mirror of
https://github.com/MariaDB/server.git
synced 2025-07-29 05:21:33 +03:00
Fixed problem with range optimization over overlapping ranges (#2448)
This commit is contained in:
monty@mysql.com
@ -449,225 +449,160 @@ uchar NEAR sort_order_tis620[]=
|
||||
(uchar) '\370',(uchar) '\371',(uchar) '\372',(uchar) '\373',(uchar) '\374',(uchar) '\375',(uchar) '\376',(uchar) '\377',
|
||||
};
|
||||
|
||||
/* Convert thai string to "Standard C String Function" sortable string
|
||||
Arg: const source string and length of converted string
|
||||
Ret: Sortable string
|
||||
|
||||
/*
|
||||
Convert thai string to "Standard C String Function" sortable string
|
||||
Arg: const source string and length of converted string
|
||||
Ret: Sortable string
|
||||
*/
|
||||
static void _thai2sortable(uchar *tstr)
|
||||
|
||||
static void _thai2sortable(uchar *tstr)
|
||||
{
|
||||
uchar *p ;
|
||||
int len,tlen ;
|
||||
uchar c,l2bias ;
|
||||
uchar *p;
|
||||
int len, tlen;
|
||||
uchar l2bias;
|
||||
|
||||
tlen= len = strlen (tstr) ;
|
||||
l2bias = 256 - 8 ;
|
||||
for (p=tstr; tlen > 0; p++,tlen--)
|
||||
{
|
||||
c = *p ;
|
||||
|
||||
if (isthai(c))
|
||||
{
|
||||
int *t_ctype0 = t_ctype[c] ;
|
||||
|
||||
if (isconsnt(c))
|
||||
l2bias -= 8 ;
|
||||
if (isldvowel(c) && isconsnt(p[1]))
|
||||
{
|
||||
/*
|
||||
simply swap between leading-vowel and consonant
|
||||
*/
|
||||
*p = p[1];
|
||||
p[1]= c ;
|
||||
tlen-- ;
|
||||
p++;
|
||||
continue ;
|
||||
}
|
||||
|
||||
// if found level 2 char (L2_GARAN,L2_TONE*,L2_TYKHU) move to last
|
||||
if (t_ctype0[1]>= L2_GARAN)
|
||||
{
|
||||
// l2bias use to control position weight of l2char
|
||||
// example (*=l2char) XX*X must come before X*XX
|
||||
strcpy (p,p+1) ;
|
||||
tstr[len-1] = l2bias + t_ctype0[1]- L2_GARAN +1 ;
|
||||
p-- ;
|
||||
continue ;
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
l2bias -= 8 ;
|
||||
*p = to_lower_tis620[c];
|
||||
}
|
||||
/*
|
||||
this routine skip non-printable char
|
||||
but not necessary, leave it like raw ascii 8 bits
|
||||
*/
|
||||
/*
|
||||
t_ctype0 = t_ctype[p[0]];
|
||||
if ((t_ctype0[0]|t_ctype0[1]|t_ctype0[2])==IGNORE)
|
||||
{
|
||||
strcpy(p,p+1);
|
||||
p-- ;
|
||||
}
|
||||
*/
|
||||
}
|
||||
}
|
||||
|
||||
/*
|
||||
NOTE: isn't it faster to alloc buffer in calling function?
|
||||
*/
|
||||
/*
|
||||
Sathit's NOTE: we don't use this function anymore
|
||||
static uchar* thai2sortable(const uchar * tstr,int len)
|
||||
{
|
||||
*/
|
||||
/* We use only 3 levels (neglect capitalization). */
|
||||
/*
|
||||
const uchar* p= tstr;
|
||||
uchar *outBuf;
|
||||
uchar *pRight1, *pRight2, *pRight3;
|
||||
uchar *pLeft1, *pLeft2, *pLeft3;
|
||||
uint bufSize;
|
||||
uint RightSize;
|
||||
|
||||
len= (int) strnlen((char*) tstr,len);
|
||||
bufSize= (uint) buffsize((char*) tstr);
|
||||
RightSize= sizeof(uchar) * (len + 1);
|
||||
if (!(outBuf= pLeft1= pRight1=
|
||||
(uchar *)malloc(sizeof(uchar) * bufSize + RightSize*2)))
|
||||
return (uchar*) tstr;
|
||||
pLeft2= pRight2= pRight1 + sizeof(uchar) * bufSize;
|
||||
pLeft3= pRight3= pRight2 + RightSize;
|
||||
|
||||
while (--len > 0)
|
||||
tlen= len= strlen (tstr);
|
||||
l2bias= 256 - 8;
|
||||
for (p= tstr; tlen > 0; p++, tlen--)
|
||||
{
|
||||
int *t_ctype0= t_ctype[p[0]];
|
||||
if (isldvowel(*p) && isconsnt(p[1]))
|
||||
uchar c= *p;
|
||||
|
||||
if (isthai(c))
|
||||
{
|
||||
int *t_ctype1= t_ctype[p[1]];
|
||||
*pRight1++= t_ctype1[0];
|
||||
*pRight2++= t_ctype1[1];
|
||||
*pRight3++= t_ctype1[2];
|
||||
*pRight1++= t_ctype0[0];
|
||||
*pRight2++= t_ctype0[1];
|
||||
*pRight3++= t_ctype0[2];
|
||||
p+= 2;
|
||||
len--;
|
||||
int *t_ctype0= t_ctype[c];
|
||||
|
||||
if (isconsnt(c))
|
||||
l2bias -= 8;
|
||||
if (isldvowel(c) && isconsnt(p[1]))
|
||||
{
|
||||
/* simply swap between leading-vowel and consonant */
|
||||
*p= p[1];
|
||||
p[1]= c;
|
||||
tlen--;
|
||||
p++;
|
||||
continue;
|
||||
}
|
||||
|
||||
/* if found level 2 char (L2_GARAN,L2_TONE*,L2_TYKHU) move to last */
|
||||
if (t_ctype0[1] >= L2_GARAN)
|
||||
{
|
||||
/*
|
||||
l2bias use to control position weight of l2char
|
||||
example (*=l2char) XX*X must come before X*XX
|
||||
*/
|
||||
strmov(p,p+1);
|
||||
tstr[len-1]= l2bias + t_ctype0[1]- L2_GARAN +1;
|
||||
p--;
|
||||
continue;
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
*pRight1= t_ctype0[0];
|
||||
if(*pRight1 != IGNORE)
|
||||
pRight1++;
|
||||
*pRight2= t_ctype0[1];
|
||||
if (*pRight2 != IGNORE)
|
||||
pRight2++;
|
||||
*pRight3= t_ctype0[2];
|
||||
if(*pRight3 != IGNORE)
|
||||
pRight3++;
|
||||
p++;
|
||||
l2bias-= 8;
|
||||
*p= to_lower_tis620[c];
|
||||
}
|
||||
}
|
||||
if (!len)
|
||||
{
|
||||
int *t_ctype0= t_ctype[p[0]];
|
||||
*pRight1= t_ctype0[0];
|
||||
if (*pRight1 != IGNORE)
|
||||
pRight1++;
|
||||
*pRight2= t_ctype0[1];
|
||||
if (*pRight2 != IGNORE)
|
||||
pRight2++;
|
||||
*pRight3= t_ctype0[2];
|
||||
if (*pRight3 != IGNORE)
|
||||
pRight3++;
|
||||
}
|
||||
*pRight1++= L2_BLANK;
|
||||
*pRight2++= L3_BLANK;
|
||||
*pRight3++= '\0';
|
||||
memcpy(pRight1, pLeft2, pRight2 - pLeft2);
|
||||
pRight1+= pRight2 - pLeft2;
|
||||
memcpy(pRight1, pLeft3, pRight3 - pLeft3);
|
||||
return outBuf;
|
||||
}
|
||||
|
||||
|
||||
/*
|
||||
strncoll() replacement, compare 2 string, both are converted to sortable
|
||||
string
|
||||
|
||||
Arg: 2 Strings and it compare length
|
||||
Ret: strcmp result
|
||||
*/
|
||||
|
||||
/* strncoll() replacement, compare 2 string, both are conveted to sortable string
|
||||
Arg: 2 Strings and it compare length
|
||||
Ret: strcmp result
|
||||
*/
|
||||
int my_strnncoll_tis620(const uchar * s1, int len1, const uchar * s2, int len2)
|
||||
{
|
||||
uchar buf[80] ;
|
||||
uchar *tc1, *tc2;
|
||||
int i;
|
||||
uchar buf[80] ;
|
||||
uchar *tc1, *tc2;
|
||||
int i;
|
||||
|
||||
len1= (int) strnlen((char*) s1,len1);
|
||||
len2= (int) strnlen((char*) s2,len2);
|
||||
