mariadb/sql/item_cmpfunc.h

1629 lines
49 KiB
C
Raw Normal View History

2003-02-04 20:52:14 +01:00
/* Copyright (C) 2000-2003 MySQL AB
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
the Free Software Foundation; version 2 of the License.
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.
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 */
/* compare and test functions */
#ifdef USE_PRAGMA_INTERFACE
2000-07-31 21:29:14 +02:00
#pragma interface /* gcc class implementation */
#endif
extern Item_result item_cmp_type(Item_result a,Item_result b);
class Item_bool_func2;
class Arg_comparator;
typedef int (Arg_comparator::*arg_cmp_func)();
typedef int (*Item_field_cmpfunc)(Item_field *f1, Item_field *f2, void *arg);
class Arg_comparator: public Sql_alloc
{
Item **a, **b;
arg_cmp_func func;
Item_bool_func2 *owner;
Arg_comparator *comparators; // used only for compare_row()
double precision;
public:
DTCollation cmp_collation;
Arg_comparator() {};
Arg_comparator(Item **a1, Item **a2): a(a1), b(a2) {};
int set_compare_func(Item_bool_func2 *owner, Item_result type);
inline int set_compare_func(Item_bool_func2 *owner_arg)
{
return set_compare_func(owner_arg, item_cmp_type((*a)->result_type(),
2006-06-14 23:48:41 +02:00
(*b)->result_type()));
}
inline int set_cmp_func(Item_bool_func2 *owner_arg,
Item **a1, Item **a2,
Item_result type)
{
a= a1;
b= a2;
return set_compare_func(owner_arg, type);
}
inline int set_cmp_func(Item_bool_func2 *owner_arg,
Item **a1, Item **a2)
{
2006-06-14 23:48:41 +02:00
return set_cmp_func(owner_arg, a1, a2,
item_cmp_type((*a1)->result_type(),
(*a2)->result_type()));
}
inline int compare() { return (this->*func)(); }
int compare_string(); // compare args[0] & args[1]
int compare_binary_string(); // compare args[0] & args[1]
int compare_real(); // compare args[0] & args[1]
2005-02-08 23:50:45 +01:00
int compare_decimal(); // compare args[0] & args[1]
int compare_int_signed(); // compare args[0] & args[1]
int compare_int_signed_unsigned();
int compare_int_unsigned_signed();
int compare_int_unsigned();
int compare_row(); // compare args[0] & args[1]
int compare_e_string(); // compare args[0] & args[1]
int compare_e_binary_string(); // compare args[0] & args[1]
int compare_e_real(); // compare args[0] & args[1]
2005-02-08 23:50:45 +01:00
int compare_e_decimal(); // compare args[0] & args[1]
int compare_e_int(); // compare args[0] & args[1]
int compare_e_int_diff_signedness();
int compare_e_row(); // compare args[0] & args[1]
int compare_real_fixed();
int compare_e_real_fixed();
2005-02-08 23:50:45 +01:00
static arg_cmp_func comparator_matrix [5][2];
friend class Item_func;
};
2000-07-31 21:29:14 +02:00
class Item_bool_func :public Item_int_func
{
public:
Item_bool_func() :Item_int_func() {}
Item_bool_func(Item *a) :Item_int_func(a) {}
Item_bool_func(Item *a,Item *b) :Item_int_func(a,b) {}
Item_bool_func(THD *thd, Item_bool_func *item) :Item_int_func(thd, item) {}
2004-08-31 20:10:57 +02:00
bool is_bool_func() { return 1; }
2000-07-31 21:29:14 +02:00
void fix_length_and_dec() { decimals=0; max_length=1; }
uint decimal_precision() const { return 1; }
2000-07-31 21:29:14 +02:00
};
Bug#24532 (The return data type of IS TRUE is different from similar operations) Before this change, the boolean predicates: - X IS TRUE, - X IS NOT TRUE, - X IS FALSE, - X IS NOT FALSE were implemented by expanding the Item tree in the parser, by using a construct like: Item_func_if(Item_func_ifnull(X, <value>), <value>, <value>) Each <value> was a constant integer, either 0 or 1. A bug in the implementation of the function IF(a, b, c), in Item_func_if::fix_length_and_dec(), would cause the following : When the arguments b and c are both unsigned, the result type of the function was signed, instead of unsigned. When the result of the if function is signed, space for the sign could be counted twice (in the max() expression for a signed argument, and in the total), causing the member max_length to be too high. An effect of this is that the final type of IF(x, int(1), int(1)) would be int(2) instead of int(1). With this fix, the problems found in Item_func_if::fix_length_and_dec() have been fixed. While it's semantically correct to represent 'X IS TRUE' with Item_func_if(Item_func_ifnull(X, <value>), <value>, <value>), there are however more problems with this construct. a) Building the parse tree involves : - creating 5 Item instances (3 ints, 1 ifnull, 1 if), - creating each Item calls my_pthread_getspecific_ptr() once in the operator new(size), and a second time in the Item::Item() constructor, resulting in a total of 10 calls to get the current thread. Evaluating the expression involves evaluating up to 4 nodes at runtime. This representation could be greatly simplified and improved. b) Transforming the parse tree internally with if(ifnull(...)) is fine as long as this transformation is internal to the server implementation. With views however, the result of the parse tree is later exposed by the ::print() functions, and stored as part of the view definition. Doing this has long term consequences: 1) The original semantic 'X IS TRUE' is lost, and replaced by the if(ifnull(...)) expression. As a result, SHOW CREATE VIEW does not restore the original code. 2) Should a future version of MySQL implement the SQL BOOLEAN data type for example, views created today using 'X IS NULL' can be exported using mysqldump, and imported again. Such views would be converted correctly and automatically to use a BOOLEAN column in the future version. With 'X IS TRUE' and the current implementations, views using these "boolean" predicates would not be converted during the export/import, and would use integer columns instead. The difference traces back to how SHOW CREATE VIEW preserves 'X IS NULL' but does not preserve the 'X IS TRUE' semantic. With this fix, internal representation of 'X IS TRUE' booleans predicates has changed, so that: - dedicated Item classes are created for each predicate, - only 1 Item is created to represent 1 predicate - my_pthread_getspecific_ptr() is invoked 1 time instead of 10 - SHOW CREATE VIEW preserves the original semantic, and prints 'X IS TRUE'. Note that, because of the fix in Item_func_if, views created before this fix will: - correctly use a int(1) type instead of int(2) for boolean predicates, - incorrectly print the if(ifnull(...), ...) expression in SHOW CREATE VIEW, since the original semantic (X IS TRUE) has been lost. - except for the syntax used in SHOW CREATE VIEW, these views will operate properly, no action is needed. Views created after this fix will operate correctly, and will preserve the original code semantic in SHOW CREATE VIEW.
2007-02-12 21:59:29 +01:00
/**
Abstract Item class, to represent <code>X IS [NOT] (TRUE | FALSE)</code>
boolean predicates.
*/
class Item_func_truth : public Item_bool_func
{
public:
virtual bool val_bool();
virtual longlong val_int();
virtual void fix_length_and_dec();
virtual void print(String *str);
protected:
Item_func_truth(Item *a, bool a_value, bool a_affirmative)
: Item_bool_func(a), value(a_value), affirmative(a_affirmative)
{}
~Item_func_truth()
{}
private:
/**
True for <code>X IS [NOT] TRUE</code>,
false for <code>X IS [NOT] FALSE</code> predicates.
*/
const bool value;
/**
True for <code>X IS Y</code>, false for <code>X IS NOT Y</code> predicates.
*/
const bool affirmative;
};
/**
This Item represents a <code>X IS TRUE</code> boolean predicate.
*/
class Item_func_istrue : public Item_func_truth
{
public:
Item_func_istrue(Item *a) : Item_func_truth(a, true, true) {}
~Item_func_istrue() {}
virtual const char* func_name() const { return "istrue"; }
};
/**
This Item represents a <code>X IS NOT TRUE</code> boolean predicate.
