database/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2026-10-23 18:49:29 +03:00
Code Activity

Add detection of partial matches for strnncoll

Browse Source 
This is achived by changing the 'is_prefix' parameter to strncoll
from a my_bool to my_bool*. This parameter is null if not specified by
the caller. If the is_prefix parameter is set to point to a my_bool
variable, this variable will be set to 1 if the second argument is
a prefix of the first.

This is needed by TO_DATE to be able to detect partial uniqiue matches of
months etc.  find_type() did this for latin1 strings, but we did not
do this for utf8mb4 strings.
This patch fixes this incompatibility.
This commit is contained in:
Monty
2026-01-13 19:47:07 +02:00
committed by Sergei Golubchik
parent bd74c8cf7c
commit 782b103925
16 changed files with 159 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
include/m_ctype.h
View File

@@ -557,7 +557,7 @@ struct my_collation_handler_st
my_bool (*init)(struct charset_info_st *, MY_CHARSET_LOADER *);
/* Collation routines */
int (*strnncoll)(CHARSET_INFO *,
const uchar *, size_t, const uchar *, size_t, my_bool);
const uchar *, size_t, const uchar *, size_t, my_bool*);
int (*strnncollsp)(CHARSET_INFO *,
const uchar *, size_t, const uchar *, size_t);
/*
@@ -716,6 +716,7 @@ struct my_charset_handler_st
size_t (*numchars)(CHARSET_INFO *, const char *b, const char *e);
size_t (*charpos)(CHARSET_INFO *, const char *b, const char *e,
size_t pos);
/* Length without trailing space */
size_t (*lengthsp)(CHARSET_INFO *, const char *ptr, size_t length);
size_t (*numcells)(CHARSET_INFO *, const char *b, const char *e);
@@ -1146,22 +1147,23 @@ struct charset_info_st
}
int strnncoll(const LEX_CSTRING a, const LEX_CSTRING b,
my_bool b_is_prefix= FALSE) const
my_bool *b_is_prefix= 0) const
{
DBUG_ASSERT(is_valid_string(a));
DBUG_ASSERT(is_valid_string(b));
return (coll->strnncoll)(this,
(const uchar *) a.str, a.length,
(const uchar *) b.str, b.length, b_is_prefix);
(const uchar *) b.str, b.length,
b_is_prefix);
}
int strnncoll(const uchar *a, size_t alen,
const uchar *b, size_t blen, my_bool b_is_prefix= FALSE) const
const uchar *b, size_t blen, my_bool *b_is_prefix= 0) const
{
return (coll->strnncoll)(this, a, alen, b, blen, b_is_prefix);
}
int strnncoll(const char *a, size_t alen,
const char *b, size_t blen, my_bool b_is_prefix= FALSE) const
const char *b, size_t blen, my_bool *b_is_prefix= 0) const
{
return (coll->strnncoll)(this,
(const uchar *) a, alen,
@@ -1478,7 +1480,7 @@ static inline int
my_ci_strnncoll(CHARSET_INFO *ci,
const uchar *a, size_t alen,
const uchar *b, size_t blen,
my_bool b_is_prefix)
my_bool *b_is_prefix)
{
return (ci->coll->strnncoll)(ci, a, alen, b, blen, b_is_prefix);
}
@@ -1635,7 +1637,7 @@ const uint16 *my_cs_contraction2_weight(CHARSET_INFO *cs, my_wc_t wc1,
/* declarations for simple charsets */
extern int my_strnncoll_simple(CHARSET_INFO *, const uchar *, size_t,
const uchar *, size_t, my_bool);
const uchar *, size_t, my_bool*);
extern int my_strnncollsp_simple(CHARSET_INFO *, const uchar *, size_t,
const uchar *, size_t);