if ((len1 + len2 +2) > (int) sizeof(buf))
|
||||
tc1 = (uchar *)malloc(len1+len2) ;
|
||||
else
|
||||
tc1 = buf ;
|
||||
tc2 = tc1 + len1+1 ;
|
||||
strncpy((char *)tc1,(char *)s1,len1) ;
|
||||
tc1[len1] = 0; // if s1's length > len1, need to put 'end of string'
|
||||
strncpy((char *)tc2,(char *)s2,len2) ;
|
||||
tc2[len2] = 0; // put end of string
|
||||
_thai2sortable(tc1);
|
||||
_thai2sortable(tc2);
|
||||
i= strcmp((char*)tc1, (char*)tc2);
|
||||
if (tc1 != buf )
|
||||
free(tc1);
|
||||
return i;
|
||||
len1= (int) strnlen((char*) s1,len1);
|
||||
len2= (int) strnlen((char*) s2,len2);
|
||||
tc1= buf;
|
||||
if ((len1 + len2 +2) > (int) sizeof(buf))
|
||||
tc1= (uchar*) malloc(len1+len2);
|
||||
tc2= tc1 + len1+1;
|
||||
memcpy((char*) tc1, (char*) s1, len1);
|
||||
tc1[len1]= 0; /* if length(s1)> len1, need to put 'end of string' */
|
||||
memcpy((char *)tc2, (char *)s2, len2);
|
||||
tc2[len2]= 0; /* put end of string */
|
||||
_thai2sortable(tc1);
|
||||
_thai2sortable(tc2);
|
||||
i= strcmp((char*)tc1, (char*)tc2);
|
||||
if (tc1 != buf)
|
||||
free(tc1);
|
||||
return i;
|
||||
}
|
||||
|
||||
/* strnxfrm replacment, convert Thai string to sortable string
|
||||
Arg: Destination buffer, source string, dest length and source length
|
||||
Ret: Conveted string size
|
||||
|
||||
/*
|
||||
strnxfrm replacment, convert Thai string to sortable string
|
||||
|
||||
Arg: Destination buffer, source string, dest length and source length
|
||||
Ret: Conveted string size
|
||||
*/
|
||||
|
||||
int my_strnxfrm_tis620(uchar * dest, const uchar * src, int len, int srclen)
|
||||
{
|
||||
if (len > srclen)
|
||||
len = srclen ;
|
||||
strncpy (dest,src,len) ;
|
||||
dest[len] = 0; // if src's length > len, need to put 'end of string'
|
||||
_thai2sortable(dest);
|
||||
return strlen(dest);
|
||||
if (len > srclen)
|
||||
len= srclen ;
|
||||
strnmov(dest, src, len) ;
|
||||
dest[len]= 0; /* if length(src) > len, need to put 'end of string' */
|
||||
_thai2sortable(dest);
|
||||
return strlen(dest);
|
||||
}
|
||||
|
||||
/* strcoll replacment, compare 2 strings
|
||||
Arg: 2 strings
|
||||
Ret: strcmp result
|
||||
|
||||
/*
|
||||
strcoll replacment, compare 2 strings
|
||||
Arg: 2 strings
|
||||
Ret: strcmp result
|
||||
*/
|
||||
|
||||
int my_strcoll_tis620(const uchar * s1, const uchar * s2)
|
||||
{
|
||||
return my_strnncoll_tis620(s1, strlen((char *)s1),s2,strlen((char *)s2));
|
||||
return my_strnncoll_tis620(s1, strlen((char *)s1),s2,strlen((char *)s2));
|
||||
}
|
||||
|
||||
/* strxfrm replacment, convert Thai string to sortable string
|
||||
Arg: Destination buffer, String and dest buffer size
|
||||
Ret: Converting string size
|
||||
|
||||
/*
|
||||
strxfrm replacment, convert Thai string to sortable string
|
||||
|
||||
Arg: Destination buffer, String and dest buffer size
|
||||
Ret: Converting string size
|
||||
*/
|
||||
|
||||
int my_strxfrm_tis620(uchar * dest, const uchar * src, int len)
|
||||
{
|
||||
return my_strnxfrm_tis620(dest,src,len,strlen((char *)src));
|
||||
}
|
||||
|
||||
/* Convert SQL like string to C string
|
||||
Arg: String, its length, escape character, resource length, minimal string and maximum string
|
||||
Ret: Alway 0
|
||||
|
||||
/*
|
||||
Convert SQL LIKE string to C string
|
||||
|
||||
Arg: String, its length, escape character, resource length,
|
||||
minimal string and maximum string
|
||||
Ret: Always 0
|
||||
*/
|
||||
|
||||
/*
|
||||
We just copy this function from opt_range.cc. No need to convert to
|
||||
thai2sortable string. min_str and max_str will be use for comparison and
|
||||
converted there.