*/
class Item_func_isnottrue : public Item_func_truth
{
public:
Item_func_isnottrue(Item *a) : Item_func_truth(a, true, false) {}
~Item_func_isnottrue() {}
virtual const char* func_name() const { return "isnottrue"; }
};
/**
This Item represents a <code>X IS FALSE</code> boolean predicate.
*/
class Item_func_isfalse : public Item_func_truth
{
public:
Item_func_isfalse(Item *a) : Item_func_truth(a, false, true) {}
~Item_func_isfalse() {}
virtual const char* func_name() const { return "isfalse"; }
};
/**
This Item represents a <code>X IS NOT FALSE</code> boolean predicate.
*/
class Item_func_isnotfalse : public Item_func_truth
{
public:
Item_func_isnotfalse(Item *a) : Item_func_truth(a, false, false) {}
~Item_func_isnotfalse() {}
virtual const char* func_name() const { return "isnotfalse"; }
};
class Item_cache;
#define UNKNOWN ((my_bool)-1)
/*
Item_in_optimizer(left_expr, Item_in_subselect(...))
Item_in_optimizer is used to wrap an instance of Item_in_subselect. This
class does the following:
- Evaluate the left expression and store it in Item_cache_* object (to
avoid re-evaluating it many times during subquery execution)
- Shortcut the evaluation of "NULL IN (...)" to NULL in the cases where we
don't care if the result is NULL or FALSE.
NOTE
It is not quite clear why the above listed functionality should be
placed into a separate class called 'Item_in_optimizer'.
*/
class Item_in_optimizer: public Item_bool_func
{
protected:
Item_cache *cache;
bool save_cache;
/*
Stores the value of "NULL IN (SELECT ...)" for uncorrelated subqueries:
UNKNOWN - "NULL in (SELECT ...)" has not yet been evaluated
FALSE - result is FALSE
TRUE - result is NULL
*/
my_bool result_for_null_param;
public:
2002-12-19 20:15:09 +01:00
Item_in_optimizer(Item *a, Item_in_subselect *b):
Item_bool_func(a, my_reinterpret_cast(Item *)(b)), cache(0),
save_cache(0), result_for_null_param(UNKNOWN)
{}
bool fix_fields(THD *, Item **);
bool fix_left(THD *thd, Item **ref);
bool is_null();
longlong val_int();
void cleanup();
const char *func_name() const { return "<in_optimizer>"; }
2002-12-19 20:15:09 +01:00
Item_cache **get_cache() { return &cache; }
void keep_top_level_cache();
};
2003-11-03 11:28:36 +01:00
class Comp_creator
{
public:
Comp_creator() {} /* Remove gcc warning */
virtual ~Comp_creator() {} /* Remove gcc warning */
2003-11-03 11:28:36 +01:00
virtual Item_bool_func2* create(Item *a, Item *b) const = 0;
virtual const char* symbol(bool invert) const = 0;
virtual bool eqne_op() const = 0;
virtual bool l_op() const = 0;
};
class Eq_creator :public Comp_creator
{
public:
Eq_creator() {} /* Remove gcc warning */
virtual ~Eq_creator() {} /* Remove gcc warning */
2003-11-03 11:28:36 +01:00
virtual Item_bool_func2* create(Item *a, Item *b) const;
virtual const char* symbol(bool invert) const { return invert? "<>" : "="; }
virtual bool eqne_op() const { return 1; }
virtual bool l_op() const { return 0; }
};
class Ne_creator :public Comp_creator
{
public:
Ne_creator() {} /* Remove gcc warning */
virtual ~Ne_creator() {} /* Remove gcc warning */
2003-11-03 11:28:36 +01:00
virtual Item_bool_func2* create(Item *a, Item *b) const;
virtual const char* symbol(bool invert) const { return invert? "=" : "<>"; }
virtual bool eqne_op() const { return 1; }
virtual bool l_op() const { return 0; }
};
class Gt_creator :public Comp_creator
{
public:
Gt_creator() {} /* Remove gcc warning */
virtual ~Gt_creator() {} /* Remove gcc warning */
2003-11-03 11:28:36 +01:00
virtual Item_bool_func2* create(Item *a, Item *b) const;
virtual const char* symbol(bool invert) const { return invert? "<=" : ">"; }
virtual bool eqne_op() const { return 0; }
virtual bool l_op() const { return 0; }
};
class Lt_creator :public Comp_creator
{
public:
Lt_creator() {} /* Remove gcc warning */
virtual ~Lt_creator() {} /* Remove gcc warning */
2003-11-03 11:28:36 +01:00
virtual Item_bool_func2* create(Item *a, Item *b) const;
virtual const char* symbol(bool invert) const { return invert? ">=" : "<"; }
virtual bool eqne_op() const { return 0; }
virtual bool l_op() const { return 1; }
};
class Ge_creator :public Comp_creator
{
public:
Ge_creator() {} /* Remove gcc warning */
virtual ~Ge_creator() {} /* Remove gcc warning */
2003-11-03 11:28:36 +01:00
virtual Item_bool_func2* create(Item *a, Item *b) const;
virtual const char* symbol(bool invert) const { return invert? "<" : ">="; }
virtual bool eqne_op() const { return 0; }
virtual bool l_op() const { return 0; }
};
class Le_creator :public Comp_creator
{
public:
Le_creator() {} /* Remove gcc warning */
virtual ~Le_creator() {} /* Remove gcc warning */
2003-11-03 11:28:36 +01:00
virtual Item_bool_func2* create(Item *a, Item *b) const;
virtual const char* symbol(bool invert) const { return invert? ">" : "<="; }
virtual bool eqne_op() const { return 0; }
virtual bool l_op() const { return 1; }
};
2000-07-31 21:29:14 +02:00
class Item_bool_func2 :public Item_int_func
{ /* Bool with 2 string args */
protected:
Arg_comparator cmp;
2000-07-31 21:29:14 +02:00
String tmp_value1,tmp_value2;
bool abort_on_null;
2000-07-31 21:29:14 +02:00
public:
Item_bool_func2(Item *a,Item *b)
:Item_int_func(a,b), cmp(tmp_arg, tmp_arg+1), abort_on_null(FALSE) {}
2000-07-31 21:29:14 +02:00
void fix_length_and_dec();
void set_cmp_func()
2002-11-15 19:32:09 +01:00
{
cmp.set_cmp_func(this, tmp_arg, tmp_arg+1);
2002-11-15 19:32:09 +01:00
}
2000-07-31 21:29:14 +02:00
optimize_type select_optimize() const { return OPTIMIZE_OP; }
virtual enum Functype rev_functype() const { return UNKNOWN_FUNC; }
bool have_rev_func() const { return rev_functype() != UNKNOWN_FUNC; }
void print(String *str) { Item_func::print_op(str); }
bool is_null() { return test(args[0]->is_null() || args[1]->is_null()); }
2004-08-31 20:10:57 +02:00
bool is_bool_func() { return 1; }
CHARSET_INFO *compare_collation() { return cmp.cmp_collation.collation; }
uint decimal_precision() const { return 1; }
void top_level_item() { abort_on_null=1; }
friend class Arg_comparator;
2000-07-31 21:29:14 +02:00
};
2002-11-15 19:32:09 +01:00
class Item_bool_rowready_func2 :public Item_bool_func2
{
public:
A fix and test case for Bug#5987 "subselect in bool function crashes server (prepared statements)": the bug was that all boolean items always recovered its original arguments at statement cleanup stage. This collided with Item_subselect::select_transformer, which tries to permanently change the item tree to use a transformed subselect instead of original one. So we had this call sequence for prepare: mysql_stmt_prepare -> JOIN::prepare -> Item_subselect::fix_fields -> the item tree gets transformed -> Item_bool_rowready_func2::cleanup, item tree is recovered to original state, while it shouldn't have been; mysql_stmt_execute -> attempts to execute a broken tree -> crash. Now instead of bluntly recovering all arguments of bool functions in Item_bool_rowready_func2::cleanup, we recover only those which were changed, and do it in one place. There still would exist a possibility for a collision with subselect tranformation, if permanent and temporary changes were performed at the same stage. But fortunately subselect transformation is always done first, so it doesn't conflict with the optimization done by propogate_cond_constants. Now we have: mysql_stmt_prepare -> JOIN::prepare -> subselect transformation permanently changes the tree -> cleanup doesn't recover anything, because nothing was registered for recovery. mysql_stmt_execute -> JOIN::prepare (the tree is already transformed, so it doesn't change), JOIN::optimize -> propogate_cond_constants -> temporary changes the item tree with constants -> JOIN::execute -> cleanup -> the changes done by propogate_cond_constants are recovered, as they were registered for recovery.
