENGINE since it
|
|
|
|
is used from the sql parser that doesn't have any #ifdef's
|
|
|
|
|
|
|
|
@retval TRUE Yes, it is part of a management partition command
|
|
|
|
@retval FALSE No, not a management partition command
|
|
|
|
*/
|
|
|
|
|
|
|
|
bool st_lex::is_partition_management() const
|
|
|
|
{
|
|
|
|
return (sql_command == SQLCOM_ALTER_TABLE &&
|
|
|
|
(alter_info.flags == ALTER_ADD_PARTITION ||
|
|
|
|
alter_info.flags == ALTER_REORGANIZE_PARTITION));
|
|
|
|
}
|
|
|
|
|