Bug#15376: Unassigned multibyte codes are converted to U+0000

Mergeing changes into 5.0


mysql-test/r/ctype_eucjpms.result:
  Adding test
mysql-test/r/ctype_ujis.result:
  After merge fix
mysql-test/t/ctype_eucjpms.test:
  Adding test
strings/ctype-cp932.c:
  After merge fix
strings/ctype-eucjpms.c:
  Applying changes to eucjpms, similar to ujis.
  Note: eucjpms appeared in 5.0 so the original patch,
  which was for for 4.1, didn't fix eucjpms.

mysql-test/r/ctype_eucjpms.result

@@ -9819,3 +9819,9 @@ eucjpms_bin	6109
 eucjpms_bin	61
 eucjpms_bin	6120
 drop table t1;
+select hex(convert(_eucjpms 0xA5FE41 using ucs2));
+hex(convert(_eucjpms 0xA5FE41 using ucs2))
+003F0041
+select hex(convert(_eucjpms 0x8FABF841 using ucs2));
+hex(convert(_eucjpms 0x8FABF841 using ucs2))
+003F0041

mysql-test/r/ctype_ujis.result

@@ -2307,6 +2307,12 @@ select c1 as c2h from t1 where c1 like 'ab#_def' escape '#';
 c2h
 ab_def
 drop table t1;
+select hex(convert(_ujis 0xA5FE41 using ucs2));
+hex(convert(_ujis 0xA5FE41 using ucs2))
+003F0041
+select hex(convert(_ujis 0x8FABF841 using ucs2));
+hex(convert(_ujis 0x8FABF841 using ucs2))
+003F0041
 DROP TABLE IF EXISTS t1, t2;
 DROP PROCEDURE IF EXISTS sp1;
 set names ujis;
@@ -2337,9 +2343,3 @@ DROP TABLE t2;
 set names default;
 set character_set_database=default;
 set character_set_server=default;
-select hex(convert(_ujis 0xA5FE41 using ucs2));
-hex(convert(_ujis 0xA5FE41 using ucs2))
-003F0041
-select hex(convert(_ujis 0x8FABF841 using ucs2));
-hex(convert(_ujis 0x8FABF841 using ucs2))
-003F0041

mysql-test/t/ctype_eucjpms.test

@@ -363,3 +363,20 @@ SET collation_connection='eucjpms_japanese_ci';
 -- source include/ctype_filesort.inc
 SET collation_connection='eucjpms_bin';
 -- source include/ctype_filesort.inc
+
+
+#
+# Bugs#15375: Unassigned multibyte codes are broken
+# into parts when converting to Unicode.
+# This query should return 0x003F0041. I.e. it should
+# scan unassigned double-byte character 0xA5FE, convert
+# it as QUESTION MARK 0x003F and then scan the next
+# character, which is a single byte character 0x41.
+#
+select hex(convert(_eucjpms 0xA5FE41 using ucs2));
+# This one should return 0x003F0041:
+# scan unassigned three-byte character 0x8FABF8,
+# convert it as QUESTION MARK 0x003F and then scan
+# the next character, which is a single byte character 0x41.
+select hex(convert(_eucjpms 0x8FABF841 using ucs2));
+

strings/ctype-cp932.c

@@ -5355,7 +5355,7 @@ my_mb_wc_cp932(CHARSET_INFO *cs  __attribute__((unused)),
   int hi=s[0];
   
   if (s >= e)
-    return MY_CS_TOOFEW(0);
+    return MY_CS_TOOSMALL;
   
   if (hi < 0x80)
   {
@@ -5370,10 +5370,10 @@ my_mb_wc_cp932(CHARSET_INFO *cs  __attribute__((unused)),
   }
   
   if (s+2>e)
-    return MY_CS_TOOFEW(0);
+    return MY_CS_TOOSMALL2;
   
   if (!(pwc[0]=func_cp932_uni_onechar((hi<<8)+s[1])))
-    return MY_CS_ILSEQ;
+    return -2;
   
   return 2;
 }

strings/ctype-eucjpms.c

@@ -243,7 +243,7 @@ my_mb_wc_jisx0201(CHARSET_INFO *cs  __attribute__((unused)),
 		  const uchar *e __attribute__((unused)))
 {
   wc[0]=tab_jisx0201_uni[*s];
-  return (!wc[0] && s[0]) ? MY_CS_ILSEQ : 1;
+  return (!wc[0] && s[0]) ? -1 : 1;
 }
 