2004-10-10 00:39:22 +02:00
Item_bool_rowready_func2(Item *a, Item *b) :Item_bool_func2(a, b)
2002-11-15 19:32:09 +01:00
{
allowed_arg_cols= 0; // Fetch this value from first argument
2002-11-15 19:32:09 +01:00
}
Item *neg_transformer(THD *thd);
virtual Item *negated_item();
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
bool subst_argument_checker(byte **arg) { return TRUE; }
2002-11-15 19:32:09 +01:00
};
2000-07-31 21:29:14 +02:00
class Item_func_not :public Item_bool_func
{
public:
Item_func_not(Item *a) :Item_bool_func(a) {}
longlong val_int();
enum Functype functype() const { return NOT_FUNC; }
2000-07-31 21:29:14 +02:00
const char *func_name() const { return "not"; }
Item *neg_transformer(THD *thd);
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
void print(String *str);
2000-07-31 21:29:14 +02:00
};
class Item_maxmin_subselect;
/*
trigcond<param>(arg) ::= param? arg : TRUE
The class Item_func_trig_cond is used for guarded predicates
which are employed only for internal purposes.
A guarded predicate is an object consisting of an a regular or
a guarded predicate P and a pointer to a boolean guard variable g.
A guarded predicate P/g is evaluated to true if the value of the
guard g is false, otherwise it is evaluated to the same value that
the predicate P: val(P/g)= g ? val(P):true.
Guarded predicates allow us to include predicates into a conjunction
conditionally. Currently they are utilized for pushed down predicates
in queries with outer join operations.
In the future, probably, it makes sense to extend this class to
the objects consisting of three elements: a predicate P, a pointer
to a variable g and a firing value s with following evaluation
rule: val(P/g,s)= g==s? val(P) : true. It will allow us to build only
one item for the objects of the form P/g1/g2...
Objects of this class are built only for query execution after
the execution plan has been already selected. That's why this
class needs only val_int out of generic methods.
Current uses of Item_func_trig_cond objects:
- To wrap selection conditions when executing outer joins
- To wrap condition that is pushed down into subquery
*/
class Item_func_trig_cond: public Item_bool_func
{
bool *trig_var;
public:
Item_func_trig_cond(Item *a, bool *f) : Item_bool_func(a) { trig_var= f; }
longlong val_int() { return *trig_var ? args[0]->val_int() : 1; }
enum Functype functype() const { return TRIG_COND_FUNC; };
const char *func_name() const { return "trigcond"; };
bool const_item() const { return FALSE; }
bool *get_trig_var() { return trig_var; }
};
class Item_func_not_all :public Item_func_not
{
2005-02-08 23:50:45 +01:00
/* allow to check presence of values in max/min optimization */
Item_sum_hybrid *test_sum_item;
Item_maxmin_subselect *test_sub_item;
bool abort_on_null;
public:
bool show;
Item_func_not_all(Item *a)
:Item_func_not(a), test_sum_item(0), test_sub_item(0), abort_on_null(0),
show(0)
{}
virtual void top_level_item() { abort_on_null= 1; }
bool top_level() { return abort_on_null; }
longlong val_int();
enum Functype functype() const { return NOT_ALL_FUNC; }
const char *func_name() const { return "<not>"; }
void print(String *str);
void set_sum_test(Item_sum_hybrid *item) { test_sum_item= item; };
void set_sub_test(Item_maxmin_subselect *item) { test_sub_item= item; };
bool empty_underlying_subquery();
Item *neg_transformer(THD *thd);
};
class Item_func_nop_all :public Item_func_not_all
{
public:
Item_func_nop_all(Item *a) :Item_func_not_all(a) {}
longlong val_int();
const char *func_name() const { return "<nop>"; }
Item *neg_transformer(THD *thd);
};
2002-11-15 19:32:09 +01:00
class Item_func_eq :public Item_bool_rowready_func2
2000-07-31 21:29:14 +02:00
{
public:
Item_func_eq(Item *a,Item *b) :Item_bool_rowready_func2(a,b) {}
2000-07-31 21:29:14 +02:00
longlong val_int();
enum Functype functype() const { return EQ_FUNC; }
enum Functype rev_functype() const { return EQ_FUNC; }
cond_result eq_cmp_result() const { return COND_TRUE; }
const char *func_name() const { return "="; }
Item *negated_item();
2000-07-31 21:29:14 +02:00
};
2002-11-28 08:11:35 +01:00
class Item_func_equal :public Item_bool_rowready_func2
2000-07-31 21:29:14 +02:00
{
public:
2002-11-28 19:58:23 +01:00
Item_func_equal(Item *a,Item *b) :Item_bool_rowready_func2(a,b) {};
2000-07-31 21:29:14 +02:00
longlong val_int();
void fix_length_and_dec();
table_map not_null_tables() const { return 0; }
2000-07-31 21:29:14 +02:00
enum Functype functype() const { return EQUAL_FUNC; }
enum Functype rev_functype() const { return EQUAL_FUNC; }
cond_result eq_cmp_result() const { return COND_TRUE; }
const char *func_name() const { return "<=>"; }
Item *neg_transformer(THD *thd) { return 0; }
2000-07-31 21:29:14 +02:00
};
2002-11-15 19:32:09 +01:00
class Item_func_ge :public Item_bool_rowready_func2
2000-07-31 21:29:14 +02:00
{
public:
Item_func_ge(Item *a,Item *b) :Item_bool_rowready_func2(a,b) {};
2000-07-31 21:29:14 +02:00
longlong val_int();
enum Functype functype() const { return GE_FUNC; }
enum Functype rev_functype() const { return LE_FUNC; }
cond_result eq_cmp_result() const { return COND_TRUE; }
const char *func_name() const { return ">="; }
Item *negated_item();
2000-07-31 21:29:14 +02:00
};
2002-11-15 19:32:09 +01:00
class Item_func_gt :public Item_bool_rowready_func2
2000-07-31 21:29:14 +02:00
{
public:
Item_func_gt(Item *a,Item *b) :Item_bool_rowready_func2(a,b) {};
2000-07-31 21:29:14 +02:00
longlong val_int();
enum Functype functype() const { return GT_FUNC; }
enum Functype rev_functype() const { return LT_FUNC; }
cond_result eq_cmp_result() const { return COND_FALSE; }
const char *func_name() const { return ">"; }
Item *negated_item();
2000-07-31 21:29:14 +02:00
};
2002-11-15 19:32:09 +01:00
class Item_func_le :public Item_bool_rowready_func2
2000-07-31 21:29:14 +02:00
{
public:
Item_func_le(Item *a,Item *b) :Item_bool_rowready_func2(a,b) {};
2000-07-31 21:29:14 +02:00
longlong val_int();
enum Functype functype() const { return LE_FUNC; }
enum Functype rev_functype() const { return GE_FUNC; }
cond_result eq_cmp_result() const { return COND_TRUE; }
const char *func_name() const { return "<="; }
Item *negated_item();
2000-07-31 21:29:14 +02:00
};
2002-11-15 19:32:09 +01:00
class Item_func_lt :public Item_bool_rowready_func2
2000-07-31 21:29:14 +02:00
{
public:
Item_func_lt(Item *a,Item *b) :Item_bool_rowready_func2(a,b) {}
2000-07-31 21:29:14 +02:00
longlong val_int();
enum Functype functype() const { return LT_FUNC; }
enum Functype rev_functype() const { return GT_FUNC; }
cond_result eq_cmp_result() const { return COND_FALSE; }
const char *func_name() const { return "<"; }
Item *negated_item();
2000-07-31 21:29:14 +02:00
};
2002-11-15 19:32:09 +01:00
class Item_func_ne :public Item_bool_rowready_func2
2000-07-31 21:29:14 +02:00
{
public:
Item_func_ne(Item *a,Item *b) :Item_bool_rowready_func2(a,b) {}
2000-07-31 21:29:14 +02:00
longlong val_int();
enum Functype functype() const { return NE_FUNC; }
cond_result eq_cmp_result() const { return COND_FALSE; }
optimize_type select_optimize() const { return OPTIMIZE_KEY; }
2000-07-31 21:29:14 +02:00
const char *func_name() const { return "<>"; }
Item *negated_item();
2000-07-31 21:29:14 +02:00
};
/*
The class Item_func_opt_neg is defined to factor out the functionality
common for the classes Item_func_between and Item_func_in. The objects
of these classes can express predicates or there negations.
