Allow IN to convert arguments into Unicode in some cases.

5 files changed, 80 insertions, 10 deletions

diff --git a/mysql-test/r/func_in.result b/mysql-test/r/func_in.result
index f66b3dea94b..374affce8c5 100644
--- a/mysql-test/r/func_in.result
+++ b/mysql-test/r/func_in.result
@@ -148,6 +148,16 @@ id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
 Warnings:
 Note	1003	select test.t1.a AS `a`,test.t1.b AS `b`,test.t1.c AS `c` from test.t1 where (_latin1'a' in (test.t1.a,test.t1.b,(test.t1.c collate _latin1'latin1_bin')))
 drop table t1;
+set names utf8;
+create table t1 (a char(10) character set utf8 not null);
+insert into t1 values ('bbbb'),(_koi8r'ÃÃÃÃ'),(_latin1'ÄÄÄÄ');
+select a from t1 where a in ('bbbb',_koi8r'ÃÃÃÃ',_latin1'ÄÄÄÄ') order by a;
+a
+ÄÄÄÄ
+bbbb
+цццц
+drop table t1;
+set names latin1;
 select '1.0' in (1,2);
 '1.0' in (1,2)
 1
diff --git a/mysql-test/t/func_in.test b/mysql-test/t/func_in.test
index 855a7cbd28f..22079377ad2 100644
--- a/mysql-test/t/func_in.test
+++ b/mysql-test/t/func_in.test
@@ -75,6 +75,13 @@ select * from t1 where 'a' in (a,b,c collate latin1_bin);
 explain extended select * from t1 where 'a' in (a,b,c collate latin1_bin);
 drop table t1;
 
+set names utf8;
+create table t1 (a char(10) character set utf8 not null);
+insert into t1 values ('bbbb'),(_koi8r'ÃÃÃÃ'),(_latin1'ÄÄÄÄ');
+select a from t1 where a in ('bbbb',_koi8r'ÃÃÃÃ',_latin1'ÄÄÄÄ') order by a;
+drop table t1;
+set names latin1;
+
 select '1.0' in (1,2);
 select 1 in ('1.0',2);
 select 1 in (1,'2.0');
diff --git a/sql/item_cmpfunc.cc b/sql/item_cmpfunc.cc
index f91bc5c4bc5..4ddb648399a 100644
--- a/sql/item_cmpfunc.cc
+++ b/sql/item_cmpfunc.cc
@@ -1735,12 +1735,58 @@ void Item_func_in::fix_length_and_dec()
   uint const_itm= 1;
   
   agg_cmp_type(&cmp_type, args, arg_count);
-  if ((cmp_type == STRING_RESULT) &&
-      (agg_arg_collations_for_comparison(cmp_collation, args, arg_count)))
-    return;
-  
+
   for (arg=args+1, arg_end=args+arg_count; arg != arg_end ; arg++)
     const_itm&= arg[0]->const_item();
+
+
+  if (cmp_type == STRING_RESULT)
+  {
+    /*
+      We allow consts character set conversion for
+
+        item IN (const1, const2, const3, ...)
+
+      if item is in a superset for all arguments,
+      and if it is a stong side according to coercibility rules.
+   
+      TODO: add covnersion for non-constant IN values
+      via creating Item_func_conv_charset().
+    */
+
+    if (agg_arg_collations_for_comparison(cmp_collation,
+                                          args, arg_count, TRUE))
+      return;
+    if ((!my_charset_same(args[0]->collation.collation, 
+                          cmp_collation.collation) || !const_itm))
+    {
+      if (agg_arg_collations_for_comparison(cmp_collation,
+                                            args, arg_count, FALSE))
+        return;
+    }
+    else
+    {
+      /* 
+         Conversion is possible:
+         All IN arguments are constants.
+      */
+      for (arg= args+1, arg_end= args+arg_count; arg < arg_end; arg++)
+      {
+        if (!my_charset_same(cmp_collation.collation,
+                             arg[0]->collation.collation))
+        {
+          Item_string *conv;
+          String tmp, cstr, *ostr= arg[0]->val_str(&tmp);
+          cstr.copy(ostr->ptr(), ostr->length(), ostr->charset(),
+                    cmp_collation.collation);
+          conv= new Item_string(cstr.ptr(),cstr.length(), cstr.charset(),
+                                arg[0]->collation.derivation);
+          conv->str_value.copy();
+          arg[0]= conv;
+        }
+      }
+    }
+  }
   
   /*
     Row item with NULLs inside can return NULL or FALSE => 
diff --git a/sql/item_func.cc b/sql/item_func.cc
index adcba34d56b..ef845bb8266 100644
--- a/sql/item_func.cc
+++ b/sql/item_func.cc
@@ -75,13 +75,16 @@ static void my_coll_agg_error(Item** args, uint count, const char *fname)
 }
 
 
-bool Item_func::agg_arg_collations(DTCollation &c, Item **av, uint count)
+bool Item_func::agg_arg_collations(DTCollation &c, Item **av, uint count,
+                                   bool allow_superset_conversion)
 {
   uint i;
+  c.nagg= 0;
+  c.strong= 0;
   c.set(av[0]->collation);
   for (i= 1; i < count; i++)
   {
-    if (c.aggregate(av[i]->collation))
+    if (c.aggregate(av[i]->collation, allow_superset_conversion))
     {
       my_coll_agg_error(av, count, func_name());
       return TRUE;
@@ -92,9 +95,10 @@ bool Item_func::agg_arg_collations(DTCollation &c, Item **av, uint count)
 
 
 bool Item_func::agg_arg_collations_for_comparison(DTCollation &c,
-						  Item **av, uint count)
+						  Item **av, uint count,
+                                                  bool allow_superset_conv)
 {
-  if (agg_arg_collations(c, av, count))
+  if (agg_arg_collations(c, av, count, allow_superset_conv))
     return TRUE;
 
   if (c.derivation == DERIVATION_NONE)
diff --git a/sql/item_func.h b/sql/item_func.h
index eaa0a044fd6..d45f7244e55 100644
--- a/sql/item_func.h
+++ b/sql/item_func.h
@@ -140,8 +140,11 @@ public:
   Field *tmp_table_field(TABLE *t_arg);
   Item *get_tmp_table_item(THD *thd);
   
-  bool agg_arg_collations(DTCollation &c, Item **items, uint nitems);
-  bool agg_arg_collations_for_comparison(DTCollation &c, Item **items, uint nitems);
+  bool agg_arg_collations(DTCollation &c, Item **items, uint nitems,
+                          bool allow_superset_conversion= FALSE);
+  bool agg_arg_collations_for_comparison(DTCollation &c,
+                                         Item **items, uint nitems,
+                                         bool allow_superset_comversion= FALSE);
 
   bool walk(Item_processor processor, byte *arg);
 };