|
||||
*/
|
||||
|
||||
/* We just copy this function from opt_range.cc. No need to convert to
|
||||
thai2sortable string. min_str and max_str will be use for comparison and
|
||||
converted there. */
|
||||
#define max_sort_chr ((char) 255)
|
||||
#define wild_one '_'
|
||||
#define wild_many '%'
|
||||
|
||||
my_bool my_like_range_tis620(const char *ptr, uint ptr_length, pchar escape,
|
||||
uint res_length, char *min_str, char *max_str,
|
||||
uint *min_length, uint *max_length)
|
||||
uint res_length, char *min_str, char *max_str,
|
||||
uint *min_length, uint *max_length)
|
||||
{
|
||||
const char *end=ptr+ptr_length;
|
||||
char *min_org=min_str;
|
||||
@ -692,7 +627,7 @@ my_bool my_like_range_tis620(const char *ptr, uint ptr_length, pchar escape,
|
||||
*min_length= (uint) (min_str - min_org);
|
||||
*max_length=res_length;
|
||||
do {
|
||||
*min_str++ = ' '; /* Because if key compression */
|
||||
*min_str++ = ' '; /* Because of key compression */
|
||||
*max_str++ = max_sort_chr;
|
||||
} while (min_str != min_end);
|
||||
return 0;
|
||||
@ -702,18 +637,21 @@ my_bool my_like_range_tis620(const char *ptr, uint ptr_length, pchar escape,
|
||||
*min_length= *max_length = (uint) (min_str - min_org);
|
||||
|
||||
while (min_str != min_end)
|
||||
*min_str++ = *max_str++ = ' '; /* Because if key compression */
|
||||
*min_str++= *max_str++ = ' '; /* Because of key compression */
|
||||
return 0;
|
||||
}
|
||||
|
||||
/* Thai normalization for input sub system
|
||||
Arg: Buffer, 's length, String, 'length
|
||||
Ret: Void
|
||||
|
||||
/*
|
||||
Thai normalization for input sub system
|
||||
|
||||
Arg: Buffer, 's length, String, 'length
|
||||
*/
|
||||
|
||||
void ThNormalize(uchar* ptr, uint field_length, const uchar* from, uint length)
|
||||
{
|
||||
const uchar* fr= from;
|
||||
uchar* p= ptr;
|
||||
const uchar *fr= from;
|
||||
uchar *p= ptr;
|
||||
uint i;
|
||||
|
||||
if (length > field_length)
|
||||
@ -721,24 +659,21 @@ void ThNormalize(uchar* ptr, uint field_length, const uchar* from, uint length)
|
||||
|
||||
for (i=0;i<length;i++,p++,fr++)
|
||||
{
|
||||
*p = *fr ;
|
||||
*p= *fr ;
|
||||
|
||||
/* Sathit's NOTE: it's better idea not to do any normalize
|
||||
*/
|
||||
if (istone(*fr) || isdiacrt1(*fr))
|
||||
{
|
||||
if (i > 0 && (islwrvowel(fr[-1]) || isuprvowel(fr[-1])))
|
||||
continue ;
|
||||
if(islwrvowel(fr[1]) || isuprvowel(fr[1]))
|
||||
{
|
||||
*p= fr[1];
|
||||
p[1]= *fr;
|
||||
fr++;
|
||||
p++;
|
||||
i++ ;
|
||||
}
|
||||
/* Sathit's NOTE: it's better idea not to do any normalize */
|
||||
if (istone(*fr) || isdiacrt1(*fr))
|
||||
{
|
||||
if (i > 0 && (islwrvowel(fr[-1]) || isuprvowel(fr[-1])))
|
||||
continue;
|
||||
if(islwrvowel(fr[1]) || isuprvowel(fr[1]))
|
||||
{
|
||||
*p= fr[1];
|
||||
p[1]= *fr;
|
||||
fr++;
|
||||
p++;
|
||||
i++;
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user