The alternative approach would be to create pairs Item_func_between,
Item_func_notbetween and Item_func_in, Item_func_notin.
*/
class Item_func_opt_neg :public Item_int_func
{
public:
bool negated; /* <=> the item represents NOT <func> */
bool pred_level; /* <=> [NOT] <func> is used on a predicate level */
public:
Item_func_opt_neg(Item *a, Item *b, Item *c)
:Item_int_func(a, b, c), negated(0), pred_level(0) {}
Item_func_opt_neg(List<Item> &list)
:Item_int_func(list), negated(0), pred_level(0) {}
public:
inline void negate() { negated= !negated; }
inline void top_level_item() { pred_level= 1; }
Item *neg_transformer(THD *thd)
{
negated= !negated;
return this;
}
bool subst_argument_checker(byte **arg) { return TRUE; }
};
class Item_func_between :public Item_func_opt_neg
2000-07-31 21:29:14 +02:00
{
DTCollation cmp_collation;
2000-07-31 21:29:14 +02:00
public:
Item_result cmp_type;
String value0,value1,value2;
Item_func_between(Item *a, Item *b, Item *c)
:Item_func_opt_neg(a, b, c) {}
2000-07-31 21:29:14 +02:00
longlong val_int();
optimize_type select_optimize() const { return OPTIMIZE_KEY; }
enum Functype functype() const { return BETWEEN; }
const char *func_name() const { return "between"; }
bool fix_fields(THD *, Item **);
2000-07-31 21:29:14 +02:00
void fix_length_and_dec();
void print(String *str);
bool is_bool_func() { return 1; }
CHARSET_INFO *compare_collation() { return cmp_collation.collation; }
uint decimal_precision() const { return 1; }
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
2000-07-31 21:29:14 +02:00
};
class Item_func_strcmp :public Item_bool_func2
{
public:
Item_func_strcmp(Item *a,Item *b) :Item_bool_func2(a,b) {}
longlong val_int();
optimize_type select_optimize() const { return OPTIMIZE_NONE; }
const char *func_name() const { return "strcmp"; }
void print(String *str) { Item_func::print(str); }
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
2000-07-31 21:29:14 +02:00
};
2005-02-08 23:50:45 +01:00
struct interval_range
{
Item_result type;
double dbl;
my_decimal dec;
};
2000-07-31 21:29:14 +02:00
class Item_func_interval :public Item_int_func
{
Item_row *row;
2005-02-08 23:50:45 +01:00
my_bool use_decimal_comparison;
interval_range *intervals;
2000-07-31 21:29:14 +02:00
public:
Item_func_interval(Item_row *a)
:Item_int_func(a),row(a),intervals(0)
{
allowed_arg_cols= 0; // Fetch this value from first argument
}
2000-07-31 21:29:14 +02:00
longlong val_int();
void fix_length_and_dec();
const char *func_name() const { return "interval"; }
uint decimal_precision() const { return 2; }
2000-07-31 21:29:14 +02:00
};
class Item_func_coalesce :public Item_func_numhybrid
2000-07-31 21:29:14 +02:00
{
2005-02-08 23:50:45 +01:00
protected:
Item_func_coalesce(Item *a, Item *b) :Item_func_numhybrid(a, b) {}
2005-02-08 23:50:45 +01:00
public:
Item_func_coalesce(List<Item> &list) :Item_func_numhybrid(list) {}
double real_op();
longlong int_op();
String *str_op(String *);
my_decimal *decimal_op(my_decimal *);
2005-02-08 23:50:45 +01:00
void fix_length_and_dec();
void find_num_type() {}
enum Item_result result_type () const { return hybrid_type; }
2005-02-08 23:50:45 +01:00
const char *func_name() const { return "coalesce"; }
table_map not_null_tables() const { return 0; }
};
class Item_func_ifnull :public Item_func_coalesce
{
protected:
enum_field_types cached_field_type;
bool field_type_defined;
2000-07-31 21:29:14 +02:00
public:
2005-02-08 23:50:45 +01:00
Item_func_ifnull(Item *a, Item *b) :Item_func_coalesce(a,b) {}
double real_op();
longlong int_op();
String *str_op(String *str);
my_decimal *decimal_op(my_decimal *);
enum_field_types field_type() const;
2000-07-31 21:29:14 +02:00
void fix_length_and_dec();
const char *func_name() const { return "ifnull"; }
Field *tmp_table_field(TABLE *table);
uint decimal_precision() const;
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
2000-07-31 21:29:14 +02:00
};
class Item_func_if :public Item_func
{
enum Item_result cached_result_type;
public:
Item_func_if(Item *a,Item *b,Item *c)
:Item_func(a,b,c), cached_result_type(INT_RESULT)
{}
2004-11-11 19:39:35 +01:00
double val_real();
2000-07-31 21:29:14 +02:00
longlong val_int();
String *val_str(String *str);
2005-02-08 23:50:45 +01:00
my_decimal *val_decimal(my_decimal *);
2000-07-31 21:29:14 +02:00
enum Item_result result_type () const { return cached_result_type; }
bool fix_fields(THD *, Item **);
2000-07-31 21:29:14 +02:00
void fix_length_and_dec();
uint decimal_precision() const;
2000-07-31 21:29:14 +02:00
const char *func_name() const { return "if"; }
};
class Item_func_nullif :public Item_bool_func2
{
enum Item_result cached_result_type;
public:
Item_func_nullif(Item *a,Item *b)
:Item_bool_func2(a,b), cached_result_type(INT_RESULT)
{}
2004-11-11 19:39:35 +01:00
double val_real();
2000-07-31 21:29:14 +02:00
longlong val_int();
String *val_str(String *str);
2005-02-08 23:50:45 +01:00
my_decimal *val_decimal(my_decimal *);
2000-07-31 21:29:14 +02:00
enum Item_result result_type () const { return cached_result_type; }
void fix_length_and_dec();
uint decimal_precision() const { return args[0]->decimal_precision(); }
2000-07-31 21:29:14 +02:00
const char *func_name() const { return "nullif"; }
void print(String *str) { Item_func::print(str); }
table_map not_null_tables() const { return 0; }
bool is_null();
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
2000-07-31 21:29:14 +02:00
};
/* Functions to handle the optimized IN */
/* A vector of values of some type */
2000-07-31 21:29:14 +02:00
class in_vector :public Sql_alloc
{
public:
2000-07-31 21:29:14 +02:00
char *base;
uint size;
qsort2_cmp compare;
CHARSET_INFO *collation;
2000-07-31 21:29:14 +02:00
uint count;
uint used_count;
in_vector() {}
in_vector(uint elements,uint element_length,qsort2_cmp cmp_func,
CHARSET_INFO *cmp_coll)
2000-07-31 21:29:14 +02:00
:base((char*) sql_calloc(elements*element_length)),
size(element_length), compare(cmp_func), collation(cmp_coll),
count(elements), used_count(elements) {}
2000-07-31 21:29:14 +02:00
virtual ~in_vector() {}
virtual void set(uint pos,Item *item)=0;
virtual byte *get_value(Item *item)=0;
void sort()
{
qsort2(base,used_count,size,compare,collation);
}
2000-07-31 21:29:14 +02:00
int find(Item *item);
/*
Create an instance of Item_{type} (e.g. Item_decimal) constant object
which type allows it to hold an element of this vector without any
conversions.
