New functions in CHARSET_INFO structure

16 files changed, 94 insertions, 2 deletions

diff --git a/strings/ctype-big5.c b/strings/ctype-big5.c
index 89a25666bd8..7827b8714ec 100644
--- a/strings/ctype-big5.c
+++ b/strings/ctype-big5.c
@@ -6254,6 +6254,8 @@ CHARSET_INFO my_charset_big5 =
     ismbchar_big5,
     ismbhead_big5,
     mbcharlen_big5,
+    my_numchars_mb,
+    my_charpos_mb,
     my_mb_wc_big5,	/* mb_wc       */
     my_wc_mb_big5,	/* wc_mb       */
     my_caseup_str_mb,
diff --git a/strings/ctype-bin.c b/strings/ctype-bin.c
index 1b9c67c10a9..4d428828826 100644
--- a/strings/ctype-bin.c
+++ b/strings/ctype-bin.c
@@ -292,6 +292,8 @@ CHARSET_INFO my_charset_bin =
     NULL,			/* ismbchar      */
     NULL,			/* ismbhead      */
     NULL,			/* mbcharlen     */
+    my_numchars_8bit,
+    my_charpos_8bit,
     my_mb_wc_bin,		/* mb_wc         */
     my_wc_mb_bin,		/* wc_mb         */
     my_caseup_str_bin,		/* caseup_str    */
diff --git a/strings/ctype-czech.c b/strings/ctype-czech.c
index 9d6b8fe2131..f42ab66fed9 100644
--- a/strings/ctype-czech.c
+++ b/strings/ctype-czech.c
@@ -628,6 +628,8 @@ CHARSET_INFO my_charset_czech =
     NULL,		/* ismbchar  */
     NULL,		/* ismbhead  */
     NULL,		/* mbcharlen */
+    my_numchars_8bit,
+    my_charpos_8bit,
     my_mb_wc_8bit,	/* mb_wc   */
     my_wc_mb_8bit,	/* wc_mb   */
     my_caseup_str_8bit,
diff --git a/strings/ctype-euc_kr.c b/strings/ctype-euc_kr.c
index 938f215e2de..e49bf6a38fd 100644
--- a/strings/ctype-euc_kr.c
+++ b/strings/ctype-euc_kr.c
@@ -8662,6 +8662,8 @@ CHARSET_INFO my_charset_euc_kr =
     ismbchar_euc_kr,
     ismbhead_euc_kr,
     mbcharlen_euc_kr,
+    my_numchars_mb,
+    my_charpos_mb,
     my_mb_wc_euc_kr,	/* mb_wc   */
     my_wc_mb_euc_kr,	/* wc_mb   */
     my_caseup_str_mb,
diff --git a/strings/ctype-extra.c b/strings/ctype-extra.c
index 6100c74860b..3e3cd38c063 100644
--- a/strings/ctype-extra.c
+++ b/strings/ctype-extra.c
@@ -2828,6 +2828,8 @@ CHARSET_INFO compiled_charsets[] = {
     NULL,		/* ismbchar   */
     NULL,		/* ismbhead   */
     NULL,		/* mbcharlen  */
+    my_numchars_8bit,
+    my_charpos_8bit,
     my_mb_wc_8bit,	/* mb_wc   */
     my_wc_mb_8bit,	/* wc_mb   */
     my_caseup_str_8bit,
@@ -2877,6 +2879,8 @@ CHARSET_INFO compiled_charsets[] = {
     NULL,		/* ismbchar   */
     NULL,		/* ismbhead   */
     NULL,		/* mbcharlen  */
+    my_numchars_8bit,
+    my_charpos_8bit,
     my_mb_wc_8bit,	/* mb_wc      */
     my_wc_mb_8bit,	/* wc_mb      */
     my_caseup_str_8bit,
@@ -2925,6 +2929,8 @@ CHARSET_INFO compiled_charsets[] = {
     NULL,		/* ismbchar  */
     NULL,		/* ismbhead  */
     NULL,		/* mbcharlen */
+    my_numchars_8bit,
+    my_charpos_8bit,
     my_mb_wc_8bit,	/* mb_wc   */
     my_wc_mb_8bit,	/* wc_mb   */
     my_caseup_str_8bit,
@@ -3937,7 +3943,8 @@ CHARSET_INFO compiled_charsets[] = {
     NULL,
     NULL,
     NULL,
-
+    NULL,
+    NULL,
     NULL,		 /* mb_wc      */
     NULL,		 /* wc_mb      */
     