 
@@ -8473,7 +8473,7 @@ my_mb_wc_euc_jp(CHARSET_INFO *cs,my_wc_t *pwc, const uchar *s, const uchar *e)
   int c1,c2,c3;
   
   if (s >= e)
-    return MY_CS_TOOFEW(0);
+    return MY_CS_TOOSMALL;
   
   c1=s[0];
   
@@ -8485,7 +8485,7 @@ my_mb_wc_euc_jp(CHARSET_INFO *cs,my_wc_t *pwc, const uchar *s, const uchar *e)
   }
   
   if (s+2>e)
-    return MY_CS_TOOFEW(0);
+    return MY_CS_TOOSMALL2;
     
   c2=s[1];
   
@@ -8500,7 +8500,7 @@ my_mb_wc_euc_jp(CHARSET_INFO *cs,my_wc_t *pwc, const uchar *s, const uchar *e)
     {
       pwc[0]=my_jisx0208_uni_onechar( ((c1-0x80) << 8) + (c2-0x80));
       if (!pwc[0])
-        return MY_CS_ILSEQ;
+        return -2;
     }
     else
     {
@@ -8520,7 +8520,7 @@ my_mb_wc_euc_jp(CHARSET_INFO *cs,my_wc_t *pwc, const uchar *s, const uchar *e)
     
     ret = my_mb_wc_jisx0201(cs,pwc,s+1,e);
     if (ret!=1)
-      return ret;
+      return -2;
     return 2;
   }
   
@@ -8531,7 +8531,7 @@ my_mb_wc_euc_jp(CHARSET_INFO *cs,my_wc_t *pwc, const uchar *s, const uchar *e)
       return MY_CS_ILSEQ;
     
     if (s+3>e)
-      return MY_CS_TOOFEW(0);
+      return MY_CS_TOOSMALL3;
     
     c3=s[2];
     if (c3 < 0xA1 || c3>=0xFF)
@@ -8540,8 +8540,8 @@ my_mb_wc_euc_jp(CHARSET_INFO *cs,my_wc_t *pwc, const uchar *s, const uchar *e)
     if (c2<0xF5)
     {
       pwc[0]=my_jisx0212_uni_onechar((c2-0x80)*256 + (c3-0x80));
-      if (!pwc)
-        return MY_CS_ILSEQ;
+      if (!pwc[0])
+        return -3;
     }
     else
     {
@@ -8572,7 +8572,7 @@ my_wc_mb_euc_jp(CHARSET_INFO *c,my_wc_t wc, unsigned char *s, unsigned char *e)
   if ((jp=my_uni_jisx0208_onechar(wc)))
   {
     if (s+2>e)
-      return MY_CS_TOOSMALL;
+      return MY_CS_TOOSMALL2;
       
     jp+=0x8080;
     s[0]=jp>>8;
@@ -8584,7 +8584,7 @@ my_wc_mb_euc_jp(CHARSET_INFO *c,my_wc_t wc, unsigned char *s, unsigned char *e)
   if (my_wc_mb_jisx0201(c,wc,s,e) == 1)
   {
     if (s+2>e)
-      return MY_CS_TOOSMALL;
+      return MY_CS_TOOSMALL2;
     s[1]= s[0];
     s[0]= 0x8E;
     return 2;
@@ -8594,7 +8594,7 @@ my_wc_mb_euc_jp(CHARSET_INFO *c,my_wc_t wc, unsigned char *s, unsigned char *e)
   if ((jp=my_uni_jisx0212_onechar(wc)))
   {
     if (s+3>e)
-      return MY_CS_TOOSMALL;
+      return MY_CS_TOOSMALL3;
       
     jp+=0x8080;
     s[0]=0x8F;
@@ -8608,7 +8608,7 @@ my_wc_mb_euc_jp(CHARSET_INFO *c,my_wc_t wc, unsigned char *s, unsigned char *e)
   if (wc>=0xE000 && wc<0xE3AC)
   { 
     if (s+2>e)
-      return MY_CS_TOOSMALL;
+      return MY_CS_TOOSMALL2;
       
     c1=((unsigned)(wc-0xE000)/94)+0xF5;
     s[0]=c1;
@@ -8622,7 +8622,7 @@ my_wc_mb_euc_jp(CHARSET_INFO *c,my_wc_t wc, unsigned char *s, unsigned char *e)
   if (wc>=0xE3AC && wc<0xE758)
   {
     if (s+3>e)
-      return MY_CS_TOOSMALL;
+      return MY_CS_TOOSMALL3;
       
     s[0]=0x8F;
     c1=((unsigned)(wc-0xE3AC)/94)+0xF5;