7
include/my_compare.h
View File

@@ -140,7 +140,7 @@ static inline int ha_compare_char_varying(CHARSET_INFO *charset_info,
b, b_length);
return charset_info->coll->strnncoll(charset_info,
a, a_length,
b, b_length, TRUE/*prefix*/);
b, b_length, &b_is_prefix);
}
@@ -190,7 +190,7 @@ static inline int ha_compare_char_fixed(CHARSET_INFO *charset_info,
MY_STRNNCOLLSP_NCHARS_EMULATE_TRIMMED_TRAILING_SPACES);
return charset_info->coll->strnncoll(charset_info,
a, a_length,
b, b_length, TRUE/*prefix*/);
b, b_length, &b_is_prefix);
}
@@ -218,10 +218,11 @@ static inline int ha_compare_word_prefix(CHARSET_INFO *charset_info,
const uchar *a, size_t a_length,
const uchar *b, size_t b_length)
{
my_bool b_is_prefix;
return charset_info->coll->strnncoll(charset_info,
a, a_length,
b, b_length,
TRUE/*b_is_prefix*/);
&b_is_prefix);
}

5
mysql-test/main/ctype_utf8mb4_uca1400_ai_ci.result
View File

@@ -618,5 +618,10 @@ Warnings:
Note 1003 select `test`.`t1`.`c1` AS `c1` from `test`.`t1` where concat(`test`.`t1`.`c1`) = 'ä' and concat(`test`.`t1`.`c1`) like 'ae'
DROP TABLE IF EXISTS t1;
#
# MDEV-38608 TO_DATE: Server crash in my_uca_strnncoll_multilevel_generic
#
CREATE TABLE t1 (f ENUM('foo','bar')) COLLATE uca1400_ai_cs;
DROP TABLE t1;
#
# End of 10.0 tests
#

7
mysql-test/main/ctype_utf8mb4_uca1400_ai_ci.test
View File

@@ -47,6 +47,13 @@ SET NAMES utf8mb4 COLLATE uca1400_ai_ci;
--source include/ctype_like_cond_propagation.inc
--source include/ctype_like_cond_propagation_utf8_german.inc
--echo #
--echo # MDEV-38608 TO_DATE: Server crash in my_uca_strnncoll_multilevel_generic
--echo #
CREATE TABLE t1 (f ENUM('foo','bar')) COLLATE uca1400_ai_cs;
DROP TABLE t1;
--echo #
--echo # End of 10.0 tests
--echo #

3
sql/mysqld.cc
View File

@@ -9441,10 +9441,11 @@ bool is_secure_file_path(char *path)
}
else
{
my_bool use_prefix;
if (files_charset_info->strnncoll(buff2, strlen(buff2),
opt_secure_file_priv,
opt_secure_file_priv_len,
TRUE))
&use_prefix))
return FALSE;
}
return TRUE;

3
storage/innobase/handler/ha_innodb.cc
View File

@@ -2383,6 +2383,7 @@ static bool is_mysql_datadir_path(const char *path)
convert_dirname(mysql_data_dir, mysql_unpacked_real_data_home, NullS);
size_t mysql_data_home_len= dirname_length(mysql_data_dir);
size_t path_len = dirname_length(path_dir);
my_bool part_matched;
if (path_len < mysql_data_home_len)
return true;
@@ -2393,7 +2394,7 @@ static bool is_mysql_datadir_path(const char *path)
return(files_charset_info->strnncoll((uchar *) path_dir, path_len,
(uchar *) mysql_data_dir,
mysql_data_home_len,
TRUE));
&part_matched));
}
/*********************************************************************//**

3
storage/innobase/include/rem0cmp.h
View File

@@ -258,9 +258,10 @@ inline bool cmp_dfield_dfield_eq_prefix(const dfield_t *dfield1,
uint cs_num= dtype_get_charset_coll(type->prtype);
CHARSET_INFO *cs= get_charset(cs_num, MYF(MY_WME));
my_bool use_prefix;
ut_a(cs);
return !cs->strnncoll(static_cast<const uchar*>(dfield_get_data(dfield1)),
dfield_get_len(dfield1),
static_cast<const uchar*>(dfield_get_data(dfield2)),
dfield_get_len(dfield2), 1);
dfield_get_len(dfield2), &use_prefix);
}

