diff options
Diffstat (limited to 'strings/ctype-gb2312.c')
| -rw-r--r-- | strings/ctype-gb2312.c | 37 |
1 files changed, 36 insertions, 1 deletions
diff --git a/strings/ctype-gb2312.c b/strings/ctype-gb2312.c index 0cbad2d1c55..52dd61a8462 100644 --- a/strings/ctype-gb2312.c +++ b/strings/ctype-gb2312.c @@ -5686,6 +5686,41 @@ my_mb_wc_gb2312(CHARSET_INFO *cs __attribute__((unused)), } +/* + Returns well formed length of a EUC-KR string. +*/ +static uint +my_well_formed_len_gb2312(CHARSET_INFO *cs __attribute__((unused)), + const char *b, const char *e, + uint pos, int *error) +{ + const char *b0= b; + const char *emb= e - 1; /* Last possible end of an MB character */ + + *error= 0; + while (pos-- && b < e) + { + if ((uchar) b[0] < 128) + { + /* Single byte ascii character */ + b++; + } + else if (b < emb && isgb2312head(*b) && isgb2312tail(b[1])) + { + /* Double byte character */ + b+= 2; + } + else + { + /* Wrong byte sequence */ + *error= 1; + break; + } + } + return (uint) (b - b0); +} + + static MY_COLLATION_HANDLER my_collation_ci_handler = { NULL, /* init */ @@ -5706,7 +5741,7 @@ static MY_CHARSET_HANDLER my_charset_handler= mbcharlen_gb2312, my_numchars_mb, my_charpos_mb, - my_well_formed_len_mb, + my_well_formed_len_gb2312, my_lengthsp_8bit, my_numcells_8bit, my_mb_wc_gb2312, /* mb_wc */ |