The purpose of this function is to be able to get elements of this
vector in form of Item_xxx constants without creating Item_xxx object
for every array element you get (i.e. this implements "FlyWeight" pattern)
*/
virtual Item* create_item() { return NULL; }
/*
Store the value at position #pos into provided item object
SYNOPSIS
value_to_item()
pos Index of value to store
item Constant item to store value into. The item must be of the same
type that create_item() returns.
*/
virtual void value_to_item(uint pos, Item *item) { }
/* Compare values number pos1 and pos2 for equality */
bool compare_elems(uint pos1, uint pos2)
{
return test(compare(collation, base + pos1*size, base + pos2*size));
}
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
virtual Item_result result_type()= 0;
2000-07-31 21:29:14 +02:00
};
class in_string :public in_vector
{
2005-02-08 23:50:45 +01:00
char buff[STRING_BUFFER_USUAL_SIZE];
2000-07-31 21:29:14 +02:00
String tmp;
public:
in_string(uint elements,qsort2_cmp cmp_func, CHARSET_INFO *cs);
2000-07-31 21:29:14 +02:00
~in_string();
void set(uint pos,Item *item);
byte *get_value(Item *item);
Item* create_item()
{
return new Item_string(collation);
}
void value_to_item(uint pos, Item *item)
{
String *str=((String*) base)+pos;
Item_string *to= (Item_string*)item;
to->str_value= *str;
}
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
Item_result result_type() { return STRING_RESULT; }
2000-07-31 21:29:14 +02:00
};
class in_longlong :public in_vector
{
/*
Here we declare a temporary variable (tmp) of the same type as the
elements of this vector. tmp is used in finding if a given value is in
the list.
*/
struct packed_longlong
{
longlong val;
longlong unsigned_flag; // Use longlong, not bool, to preserve alignment
} tmp;
2000-07-31 21:29:14 +02:00
public:
in_longlong(uint elements);
void set(uint pos,Item *item);
byte *get_value(Item *item);
Item* create_item()
{
/*
We're created a signed INT, this may not be correct in
general case (see BUG#19342).
*/
2006-04-25 23:21:33 +02:00
return new Item_int((longlong)0);
}
void value_to_item(uint pos, Item *item)
{
((Item_int*) item)->value= ((packed_longlong*) base)[pos].val;
2007-03-02 18:32:46 +01:00
((Item_int*) item)->unsigned_flag= (my_bool)
((packed_longlong*) base)[pos].unsigned_flag;
}
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
Item_result result_type() { return INT_RESULT; }
friend int cmp_longlong(void *cmp_arg, packed_longlong *a,packed_longlong *b);
2000-07-31 21:29:14 +02:00
};
class in_double :public in_vector
{
double tmp;
public:
in_double(uint elements);
void set(uint pos,Item *item);
byte *get_value(Item *item);
Item *create_item()
{
return new Item_float(0.0);
}
void value_to_item(uint pos, Item *item)
{
((Item_float*)item)->value= ((double*) base)[pos];
}
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
Item_result result_type() { return REAL_RESULT; }
2000-07-31 21:29:14 +02:00
};
2005-02-08 23:50:45 +01:00
class in_decimal :public in_vector
{
my_decimal val;
public:
in_decimal(uint elements);
void set(uint pos, Item *item);
byte *get_value(Item *item);
Item *create_item()
{
return new Item_decimal(0, FALSE);
}
void value_to_item(uint pos, Item *item)
{
my_decimal *dec= ((my_decimal *)base) + pos;
Item_decimal *item_dec= (Item_decimal*)item;
item_dec->set_decimal_value(dec);
}
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
Item_result result_type() { return DECIMAL_RESULT; }
2005-02-08 23:50:45 +01:00
};
2000-07-31 21:29:14 +02:00
/*
** Classes for easy comparing of non const items
*/
class cmp_item :public Sql_alloc
{
public:
CHARSET_INFO *cmp_charset;
cmp_item() { cmp_charset= &my_charset_bin; }
2000-07-31 21:29:14 +02:00
virtual ~cmp_item() {}
virtual void store_value(Item *item)= 0;
virtual int cmp(Item *item)= 0;
// for optimized IN with row
virtual int compare(cmp_item *item)= 0;
2005-02-08 23:50:45 +01:00
static cmp_item* get_comparator(Item_result type, CHARSET_INFO *cs);
virtual cmp_item *make_same()= 0;
virtual void store_value_by_template(cmp_item *tmpl, Item *item)
{
store_value(item);
}
2000-07-31 21:29:14 +02:00
};
class cmp_item_string :public cmp_item
{
protected:
String *value_res;
public:
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
cmp_item_string () {}
cmp_item_string (CHARSET_INFO *cs) { cmp_charset= cs; }
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
void set_charset(CHARSET_INFO *cs) { cmp_charset= cs; }
friend class cmp_item_sort_string;
friend class cmp_item_sort_string_in_static;
};
2000-07-31 21:29:14 +02:00
class cmp_item_sort_string :public cmp_item_string
{
protected:
2005-02-08 23:50:45 +01:00
char value_buff[STRING_BUFFER_USUAL_SIZE];
String value;
2000-07-31 21:29:14 +02:00
public:
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
cmp_item_sort_string():
cmp_item_string() {}
cmp_item_sort_string(CHARSET_INFO *cs):
cmp_item_string(cs),
value(value_buff, sizeof(value_buff), cs) {}
2000-07-31 21:29:14 +02:00
void store_value(Item *item)
{
value_res= item->val_str(&value);
}
2000-07-31 21:29:14 +02:00
int cmp(Item *arg)
{
2005-02-08 23:50:45 +01:00
char buff[STRING_BUFFER_USUAL_SIZE];
String tmp(buff, sizeof(buff), cmp_charset), *res;
res= arg->val_str(&tmp);
return (value_res ? (res ? sortcmp(value_res, res, cmp_charset) : 1) :
(res ? -1 : 0));
}
int compare(cmp_item *ci)
{
cmp_item_string *l_cmp= (cmp_item_string *) ci;
return sortcmp(value_res, l_cmp->value_res, cmp_charset);
}
cmp_item *make_same();
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
void set_charset(CHARSET_INFO *cs)
{
cmp_charset= cs;
value.set_quick(value_buff, sizeof(value_buff), cs);
}
2000-07-31 21:29:14 +02:00
};
class cmp_item_int :public cmp_item
{
longlong value;
public:
cmp_item_int() {} /* Remove gcc warning */
2000-07-31 21:29:14 +02:00
void store_value(Item *item)
{
value= item->val_int();
}
2000-07-31 21:29:14 +02:00
int cmp(Item *arg)
{
return value != arg->val_int();
}
int compare(cmp_item *ci)
{
cmp_item_int *l_cmp= (cmp_item_int *)ci;
return (value < l_cmp->value) ? -1 : ((value == l_cmp->value) ? 0 : 1);
}
cmp_item *make_same();
2000-07-31 21:29:14 +02:00
};
class cmp_item_real :public cmp_item
{
double value;
public:
cmp_item_real() {} /* Remove gcc warning */
2000-07-31 21:29:14 +02:00
void store_value(Item *item)
{
2004-11-11 19:39:35 +01:00
value= item->val_real();
}
2000-07-31 21:29:14 +02:00
int cmp(Item *arg)
{
2004-11-11 19:39:35 +01:00
return value != arg->val_real();
}
int compare(cmp_item *ci)
{
cmp_item_real *l_cmp= (cmp_item_real *) ci;
return (value < l_cmp->value)? -1 : ((value == l_cmp->value) ? 0 : 1);
}
cmp_item *make_same();
2000-07-31 21:29:14 +02:00
};
2005-02-08 23:50:45 +01:00
class cmp_item_decimal :public cmp_item
{
my_decimal value;
public:
cmp_item_decimal() {} /* Remove gcc warning */
2005-02-08 23:50:45 +01:00
void store_value(Item *item);
int cmp(Item *arg);
int compare(cmp_item *c);
cmp_item *make_same();
};
class cmp_item_row :public cmp_item
{
cmp_item **comparators;
uint n;
public:
cmp_item_row(): comparators(0), n(0) {}
~cmp_item_row();
void store_value(Item *item);
int cmp(Item *arg);
int compare(cmp_item *arg);
cmp_item *make_same();
void store_value_by_template(cmp_item *tmpl, Item *);
};
class in_row :public in_vector
{
cmp_item_row tmp;
public:
in_row(uint elements, Item *);
~in_row();
void set(uint pos,Item *item);
byte *get_value(Item *item);
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
Item_result result_type() { return ROW_RESULT; }
};
/*
cmp_item for optimized IN with row (right part string, which never
be changed)
*/
class cmp_item_sort_string_in_static :public cmp_item_string
{
protected:
String value;
public:
cmp_item_sort_string_in_static(CHARSET_INFO *cs):
cmp_item_string(cs) {}
void store_value(Item *item)
{
value_res= item->val_str(&value);
}
int cmp(Item *item)
{
// Should never be called
DBUG_ASSERT(0);
return 1;
}
int compare(cmp_item *ci)
{
cmp_item_string *l_cmp= (cmp_item_string *) ci;
return sortcmp(value_res, l_cmp->value_res, cmp_charset);
}
cmp_item *make_same()
{
return new cmp_item_sort_string_in_static(cmp_charset);
}
};
2000-07-31 21:29:14 +02:00
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
/*
The class Item_func_case is the CASE ... WHEN ... THEN ... END function
implementation.
