LOCATE(), INSTR(), POSITION(), FIND_IN_SET(), SUBSTRING_INDEX()

now process arguments according to standard SQL collation determation
rules.

mysql-test/r/func_str.result

@@ -249,6 +249,33 @@ INSERT INTO t1 VALUES (1, 'a545f661efdd1fb66fdee3aab79945bf');
 SELECT 1 FROM t1 WHERE tmp=AES_DECRYPT(tmp,"password");
 1
 DROP TABLE t1;
+select POSITION(_latin1'B' IN _latin1'abcd');
+POSITION(_latin1'B' IN _latin1'abcd')
+2
+select POSITION(_latin1'B' IN _latin1'abcd' COLLATE latin1_bin);
+POSITION(_latin1'B' IN _latin1'abcd' COLLATE latin1_bin)
+0
+select POSITION(_latin1'B' COLLATE latin1_bin IN _latin1'abcd');
+POSITION(_latin1'B' COLLATE latin1_bin IN _latin1'abcd')
+0
+select POSITION(_latin1'B' COLLATE latin1_general_ci IN _latin1'abcd' COLLATE latin1_bin);
+ERROR HY000: Illegal mix of collations (latin1_bin,EXPLICIT) and (latin1_general_ci,EXPLICIT) for operation 'locate'
+select POSITION(_latin1'B' IN _latin2'abcd');
+ERROR HY000: Illegal mix of collations (latin2_general_ci,COERCIBLE) and (latin1_swedish_ci,COERCIBLE) for operation 'locate'
+select FIND_IN_SET(_latin1'B',_latin1'a,b,c,d');
+FIND_IN_SET(_latin1'B',_latin1'a,b,c,d')
+2
+select FIND_IN_SET(_latin1'B' COLLATE latin1_general_ci,_latin1'a,b,c,d' COLLATE latin1_bin);
+ERROR HY000: Illegal mix of collations (latin1_general_ci,EXPLICIT) and (latin1_bin,EXPLICIT) for operation 'find_in_set'
+select FIND_IN_SET(_latin1'B',_latin2'a,b,c,d');
+ERROR HY000: Illegal mix of collations (latin1_swedish_ci,COERCIBLE) and (latin2_general_ci,COERCIBLE) for operation 'find_in_set'
+select SUBSTRING_INDEX(_latin1'abcdabcdabcd',_latin1'd',2);
+SUBSTRING_INDEX(_latin1'abcdabcdabcd',_latin1'd',2)
+abcdabc
+select SUBSTRING_INDEX(_latin1'abcdabcdabcd',_latin2'd',2);
+ERROR HY000: Illegal mix of collations (latin1_swedish_ci,COERCIBLE) and (latin2_general_ci,COERCIBLE) for operation 'substr_index'
+select SUBSTRING_INDEX(_latin1'abcdabcdabcd' COLLATE latin1_general_ci,_latin1'd' COLLATE latin1_bin,2);
+ERROR HY000: Illegal mix of collations (latin1_general_ci,EXPLICIT) and (latin1_bin,EXPLICIT) for operation 'substr_index'
 select collation(bin(130)), coercibility(bin(130));
 collation(bin(130))	coercibility(bin(130))
 latin1_swedish_ci	3

mysql-test/t/func_str.test

@@ -136,6 +136,33 @@ DROP TABLE t1;
 #
 # Test collation and coercibility
 #
+select POSITION(_latin1'B' IN _latin1'abcd');
+select POSITION(_latin1'B' IN _latin1'abcd' COLLATE latin1_bin);
+select POSITION(_latin1'B' COLLATE latin1_bin IN _latin1'abcd');
+--error 1265
+select POSITION(_latin1'B' COLLATE latin1_general_ci IN _latin1'abcd' COLLATE latin1_bin);
+--error 1265
+select POSITION(_latin1'B' IN _latin2'abcd');
+
+select FIND_IN_SET(_latin1'B',_latin1'a,b,c,d');
+--fix this:
+--select FIND_IN_SET(_latin1'B',_latin1'a,b,c,d' COLLATE latin1_bin);
+--select FIND_IN_SET(_latin1'B' COLLATE latin1_bin,_latin1'a,b,c,d');
+--error 1265
+select FIND_IN_SET(_latin1'B' COLLATE latin1_general_ci,_latin1'a,b,c,d' COLLATE latin1_bin);
+--error 1265
+select FIND_IN_SET(_latin1'B',_latin2'a,b,c,d');
+
+select SUBSTRING_INDEX(_latin1'abcdabcdabcd',_latin1'd',2);
+--fix this:
+--select SUBSTRING_INDEX(_latin1'abcdabcdabcd' COLLATE latin1_bin,_latin1'd',2);
+--select SUBSTRING_INDEX(_latin1'abcdabcdabcd',_latin1'd' COLLATE latin1_bin,2);
+--error 1265
+select SUBSTRING_INDEX(_latin1'abcdabcdabcd',_latin2'd',2);
+--error 1265
+select SUBSTRING_INDEX(_latin1'abcdabcdabcd' COLLATE latin1_general_ci,_latin1'd' COLLATE latin1_bin,2);
+
+
 select collation(bin(130)), coercibility(bin(130));
 select collation(oct(130)), coercibility(oct(130));
 select collation(conv(130,16,10)), coercibility(conv(130,16,10));

sql/item_func.cc

@@ -1045,10 +1045,18 @@ longlong Item_func_coercibility::val_int()
   return (longlong) args[0]->derivation();
 }
 
