Commit 59f3a958 authored by serg@serg.mysql.com's avatar serg@serg.mysql.com

ft_boolean_syntax variable

parent 307fb4a3
...@@ -49,6 +49,7 @@ extern const char *ft_precompiled_stopwords[]; ...@@ -49,6 +49,7 @@ extern const char *ft_precompiled_stopwords[];
extern ulong ft_min_word_len; extern ulong ft_min_word_len;
extern ulong ft_max_word_len; extern ulong ft_max_word_len;
extern ulong ft_max_word_len_for_sort; extern ulong ft_max_word_len_for_sort;
extern char *ft_boolean_syntax;
int ft_init_stopwords(const char **); int ft_init_stopwords(const char **);
void ft_free_stopwords(void); void ft_free_stopwords(void);
......
...@@ -36,7 +36,7 @@ static double _wghts[11]={ ...@@ -36,7 +36,7 @@ static double _wghts[11]={
3.375000000000000, 3.375000000000000,
5.062500000000000, 5.062500000000000,
7.593750000000000}; 7.593750000000000};
static double *wghts=_wghts+5; // wghts[i] = 1.5**i static double *wghts=_wghts+5; /* wghts[i] = 1.5**i */
static double _nwghts[11]={ static double _nwghts[11]={
-0.065843621399177, -0.065843621399177,
...@@ -50,7 +50,7 @@ static double _nwghts[11]={ ...@@ -50,7 +50,7 @@ static double _nwghts[11]={
-1.687500000000000, -1.687500000000000,
-2.531250000000000, -2.531250000000000,
-3.796875000000000}; -3.796875000000000};
static double *nwghts=_nwghts+5; // nwghts[i] = -0.5*1.5**i static double *nwghts=_nwghts+5; /* nwghts[i] = -0.5*1.5**i */
typedef struct st_ftb_expr FTB_EXPR; typedef struct st_ftb_expr FTB_EXPR;
struct st_ftb_expr { struct st_ftb_expr {
...@@ -114,20 +114,7 @@ void _ftb_parse_query(FTB *ftb, byte **start, byte *end, ...@@ -114,20 +114,7 @@ void _ftb_parse_query(FTB *ftb, byte **start, byte *end,
byte r=param.plusminus; byte r=param.plusminus;
float weight=(param.pmsign ? nwghts : wghts)[(r>5)?5:((r<-5)?-5:r)]; float weight=(param.pmsign ? nwghts : wghts)[(r>5)?5:((r<-5)?-5:r)];
switch (res) { switch (res) {
case FTB_LBR: case 1: /* word found */
ftbe=(FTB_EXPR *)alloc_root(&ftb->mem_root, sizeof(FTB_EXPR));
ftbe->yesno=param.yesno;
ftbe->weight=weight;
ftbe->up=up;
ftbe->ythresh=0;
ftbe->docid=HA_POS_ERROR;
if (ftbe->yesno > 0) up->ythresh++;
_ftb_parse_query(ftb, start, end, ftbe, depth+1,
(param.yesno<0 ? depth+1 : ndepth));
break;
case FTB_RBR:
return;
case 1:
ftbw=(FTB_WORD *)alloc_root(&ftb->mem_root, ftbw=(FTB_WORD *)alloc_root(&ftb->mem_root,
sizeof(FTB_WORD) + (param.trunc ? MI_MAX_KEY_BUFF : w.len+extra)); sizeof(FTB_WORD) + (param.trunc ? MI_MAX_KEY_BUFF : w.len+extra));
ftbw->len=w.len+1; ftbw->len=w.len+1;
...@@ -142,6 +129,19 @@ void _ftb_parse_query(FTB *ftb, byte **start, byte *end, ...@@ -142,6 +129,19 @@ void _ftb_parse_query(FTB *ftb, byte **start, byte *end,
if (ftbw->yesno > 0) up->ythresh++; if (ftbw->yesno > 0) up->ythresh++;
queue_insert(& ftb->queue, (byte *)ftbw); queue_insert(& ftb->queue, (byte *)ftbw);
break; break;
case 2: /* left bracket */
ftbe=(FTB_EXPR *)alloc_root(&ftb->mem_root, sizeof(FTB_EXPR));
ftbe->yesno=param.yesno;
ftbe->weight=weight;
ftbe->up=up;
ftbe->ythresh=0;
ftbe->docid=HA_POS_ERROR;
if (ftbe->yesno > 0) up->ythresh++;
_ftb_parse_query(ftb, start, end, ftbe, depth+1,
(param.yesno<0 ? depth+1 : ndepth));
break;
case 3: /* right bracket */
return;
} }
} }
return; return;
......