When there is no expression between CASE and the first WHEN
(the CASE expression) then this function simple checks all WHEN expressions
one after another. When some WHEN expression evaluated to TRUE then the
value of the corresponding THEN expression is returned.
When the CASE expression is specified then it is compared to each WHEN
expression individually. When an equal WHEN expression is found
corresponding THEN expression is returned.
In order to do correct comparisons several comparators are used. One for
each result type. Different result types that are used in particular
CASE ... END expression are collected in the fix_length_and_dec() member
function and only comparators for there result types are used.
*/
class Item_func_case :public Item_func
{
int first_expr_num, else_expr_num;
enum Item_result cached_result_type, left_result_type;
String tmp_value;
uint ncases;
Item_result cmp_type;
DTCollation cmp_collation;
cmp_item *cmp_items[5]; /* For all result types */
cmp_item *case_item;
public:
Item_func_case(List<Item> &list, Item *first_expr_arg, Item *else_expr_arg)
:Item_func(), first_expr_num(-1), else_expr_num(-1),
cached_result_type(INT_RESULT), left_result_type(INT_RESULT), case_item(0)
{
ncases= list.elements;
if (first_expr_arg)
{
first_expr_num= list.elements;
list.push_back(first_expr_arg);
}
if (else_expr_arg)
{
else_expr_num= list.elements;
list.push_back(else_expr_arg);
}
set_arguments(list);
bzero(&cmp_items, sizeof(cmp_items));
}
double val_real();
longlong val_int();
String *val_str(String *);
my_decimal *val_decimal(my_decimal *);
bool fix_fields(THD *thd, Item **ref);
void fix_length_and_dec();
uint decimal_precision() const;
table_map not_null_tables() const { return 0; }
enum Item_result result_type () const { return cached_result_type; }
const char *func_name() const { return "case"; }
void print(String *str);
Item *find_item(String *str);
CHARSET_INFO *compare_collation() { return cmp_collation.collation; }
bool check_partition_func_processor(byte *bool_arg) { return FALSE;}
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
void cleanup()
{
uint i;
DBUG_ENTER("Item_func_case::cleanup");
Item_func::cleanup();
for (i= 0; i <= (uint)DECIMAL_RESULT; i++)
{
delete cmp_items[i];
cmp_items[i]= 0;
}
DBUG_VOID_RETURN;
}
};
/*
The Item_func_in class implements the in_expr IN(values_list) function.
The current implementation distinguishes 2 cases:
1) all items in the value_list are constants and have the same
result type. This case is handled by in_vector class.
2) items in the value_list have different result types or there is some
non-constant items.
In this case Item_func_in employs several cmp_item objects to performs
comparisons of in_expr and an item from the values_list. One cmp_item
object for each result type. Different result types are collected in the
fix_length_and_dec() member function by means of collect_cmp_types()
function.
*/
class Item_func_in :public Item_func_opt_neg
2000-07-31 21:29:14 +02:00
{
public:
Item_result cmp_type;
/*
an array of values when the right hand arguments of IN
are all SQL constant and there are no nulls
*/
2000-07-31 21:29:14 +02:00
in_vector *array;
bool have_null;
/*
true when all arguments of the IN clause are of compatible types
and can be used safely as comparisons for key conditions
*/
bool arg_types_compatible;
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
Item_result left_result_type;
cmp_item *cmp_items[5]; /* One cmp_item for each result type */
DTCollation cmp_collation;
Item_func_in(List<Item> &list)
:Item_func_opt_neg(list), array(0), have_null(0),
arg_types_compatible(FALSE)
{
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
bzero(&cmp_items, sizeof(cmp_items));
allowed_arg_cols= 0; // Fetch this value from first argument
}
2000-07-31 21:29:14 +02:00
longlong val_int();
bool fix_fields(THD *, Item **);
2000-07-31 21:29:14 +02:00
void fix_length_and_dec();
uint decimal_precision() const { return 1; }
void cleanup()
{
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
uint i;
DBUG_ENTER("Item_func_in::cleanup");
Item_int_func::cleanup();
delete array;
array= 0;
Fixed bug #18360: Type aggregation for IN and CASE may lead to a wrong result The IN function aggregates result types of all expressions. It uses that type in comparison of left expression and expressions in right part. This approach works in most cases. But let's consider the case when the right part contains both strings and integers. In that case this approach may cause wrong results because all strings which do not start with a digit are evaluated as 0. CASE uses the same approach when a CASE expression is given thus it's also affected. The idea behind this fix is to make IN function to compare expressions with different result types differently. For example a string in the left part will be compared as string with strings specified in right part and will be converted to real for comparison to int or real items in the right part. A new function called collect_cmp_types() is added. It collects different result types for comparison of first item in the provided list with each other item in the list. The Item_func_in class now can refer up to 5 cmp_item objects: 1 for each result type for comparison purposes. cmp_item objects are allocated according to found result types. The comparison of the left expression with any right part expression is now based only on result types of these expressions. The Item_func_case class is modified in the similar way when a CASE expression is specified. Now it can allocate up to 5 cmp_item objects to compare CASE expression with WHEN expressions of different types. The comparison of the CASE expression with any WHEN expression now based only on result types of these expressions.
2006-09-26 18:52:54 +02:00
for (i= 0; i <= (uint)DECIMAL_RESULT; i++)
{
delete cmp_items[i];
cmp_items[i]= 0;
}
DBUG_VOID_RETURN;
}
2000-07-31 21:29:14 +02:00
optimize_type select_optimize() const
{ return OPTIMIZE_KEY; }
2000-07-31 21:29:14 +02:00
void print(String *str);
enum Functype functype() const { return IN_FUNC; }
const char *func_name() const { return " IN "; }
bool nulls_in_row();
2004-08-31 20:10:57 +02:00
bool is_bool_func() { return 1; }
CHARSET_INFO *compare_collation() { return cmp_collation.collation; }
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
2000-07-31 21:29:14 +02:00
};
/* Functions used by where clause */
class Item_func_isnull :public Item_bool_func
{
protected:
2002-05-07 18:08:56 +02:00
longlong cached_value;
2000-07-31 21:29:14 +02:00
public:
Item_func_isnull(Item *a) :Item_bool_func(a) {}
longlong val_int();
enum Functype functype() const { return ISNULL_FUNC; }
void fix_length_and_dec()
{
decimals=0; max_length=1; maybe_null=0;
update_used_tables();
2000-07-31 21:29:14 +02:00
}
const char *func_name() const { return "isnull"; }
/* Optimize case of not_null_column IS NULL */
virtual void update_used_tables()
2000-07-31 21:29:14 +02:00
{
if (!args[0]->maybe_null)
{
used_tables_cache= 0; /* is always false */
const_item_cache= 1;
cached_value= (longlong) 0;
2000-07-31 21:29:14 +02:00
}
else
2002-05-07 18:08:56 +02:00
{
args[0]->update_used_tables();
if ((const_item_cache= !(used_tables_cache= args[0]->used_tables())) &&
!with_subselect)
{
/* Remember if the value is always NULL or never NULL */
cached_value= (longlong) args[0]->is_null();
}
2002-05-07 18:08:56 +02:00
}
2000-07-31 21:29:14 +02:00
}
table_map not_null_tables() const { return 0; }
2000-07-31 21:29:14 +02:00
optimize_type select_optimize() const { return OPTIMIZE_NULL; }
Item *neg_transformer(THD *thd);
CHARSET_INFO *compare_collation() { return args[0]->collation.collation; }
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
2000-07-31 21:29:14 +02:00
};
/* Functions used by HAVING for rewriting IN subquery */
class Item_in_subselect;
/*
This is like IS NOT NULL but it also remembers if it ever has
encountered a NULL.
*/
class Item_is_not_null_test :public Item_func_isnull
{
Item_in_subselect* owner;
public:
Item_is_not_null_test(Item_in_subselect* ow, Item *a)
:Item_func_isnull(a), owner(ow)
{}
enum Functype functype() const { return ISNOTNULLTEST_FUNC; }
longlong val_int();
const char *func_name() const { return "<is_not_null_test>"; }
void update_used_tables();
/*
we add RAND_TABLE_BIT to prevent moving this item from HAVING to WHERE
*/
table_map used_tables() const
{ return used_tables_cache | RAND_TABLE_BIT; }
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return TRUE;}
};
2000-07-31 21:29:14 +02:00
class Item_func_isnotnull :public Item_bool_func
{
bool abort_on_null;
2000-07-31 21:29:14 +02:00
public:
Item_func_isnotnull(Item *a) :Item_bool_func(a), abort_on_null(0) {}
2000-07-31 21:29:14 +02:00
longlong val_int();
enum Functype functype() const { return ISNOTNULL_FUNC; }
void fix_length_and_dec()
{
decimals=0; max_length=1; maybe_null=0;
}
2000-07-31 21:29:14 +02:00
const char *func_name() const { return "isnotnull"; }
optimize_type select_optimize() const { return OPTIMIZE_NULL; }
table_map not_null_tables() const
{ return abort_on_null ? not_null_tables_cache : 0; }
Item *neg_transformer(THD *thd);
void print(String *str);
CHARSET_INFO *compare_collation() { return args[0]->collation.collation; }
void top_level_item() { abort_on_null=1; }
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
2000-07-31 21:29:14 +02:00
};
2000-07-31 21:29:14 +02:00
class Item_func_like :public Item_bool_func2
{
// Turbo Boyer-Moore data
bool canDoTurboBM; // pattern is '%abcd%' case
const char* pattern;
int pattern_len;
// TurboBM buffers, *this is owner
int* bmGs; // good suffix shift table, size is pattern_len + 1
int* bmBc; // bad character shift table, size is alphabet_size
void turboBM_compute_suffixes(int* suff);
void turboBM_compute_good_suffix_shifts(int* suff);
void turboBM_compute_bad_character_shifts();
bool turboBM_matches(const char* text, int text_len) const;
enum { alphabet_size = 256 };
Item *escape_item;
bool escape_used_in_parsing;
2000-07-31 21:29:14 +02:00
public:
int escape;
Item_func_like(Item *a,Item *b, Item *escape_arg, bool escape_used)
:Item_bool_func2(a,b), canDoTurboBM(FALSE), pattern(0), pattern_len(0),
bmGs(0), bmBc(0), escape_item(escape_arg),
escape_used_in_parsing(escape_used) {}
2000-07-31 21:29:14 +02:00
longlong val_int();
enum Functype functype() const { return LIKE_FUNC; }
optimize_type select_optimize() const;
cond_result eq_cmp_result() const { return COND_TRUE; }
const char *func_name() const { return "like"; }
bool fix_fields(THD *thd, Item **ref);
void cleanup();
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
2000-07-31 21:29:14 +02:00
};
#ifdef USE_REGEX
#include "my_regex.h"
2000-07-31 21:29:14 +02:00
class Item_func_regex :public Item_bool_func
{
my_regex_t preg;
2000-07-31 21:29:14 +02:00
bool regex_compiled;
bool regex_is_const;
String prev_regexp;
DTCollation cmp_collation;
2000-07-31 21:29:14 +02:00
public:
Item_func_regex(Item *a,Item *b) :Item_bool_func(a,b),
regex_compiled(0),regex_is_const(0) {}
void cleanup();
2000-07-31 21:29:14 +02:00
longlong val_int();
bool fix_fields(THD *thd, Item **ref);
const char *func_name() const { return "regexp"; }
void print(String *str) { print_op(str); }
CHARSET_INFO *compare_collation() { return cmp_collation.collation; }
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
2000-07-31 21:29:14 +02:00
};
#else
class Item_func_regex :public Item_bool_func
{
public:
Item_func_regex(Item *a,Item *b) :Item_bool_func(a,b) {}
longlong val_int() { return 0;}
const char *func_name() const { return "regex"; }
void print(String *str) { print_op(str); }
2000-07-31 21:29:14 +02:00
};
#endif /* USE_REGEX */
typedef class Item COND;
class Item_cond :public Item_bool_func
{
protected:
List<Item> list;
bool abort_on_null;
table_map and_tables_cache;
2000-07-31 21:29:14 +02:00
public:
/* Item_cond() is only used to create top level items */
Item_cond(): Item_bool_func(), abort_on_null(1)
{ const_item_cache=0; }
Item_cond(Item *i1,Item *i2)
:Item_bool_func(), abort_on_null(0)
{
list.push_back(i1);
list.push_back(i2);
}
Item_cond(THD *thd, Item_cond *item);
Item_cond(List<Item> &nlist)
:Item_bool_func(), list(nlist), abort_on_null(0) {}
2000-07-31 21:29:14 +02:00
bool add(Item *item) { return list.push_back(item); }
void add_at_head(List<Item> *nlist) { list.prepand(nlist); }
bool fix_fields(THD *, Item **ref);
2000-07-31 21:29:14 +02:00
enum Type type() const { return COND_ITEM; }
List<Item>* argument_list() { return &list; }
table_map used_tables() const;
void update_used_tables();
void print(String *str);
void split_sum_func(THD *thd, Item **ref_pointer_array, List<Item> &fields);
friend int setup_conds(THD *thd, TABLE_LIST *tables, TABLE_LIST *leaves,
COND **conds);
void top_level_item() { abort_on_null=1; }
void copy_andor_arguments(THD *thd, Item_cond *item);
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this.
2006-06-04 17:52:22 +02:00
bool walk(Item_processor processor, bool walk_subquery, byte *arg);
Item *transform(Item_transformer transformer, byte *arg);
void traverse_cond(Cond_traverser, void *arg, traverse_order order);
void neg_arguments(THD *thd);
2006-09-26 08:25:23 +02:00
bool check_partition_func_processor(byte *int_arg) {return FALSE;}
bool subst_argument_checker(byte **arg) { return TRUE; }
Item *compile(Item_analyzer analyzer, byte **arg_p,
Item_transformer transformer, byte *arg_t);
2000-07-31 21:29:14 +02:00
};
/*
2005-02-08 23:50:45 +01:00
The class Item_equal is used to represent conjunctions of equality
predicates of the form field1 = field2, and field=const in where
conditions and on expressions.
All equality predicates of the form field1=field2 contained in a
2005-02-08 23:50:45 +01:00
conjunction are substituted for a sequence of items of this class.
An item of this class Item_equal(f1,f2,...fk) represents a
multiple equality f1=f2=...=fk.
2005-02-08 23:50:45 +01:00
If a conjunction contains predicates f1=f2 and f2=f3, a new item of
this class is created Item_equal(f1,f2,f3) representing the multiple
equality f1=f2=f3 that substitutes the above equality predicates in
2005-02-08 23:50:45 +01:00
the conjunction.
A conjunction of the predicates f2=f1 and f3=f1 and f3=f2 will be
substituted for the item representing the same multiple equality
f1=f2=f3.