diff --git a/strings/ctype-gb2312.c b/strings/ctype-gb2312.c
index ff2dcd381c1..9cc19dc46ef 100644
--- a/strings/ctype-gb2312.c
+++ b/strings/ctype-gb2312.c
@@ -5712,6 +5712,8 @@ CHARSET_INFO my_charset_gb2312 =
     ismbchar_gb2312,
     ismbhead_gb2312,
     mbcharlen_gb2312,
+    my_numchars_mb,
+    my_charpos_mb,
     my_mb_wc_gb2312,	/* mb_wc      */
     my_wc_mb_gb2312,	/* wc_mb      */
     my_caseup_str_mb,
diff --git a/strings/ctype-gbk.c b/strings/ctype-gbk.c
index a7a29a4af7a..42a3686475a 100644
--- a/strings/ctype-gbk.c
+++ b/strings/ctype-gbk.c
@@ -9909,6 +9909,8 @@ CHARSET_INFO my_charset_gbk =
     ismbchar_gbk,
     ismbhead_gbk,
     mbcharlen_gbk,
+    my_numchars_mb,
+    my_charpos_mb,
     my_mb_wc_gbk,	/* mb_wc      */
     my_wc_mb_gbk,	/* wc_mb      */
     my_caseup_str_mb,
diff --git a/strings/ctype-latin1.c b/strings/ctype-latin1.c
index 588eff0cab1..518418a9050 100644
--- a/strings/ctype-latin1.c
+++ b/strings/ctype-latin1.c
@@ -198,6 +198,8 @@ CHARSET_INFO my_charset_latin1 =
     NULL,		/* ismbchar  */
     NULL,		/* ismbhead  */
     NULL,		/* mbcharlen */
+    my_numchars_8bit,
+    my_charpos_8bit,
     my_mb_wc_latin1,	/* mb_wc     */
     my_wc_mb_latin1,	/* wc_mb     */
     my_caseup_str_8bit,
diff --git a/strings/ctype-latin1_de.c b/strings/ctype-latin1_de.c
index 0a6de1ca86b..93c8ba54bc0 100644
--- a/strings/ctype-latin1_de.c
+++ b/strings/ctype-latin1_de.c
@@ -369,6 +369,8 @@ CHARSET_INFO my_charset_latin1_de =
     NULL,		/* ismbchar  */
     NULL,		/* ismbhead  */
     NULL,		/* mbcharlen */
+    my_numchars_8bit,
+    my_charpos_8bit,
     my_mb_wc_8bit,	/* mb_wc     */
     my_wc_mb_8bit,	/* wc_mb     */
     my_caseup_str_8bit,
diff --git a/strings/ctype-mb.c b/strings/ctype-mb.c
index 4c8471d4217..f5b7000cb18 100644
--- a/strings/ctype-mb.c
+++ b/strings/ctype-mb.c
@@ -276,5 +276,33 @@ int my_wildcmp_mb(CHARSET_INFO *cs,
   return (str != str_end ? 1 : 0);
 }
 
+uint my_numchars_mb(CHARSET_INFO *cs __attribute__((unused)),
+		      const char *b, const char *e)
+{
+  register uint32 n=0,mblen;
+  while (b < e) 
+  {
+    b+= (mblen= my_ismbchar(cs,b,e)) ? mblen : 1;
+    ++n;
+  }
+  return n;
+}
+
+uint my_charpos_mb(CHARSET_INFO *cs __attribute__((unused)),
+		     const char *b  __attribute__((unused)),
+		     const char *e  __attribute__((unused)),
+		     uint pos)
+{
+  uint res=0, mblen;
+  const char *b0;
+  
+  while (pos && b<e)
+  {
+    b+= (mblen= my_ismbchar(cs,b,e)) ? mblen : 1;
+    pos--;
+  }
+  return b-b0;
+}
+
 
