Problem fix:

http://bugs.mysql.com/bug.php?id=2366
Wrong utf8 behaviour when data is trancated

13 files changed, 52 insertions, 2 deletions

diff --git a/strings/ctype-big5.c b/strings/ctype-big5.c
index 8d4081fb2aa..574156a99ed 100644
--- a/strings/ctype-big5.c
+++ b/strings/ctype-big5.c
@@ -6245,6 +6245,7 @@ static MY_CHARSET_HANDLER my_charset_big5_handler=
   mbcharlen_big5,
   my_numchars_mb,
   my_charpos_mb,
+  my_wellformedlen_mb,
   my_lengthsp_8bit,
   my_mb_wc_big5,	/* mb_wc       */
   my_wc_mb_big5,	/* wc_mb       */
diff --git a/strings/ctype-bin.c b/strings/ctype-bin.c
index 43575bbc277..fc22938d46e 100644
--- a/strings/ctype-bin.c
+++ b/strings/ctype-bin.c
@@ -347,6 +347,7 @@ static MY_CHARSET_HANDLER my_charset_handler=
     my_mbcharlen_8bit,		/* mbcharlen     */
     my_numchars_8bit,
     my_charpos_8bit,
+    my_wellformedlen_8bit,
     my_lengthsp_8bit,
     my_mb_wc_bin,
     my_wc_mb_bin,
diff --git a/strings/ctype-euc_kr.c b/strings/ctype-euc_kr.c
index 366a5d500ed..278e8529e83 100644
--- a/strings/ctype-euc_kr.c
+++ b/strings/ctype-euc_kr.c
@@ -8653,6 +8653,7 @@ static MY_CHARSET_HANDLER my_charset_handler=
   mbcharlen_euc_kr,
   my_numchars_mb,
   my_charpos_mb,
+  my_wellformedlen_mb,
   my_lengthsp_8bit,
   my_mb_wc_euc_kr,	/* mb_wc   */
   my_wc_mb_euc_kr,	/* wc_mb   */
diff --git a/strings/ctype-gb2312.c b/strings/ctype-gb2312.c
index 44a58b2b906..722f00f0f7a 100644
--- a/strings/ctype-gb2312.c
+++ b/strings/ctype-gb2312.c
@@ -5704,6 +5704,7 @@ static MY_CHARSET_HANDLER my_charset_handler=
   mbcharlen_gb2312,
   my_numchars_mb,
   my_charpos_mb,
+  my_wellformedlen_mb,
   my_lengthsp_8bit,
   my_mb_wc_gb2312,	/* mb_wc      */
   my_wc_mb_gb2312,	/* wc_mb      */
diff --git a/strings/ctype-gbk.c b/strings/ctype-gbk.c
index 5475c3bd363..9e71a18e531 100644
--- a/strings/ctype-gbk.c
+++ b/strings/ctype-gbk.c
@@ -9900,6 +9900,7 @@ static MY_CHARSET_HANDLER my_charset_handler=
   mbcharlen_gbk,
   my_numchars_mb,
   my_charpos_mb,
+  my_wellformedlen_mb,
   my_lengthsp_8bit,
   my_mb_wc_gbk,
   my_wc_mb_gbk,
diff --git a/strings/ctype-latin1.c b/strings/ctype-latin1.c
index c00ded21575..933737b5f61 100644
--- a/strings/ctype-latin1.c
+++ b/strings/ctype-latin1.c
@@ -180,6 +180,7 @@ static MY_CHARSET_HANDLER my_charset_handler=
     my_mbcharlen_8bit,
     my_numchars_8bit,
     my_charpos_8bit,
+    my_wellformedlen_8bit,
     my_lengthsp_8bit,
     my_mb_wc_latin1,
     my_wc_mb_latin1,
diff --git a/strings/ctype-mb.c b/strings/ctype-mb.c
index b5e8c4598a0..377bf311d38 100644
--- a/strings/ctype-mb.c
+++ b/strings/ctype-mb.c
@@ -274,6 +274,22 @@ uint my_charpos_mb(CHARSET_INFO *cs __attribute__((unused)),
   return pos ? e+2-b0 : b-b0;
 }
 
+uint my_wellformedlen_mb(CHARSET_INFO *cs __attribute__((unused)),
+		     const char *b, const char *e, uint pos)
+{
+  uint mblen;
+  const char *b0=b;
+  
+  while (pos && b<e)
+  {
+    b+= (mblen= my_ismbchar(cs,b,e)) ? mblen : 1;
+    pos--;
+  }
+  return b-b0;
+}
+
+
+
 uint my_instr_mb(CHARSET_INFO *cs,
                  const char *b, uint b_length, 
                  const char *s, uint s_length,
diff --git a/strings/ctype-simple.c b/strings/ctype-simple.c
index 62eb1418970..fb2dee99ba5 100644
--- a/strings/ctype-simple.c
+++ b/strings/ctype-simple.c
@@ -1020,6 +1020,15 @@ uint my_charpos_8bit(CHARSET_INFO *cs __attribute__((unused)),
   return pos;
 }
 
