UCS-2 aligning 0xAA -> 0x00AA

include/m_ctype.h

@@ -206,8 +206,9 @@ typedef struct charset_info_st
   uchar state_map[256];
   uchar ident_map[256];
   uint      strxfrm_multiply;
+  uint      mbminlen;
   uint      mbmaxlen;
-  char    max_sort_char; /* For LIKE optimization */
+  char      max_sort_char; /* For LIKE optimization */
   
   MY_CHARSET_HANDLER *cset;
   MY_COLLATION_HANDLER *coll;

mysql-test/r/ctype_ucs.result

@@ -276,3 +276,51 @@ aardvara
 aardvark
 aardvarz
 DROP TABLE t1;
+SELECT HEX(_ucs2 0x0);
+HEX(_ucs2 0x0)
+0000
+SELECT HEX(_ucs2 0x01);
+HEX(_ucs2 0x01)
+0001
+SELECT HEX(_ucs2 0x012);
+HEX(_ucs2 0x012)
+0012
+SELECT HEX(_ucs2 0x0123);
+HEX(_ucs2 0x0123)
+0123
+SELECT HEX(_ucs2 0x01234);
+HEX(_ucs2 0x01234)
+00001234
+SELECT HEX(_ucs2 0x012345);
+HEX(_ucs2 0x012345)
+00012345
+SELECT HEX(_ucs2 0x0123456);
+HEX(_ucs2 0x0123456)
+00123456
+SELECT HEX(_ucs2 0x01234567);
+HEX(_ucs2 0x01234567)
+01234567
+SELECT HEX(_ucs2 0x012345678);
+HEX(_ucs2 0x012345678)
+000012345678
+SELECT HEX(_ucs2 0x0123456789);
+HEX(_ucs2 0x0123456789)
+000123456789
+SELECT HEX(_ucs2 0x0123456789A);
+HEX(_ucs2 0x0123456789A)
+00123456789A
+SELECT HEX(_ucs2 0x0123456789AB);
+HEX(_ucs2 0x0123456789AB)
+0123456789AB
+SELECT HEX(_ucs2 0x0123456789ABC);
+HEX(_ucs2 0x0123456789ABC)
+0000123456789ABC
+SELECT HEX(_ucs2 0x0123456789ABCD);
+HEX(_ucs2 0x0123456789ABCD)
+000123456789ABCD
+SELECT HEX(_ucs2 0x0123456789ABCDE);
+HEX(_ucs2 0x0123456789ABCDE)
+00123456789ABCDE
+SELECT HEX(_ucs2 0x0123456789ABCDEF);
+HEX(_ucs2 0x0123456789ABCDEF)
+0123456789ABCDEF

mysql-test/t/ctype_ucs.test

@@ -197,3 +197,24 @@ DROP TABLE t1;
 # END OF Bug 1264 test
 #
 ########################################################
+
+
+# Bug #2390
+# Check alignment
+#
+SELECT HEX(_ucs2 0x0);
+SELECT HEX(_ucs2 0x01);
+SELECT HEX(_ucs2 0x012);
+SELECT HEX(_ucs2 0x0123);
+SELECT HEX(_ucs2 0x01234);
+SELECT HEX(_ucs2 0x012345);
+SELECT HEX(_ucs2 0x0123456);
+SELECT HEX(_ucs2 0x01234567);
+SELECT HEX(_ucs2 0x012345678);
+SELECT HEX(_ucs2 0x0123456789);
+SELECT HEX(_ucs2 0x0123456789A);
+SELECT HEX(_ucs2 0x0123456789AB);
+SELECT HEX(_ucs2 0x0123456789ABC);
+SELECT HEX(_ucs2 0x0123456789ABCD);
+SELECT HEX(_ucs2 0x0123456789ABCDE);
+SELECT HEX(_ucs2 0x0123456789ABCDEF);

mysys/charset.c

@@ -131,7 +131,8 @@ static void simple_cs_init_functions(CHARSET_INFO *cs)
     cs->coll= &my_collation_8bit_simple_ci_handler;
   
   cs->cset= &my_charset_8bit_handler;
-  cs->mbmaxlen    = 1;
+  cs->mbminlen= 1;
+  cs->mbmaxlen= 1;
 }
 