5
strings/CHARSET_INFO.txt
View File

@@ -259,6 +259,11 @@ Simple scanner routines
-----------------------
scan() - to skip leading spaces in the given string.
Used when a string value is inserted into a numeric field.
the last parameter is one of:
MY_SEQ_INTTAIL ; Ignore '.' and a set of '0'
MY_SEQ_SPACES ; Skip over pre-space
MY_SEQ_NOSPACES; ; Skip over non space characters

38
strings/ctype-bin.c
View File

@@ -83,11 +83,23 @@ my_coll_init_8bit_bin(struct charset_info_st *cs,
static int my_strnncoll_binary(CHARSET_INFO * cs __attribute__((unused)),
const uchar *s, size_t slen,
const uchar *t, size_t tlen,
my_bool t_is_prefix)
my_bool *t_is_prefix)
{
size_t len=MY_MIN(slen,tlen);
size_t len= MY_MIN(slen,tlen);
int cmp= len ? memcmp(s, t, len) : 0;
return cmp ? cmp : (int)((t_is_prefix ? len : slen) - tlen);
if (!t_is_prefix)
{
if (cmp)
return cmp;
return (int) (slen - tlen);
}
if (cmp)
{
*t_is_prefix= 0;
return cmp;
}
return !(*t_is_prefix= tlen <= slen);
}
@@ -144,11 +156,23 @@ static int my_strnncollsp_nchars_binary(CHARSET_INFO * cs __attribute__((unused)
static int my_strnncoll_8bit_bin(CHARSET_INFO * cs __attribute__((unused)),
const uchar *s, size_t slen,
const uchar *t, size_t tlen,
my_bool t_is_prefix)
my_bool *t_is_prefix)
{
size_t len=MY_MIN(slen,tlen);
size_t len= MY_MIN(slen,tlen);
int cmp= len ? memcmp(s, t, len) : 0;
return cmp ? cmp : (int)((t_is_prefix ? len : slen) - tlen);
if (!t_is_prefix)
{
if (cmp)
return cmp;
return (int) (slen - tlen);
}
if (cmp)
{
*t_is_prefix= 0;
return cmp;
}
return !(*t_is_prefix= tlen <= slen);
}
@@ -232,7 +256,7 @@ static int my_strnncollsp_8bit_nopad_bin(CHARSET_INFO * cs
const uchar *a, size_t a_length,
const uchar *b, size_t b_length)
{
return my_strnncoll_8bit_bin(cs, a, a_length, b, b_length, FALSE);
return my_strnncoll_8bit_bin(cs, a, a_length, b, b_length, 0);
}

18
strings/ctype-czech.c
View File

@@ -218,14 +218,20 @@ while (1) \
static int my_strnncoll_czech(CHARSET_INFO *cs __attribute__((unused)),
const uchar *s1, size_t len1,
const uchar *s2, size_t len2,
my_bool s2_is_prefix)
my_bool *s2_is_prefix)
{
int v1, v2;
const uchar *p1, * p2, * store1, * store2;
int pass1 = 0, pass2 = 0;
if (s2_is_prefix && len1 > len2)
len1=len2;
if (s2_is_prefix)
{
*s2_is_prefix= 0;
if (len1 > len2)
{
*s2_is_prefix= 1;
len1= len2;
}
}
p1 = s1; p2 = s2;
store1 = s1; store2 = s2;
@@ -236,7 +242,11 @@ static int my_strnncoll_czech(CHARSET_INFO *cs __attribute__((unused)),
NEXT_CMP_VALUE(s1, p1, store1, pass1, v1, (int)len1);
NEXT_CMP_VALUE(s2, p2, store2, pass2, v2, (int)len2);
if ((diff = v1 - v2))
{
if (s2_is_prefix)
*s2_is_prefix= 0;
return diff;
}
}
while (v1);
return 0;