+uint my_wellformedlen_8bit(CHARSET_INFO *cs __attribute__((unused)),
+			const char *b,
+			const char *e,
+			uint nchars)
+{
+  uint nbytes= e-b;
+  return nbytes < nchars ? nbytes : nchars;
+}
+
 uint my_lengthsp_8bit(CHARSET_INFO *cs __attribute__((unused)),
 		      const char *ptr, uint length)
 {
@@ -1096,6 +1105,7 @@ MY_CHARSET_HANDLER my_charset_8bit_handler=
     my_mbcharlen_8bit,		/* mbcharlen     */
     my_numchars_8bit,
     my_charpos_8bit,
+    my_wellformedlen_8bit,
     my_lengthsp_8bit,
     my_mb_wc_8bit,
     my_wc_mb_8bit,
diff --git a/strings/ctype-sjis.c b/strings/ctype-sjis.c
index 42f32fe739b..feff0fff227 100644
--- a/strings/ctype-sjis.c
+++ b/strings/ctype-sjis.c
@@ -4489,6 +4489,7 @@ static MY_CHARSET_HANDLER my_charset_handler=
   mbcharlen_sjis,
   my_numchars_mb,
   my_charpos_mb,
+  my_wellformedlen_mb,
   my_lengthsp_8bit,
   my_mb_wc_sjis,	/* mb_wc */
   my_wc_mb_sjis,	/* wc_mb */
diff --git a/strings/ctype-tis620.c b/strings/ctype-tis620.c
index 09552a0dc23..d577b964405 100644
--- a/strings/ctype-tis620.c
+++ b/strings/ctype-tis620.c
@@ -874,7 +874,8 @@ NULL,NULL,NULL,NULL,NULL,NULL,NULL,plFF
 
 
 static
-int my_mb_wc_tis620(CHARSET_INFO *cs,my_wc_t *wc,
+int my_mb_wc_tis620(CHARSET_INFO *cs  __attribute__((unused)),
+		  my_wc_t *wc,
 		  const unsigned char *str,
 		  const unsigned char *end __attribute__((unused)))
 {
@@ -886,7 +887,8 @@ int my_mb_wc_tis620(CHARSET_INFO *cs,my_wc_t *wc,
 }
 
 static
-int my_wc_mb_tis620(CHARSET_INFO *cs,my_wc_t wc,
+int my_wc_mb_tis620(CHARSET_INFO *cs  __attribute__((unused)),
+		  my_wc_t wc,
 		  unsigned char *str,
 		  unsigned char *end __attribute__((unused)))
 {
@@ -919,6 +921,7 @@ static MY_CHARSET_HANDLER my_charset_handler=
     my_mbcharlen_8bit,	/* mbcharlen */
     my_numchars_8bit,
     my_charpos_8bit,
+    my_wellformedlen_8bit,
     my_lengthsp_8bit,
     my_mb_wc_tis620,	/* mb_wc     */
     my_wc_mb_tis620,	/* wc_mb     */
diff --git a/strings/ctype-ucs2.c b/strings/ctype-ucs2.c
index 775501027d0..bb74e0cf56b 100644
--- a/strings/ctype-ucs2.c
+++ b/strings/ctype-ucs2.c
@@ -1002,6 +1002,17 @@ uint my_charpos_ucs2(CHARSET_INFO *cs __attribute__((unused)),
 }
 
 static
+uint my_wellformedlen_ucs2(CHARSET_INFO *cs __attribute__((unused)),
+		     const char *b,
+		     const char *e,
+		     uint nchars)
+{
+  uint nbytes= (e-b) & ~ (uint)1;
+  nchars*= 2;
+  return nbytes < nchars ? nbytes : nchars;
+}
+
+static
 void my_fill_ucs2(CHARSET_INFO *cs __attribute__((unused)),
 		   char *s, uint l, int fill)
 {
@@ -1287,6 +1298,7 @@ static MY_CHARSET_HANDLER my_charset_ucs2_handler=
     my_mbcharlen_ucs2,	/* mbcharlen    */
     my_numchars_ucs2,
     my_charpos_ucs2,
+    my_wellformedlen_ucs2,
     my_lengthsp_ucs2,
     my_ucs2_uni,	/* mb_wc        */
     my_uni_ucs2,	/* wc_mb        */
diff --git a/strings/ctype-ujis.c b/strings/ctype-ujis.c
index f6928e9426e..f27ddcf3e30 100644
--- a/strings/ctype-ujis.c
+++ b/strings/ctype-ujis.c
@@ -8444,6 +8444,7 @@ static MY_CHARSET_HANDLER my_charset_handler=
     mbcharlen_ujis,
     my_numchars_mb,
     my_charpos_mb,
+    my_wellformedlen_mb,
     my_lengthsp_8bit,
     my_mb_wc_euc_jp,	 /* mb_wc       */
     my_wc_mb_euc_jp,	 /* wc_mb       */
diff --git a/strings/ctype-utf8.c b/strings/ctype-utf8.c
index 8004fba75b7..ef9719bf040 100644
--- a/strings/ctype-utf8.c
+++ b/strings/ctype-utf8.c
@@ -1969,6 +1969,7 @@ static MY_CHARSET_HANDLER my_charset_handler=
     my_mbcharlen_utf8,
     my_numchars_mb,
     my_charpos_mb,
+    my_wellformedlen_mb,
     my_lengthsp_8bit,
     my_utf8_uni,
     my_uni_utf8,