1
by brian
clean slate |
1 |
/* Copyright (C) 2000-2005 MySQL AB
|
2 |
||
3 |
This program is free software; you can redistribute it and/or modify
|
|
4 |
it under the terms of the GNU General Public License as published by
|
|
5 |
the Free Software Foundation; version 2 of the License.
|
|
6 |
||
7 |
This program is distributed in the hope that it will be useful,
|
|
8 |
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
9 |
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
10 |
GNU General Public License for more details.
|
|
11 |
||
12 |
You should have received a copy of the GNU General Public License
|
|
13 |
along with this program; if not, write to the Free Software
|
|
14 |
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */
|
|
15 |
||
16 |
/* Written by Sergei A. Golubchik, who has a shared copyright to this code */
|
|
17 |
||
18 |
#include "ftdefs.h" |
|
19 |
||
20 |
typedef struct st_ft_docstat { |
|
21 |
FT_WORD *list; |
|
22 |
uint uniq; |
|
23 |
double sum; |
|
24 |
} FT_DOCSTAT; |
|
25 |
||
26 |
typedef struct st_my_ft_parser_param |
|
27 |
{
|
|
28 |
TREE *wtree; |
|
29 |
MEM_ROOT *mem_root; |
|
30 |
} MY_FT_PARSER_PARAM; |
|
31 |
||
32 |
static int FT_WORD_cmp(CHARSET_INFO* cs, FT_WORD *w1, FT_WORD *w2) |
|
33 |
{
|
|
34 |
return ha_compare_text(cs, (uchar*) w1->pos, w1->len, |
|
35 |
(uchar*) w2->pos, w2->len, 0, 0); |
|
36 |
}
|
|
37 |
||
38 |
static int walk_and_copy(FT_WORD *word,uint32 count,FT_DOCSTAT *docstat) |
|
39 |
{
|
|
40 |
word->weight=LWS_IN_USE; |
|
41 |
docstat->sum+=word->weight; |
|
42 |
memcpy_fixed((docstat->list)++,word,sizeof(FT_WORD)); |
|
43 |
return 0; |
|
44 |
}
|
|
45 |
||
46 |
/* transforms tree of words into the array, applying normalization */
|
|
47 |
||
48 |
FT_WORD * ft_linearize(TREE *wtree, MEM_ROOT *mem_root) |
|
49 |
{
|
|
50 |
FT_WORD *wlist,*p; |
|
51 |
FT_DOCSTAT docstat; |
|
52 |
DBUG_ENTER("ft_linearize"); |
|
53 |
||
54 |
if ((wlist=(FT_WORD *) alloc_root(mem_root, sizeof(FT_WORD)* |
|
55 |
(1+wtree->elements_in_tree)))) |
|
56 |
{
|
|
57 |
docstat.list=wlist; |
|
58 |
docstat.uniq=wtree->elements_in_tree; |
|
59 |
docstat.sum=0; |
|
60 |
tree_walk(wtree,(tree_walk_action)&walk_and_copy,&docstat,left_root_right); |
|
61 |
}
|
|
62 |
delete_tree(wtree); |
|
63 |
if (!wlist) |
|
64 |
DBUG_RETURN(NULL); |
|
65 |
||
66 |
docstat.list->pos=NULL; |
|
67 |
||
68 |
for (p=wlist;p->pos;p++) |
|
69 |
{
|
|
70 |
p->weight=PRENORM_IN_USE; |
|
71 |
}
|
|
72 |
||
73 |
for (p=wlist;p->pos;p++) |
|
74 |
{
|
|
75 |
p->weight/=NORM_IN_USE; |
|
76 |
}
|
|
77 |
||
78 |
DBUG_RETURN(wlist); |
|
79 |
}
|
|
80 |
||
81 |
my_bool ft_boolean_check_syntax_string(const uchar *str) |
|
82 |
{
|
|
83 |
uint i, j; |
|
84 |
||
85 |
if (!str || |
|
86 |
(strlen((char*) str)+1 != sizeof(ft_boolean_syntax)) || |
|
87 |
(str[0] != ' ' && str[1] != ' ')) |
|
88 |
return 1; |
|
89 |