6
strings/ctype-latin1.c
View File

@@ -589,11 +589,13 @@ static const uchar combo2map[]={
static int my_strnncoll_latin1_de(CHARSET_INFO *cs __attribute__((unused)),
const uchar *a, size_t a_length,
const uchar *b, size_t b_length,
my_bool b_is_prefix)
my_bool *b_is_prefix)
{
const uchar *a_end= a + a_length;
const uchar *b_end= b + b_length;
uchar a_char, a_extend= 0, b_char, b_extend= 0;
if (b_is_prefix)
*b_is_prefix= 0;
while ((a < a_end || a_extend) && (b < b_end || b_extend))
{
@@ -622,7 +624,7 @@ static int my_strnncoll_latin1_de(CHARSET_INFO *cs __attribute__((unused)),
A simple test of string lengths won't work -- we test to see
which string ran out first
*/
return ((a < a_end || a_extend) ? (b_is_prefix ? 0 : 1) :
return ((a < a_end || a_extend) ? (b_is_prefix ? (*b_is_prefix= 1, 0) : 1) :
(b < b_end || b_extend) ? -1 : 0);
}

26
strings/ctype-simple.c
View File

@@ -142,22 +142,40 @@ my_strnxfrm_simple_nopad(CHARSET_INFO * cs,
int my_strnncoll_simple(CHARSET_INFO * cs, const uchar *s, size_t slen,
const uchar *t, size_t tlen,
my_bool t_is_prefix)
my_bool *t_is_prefix)
{
size_t len = ( slen > tlen ) ? tlen : slen;
const uchar *map= cs->sort_order;
if (t_is_prefix && slen > tlen)
slen=tlen;
my_bool dummy;
if (t_is_prefix)
{
*t_is_prefix= 0;
if (slen > tlen)
{
*t_is_prefix= 1;
slen= tlen;
}
}
else
t_is_prefix= &dummy;
while (len--)
{
if (map[*s++] != map[*t++])
{
*t_is_prefix= 0;
return ((int) map[s[-1]] - (int) map[t[-1]]);
}
}
/*
We can't use (slen - tlen) here as the result may be outside of the
precision of a signed int
*/
return slen > tlen ? 1 : slen < tlen ? -1 : 0 ;
if (tlen == slen)
return 0;
*t_is_prefix= 0;
return slen > tlen ? 1 : -1;
}

15
strings/ctype-tis620.c
View File

@@ -516,14 +516,21 @@ static
int my_strnncoll_tis620(CHARSET_INFO *cs __attribute__((unused)),
const uchar *s1, size_t len1,
const uchar *s2, size_t len2,
my_bool s2_is_prefix)
my_bool *s2_is_prefix)
{
uchar buf[80] ;
uchar *tc1, *tc2;
int i;
if (s2_is_prefix && len1 > len2)
len1= len2;
if (s2_is_prefix)
{
*s2_is_prefix= 0;
if (len1 > len2)
{
*s2_is_prefix= 1;
len1= len2;
}
}
tc1= buf;
if ((len1 + len2 +2) > (int) sizeof(buf))
@@ -538,6 +545,8 @@ int my_strnncoll_tis620(CHARSET_INFO *cs __attribute__((unused)),
i= strcmp((char*)tc1, (char*)tc2);
if (tc1 != buf)
my_free(tc1);
if (i && s2_is_prefix)
*s2_is_prefix= 0;
return i;
}