 #endif
diff --git a/strings/ctype-simple.c b/strings/ctype-simple.c
index c4c89d02d58..5fb62d3056f 100644
--- a/strings/ctype-simple.c
+++ b/strings/ctype-simple.c
@@ -1035,4 +1035,18 @@ void my_fill_8bit(CHARSET_INFO *cs __attribute__((unused)),
 		   char *s, uint l, int fill)
 {
   bfill(s,l,fill);
-}
\ No newline at end of file
+}
+
+uint my_numchars_8bit(CHARSET_INFO *cs __attribute__((unused)),
+		      const char *b, const char *e)
+{
+  return e-b;
+}
+
+uint my_charpos_8bit(CHARSET_INFO *cs __attribute__((unused)),
+		     const char *b  __attribute__((unused)),
+		     const char *e  __attribute__((unused)),
+		     uint pos)
+{
+  return pos;
+}
diff --git a/strings/ctype-sjis.c b/strings/ctype-sjis.c
index 4d9bb4be30c..a266fa71199 100644
--- a/strings/ctype-sjis.c
+++ b/strings/ctype-sjis.c
@@ -4496,6 +4496,8 @@ CHARSET_INFO my_charset_sjis =
     ismbchar_sjis,
     ismbhead_sjis,
     mbcharlen_sjis,
+    my_numchars_mb,
+    my_charpos_mb,
     my_mb_wc_sjis,	/* mb_wc */
     my_wc_mb_sjis,	/* wc_mb */
     my_caseup_str_8bit,
diff --git a/strings/ctype-tis620.c b/strings/ctype-tis620.c
index d43361a1723..bf3d332880d 100644
--- a/strings/ctype-tis620.c
+++ b/strings/ctype-tis620.c
@@ -719,6 +719,8 @@ CHARSET_INFO my_charset_tis620 =
     NULL,		/* ismbchar  */
     NULL,		/* ismbhead  */
     NULL,		/* mbcharlen */
+    my_numchars_8bit,
+    my_charpos_8bit,
     my_mb_wc_8bit,	/* mb_wc     */
     my_wc_mb_8bit,	/* wc_mb     */
     my_caseup_str_8bit,
diff --git a/strings/ctype-ujis.c b/strings/ctype-ujis.c
index 19f89f1b755..d39be825ef0 100644
--- a/strings/ctype-ujis.c
+++ b/strings/ctype-ujis.c
@@ -8453,6 +8453,8 @@ CHARSET_INFO my_charset_ujis =
     ismbchar_ujis,
     ismbhead_ujis,
     mbcharlen_ujis,
+    my_numchars_mb,
+    my_charpos_mb,
     my_mb_wc_euc_jp,	 /* mb_wc       */
     my_wc_mb_euc_jp,	 /* wc_mb       */
     my_caseup_str_mb,
diff --git a/strings/ctype-utf8.c b/strings/ctype-utf8.c
index fc8a4c0fb64..6eebcb022a5 100644
--- a/strings/ctype-utf8.c
+++ b/strings/ctype-utf8.c
@@ -1998,6 +1998,8 @@ CHARSET_INFO my_charset_utf8 =
     my_ismbchar_utf8,	/* ismbchar     */
     my_ismbhead_utf8,	/* ismbhead     */
     my_mbcharlen_utf8,	/* mbcharlen    */
+    my_numchars_mb,
+    my_charpos_mb,
     my_utf8_uni,	/* mb_wc        */
     my_uni_utf8,	/* wc_mb        */
     my_caseup_str_utf8,
@@ -3064,6 +3066,21 @@ cnv:
   return (int) (dst-db);
 }
 
+static
+uint my_numchars_ucs2(CHARSET_INFO *cs __attribute__((unused)),
+		      const char *b, const char *e)
+{
+  return (e-b)/2;
+}
+
+static
+uint my_charpos_ucs2(CHARSET_INFO *cs __attribute__((unused)),
+		     const char *b  __attribute__((unused)),
+		     const char *e  __attribute__((unused)),
+		     uint pos)
+{
+  return pos/2;
+}
 
 CHARSET_INFO my_charset_ucs2 =
 {
@@ -3088,6 +3105,8 @@ CHARSET_INFO my_charset_ucs2 =
     my_ismbchar_ucs2,	/* ismbchar     */
     my_ismbhead_ucs2,	/* ismbhead     */
     my_mbcharlen_ucs2,	/* mbcharlen    */
+    my_numchars_ucs2,
+    my_charpos_ucs2,
     my_ucs2_uni,	/* mb_wc        */
     my_uni_ucs2,	/* wc_mb        */
     my_caseup_str_ucs2,
diff --git a/strings/ctype-win1250ch.c b/strings/ctype-win1250ch.c
index 3ed48f13913..467d1ba4c91 100644
--- a/strings/ctype-win1250ch.c
+++ b/strings/ctype-win1250ch.c
@@ -663,6 +663,8 @@ CHARSET_INFO my_charset_win1250ch =
     NULL,			/* ismbchar  */
     NULL,			/* ismbhead  */
     NULL,			/* mbcharlen */
+    my_numchars_8bit,
+    my_charpos_8bit,
     my_mb_wc_8bit,		/* mb_wc     */
     my_wc_mb_8bit,		/* wc_mb     */
     my_caseup_str_8bit,