for (i=0; i<sizeof(ft_boolean_syntax); i++) |
|
90 |
{
|
|
91 |
/* limiting to 7-bit ascii only */
|
|
92 |
if ((unsigned char)(str[i]) > 127 || my_isalnum(default_charset_info, str[i])) |
|
93 |
return 1; |
|
94 |
for (j=0; j<i; j++) |
|
95 |
if (str[i] == str[j] && (i != 11 || j != 10)) |
|
96 |
return 1; |
|
97 |
}
|
|
98 |
return 0; |
|
99 |
}
|
|
100 |
||
101 |
/*
|
|
102 |
RETURN VALUE
|
|
103 |
0 - eof
|
|
104 |
1 - word found
|
|
105 |
2 - left bracket
|
|
106 |
3 - right bracket
|
|
107 |
4 - stopword found
|
|
108 |
*/
|
|
109 |
uchar ft_get_word(CHARSET_INFO *cs, uchar **start, uchar *end, |
|
110 |
FT_WORD *word, MYSQL_FTPARSER_BOOLEAN_INFO *param) |
|
111 |
{
|
|
112 |
uchar *doc=*start; |
|
113 |
int ctype; |
|
114 |
uint mwc, length; |
|
115 |
int mbl; |
|
116 |
||
117 |
param->yesno=(FTB_YES==' ') ? 1 : (param->quot != 0); |
|
118 |
param->weight_adjust= param->wasign= 0; |
|
119 |
param->type= FT_TOKEN_EOF; |
|
120 |
||
121 |
while (doc<end) |
|
122 |
{
|
|
123 |
for (; doc < end; doc+= (mbl > 0 ? mbl : (mbl < 0 ? -mbl : 1))) |
|
124 |
{
|
|
125 |
mbl= cs->cset->ctype(cs, &ctype, (uchar*)doc, (uchar*)end); |
|
126 |
if (true_word_char(ctype, *doc)) |
|
127 |
break; |
|
128 |
if (*doc == FTB_RQUOT && param->quot) |
|
129 |
{
|
|
130 |
param->quot= (char*) doc; |
|
131 |
*start=doc+1; |
|
132 |
param->type= FT_TOKEN_RIGHT_PAREN; |
|
133 |
goto ret; |
|
134 |
}
|
|
135 |
if (!param->quot) |
|
136 |
{
|
|
137 |
if (*doc == FTB_LBR || *doc == FTB_RBR || *doc == FTB_LQUOT) |
|
138 |
{
|
|
139 |
/* param->prev=' '; */
|
|
140 |
*start=doc+1; |
|
141 |
if (*doc == FTB_LQUOT) |
|
142 |
param->quot= (char*) *start; |
|
143 |
param->type= (*doc == FTB_RBR ? FT_TOKEN_RIGHT_PAREN : FT_TOKEN_LEFT_PAREN); |
|
144 |
goto ret; |
|
145 |
}
|
|
146 |
if (param->prev == ' ') |
|
147 |
{
|
|
148 |
if (*doc == FTB_YES ) { param->yesno=+1; continue; } else |
|
149 |
if (*doc == FTB_EGAL) { param->yesno= 0; continue; } else |
|
150 |
if (*doc == FTB_NO ) { param->yesno=-1; continue; } else |
|
151 |
if (*doc == FTB_INC ) { param->weight_adjust++; continue; } else |
|
152 |
if (*doc == FTB_DEC ) { param->weight_adjust--; continue; } else |
|
153 |
if (*doc == FTB_NEG ) { param->wasign= !param->wasign; continue; } |
|
154 |
}
|
|
155 |
}
|
|
156 |
param->prev=*doc; |
|
157 |
param->yesno=(FTB_YES==' ') ? 1 : (param->quot != 0); |
|
158 |
param->weight_adjust= param->wasign= 0; |
|
159 |
}
|
|
160 |
||
161 |
mwc=length=0; |
|
162 |
for (word->pos= doc; doc < end; length++, |
|
163 |
doc+= (mbl > 0 ? mbl : (mbl < 0 ? -mbl : 1))) |
|
164 |
{
|
|
165 |
mbl= cs->cset->ctype(cs, &ctype, (uchar*)doc, (uchar*)end); |
|
166 |
if (true_word_char(ctype, *doc)) |
|
167 |
mwc=0; |
|
168 |
else if (!misc_word_char(*doc) || mwc) |
|
169 |
break; |
|
170 |
else
|
|
171 |
mwc++; |
|
172 |
}
|
|
173 |
param->prev='A'; /* be sure *prev is true_word_char */ |
|
174 |
word->len= (uint)(doc-word->pos) - mwc; |
|
175 |
if ((param->trunc=(doc<end && *doc == FTB_TRUNC))) |
|
176 |
doc++; |
|
177 |
||
178 |
if (((length >= ft_min_word_len && !is_stopword((char*) word->pos, |
|
179 |
word->len)) |
|
180 |
|| param->trunc) && length < ft_max_word_len) |
|
181 |
{
|
|
182 |
*start=doc; |
|
183 |
param->type= FT_TOKEN_WORD; |
|
184 |
goto ret; |
|
185 |
}
|
|
186 |
else if (length) /* make sure length > 0 (if start contains spaces only) */ |
|
187 |
{
|
|
188 |
*start= doc; |
|
189 |
param->type= FT_TOKEN_STOPWORD; |
|
190 |
goto ret; |
|
191 |
}
|
|
192 |
}
|
|
193 |
if (param->quot) |
|
194 |
{
|
|
195 |
*start= doc; |
|
196 |
param->quot= (char*) doc; |
|
197 |
param->type= 3; /* FT_RBR */ |
|
198 |
goto ret; |
|
199 |
}
|
|
200 |
ret: |
|
201 |
return param->type; |
|
202 |
}
|
|
203 |
||
204 |
uchar ft_simple_get_word(CHARSET_INFO *cs, uchar **start, const uchar *end, |
|
205 |
FT_WORD *word, my_bool skip_stopwords) |
|
206 |
{
|
|
207 |
uchar *doc= *start; |
|
208 |
uint mwc, length; |
|
209 |
int mbl; |
|
210 |
int ctype; |
|
211 |
DBUG_ENTER("ft_simple_get_word"); |
|
212 |
||
213 |
do
|
|
214 |
{
|
|
215 |
for (;; doc+= (mbl > 0 ? mbl : (mbl < 0 ? -mbl : 1))) |
|
216 |
{
|
|
217 |
if (doc >= end) |
|
218 |
DBUG_RETURN(0); |
|
219 |
mbl= cs->cset->ctype(cs, &ctype, (uchar*)doc, (uchar*)end); |
|
220 |
if (true_word_char(ctype, *doc)) |
|
221 |
break; |
|
222 |
}
|
|
223 |
||
224 |
mwc= length= 0; |
|
225 |
for (word->pos= doc; doc < end; length++, |
|
226 |
doc+= (mbl > 0 ? mbl : (mbl < 0 ? -mbl : 1))) |
|
227 |
{
|
|
228 |
mbl= cs->cset->ctype(cs, &ctype, (uchar*)doc, (uchar*)end); |
|
229 |
if (true_word_char(ctype, *doc)) |
|
230 |
mwc= 0; |
|
231 |
else if (!misc_word_char(*doc) || mwc) |
|
232 |
break; |
|
233 |
else
|
|
234 |
mwc++; |
|
235 |
}
|
|
236 |
||
237 |
word->len= (uint)(doc-word->pos) - mwc; |
|
238 |
||
239 |
if (skip_stopwords == FALSE || |
|
240 |
(length >= ft_min_word_len && length < ft_max_word_len && |
|
241 |
!is_stopword((char*) word->pos, word->len))) |
|
242 |
{
|
|
243 |
*start= doc; |
|
244 |
DBUG_RETURN(1); |
|
245 |
}
|
|
246 |
} while (doc < end); |
|
247 |
DBUG_RETURN(0); |
|
248 |
}
|
|
249 |
||
250 |
void ft_parse_init(TREE *wtree, CHARSET_INFO *cs) |
|
251 |
{
|
|
252 |
DBUG_ENTER("ft_parse_init"); |
|
253 |
if (!is_tree_inited(wtree)) |
|
254 |
init_tree(wtree,0,0,sizeof(FT_WORD),(qsort_cmp2)&FT_WORD_cmp,0,NULL, cs); |
|
255 |
DBUG_VOID_RETURN; |
|
256 |
}
|
|
257 |
||
258 |
||
259 |
static int ft_add_word(MYSQL_FTPARSER_PARAM *param, |
|
260 |
char *word, int word_len, |
|
261 |
MYSQL_FTPARSER_BOOLEAN_INFO *boolean_info __attribute__((unused))) |
|
262 |
{
|
|
263 |
TREE *wtree; |
|
264 |
FT_WORD w; |
|
265 |
MY_FT_PARSER_PARAM *ft_param=param->mysql_ftparam; |
|
266 |
DBUG_ENTER("ft_add_word"); |
|
267 |
wtree= ft_param->wtree; |
|
268 |
if (param->flags & MYSQL_FTFLAGS_NEED_COPY) |
|
269 |
{
|
|
270 |
uchar *ptr; |
|
271 |
DBUG_ASSERT(wtree->with_delete == 0); |
|
272 |
ptr= (uchar *)alloc_root(ft_param->mem_root, word_len); |
|
273 |
memcpy(ptr, word, word_len); |
|
274 |
w.pos= ptr; |
|
275 |
}
|
|
276 |
else
|
|
277 |
w.pos= (uchar*) word; |
|
278 |
w.len= word_len; |
|
279 |
if (!tree_insert(wtree, &w, 0, wtree->custom_arg)) |
|
280 |
{
|
|
281 |
delete_tree(wtree); |
|
282 |
DBUG_RETURN(1); |
|
283 |
}
|
|
284 |
DBUG_RETURN(0); |
|
285 |
}
|
|
286 |
||
287 |
||
288 |
static int ft_parse_internal(MYSQL_FTPARSER_PARAM *param, |
|
289 |
char *doc_arg, int doc_len) |
|
290 |
{
|
|
291 |
uchar *doc= (uchar*) doc_arg; |
|
292 |
uchar *end= doc + doc_len; |
|
293 |
MY_FT_PARSER_PARAM *ft_param=param->mysql_ftparam; |
|
294 |
TREE *wtree= ft_param->wtree; |
|
295 |
FT_WORD w; |
|
296 |
DBUG_ENTER("ft_parse_internal"); |
|
297 |
||
298 |
while (ft_simple_get_word(wtree->custom_arg, &doc, end, &w, TRUE)) |
|
299 |
if (param->mysql_add_word(param, (char*) w.pos, w.len, 0)) |
|
300 |
DBUG_RETURN(1); |
|
301 |
DBUG_RETURN(0); |
|
302 |
}
|
|
303 |
||
304 |
||
305 |
int ft_parse(TREE *wtree, uchar *doc, int doclen, |
|
306 |
struct st_mysql_ftparser *parser, |
|
307 |
MYSQL_FTPARSER_PARAM *param, MEM_ROOT *mem_root) |
|
308 |
{
|
|
309 |
MY_FT_PARSER_PARAM my_param; |
|
310 |
DBUG_ENTER("ft_parse"); |
|
311 |
DBUG_ASSERT(parser); |
|
312 |
||
313 |
my_param.wtree= wtree; |
|
314 |
my_param.mem_root= mem_root; |
|
315 |
||
316 |
param->mysql_parse= ft_parse_internal; |
|
317 |
param->mysql_add_word= ft_add_word; |
|
318 |
param->mysql_ftparam= &my_param; |
|
319 |
param->cs= wtree->custom_arg; |
|
320 |
param->doc= (char*) doc; |
|
321 |
param->length= doclen; |
|
322 |
param->mode= MYSQL_FTPARSER_SIMPLE_MODE; |
|
323 |
DBUG_RETURN(parser->parse(param)); |
|
324 |
}
|
|
325 |
||
326 |
#define MAX_PARAM_NR 2
|
|
327 |
MYSQL_FTPARSER_PARAM *ftparser_call_initializer(MI_INFO *info, |
|
328 |
uint keynr, uint paramnr) |
|
329 |
{
|
|
330 |
uint32 ftparser_nr; |
|
331 |
struct st_mysql_ftparser *parser; |
|
332 |
if (! info->ftparser_param) |
|
333 |
{
|
|
334 |
/* info->ftparser_param can not be zero after the initialization,
|
|
335 |
because it always includes built-in fulltext parser. And built-in
|
|
336 |
parser can be called even if the table has no fulltext indexes and
|
|
337 |
no varchar/text fields. */
|
|
338 |
if (! info->s->ftparsers) |
|
339 |
{
|
|
340 |
/* It's ok that modification to shared structure is done w/o mutex
|
|
341 |
locks, because all threads would set the same variables to the
|
|
342 |
same values. */
|
|
343 |
uint i, j, keys= info->s->state.header.keys, ftparsers= 1; |
|
344 |
for (i= 0; i < keys; i++) |
|
345 |
{
|
|
346 |
MI_KEYDEF *keyinfo= &info->s->keyinfo[i]; |
|
347 |
if (keyinfo->flag & HA_FULLTEXT) |
|
348 |
{
|
|
349 |
for (j= 0;; j++) |
|
350 |
{
|
|
351 |
if (j == i) |
|
352 |
{
|
|
353 |
keyinfo->ftparser_nr= ftparsers++; |
|
354 |
break; |
|
355 |
}
|
|
356 |
if (info->s->keyinfo[j].flag & HA_FULLTEXT && |
|
357 |
keyinfo->parser == info->s->keyinfo[j].parser) |
|
358 |
{
|
|
359 |
keyinfo->ftparser_nr= info->s->keyinfo[j].ftparser_nr; |
|
360 |
break; |
|
361 |
}
|
|
362 |
}
|
|
363 |
}
|
|
364 |
}
|
|
365 |
info->s->ftparsers= ftparsers; |
|
366 |
}
|
|
367 |
/*
|
|
368 |
We have to allocate two MYSQL_FTPARSER_PARAM structures per plugin
|
|
369 |
because in a boolean search a parser is called recursively
|
|
370 |
ftb_find_relevance* calls ftb_check_phrase*
|
|
371 |
(MAX_PARAM_NR=2)
|
|
372 |
*/
|
|
373 |
info->ftparser_param= (MYSQL_FTPARSER_PARAM *) |
|
374 |
my_malloc(MAX_PARAM_NR * sizeof(MYSQL_FTPARSER_PARAM) * |
|
375 |
info->s->ftparsers, MYF(MY_WME|MY_ZEROFILL)); |
|
376 |
init_alloc_root(&info->ft_memroot, FTPARSER_MEMROOT_ALLOC_SIZE, 0); |
|
377 |
if (! info->ftparser_param) |
|
378 |
return 0; |
|
379 |
}
|
|
380 |
if (keynr == NO_SUCH_KEY) |
|
381 |
{
|
|
382 |
ftparser_nr= 0; |
|
383 |
parser= &ft_default_parser; |
|
384 |
}
|
|
385 |
else
|
|
386 |
{
|
|
387 |
ftparser_nr= info->s->keyinfo[keynr].ftparser_nr; |
|
388 |
parser= info->s->keyinfo[keynr].parser; |
|
389 |
}
|
|
390 |
DBUG_ASSERT(paramnr < MAX_PARAM_NR); |
|
391 |
ftparser_nr= ftparser_nr*MAX_PARAM_NR + paramnr; |
|
392 |
if (! info->ftparser_param[ftparser_nr].mysql_add_word) |
|
393 |
{
|
|
394 |
/* Note, that mysql_add_word is used here as a flag:
|
|
395 |
mysql_add_word == 0 - parser is not initialized
|
|
396 |
mysql_add_word != 0 - parser is initialized, or no
|
|
397 |
initialization needed. */
|
|
398 |
info->ftparser_param[ftparser_nr].mysql_add_word= |
|
399 |
(int (*)(struct st_mysql_ftparser_param *, char *, int, |
|
400 |
MYSQL_FTPARSER_BOOLEAN_INFO *)) 1; |
|
401 |
if (parser->init && parser->init(&info->ftparser_param[ftparser_nr])) |
|
402 |
return 0; |
|
403 |
}
|
|
404 |
return &info->ftparser_param[ftparser_nr]; |
|
405 |
}
|
|
406 |
||
407 |
void ftparser_call_deinitializer(MI_INFO *info) |
|
408 |
{
|
|
409 |
uint i, j, keys= info->s->state.header.keys; |
|
410 |
free_root(&info->ft_memroot, MYF(0)); |
|
411 |
if (! info->ftparser_param) |
|
412 |
return; |
|
413 |
for (i= 0; i < keys; i++) |
|
414 |
{
|
|
415 |
MI_KEYDEF *keyinfo= &info->s->keyinfo[i]; |
|
416 |
for (j=0; j < MAX_PARAM_NR; j++) |
|
417 |
{
|
|
418 |
MYSQL_FTPARSER_PARAM *ftparser_param= |
|
419 |
&info->ftparser_param[keyinfo->ftparser_nr*MAX_PARAM_NR + j]; |
|
420 |
if (keyinfo->flag & HA_FULLTEXT && ftparser_param->mysql_add_word) |
|
421 |
{
|
|
422 |
if (keyinfo->parser->deinit) |
|
423 |
keyinfo->parser->deinit(ftparser_param); |
|
424 |
ftparser_param->mysql_add_word= 0; |
|
425 |
}
|
|
426 |
else
|
|
427 |
break; |
|
428 |
}
|
|
429 |
}
|
|
430 |
}
|
|
431 |