 
@@ -273,6 +274,7 @@ static int simple_cs_copy_data(CHARSET_INFO *to, CHARSET_INFO *from)
     if (create_fromuni(to))
       goto err;
   }
+  to->mbminlen= 1;
   to->mbmaxlen= 1;
 
   return 0;

sql/item.h

@@ -477,7 +477,7 @@ public:
   	      CHARSET_INFO *cs, Derivation dv= DERIVATION_COERCIBLE)
   {
     collation.set(cs, dv);
-    str_value.set(str,length,cs);
+    str_value.set_or_copy_aligned(str,length,cs);
     /*
       We have to have a different max_length than 'length' here to
       ensure that we get the right length if we do use the item
@@ -493,7 +493,7 @@ public:
 	      CHARSET_INFO *cs, Derivation dv= DERIVATION_COERCIBLE)
   {
     collation.set(cs, dv);
-    str_value.set(str,length,cs);
+    str_value.set_or_copy_aligned(str,length,cs);
     max_length= str_value.numchars()*cs->mbmaxlen;
     set_name(name_par,0,cs);
     decimals=NOT_FIXED_DEC;

sql/sql_string.cc

@@ -228,6 +228,52 @@ bool String::copy(const char *str,uint32 arg_length, CHARSET_INFO *cs)
   return FALSE;
 }
 
+/*
+** For real multi-byte, ascii incompatible charactser sets,
+** like UCS-2, add leading zeros if we have an incomplete character.
+** Thus, 
+**   SELECT _ucs2 0xAA 
+** will automatically be converted into
+**   SELECT _ucs2 0x00AA
+*/
+
+bool String::set_or_copy_aligned(const char *str,uint32 arg_length,
+				 CHARSET_INFO *cs)
+{
+  /* How many bytes are in incomplete character */
+  uint32 offs= (arg_length % cs->mbminlen); 
+
+  if (!offs) /* All characters are complete, just copy */
+  {
+    set(str, arg_length, cs);
+    return FALSE;
+  }
+  
+  offs= cs->mbmaxlen - offs; /* How many zeros we should prepend */
+  uint32 aligned_length= arg_length + offs;
+  if (alloc(aligned_length))
+    return TRUE;
+  
+  /*
+    Probably this condition is not really necessary
+    because if aligned_length is 0 then offs is 0 too
+    and we'll return after calling set().
+  */
+  if ((str_length= aligned_length))
+  {
+    /*
+      Note, this is only safe for little-endian UCS-2.
+      If we add big-endian UCS-2 sometimes, this code
+      will be more complicated. But it's OK for now.
+    */
+    bzero((char*)Ptr, offs);
+    memcpy(Ptr + offs, str, arg_length);
+  }
+  Ptr[aligned_length]=0;
+  str_charset=cs;
+  return FALSE;
+}
+
 	/* Copy with charset convertion */
 
 bool String::copy(const char *str, uint32 arg_length,

sql/sql_string.h

@@ -183,6 +183,7 @@ public:
   bool copy();					// Alloc string if not alloced
   bool copy(const String &s);			// Allocate new string
   bool copy(const char *s,uint32 arg_length, CHARSET_INFO *cs);	// Allocate new string
+  bool set_or_copy_aligned(const char *s, uint32 arg_length, CHARSET_INFO *cs);
   bool copy(const char*s,uint32 arg_length, CHARSET_INFO *csfrom,
 	    CHARSET_INFO *csto);
   bool append(const String &s);

strings/ctype-big5.c

@@ -6281,6 +6281,7 @@ CHARSET_INFO my_charset_big5_chinese_ci=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     2,			/* mbmaxlen   */
     0,
     &my_charset_big5_handler,
@@ -6304,6 +6305,7 @@ CHARSET_INFO my_charset_big5_bin=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     2,			/* mbmaxlen   */
     0,
     &my_charset_big5_handler,

strings/ctype-bin.c

@@ -381,6 +381,7 @@ CHARSET_INFO my_charset_bin =
     NULL,			/* tab_from_uni  */
     "","",
     1,				/* strxfrm_multiply */
+    1,				/* mbminlen      */
     1,				/* mbmaxlen      */
     (char) 255,			/* max_sort_char */
     &my_charset_handler,

strings/ctype-czech.c

@@ -631,6 +631,7 @@ CHARSET_INFO my_charset_latin2_czech_ci =
     idx_uni_8859_2,	/* tab_from_uni */
     "","",
     4,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     1,			/* mbmaxlen  */
     0,
     &my_charset_8bit_handler,

strings/ctype-euc_kr.c

@@ -8689,6 +8689,7 @@ CHARSET_INFO my_charset_euckr_korean_ci=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     2,			/* mbmaxlen   */
     0,
     &my_charset_handler,
@@ -8712,6 +8713,7 @@ CHARSET_INFO my_charset_euckr_bin=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     2,			/* mbmaxlen   */
     0,
     &my_charset_handler,