+void Item_func_locate::fix_length_and_dec()
+{
+  maybe_null=0; max_length=11;
+  if (cmp_collation.set(args[0]->collation, args[1]->collation))
+    my_coll_agg_error(args[0]->collation, args[1]->collation, func_name());
+}
+
 longlong Item_func_locate::val_int()
 {
   String *a=args[0]->val_str(&value1);
   String *b=args[1]->val_str(&value2);
+  bool binary_cmp= (cmp_collation.collation->state & MY_CS_BINSORT) ? 1 : 0;
   if (!a || !b)
   {
     null_value=1;
@@ -1063,7 +1071,7 @@ longlong Item_func_locate::val_int()
   {
     start=(uint) args[2]->val_int()-1;
 #ifdef USE_MB
-    if (use_mb(a->charset()))
+    if (use_mb(cmp_collation.collation))
     {
       start0=start;
       if (!binary_cmp)
@@ -1076,7 +1084,7 @@ longlong Item_func_locate::val_int()
   if (!b->length())				// Found empty string at start
     return (longlong) (start+1);
 #ifdef USE_MB
-  if (use_mb(a->charset()) && !binary_cmp)
+  if (use_mb(cmp_collation.collation) && !binary_cmp)
   {
     const char *ptr=a->ptr()+start;
     const char *search=b->ptr();
@@ -1095,7 +1103,7 @@ longlong Item_func_locate::val_int()
         return (longlong) start0+1;
       }
   skipp:
-      if ((l=my_ismbchar(a->charset(),ptr,strend)))
+      if ((l=my_ismbchar(cmp_collation.collation,ptr,strend)))
 	ptr+=l;
       else ++ptr;
       ++start0;
@@ -1201,6 +1209,8 @@ void Item_func_find_in_set::fix_length_and_dec()
       }
     }
   }
+  if (cmp_collation.set(args[0]->collation, args[1]->collation))
+    my_coll_agg_error(args[0]->collation, args[1]->collation, func_name());
 }
 
 static const char separator=',';
@@ -1228,7 +1238,6 @@ longlong Item_func_find_in_set::val_int()
   null_value=0;
 
   int diff;
-  CHARSET_INFO *charset= find->charset();
   if ((diff=buffer->length() - find->length()) >= 0)
   {
     const char *f_pos=find->ptr();
@@ -1242,7 +1251,8 @@ longlong Item_func_find_in_set::val_int()
       const char *pos= f_pos;
       while (pos != f_end)
       {
-	if (my_toupper(charset,*str) != my_toupper(charset,*pos))
+	if (my_toupper(cmp_collation.collation,*str) != 
+	    my_toupper(cmp_collation.collation,*pos))
 	  goto not_found;
 	str++;
 	pos++;

sql/item_func.h

@@ -610,17 +610,13 @@ class Item_func_coercibility :public Item_int_func
 class Item_func_locate :public Item_int_func
 {
   String value1,value2;
-  bool binary_cmp;
+  DTCollation cmp_collation;
 public:
   Item_func_locate(Item *a,Item *b) :Item_int_func(a,b) {}
   Item_func_locate(Item *a,Item *b,Item *c) :Item_int_func(a,b,c) {}
   const char *func_name() const { return "locate"; }
   longlong val_int();
-  void fix_length_and_dec()
-  {
-    maybe_null=0; max_length=11;
-    binary_cmp = args[0]->binary() || args[1]->binary();
-  }
+  void fix_length_and_dec();
 };
 
 
@@ -684,6 +680,7 @@ class Item_func_find_in_set :public Item_int_func
   String value,value2;
   uint enum_value;
   ulonglong enum_bit;
+  DTCollation cmp_collation;
 public:
   Item_func_find_in_set(Item *a,Item *b) :Item_int_func(a,b),enum_value(0) {}
   longlong val_int();

sql/item_strfunc.cc

@@ -1037,6 +1037,15 @@ void Item_func_substr::fix_length_and_dec()
 }
 
 
+void Item_func_substr_index::fix_length_and_dec()
+{ 
+  max_length= args[0]->max_length;
+  if (collation.set(args[0]->collation, args[1]->collation) ||
+      (collation.derivation == DERIVATION_NONE))
+    my_coll_agg_error(args[0]->collation, args[1]->collation, func_name());
+}
+
+
 String *Item_func_substr_index::val_str(String *str)
 {
   String *res =args[0]->val_str(str);
@@ -1054,6 +1063,8 @@ String *Item_func_substr_index::val_str(String *str)
   if (!res->length() || !delimeter_length || !count)
     return &empty_string;		// Wrong parameters
 
+  res->set_charset(collation.collation);
+
 #ifdef USE_MB
   if (use_mb(res->charset()))
   {

sql/item_strfunc.h

@@ -215,7 +215,7 @@ class Item_func_substr_index :public Item_str_func
 public:
   Item_func_substr_index(Item *a,Item *b,Item *c) :Item_str_func(a,b,c) {}
   String *val_str(String *);
-  void fix_length_and_dec() { max_length= args[0]->max_length; }
+  void fix_length_and_dec();
   const char *func_name() const { return "substr_index"; }
 };