...@@ -253,7 +253,7 @@ float ft_nlq_find_relevance(FT_INFO *handler, my_off_t docid, ...@@ -253,7 +253,7 @@ float ft_nlq_find_relevance(FT_INFO *handler, my_off_t docid,
int a,b,c; int a,b,c;
FT_DOC *docs=handler->doc; FT_DOC *docs=handler->doc;
// Assuming docs[] is sorted by dpos... /* Assuming docs[] is sorted by dpos... */
for (a=0, b=handler->ndocs, c=(a+b)/2; b-a>1; c=(a+b)/2) for (a=0, b=handler->ndocs, c=(a+b)/2; b-a>1; c=(a+b)/2)
{ {
......
...@@ -33,9 +33,6 @@ typedef struct st_ft_docstat { ...@@ -33,9 +33,6 @@ typedef struct st_ft_docstat {
double max, nsum, nsum2; double max, nsum, nsum2;
#endif /* EVAL_RUN */ #endif /* EVAL_RUN */
// MI_INFO *info;
// uint keynr;
// byte *keybuf;
} FT_DOCSTAT; } FT_DOCSTAT;
static int FT_WORD_cmp(void* cmp_arg, FT_WORD *w1, FT_WORD *w2) static int FT_WORD_cmp(void* cmp_arg, FT_WORD *w1, FT_WORD *w2)
...@@ -63,9 +60,7 @@ static int walk_and_copy(FT_WORD *word,uint32 count,FT_DOCSTAT *docstat) ...@@ -63,9 +60,7 @@ static int walk_and_copy(FT_WORD *word,uint32 count,FT_DOCSTAT *docstat)
/* transforms tree of words into the array, applying normalization */ /* transforms tree of words into the array, applying normalization */
FT_WORD * ft_linearize(//MI_INFO *info, uint keynr, FT_WORD * ft_linearize(TREE *wtree)
//byte *keybuf,
TREE *wtree)
{ {
FT_WORD *wlist,*p; FT_WORD *wlist,*p;
FT_DOCSTAT docstat; FT_DOCSTAT docstat;
...@@ -74,9 +69,6 @@ FT_WORD * ft_linearize(//MI_INFO *info, uint keynr, ...@@ -74,9 +69,6 @@ FT_WORD * ft_linearize(//MI_INFO *info, uint keynr,
if ((wlist=(FT_WORD *) my_malloc(sizeof(FT_WORD)* if ((wlist=(FT_WORD *) my_malloc(sizeof(FT_WORD)*
(1+wtree->elements_in_tree),MYF(0)))) (1+wtree->elements_in_tree),MYF(0))))
{ {
// docstat.info=info;
// docstat.keynr=keynr;
// docstat.keybuf=keybuf;
docstat.list=wlist; docstat.list=wlist;
docstat.uniq=wtree->elements_in_tree; docstat.uniq=wtree->elements_in_tree;
#ifdef EVAL_RUN #ifdef EVAL_RUN
...@@ -122,12 +114,20 @@ FT_WORD * ft_linearize(//MI_INFO *info, uint keynr, ...@@ -122,12 +114,20 @@ FT_WORD * ft_linearize(//MI_INFO *info, uint keynr,
#endif #endif
#define word_char(X) (true_word_char(X) || misc_word_char(X)) #define word_char(X) (true_word_char(X) || misc_word_char(X))
/* returns:
* 0 - eof
* 1 - word found
* 2 - left bracket
* 3 - right bracket
*/
byte ft_get_word(byte **start, byte *end, FT_WORD *word, FTB_PARAM *param) byte ft_get_word(byte **start, byte *end, FT_WORD *word, FTB_PARAM *param)
{ {
byte *doc=*start; byte *doc=*start;
int mwc; int mwc;
param->yesno=param->plusminus=param->pmsign=0; param->yesno=(FTB_YES==' ')?1:0;
param->plusminus=param->pmsign=0;
while (doc<end) while (doc<end)
{ {
...@@ -138,18 +138,16 @@ byte ft_get_word(byte **start, byte *end, FT_WORD *word, FTB_PARAM *param) ...@@ -138,18 +138,16 @@ byte ft_get_word(byte **start, byte *end, FT_WORD *word, FTB_PARAM *param)
{ {
/* param->prev=' '; */ /* param->prev=' '; */
*start=doc+1; *start=doc+1;
return *doc; return (*doc == FTB_RBR)+2;
} }
if (param->prev == ' ') if (param->prev == ' ')
{ {
switch (*doc) { if (*doc == FTB_YES ) { param->yesno=+1; continue; } else
case FTB_YES: param->yesno=+1; continue; if (*doc == FTB_EGAL) { param->yesno= 0; continue; } else
case FTB_NO: param->yesno=-1; continue; if (*doc == FTB_NO ) { param->yesno=-1; continue; } else
case FTB_INC: param->plusminus++; continue; if (*doc == FTB_INC ) { param->plusminus++; continue; } else
case FTB_DEC: param->plusminus--; continue; if (*doc == FTB_DEC ) { param->plusminus--; continue; } else
case FTB_NEG: param->pmsign=!param->pmsign; continue; if (*doc == FTB_NEG ) { param->pmsign=!param->pmsign; continue; }
default: break;
}
} }
param->prev=*doc; param->prev=*doc;
param->yesno=param->plusminus=param->pmsign=0; param->yesno=param->plusminus=param->pmsign=0;
...@@ -162,7 +160,7 @@ byte ft_get_word(byte **start, byte *end, FT_WORD *word, FTB_PARAM *param) ...@@ -162,7 +160,7 @@ byte ft_get_word(byte **start, byte *end, FT_WORD *word, FTB_PARAM *param)
else if (!misc_word_char(*doc) || mwc++) else if (!misc_word_char(*doc) || mwc++)
break; break;
param->prev='A'; // be sure *prev is true_word_char param->prev='A'; /* be sure *prev is true_word_char */
word->len= (uint)(doc-word->pos) - mwc; word->len= (uint)(doc-word->pos) - mwc;
if ((param->trunc=(doc<end && *doc == FTB_TRUNC))) if ((param->trunc=(doc<end && *doc == FTB_TRUNC)))
doc++; doc++;
......