strings/ctype-extra.c

@@ -34,6 +34,7 @@ CHARSET_INFO compiled_charsets[] = {
     0,
     0,
     0,
+    0,
     NULL,
     NULL
   }

strings/ctype-gb2312.c

@@ -5740,6 +5740,7 @@ CHARSET_INFO my_charset_gb2312_chinese_ci=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     2,			/* mbmaxlen   */
     0,
     &my_charset_handler,
@@ -5762,6 +5763,7 @@ CHARSET_INFO my_charset_gb2312_bin=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     2,			/* mbmaxlen   */
     0,
     &my_charset_handler,

strings/ctype-gbk.c

@@ -9936,6 +9936,7 @@ CHARSET_INFO my_charset_gbk_chinese_ci=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     2,			/* mbmaxlen */
     0,
     &my_charset_handler,
@@ -9958,6 +9959,7 @@ CHARSET_INFO my_charset_gbk_bin=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     2,			/* mbmaxlen */
     0,
     &my_charset_handler,

strings/ctype-latin1.c

@@ -215,6 +215,7 @@ CHARSET_INFO my_charset_latin1=
     NULL,		/* tab_from_uni */
     "","",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     1,			/* mbmaxlen  */
     0,
     &my_charset_handler,
@@ -410,6 +411,7 @@ CHARSET_INFO my_charset_latin1_german2_ci=
   NULL,					/* tab_from_uni */
   "","",
   2,					/* strxfrm_multiply */
+  1,					/* mbminlen   */
   1,					/* mbmaxlen  */
   0,
   &my_charset_handler,
@@ -433,6 +435,7 @@ CHARSET_INFO my_charset_latin1_bin=
   "",
   "",
   1,					/* strxfrm_multiply */
+  1,					/* mbminlen   */
   1,					/* mbmaxlen  */
   0,
   &my_charset_handler,

strings/ctype-sjis.c

@@ -4525,6 +4525,7 @@ CHARSET_INFO my_charset_sjis_japanese_ci=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     2,			/* mbmaxlen */
     0,
     &my_charset_handler,
@@ -4547,6 +4548,7 @@ CHARSET_INFO my_charset_sjis_bin=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     2,			/* mbmaxlen */
     0,
     &my_charset_handler,

strings/ctype-tis620.c

@@ -956,6 +956,7 @@ CHARSET_INFO my_charset_tis620_thai_ci=
     "",
     "",
     4,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     1,			/* mbmaxlen  */
     0,
     &my_charset_handler,
@@ -978,6 +979,7 @@ CHARSET_INFO my_charset_tis620_bin=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen   */
     1,			/* mbmaxlen  */
     0,
     &my_charset_handler,

strings/ctype-ucs2.c

@@ -1322,6 +1322,7 @@ CHARSET_INFO my_charset_ucs2_general_ci=
     "",
     "",
     1,			/* strxfrm_multiply */
+    2,			/* mbminlen     */
     2,			/* mbmaxlen     */
     0,
     &my_charset_ucs2_handler,
@@ -1345,6 +1346,7 @@ CHARSET_INFO my_charset_ucs2_bin=
     "",
     "",
     1,			/* strxfrm_multiply */
+    2,			/* mbminlen     */
     2,			/* mbmaxlen     */
     0,
     &my_charset_ucs2_handler,

strings/ctype-ujis.c

@@ -8480,6 +8480,7 @@ CHARSET_INFO my_charset_ujis_japanese_ci=
     NULL,		/* tab_from_uni */
     "","",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen     */
     3,			/* mbmaxlen     */
     0,
     &my_charset_handler,
@@ -8502,6 +8503,7 @@ CHARSET_INFO my_charset_ujis_bin=
     NULL,		/* tab_from_uni */
     "","",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen     */
     3,			/* mbmaxlen     */
     0,
     &my_charset_handler,

strings/ctype-utf8.c

@@ -2006,6 +2006,7 @@ CHARSET_INFO my_charset_utf8_general_ci=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen     */
     3,			/* mbmaxlen     */
     0,
     &my_charset_handler,
@@ -2029,6 +2030,7 @@ CHARSET_INFO my_charset_utf8_bin=
     "",
     "",
     1,			/* strxfrm_multiply */
+    1,			/* mbminlen     */
     3,			/* mbmaxlen     */
     0,
     &my_charset_handler,

strings/ctype-win1250ch.c

@@ -671,6 +671,7 @@ CHARSET_INFO my_charset_cp1250_czech_ci =
     idx_uni_cp1250,		/* tab_from_uni */
     "","",
     2,				/* strxfrm_multiply */
+    1,				/* mbminlen  */
     1,				/* mbmaxlen  */
     0,
     &my_charset_8bit_handler,