18
strings/ctype-uca.inl
View File

@@ -89,7 +89,7 @@ MY_FUNCTION_NAME(strnncoll_onelevel)(CHARSET_INFO *cs,
const MY_UCA_WEIGHT_LEVEL *level,
const uchar *s, size_t slen,
const uchar *t, size_t tlen,
my_bool t_is_prefix)
my_bool *t_is_prefix)
{
my_uca_scanner sscanner;
my_uca_scanner tscanner;
@@ -119,7 +119,12 @@ MY_FUNCTION_NAME(strnncoll_onelevel)(CHARSET_INFO *cs,
t_res= MY_FUNCTION_NAME(scanner_next)(&tscanner, &param);
} while ( s_res == t_res && s_res >0);
return (t_is_prefix && t_res < 0) ? 0 : (s_res - t_res);
if (t_is_prefix)
{
if ((*t_is_prefix= s_res != t_res && t_res < 0))
return 0;
}
return (s_res - t_res);
}
@@ -130,7 +135,7 @@ static int
MY_FUNCTION_NAME(strnncoll)(CHARSET_INFO *cs,
const uchar *s, size_t slen,
const uchar *t, size_t tlen,
my_bool t_is_prefix)
my_bool *t_is_prefix)
{
return MY_FUNCTION_NAME(strnncoll_onelevel)(cs, &cs->uca->level[0],
s, slen, t, tlen, t_is_prefix);
@@ -139,14 +144,19 @@ MY_FUNCTION_NAME(strnncoll)(CHARSET_INFO *cs,
/*
Multi-level, PAD SPACE.
Note that t_is_prefix is set from the last call to strnncoll_onelevel
*/
static int
MY_FUNCTION_NAME(strnncoll_multilevel)(CHARSET_INFO *cs,
const uchar *s, size_t slen,
const uchar *t, size_t tlen,
my_bool t_is_prefix)
my_bool *t_is_prefix)
{
uint i, level_flags= cs->levels_for_order;
if (t_is_prefix && *t_is_prefix)
*t_is_prefix= 0;
for (i= 0; level_flags; i++, level_flags>>= 1)
{
int ret;

17
strings/ctype-win1250ch.c
View File

@@ -452,15 +452,22 @@ static const struct wordvalue doubles[] = {
static int my_strnncoll_win1250ch(CHARSET_INFO *cs __attribute__((unused)),
const uchar *s1, size_t len1,
const uchar *s2, size_t len2,
my_bool s2_is_prefix)
my_bool *s2_is_prefix)
{
int v1, v2;
const uchar *p1, * p2;
int pass1 = 0, pass2 = 0;
int diff;
if (s2_is_prefix && len1 > len2)
len1=len2;
if (s2_is_prefix)
{
*s2_is_prefix= 0;
if (len1 > len2)
{
*s2_is_prefix= 1;
len1= len2;
}
}
p1 = s1; p2 = s2;
@@ -469,7 +476,11 @@ static int my_strnncoll_win1250ch(CHARSET_INFO *cs __attribute__((unused)),
NEXT_CMP_VALUE(s1, p1, pass1, v1, (int)len1);
NEXT_CMP_VALUE(s2, p2, pass2, v2, (int)len2);
if ((diff = v1 - v2))
{
if (s2_is_prefix)
*s2_is_prefix= 0;
return diff;
}
} while (v1);
return 0;
}

14
strings/strcoll.inl
View File

@@ -218,16 +218,19 @@ bad:
@param b - the right string
@param b_length - the length of the right string
@param b_is_prefix - if the caller wants to check if "b" is a prefix of "a"
*b_is_prefix is set to 1 if it was a prefix match
@return - the comparison result
*/
static int
MY_FUNCTION_NAME(strnncoll)(CHARSET_INFO *cs __attribute__((unused)),
const uchar *a, size_t a_length,
const uchar *b, size_t b_length,
my_bool b_is_prefix)
my_bool *b_is_prefix)
{
const uchar *a_end= a + a_length;
const uchar *b_end= b + b_length;
if (b_is_prefix)
*b_is_prefix= 0;
for ( ; ; )
{
int a_weight, b_weight, res;
@@ -271,7 +274,14 @@ MY_FUNCTION_NAME(strnncoll)(CHARSET_INFO *cs __attribute__((unused)),
return b_wlen ? -1 : 0;
if (!b_wlen)
return b_is_prefix ? 0 : +1;
{
if (b_is_prefix)
{
*b_is_prefix= 1; /* Prefix match */
return 0;
}
return +1;
}
if ((res= (a_weight - b_weight)))
return res;