mariadb/sql/sql_trigger.h
unknown c3da2d12f7 Fix for bug #5888 "Triggers with nonexistent columns cause packets
out of order". (final version)

Now instead of binding Item_trigger_field to TABLE objects during
trigger definition parsing at table open, we perform pass through
special list of all such objects in trigger. This allows easily check
all references to fields in old/new version of row in trigger during
execution of CREATE TRIGGER statement (this is more courtesy for users
since we can't check everything anyway).
We also report that such reference is bad by returning error from
Item_trigger_field::fix_fields() method (instead of setup_field())
This means that if trigger is broken we will bark during trigger
execution instead of trigger definition parsing at table open.
(i.e. now we allow to open tables with broken triggers).


mysql-test/r/trigger.result:
  Added test which attempts to create trigger for table referencing to
  field which does not exist in this table.
mysql-test/t/trigger.test:
  Added test which attempts to create trigger for table referencing to
  field which does not exist in this table.
sql/item.cc:
  Item_trigger_field::setup_field() now returns void. If any error
  will occur we will report it at fix_fields() stage.
sql/item.h:
  Item_trigger_field:
  - Added next_trg_field member for linking all such objects in trigger
    in one list.
  - Also setup_field() now returns void. If any error will occur we will
    report it at fix_fields() stage.
sql/mysql_priv.h:
  Added SQL_LIST::push_back() method which allows to add another SQL_LIST
  to the end of this SQL_LIST.
sql/sp_head.cc:
  sp_head::init()/reset_lex()/restore_lex():
   In order to fill global LEX::trg_table_fields (list of all 
   Item_trigger_field objects for trigger) we should init the same list
   in LEX of substatement before its parsing and merge it to global list
   after parsing.
sql/sp_head.h:
  sp_instr_trigger_field:
    Made trigger_field member public to be able to add it more easily to
    global list of all Item_trigger_field objects in trigger.
sql/sql_lex.cc:
  LEX::trg_table was removed.
sql/sql_lex.h:
  Now we are binding Item_trigger_field's to TABLE object by passing
  through specially constructed list of all such objects in this trigger
  instead of doing this during trigger definition parsing at table open.
  So we no longer need LEX::trg_table, we use LEX::trg_table_fields list
  instead.
sql/sql_parse.cc:
  mysql_execute_command():
    Since now we use trigger body for some checks in
    mysql_create_or_drop_trigger() we should destroy it only
    after calling this function.
sql/sql_trigger.cc:
  Now instead of binding Item_trigger_field to TABLE objects during
  trigger definition parsing at table open, we perform pass through
  special list of all such objects in trigger. This allows easily check
  all references to fields in old/new version of row in trigger during
  execution of CREATE TRIGGER statement (this is more courtesy for users
  since we can't check everything anyway).
  We also report that such reference is bad by returning error from
  Item_trigger_field::fix_fields() method (instead of setup_field())
  This means that if trigger is broken we will bark during trigger
  execution instead of trigger definition parsing at table open.
  (i.e. now we allow to open tables with broken triggers).
  
  Table_triggers_list::prepare_old_row_accessors() method was added to be
  able to reuse code creating Field objects referencing TABLE::record[1]
  buffer instead of TABLE::record[0].
sql/sql_trigger.h:
  Added Table_triggers_list::prepare_old_row_accessors() method to be
  able to reuse code creating Field objects referencing to TABLE::record[1]
  instead of record[0].
sql/sql_yacc.yy:
  Now instead of performing binding of Item_trigger_field objects
  to TABLE object during trigger definition parsing at table open,
  we perform this binding by passing through specially constructed
  list of all such items in trigger.
  We also check value returned from memory allocation functions.
2004-11-24 12:24:02 +03:00

71 lines
1.9 KiB
C++

/*
This class holds all information about triggers of table.
QQ: Will it be merged into TABLE in future ?
*/
class Table_triggers_list: public Sql_alloc
{
/* Triggers as SPs grouped by event, action_time */
sp_head *bodies[3][2];
/*
Copy of TABLE::Field array with field pointers set to old version
of record, used for OLD values in trigger on UPDATE.
*/
Field **old_field;
/*
Names of triggers.
Should correspond to order of triggers on definitions_list,
used in CREATE/DROP TRIGGER for looking up trigger by name.
*/
List<LEX_STRING> names_list;
public:
/*
Field responsible for storing triggers definitions in file.
It have to be public because we are using it directly from parser.
*/
List<LEX_STRING> definitions_list;
Table_triggers_list():
old_field(0)
{
bzero((char *)bodies, sizeof(bodies));
}
~Table_triggers_list();
bool create_trigger(THD *thd, TABLE_LIST *table);
bool drop_trigger(THD *thd, TABLE_LIST *table);
bool process_triggers(THD *thd, trg_event_type event,
trg_action_time_type time_type)
{
int res= 0;
if (bodies[event][time_type])
{
/*
Similar to function invocation we don't need to surpress sending of
ok packets here because don't allow execute statements from trigger.
FIXME: We should juggle with security context here (because trigger
should be invoked with creator rights).
*/
res= bodies[event][time_type]->execute_function(thd, 0, 0, 0);
}
return res;
}
static bool check_n_load(THD *thd, const char *db, const char *table_name,
TABLE *table);
bool has_delete_triggers()
{
return (bodies[TRG_EVENT_DELETE][TRG_ACTION_BEFORE] ||
bodies[TRG_EVENT_DELETE][TRG_ACTION_AFTER]);
}
friend class Item_trigger_field;
private:
bool prepare_old_row_accessors(TABLE *table);
};