2005-02-08 23:50:45 +01:00
An item Item_equal(f1,f2) can appear instead of a conjunction of
f2=f1 and f1=f2, or instead of just the predicate f1=f2.
2005-02-08 23:50:45 +01:00
An item of the class Item_equal inherits equalities from outer
conjunctive levels.
Suppose we have a where condition of the following form:
WHERE f1=f2 AND f3=f4 AND f3=f5 AND ... AND (...OR (f1=f3 AND ...)).
In this case:
f1=f2 will be substituted for Item_equal(f1,f2);
f3=f4 and f3=f5 will be substituted for Item_equal(f3,f4,f5);
f1=f3 will be substituted for Item_equal(f1,f2,f3,f4,f5);
An object of the class Item_equal can contain an optional constant
2005-02-08 23:50:45 +01:00
item c. Then it represents a multiple equality of the form
c=f1=...=fk.
Objects of the class Item_equal are used for the following:
1. An object Item_equal(t1.f1,...,tk.fk) allows us to consider any
pair of tables ti and tj as joined by an equi-condition.
Thus it provide us with additional access paths from table to table.
2. An object Item_equal(t1.f1,...,tk.fk) is applied to deduce new
SARGable predicates:
f1=...=fk AND P(fi) => f1=...=fk AND P(fi) AND P(fj).
It also can give us additional index scans and can allow us to
improve selectivity estimates.
3. An object Item_equal(t1.f1,...,tk.fk) is used to optimize the
selected execution plan for the query: if table ti is accessed
before the table tj then in any predicate P in the where condition
2005-02-08 23:50:45 +01:00
the occurrence of tj.fj is substituted for ti.fi. This can allow
an evaluation of the predicate at an earlier step.
When feature 1 is supported they say that join transitive closure
is employed.
When feature 2 is supported they say that search argument transitive
closure is employed.
Both features are usually supported by preprocessing original query and
adding additional predicates.
We do not just add predicates, we rather dynamically replace some
predicates that can not be used to access tables in the investigated
plan for those, obtained by substitution of some fields for equal fields,
that can be used.
Prepared Statements/Stored Procedures note: instances of class
Item_equal are created only at the time a PS/SP is executed and
are deleted in the end of execution. All changes made to these
objects need not be registered in the list of changes of the parse
tree and do not harm PS/SP re-execution.
Item equal objects are employed only at the optimize phase. Usually they are
not supposed to be evaluated. Yet in some cases we call the method val_int()
for them. We have to take care of restricting the predicate such an
object represents f1=f2= ...=fn to the projection of known fields fi1=...=fik.
*/
class Item_equal: public Item_bool_func
{
List<Item_field> fields; /* list of equal field items */
Item *const_item; /* optional constant item equal to fields items */
cmp_item *eval_item;
bool cond_false;
public:
inline Item_equal()
: Item_bool_func(), const_item(0), eval_item(0), cond_false(0)
{ const_item_cache=0 ;}
Item_equal(Item_field *f1, Item_field *f2);
Item_equal(Item *c, Item_field *f);
Item_equal(Item_equal *item_equal);
inline Item* get_const() { return const_item; }
void add(Item *c);
void add(Item_field *f);
uint members();
bool contains(Field *field);
Item_field* get_first() { return fields.head(); }
void merge(Item_equal *item);
void update_const();
enum Functype functype() const { return MULT_EQUAL_FUNC; }
longlong val_int();
const char *func_name() const { return "multiple equal"; }
optimize_type select_optimize() const { return OPTIMIZE_EQUAL; }
void sort(Item_field_cmpfunc cmp, void *arg);
friend class Item_equal_iterator;
void fix_length_and_dec();
bool fix_fields(THD *thd, Item **ref);
void update_used_tables();
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this.
2006-06-04 17:52:22 +02:00
bool walk(Item_processor processor, bool walk_subquery, byte *arg);
Item *transform(Item_transformer transformer, byte *arg);
void print(String *str);
CHARSET_INFO *compare_collation()
{ return fields.head()->collation.collation; }
};
class COND_EQUAL: public Sql_alloc
{
public:
uint max_members; /* max number of members the current level
list and all lower level lists */
COND_EQUAL *upper_levels; /* multiple equalities of upper and levels */
List<Item_equal> current_level; /* list of multiple equalities of
the current and level */
COND_EQUAL()
{
max_members= 0;
upper_levels= 0;
}
};
2004-11-29 22:47:50 +01:00
class Item_equal_iterator : public List_iterator_fast<Item_field>
{
public:
inline Item_equal_iterator(Item_equal &item_equal)
:List_iterator_fast<Item_field> (item_equal.fields)
{}
inline Item_field* operator++(int)
{
Item_field *item= (*(List_iterator_fast<Item_field> *) this)++;
return item;
}
inline void rewind(void)
{
List_iterator_fast<Item_field>::rewind();
}
};
2000-07-31 21:29:14 +02:00
class Item_cond_and :public Item_cond
{
public:
COND_EQUAL cond_equal; /* contains list of Item_equal objects for
the current and level and reference
to multiple equalities of upper and levels */
2000-07-31 21:29:14 +02:00
Item_cond_and() :Item_cond() {}
Item_cond_and(Item *i1,Item *i2) :Item_cond(i1,i2) {}
Item_cond_and(THD *thd, Item_cond_and *item) :Item_cond(thd, item) {}
Item_cond_and(List<Item> &list_arg): Item_cond(list_arg) {}
2000-07-31 21:29:14 +02:00
enum Functype functype() const { return COND_AND_FUNC; }
longlong val_int();
const char *func_name() const { return "and"; }
table_map not_null_tables() const
{ return abort_on_null ? not_null_tables_cache: and_tables_cache; }
Item* copy_andor_structure(THD *thd)
{
Item_cond_and *item;
2004-02-05 18:14:48 +01:00
if ((item= new Item_cond_and(thd, this)))
item->copy_andor_arguments(thd, this);
return item;
}
Item *neg_transformer(THD *thd);
2000-07-31 21:29:14 +02:00
};
class Item_cond_or :public Item_cond
{
public:
Item_cond_or() :Item_cond() {}
Item_cond_or(Item *i1,Item *i2) :Item_cond(i1,i2) {}
Item_cond_or(THD *thd, Item_cond_or *item) :Item_cond(thd, item) {}
Item_cond_or(List<Item> &list_arg): Item_cond(list_arg) {}
2000-07-31 21:29:14 +02:00
enum Functype functype() const { return COND_OR_FUNC; }
longlong val_int();
const char *func_name() const { return "or"; }
table_map not_null_tables() const { return and_tables_cache; }
Item* copy_andor_structure(THD *thd)
{
Item_cond_or *item;
2004-02-05 18:14:48 +01:00
if ((item= new Item_cond_or(thd, this)))
item->copy_andor_arguments(thd, this);
return item;
}
Item *neg_transformer(THD *thd);
2000-07-31 21:29:14 +02:00
};
/*
2005-02-08 23:50:45 +01:00
XOR is Item_cond, not an Item_int_func because we could like to
optimize (a XOR b) later on. It's low prio, though
*/
2003-08-11 21:44:43 +02:00
class Item_cond_xor :public Item_cond
{
public:
Item_cond_xor() :Item_cond() {}
Item_cond_xor(Item *i1,Item *i2) :Item_cond(i1,i2) {}
enum Functype functype() const { return COND_XOR_FUNC; }
/* TODO: remove the next line when implementing XOR optimization */
enum Type type() const { return FUNC_ITEM; }
longlong val_int();
const char *func_name() const { return "xor"; }
void top_level_item() {}
};
2005-02-08 23:50:45 +01:00
/* Some useful inline functions */
2000-07-31 21:29:14 +02:00
inline Item *and_conds(Item *a, Item *b)
2000-07-31 21:29:14 +02:00
{
if (!b) return a;
if (!a) return b;
return new Item_cond_and(a, b);
2000-07-31 21:29:14 +02:00
}
2002-11-21 14:56:48 +01:00
Item *and_expressions(Item *a, Item *b, Item **org_item);