...@@ -21,6 +21,7 @@ ...@@ -21,6 +21,7 @@
ulong ft_min_word_len=4; ulong ft_min_word_len=4;
ulong ft_max_word_len=HA_FT_MAXLEN; ulong ft_max_word_len=HA_FT_MAXLEN;
ulong ft_max_word_len_for_sort=20; ulong ft_max_word_len_for_sort=20;
char *ft_boolean_syntax="+ -><()~*";
const MI_KEYSEG ft_keysegs[FT_SEGS]={ const MI_KEYSEG ft_keysegs[FT_SEGS]={
{ {
......
...@@ -86,14 +86,15 @@ extern ulong collstat; ...@@ -86,14 +86,15 @@ extern ulong collstat;
/*=================================================================*/ /*=================================================================*/
/* Boolean search operators */ /* Boolean search operators */
#define FTB_YES '+' #define FTB_YES (ft_boolean_syntax[0])
#define FTB_NO '-' #define FTB_EGAL (ft_boolean_syntax[1])
#define FTB_INC '>' #define FTB_NO (ft_boolean_syntax[2])
#define FTB_DEC '<' #define FTB_INC (ft_boolean_syntax[3])
#define FTB_LBR '(' #define FTB_DEC (ft_boolean_syntax[4])
#define FTB_RBR ')' #define FTB_LBR (ft_boolean_syntax[5])
#define FTB_NEG '~' #define FTB_RBR (ft_boolean_syntax[6])
#define FTB_TRUNC '*' #define FTB_NEG (ft_boolean_syntax[7])
#define FTB_TRUNC (ft_boolean_syntax[8])
typedef struct st_ft_word { typedef struct st_ft_word {
byte * pos; byte * pos;
......
...@@ -3,3 +3,4 @@ Variable_name Value ...@@ -3,3 +3,4 @@ Variable_name Value
ft_min_word_len 4 ft_min_word_len 4
ft_max_word_len 254 ft_max_word_len 254
ft_max_word_len_for_sort 20 ft_max_word_len_for_sort 20
ft_boolean_syntax + -><()~*
...@@ -1883,7 +1883,7 @@ The server will not act as a slave."); ...@@ -1883,7 +1883,7 @@ The server will not act as a slave.");
if (opt_myisam_log) if (opt_myisam_log)
(void) mi_log( 1 ); (void) mi_log( 1 );
ft_init_stopwords(ft_precompiled_stopwords); /* SerG */ ft_init_stopwords(ft_precompiled_stopwords);
#ifdef __WIN__ #ifdef __WIN__
#define MYSQL_ERR_FILE "mysql.err" #define MYSQL_ERR_FILE "mysql.err"
...@@ -2986,6 +2986,7 @@ struct show_var_st init_vars[]= { ...@@ -2986,6 +2986,7 @@ struct show_var_st init_vars[]= {
{"ft_min_word_len", (char*) &ft_min_word_len, SHOW_LONG}, {"ft_min_word_len", (char*) &ft_min_word_len, SHOW_LONG},
{"ft_max_word_len", (char*) &ft_max_word_len, SHOW_LONG}, {"ft_max_word_len", (char*) &ft_max_word_len, SHOW_LONG},
{"ft_max_word_len_for_sort",(char*) &ft_max_word_len_for_sort, SHOW_LONG}, {"ft_max_word_len_for_sort",(char*) &ft_max_word_len_for_sort, SHOW_LONG},
{"ft_boolean_syntax", ft_boolean_syntax, SHOW_CHAR},
{"have_bdb", (char*) &have_berkeley_db, SHOW_HAVE}, {"have_bdb", (char*) &have_berkeley_db, SHOW_HAVE},
{"have_innodb", (char*) &have_innodb, SHOW_HAVE}, {"have_innodb", (char*) &have_innodb, SHOW_HAVE},
{"have_isam", (char*) &have_isam, SHOW_HAVE}, {"have_isam", (char*) &have_isam, SHOW_HAVE},
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment