MDEV-6353 my_ismbchar() and my_mbcharlen() refactoring

17 files changed, 17 insertions, 186 deletions

diff --git a/strings/ctype-big5.c b/strings/ctype-big5.c
index 962931913a2..9ae394eeedc 100644
--- a/strings/ctype-big5.c
+++ b/strings/ctype-big5.c
@@ -848,12 +848,6 @@ static uint16 big5strokexfrm(uint16 i)
 }
 
 
-static uint mbcharlen_big5(CHARSET_INFO *cs __attribute__((unused)), uint c)
-{
-  return (isbig5head(c)? 2 : 1);
-}
-
-
 /* page 0 0xA140-0xC7FC */
 static const uint16 tab_big5_uni0[]={
 0x3000,0xFF0C,0x3001,0x3002,0xFF0E,0x2022,0xFF1B,0xFF1A,
@@ -6731,7 +6725,6 @@ static MY_COLLATION_HANDLER my_collation_handler_big5_bin=
 static MY_CHARSET_HANDLER my_charset_big5_handler=
 {
   NULL,			/* init */
-  mbcharlen_big5,
   my_numchars_mb,
   my_charpos_mb,
   my_well_formed_len_big5,
diff --git a/strings/ctype-bin.c b/strings/ctype-bin.c
index 8331de34ee4..aab7f2b7448 100644
--- a/strings/ctype-bin.c
+++ b/strings/ctype-bin.c
@@ -225,13 +225,6 @@ static int my_strcasecmp_bin(CHARSET_INFO * cs __attribute__((unused)),
 }
 
 
-uint my_mbcharlen_8bit(CHARSET_INFO *cs __attribute__((unused)),
-                      uint c __attribute__((unused)))
-{
-  return 1;
-}
-
-
 static int my_mb_wc_bin(CHARSET_INFO *cs __attribute__((unused)),
 			my_wc_t *wc,
 			const uchar *str,
@@ -510,7 +503,6 @@ static MY_COLLATION_HANDLER my_collation_binary_handler =
 static MY_CHARSET_HANDLER my_charset_handler=
 {
   NULL,			/* init */
-  my_mbcharlen_8bit,	/* mbcharlen     */
   my_numchars_8bit,
   my_charpos_8bit,
   my_well_formed_len_8bit,
diff --git a/strings/ctype-cp932.c b/strings/ctype-cp932.c
index 2163662269d..151fac8a052 100644
--- a/strings/ctype-cp932.c
+++ b/strings/ctype-cp932.c
@@ -191,12 +191,6 @@ static const uchar sort_order_cp932[]=
 #include "ctype-mb.ic"
 
 
-static uint mbcharlen_cp932(CHARSET_INFO *cs __attribute__((unused)),uint c)
-{
-  return (iscp932head((uchar) c) ? 2 : 1);
-}
-
-
 #define cp932code(c,d)	((((uint) (uchar)(c)) << 8) | (uint) (uchar) (d))
 
 
@@ -34687,7 +34681,6 @@ static MY_COLLATION_HANDLER my_collation_handler_cp932_bin=
 static MY_CHARSET_HANDLER my_charset_handler=
 {
   NULL,			/* init */
-  mbcharlen_cp932,
   my_numchars_mb,
   my_charpos_mb,
   my_well_formed_len_cp932,
diff --git a/strings/ctype-euc_kr.c b/strings/ctype-euc_kr.c
index 19ed586ea49..d2389136c49 100644
--- a/strings/ctype-euc_kr.c
+++ b/strings/ctype-euc_kr.c
@@ -210,12 +210,6 @@ static const uchar sort_order_euc_kr[]=
 #include "ctype-mb.ic"
 
 
-static uint mbcharlen_euc_kr(CHARSET_INFO *cs __attribute__((unused)),uint c)
-{
-  return (iseuc_kr_head(c) ? 2 : 1);
-}
-
-
 static MY_UNICASE_CHARACTER cA3[256]=
 {
   {0,0,0},{0,0,0},{0,0,0},{0,0,0},{0,0,0},{0,0,0},{0,0,0},{0,0,0}, /* xx00 */
@@ -9979,7 +9973,6 @@ static MY_COLLATION_HANDLER my_collation_handler_euckr_bin=
 static MY_CHARSET_HANDLER my_charset_handler=
 {
   NULL,			/* init */
-  mbcharlen_euc_kr,
   my_numchars_mb,
   my_charpos_mb,
   my_well_formed_len_euckr,
diff --git a/strings/ctype-eucjpms.c b/strings/ctype-eucjpms.c
index 469d3a5be6c..caafd1d13b4 100644
--- a/strings/ctype-eucjpms.c
+++ b/strings/ctype-eucjpms.c
@@ -221,12 +221,6 @@ static const uchar sort_order_eucjpms[]=
 #include "strcoll.ic"
 
 
-static uint mbcharlen_eucjpms(CHARSET_INFO *cs __attribute__((unused)),uint c)
-{
-  return (iseucjpms(c)? 2: iseucjpms_ss2(c)? 2: iseucjpms_ss3(c)? 3: 1);
-}
-
-
 /* Case info pages for JIS-X-0208 range */
 
 static MY_UNICASE_CHARACTER cA2[256]=
@@ -67511,7 +67505,6 @@ static MY_COLLATION_HANDLER my_collation_eucjpms_bin_handler =
 static MY_CHARSET_HANDLER my_charset_handler=
 {
     NULL,		/* init */
-    mbcharlen_eucjpms,
     my_numchars_mb,
     my_charpos_mb,
     my_well_formed_len_eucjpms,
diff --git a/strings/ctype-gb2312.c b/strings/ctype-gb2312.c
index a77237c1791..dbb92fabc4b 100644
--- a/strings/ctype-gb2312.c
+++ b/strings/ctype-gb2312.c
@@ -173,12 +173,6 @@ static const uchar sort_order_gb2312[]=
 #include "ctype-mb.ic"
 
 
-static uint mbcharlen_gb2312(CHARSET_INFO *cs __attribute__((unused)),uint c)
-{
-  return (isgb2312head(c)? 2 : 1);
-}
-
-
 static MY_UNICASE_CHARACTER cA2[256]=
 {
   {0,0,0},{0,0,0},{0,0,0},{0,0,0},{0,0,0},{0,0,0},{0,0,0},{0,0,0}, /* xx00 */
@@ -6385,7 +6379,6 @@ static MY_COLLATION_HANDLER my_collation_handler_gb2312_bin=
 static MY_CHARSET_HANDLER my_charset_handler=
 {
   NULL,			/* init */
-  mbcharlen_gb2312,
   my_numchars_mb,
   my_charpos_mb,
   my_well_formed_len_gb2312,
diff --git a/strings/ctype-gbk.c b/strings/ctype-gbk.c
index e4e015a59d2..617d72dc4d9 100644
--- a/strings/ctype-gbk.c
+++ b/strings/ctype-gbk.c
@@ -3451,11 +3451,6 @@ static uint16 gbksortorder(uint16 i)
 }
 
 
-static uint mbcharlen_gbk(CHARSET_INFO *cs __attribute__((unused)),uint c)
-{
-  return (isgbkhead(c)? 2 : 1);
-}
-
 /* page 0 0x8140-0xFE4F */
 static const uint16 tab_gbk_uni0[]={
 0x4E02,0x4E04,0x4E05,0x4E06,0x4E0F,0x4E12,0x4E17,0x4E1F,
@@ -10666,7 +10661,6 @@ static MY_COLLATION_HANDLER my_collation_handler_gbk_bin=
 static MY_CHARSET_HANDLER my_charset_handler=
 {
   NULL,			/* init */
-  mbcharlen_gbk,
   my_numchars_mb,
   my_charpos_mb,
   my_well_formed_len_gbk,
diff --git a/strings/ctype-latin1.c b/strings/ctype-latin1.c
index aba63d97abb..fe255349530 100644
--- a/strings/ctype-latin1.c
+++ b/strings/ctype-latin1.c
@@ -396,7 +396,6 @@ int my_wc_mb_latin1(CHARSET_INFO *cs  __attribute__((unused)),
 static MY_CHARSET_HANDLER my_charset_handler=
 {
     NULL,			/* init */
-    my_mbcharlen_8bit,
     my_numchars_8bit,
     my_charpos_8bit,
     my_well_formed_len_8bit,
diff --git a/strings/ctype-mb.c b/strings/ctype-mb.c
index 3fa66cb0b2f..56b3309ca10 100644
--- a/strings/ctype-mb.c
+++ b/strings/ctype-mb.c
@@ -230,7 +230,7 @@ int my_strcasecmp_mb(CHARSET_INFO * cs,const char *s, const char *t)
         if (*s++ != *t++) 
           return 1;
     }
-    else if (my_mbcharlen(cs, *t) > 1)
+    else if (my_charlen(cs, t, t + cs->mbmaxlen) > 1)
       return 1;
     else if (map[(uchar) *s++] != map[(uchar) *t++])
       return 1;
diff --git a/strings/ctype-simple.c b/strings/ctype-simple.c
index 5e5a345a638..f405c4f327b 100644
--- a/strings/ctype-simple.c
+++ b/strings/ctype-simple.c
@@ -1059,6 +1059,13 @@ size_t my_scan_8bit(CHARSET_INFO *cs, const char *str, const char *end, int sq)
         break;
     }
     return (size_t) (str - str0);
+  case MY_SEQ_NONSPACES:
+    for ( ; str < end ; str++)
+    {
+      if (my_isspace(cs, *str))
+        break;
+    }
+    return (size_t) (str - str0);
   default:
     return 0;
   }
@@ -1916,7 +1923,6 @@ my_strxfrm_pad_desc_and_reverse(CHARSET_INFO *cs,
 MY_CHARSET_HANDLER my_charset_8bit_handler=
 {
     my_cset_init_8bit,
-    my_mbcharlen_8bit,		/* mbcharlen     */
     my_numchars_8bit,
     my_charpos_8bit,
     my_well_formed_len_8bit,
diff --git a/strings/ctype-sjis.c b/strings/ctype-sjis.c
index ebcea22d242..e0546146008 100644
--- a/strings/ctype-sjis.c
+++ b/strings/ctype-sjis.c
@@ -192,12 +192,6 @@ static const uchar sort_order_sjis[]=
 #include "ctype-mb.ic"
 
 
-static uint mbcharlen_sjis(CHARSET_INFO *cs __attribute__((unused)),uint c)
-{
-  return (issjishead((uchar) c) ? 2 : 1);
-}
-
-
 #define sjiscode(c,d)	((((uint) (uchar)(c)) << 8) | (uint) (uchar) (d))
 
 
@@ -34066,7 +34060,6 @@ static MY_COLLATION_HANDLER my_collation_handler_sjis_bin=
 static MY_CHARSET_HANDLER my_charset_handler=
 {
   NULL,			/* init */
-  mbcharlen_sjis,
   my_numchars_mb,
   my_charpos_mb,
   my_well_formed_len_sjis,
diff --git a/strings/ctype-tis620.c b/strings/ctype-tis620.c
index 711bb21773e..82fd864b05d 100644
--- a/strings/ctype-tis620.c
+++ b/strings/ctype-tis620.c
@@ -834,7 +834,6 @@ static MY_COLLATION_HANDLER my_collation_ci_handler =
 static MY_CHARSET_HANDLER my_charset_handler=
 {
     NULL,		/* init */
-    my_mbcharlen_8bit,	/* mbcharlen */
     my_numchars_8bit,
     my_charpos_8bit,
     my_well_formed_len_8bit,
diff --git a/strings/ctype-ucs2.c b/strings/ctype-ucs2.c
index 74e474cc28c..06dab08a772 100644
--- a/strings/ctype-ucs2.c
+++ b/strings/ctype-ucs2.c
@@ -1049,6 +1049,9 @@ my_scan_mb2(CHARSET_INFO *cs __attribute__((unused)),
     {
     }
     return (size_t) (str - str0);
+  case MY_SEQ_NONSPACES:
+    DBUG_ASSERT(0); /* Not implemented */
+    /* pass through */
   default:
     return 0;
   }
@@ -1431,15 +1434,6 @@ my_charlen_utf16(CHARSET_INFO *cs, const uchar *str, const uchar *end)
 /* Defines my_well_formed_char_length_utf16 */
 
 
-static uint
-my_mbcharlen_utf16(CHARSET_INFO *cs  __attribute__((unused)),
-                   uint c __attribute__((unused)))
-{
-  DBUG_ASSERT(0);
-  return MY_UTF16_HIGH_HEAD(c) ? 4 : 2;
-}
-
-
 static size_t
 my_numchars_utf16(CHARSET_INFO *cs,
                   const char *b, const char *e)
@@ -1567,7 +1561,6 @@ static MY_COLLATION_HANDLER my_collation_utf16_bin_handler =
 MY_CHARSET_HANDLER my_charset_utf16_handler=
 {
   NULL,                /* init         */
-  my_mbcharlen_utf16,  /* mbcharlen    */
   my_numchars_utf16,
   my_charpos_utf16,
   my_well_formed_len_utf16,
@@ -1789,7 +1782,6 @@ static MY_COLLATION_HANDLER my_collation_utf16le_bin_handler =
 static MY_CHARSET_HANDLER my_charset_utf16le_handler=
 {
   NULL,                /* init         */
-  my_mbcharlen_utf16,
   my_numchars_utf16,
   my_charpos_utf16,
   my_well_formed_len_utf16,
@@ -2083,14 +2075,6 @@ my_charlen_utf32(CHARSET_INFO *cs __attribute__((unused)),
 /* Defines my_well_formed_char_length_utf32 */
 
 
-static uint
-my_mbcharlen_utf32(CHARSET_INFO *cs  __attribute__((unused)) , 
-                   uint c __attribute__((unused)))
-{
-  return 4;
-}
-
-
 static int
 my_vsnprintf_utf32(char *dst, size_t n, const char* fmt, va_list ap)
 {
@@ -2484,6 +2468,9 @@ my_scan_utf32(CHARSET_INFO *cs,
       str+= res;
     }
     return (size_t) (str - str0);
+  case MY_SEQ_NONSPACES:
+    DBUG_ASSERT(0); /* Not implemented */
+    /* pass through */
   default:
     return 0;
   }
@@ -2525,7 +2512,6 @@ static MY_COLLATION_HANDLER my_collation_utf32_bin_handler =
 MY_CHARSET_HANDLER my_charset_utf32_handler=
 {
   NULL, /* init */
-  my_mbcharlen_utf32,
   my_numchars_utf32,
   my_charpos_utf32,
   my_well_formed_len_utf32,
@@ -2862,13 +2848,6 @@ my_fill_ucs2(CHARSET_INFO *cs __attribute__((unused)),
 }
 
 
-static uint my_mbcharlen_ucs2(CHARSET_INFO *cs  __attribute__((unused)) , 
-                              uint c __attribute__((unused)))
-{
-  return 2;
-}
-
-
 static
 size_t my_numchars_ucs2(CHARSET_INFO *cs __attribute__((unused)),
                         const char *b, const char *e)
@@ -3003,7 +2982,6 @@ static MY_COLLATION_HANDLER my_collation_ucs2_bin_handler =
 MY_CHARSET_HANDLER my_charset_ucs2_handler=
 {
     NULL,		/* init */
-    my_mbcharlen_ucs2,	/* mbcharlen    */
     my_numchars_ucs2,
     my_charpos_ucs2,
     my_well_formed_len_ucs2,
diff --git a/strings/ctype-ujis.c b/strings/ctype-ujis.c
index b24fdb3075f..786ae991aef 100644
--- a/strings/ctype-ujis.c
+++ b/strings/ctype-ujis.c
@@ -220,12 +220,6 @@ static const uchar sort_order_ujis[]=
 #include "strcoll.ic"
 
 
-static uint mbcharlen_ujis(CHARSET_INFO *cs __attribute__((unused)),uint c)
-{
-  return (isujis(c)? 2: isujis_ss2(c)? 2: isujis_ss3(c)? 3: 1);
-}
-
-
 static
 size_t my_numcells_eucjp(CHARSET_INFO *cs __attribute__((unused)),
                        const char *str, const char *str_end)
@@ -67255,7 +67249,6 @@ static MY_COLLATION_HANDLER my_collation_ujis_bin_handler =
 static MY_CHARSET_HANDLER my_charset_handler=
 {
     NULL,		/* init */
-    mbcharlen_ujis,
     my_numchars_mb,
     my_charpos_mb,
     my_well_formed_len_ujis,
diff --git a/strings/ctype-utf8.c b/strings/ctype-utf8.c
index 3a5616b7323..b6a7a0d0347 100644
--- a/strings/ctype-utf8.c
+++ b/strings/ctype-utf8.c
@@ -5426,21 +5426,6 @@ my_weight_mb3_utf8_general_mysql500_ci(uchar b0, uchar b1, uchar b2)
 #include "strcoll.ic"
 
 
-static uint my_mbcharlen_utf8(CHARSET_INFO *cs  __attribute__((unused)),
-                              uint c)
-{
-  if (c < 0x80)
-    return 1;
-  else if (c < 0xc2)
-    return 0; /* Illegal mb head */
-  else if (c < 0xe0)
-    return 2;
-  else if (c < 0xf0)
-    return 3;
-  return 0; /* Illegal mb head */;
-}
-
-
 static MY_COLLATION_HANDLER my_collation_utf8_general_ci_handler =
 {
     NULL,               /* init */
@@ -5491,7 +5476,6 @@ static MY_COLLATION_HANDLER my_collation_utf8_bin_handler =
 MY_CHARSET_HANDLER my_charset_utf8_handler=
 {
     NULL,               /* init */
-    my_mbcharlen_utf8,
     my_numchars_mb,
     my_charpos_mb,
     my_well_formed_len_utf8,
@@ -7045,7 +7029,6 @@ static MY_COLLATION_HANDLER my_collation_filename_handler =
 static MY_CHARSET_HANDLER my_charset_filename_handler=
 {
     NULL,               /* init */
-    my_mbcharlen_utf8,
     my_numchars_mb,
     my_charpos_mb,
     my_well_formed_len_mb,
@@ -7111,57 +7094,6 @@ struct charset_info_st my_charset_filename=
 };
 
 
-#ifdef MY_TEST_UTF8
-#include <stdio.h>
-
-static void test_mb(CHARSET_INFO *cs, uchar *s)
-{
-  while(*s)
-  {
-    if (my_ismbhead_utf8(cs,*s))
-    {
-      uint len=my_mbcharlen_utf8(cs,*s);
-      while(len--)
-      {
-        printf("%c",*s);
-        s++;
-      }
-      printf("\n");
-    }
-    else
-    {
-      printf("%c\n",*s);
-      s++;
-    }
-  }
-}
-
-int main()
-{
-  char str[1024]=" utf8 test проба ПЕРА по-РУССКИ";
-  CHARSET_INFO *cs;
-
-  test_mb(cs,(uchar*)str);
-
-  printf("orig      :'%s'\n",str);
-
-  my_caseup_utf8(cs,str,15);
-  printf("caseup    :'%s'\n",str);
-
-  my_caseup_str_utf8(cs,str);
-  printf("caseup_str:'%s'\n",str);
-
-  my_casedn_utf8(cs,str,15);
-  printf("casedn    :'%s'\n",str);
-
-  my_casedn_str_utf8(cs,str);
-  printf("casedn_str:'%s'\n",str);
-
-  return 0;
-}
-
-#endif
-
 #endif /* HAVE_CHARSET_UTF8 */
 
 
@@ -7755,23 +7687,6 @@ size_t my_well_formed_len_utf8mb4(CHARSET_INFO *cs,
 #include "strcoll.ic"
 
 
-static uint
-my_mbcharlen_utf8mb4(CHARSET_INFO *cs  __attribute__((unused)), uint c)
-{
-  if (c < 0x80)
-    return 1;
-  if (c < 0xc2)
-    return 0; /* Illegal mb head */
-  if (c < 0xe0)
-    return 2;
-  if (c < 0xf0)
-    return 3;
-  if (c < 0xf8)
-    return 4;
-  return 0; /* Illegal mb head */;
-}
-
-
 static MY_COLLATION_HANDLER my_collation_utf8mb4_general_ci_handler=
 {
   NULL,               /* init */
@@ -7807,7 +7722,6 @@ static MY_COLLATION_HANDLER my_collation_utf8mb4_bin_handler =
 MY_CHARSET_HANDLER my_charset_utf8mb4_handler=
 {
   NULL,               /* init */
-  my_mbcharlen_utf8mb4,
   my_numchars_mb,
   my_charpos_mb,
   my_well_formed_len_utf8mb4,
diff --git a/strings/my_strchr.c b/strings/my_strchr.c
index 0305ef89542..2365731fa03 100644
--- a/strings/my_strchr.c
+++ b/strings/my_strchr.c
@@ -38,7 +38,7 @@
     const char *acc_end= (ACC) + (LEN);                                 \
     for (ptr_str= (STR) ; ptr_str < (END) ; ptr_str+= mbl)              \
     {                                                                   \
-      mbl= my_mbcharlen((CS), *(uchar*)ptr_str);                        \
+      mbl= my_charlen_fix((CS), ptr_str, (END));                        \
       if (mbl < 2)                                                      \
       {                                                                 \
         DBUG_ASSERT(mbl == 1);                                          \
@@ -63,10 +63,9 @@ end:                                                                    \
 char *my_strchr(CHARSET_INFO *cs, const char *str, const char *end,
                 pchar c)
 {
-  uint mbl;
   while (str < end)
   {
-    mbl= my_mbcharlen(cs, *(uchar *)str);
+    uint mbl= my_ismbchar(cs, str, end);
     if (mbl < 2)
     {
       if (*str == c)
diff --git a/strings/my_vsnprintf.c b/strings/my_vsnprintf.c
index 4178b20789d..75514a90925 100644
--- a/strings/my_vsnprintf.c
+++ b/strings/my_vsnprintf.c
@@ -168,8 +168,7 @@ static char *backtick_string(CHARSET_INFO *cs, char *to, const char *end,
   for ( ; par < par_end; par+= char_len)
   {
     uchar c= *(uchar *) par;
-    if (!(char_len= my_mbcharlen(cs, c)))
-      char_len= 1;
+    char_len= my_charlen_fix(cs, par, par_end);
     if (char_len == 1 && c == (uchar) quote_char )
     {
       if (start + 1 >= end)