4 * This module handles fulltext indexing of the message base.
7 * Copyright (c) 2005-2009 by the citadel.org team
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 3 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
19 * You should have received a copy of the GNU General Public License
20 * along with this program; if not, write to the Free Software
21 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
33 #include <sys/types.h>
35 #if TIME_WITH_SYS_TIME
36 # include <sys/time.h>
40 # include <sys/time.h>
49 #include <libcitadel.h>
52 #include "citserver.h"
58 #include "serv_fulltext.h"
59 #include "ft_wordbreaker.h"
62 #include "ctdl_module.h"
66 long ft_newhighest = 0L;
67 long *ft_newmsgs = NULL;
72 int ftc_num_msgs[65536];
73 long *ftc_msgs[65536];
79 int longcmp(const void *rec1, const void *rec2) {
82 i1 = *(const long *)rec1;
83 i2 = *(const long *)rec2;
85 if (i1 > i2) return(1);
86 if (i1 < i2) return(-1);
91 * Flush our index cache out to disk.
93 void ft_flush_cache(void) {
95 time_t last_update = 0;
97 for (i=0; i<65536; ++i) {
98 if ((time(NULL) - last_update) >= 10) {
99 CtdlLogPrintf(CTDL_INFO,
100 "Flushing index cache to disk (%d%% complete)\n",
103 last_update = time(NULL);
105 if (ftc_msgs[i] != NULL) {
106 cdb_store(CDB_FULLTEXT, &i, sizeof(int), ftc_msgs[i],
107 (ftc_num_msgs[i] * sizeof(long)));
113 CtdlLogPrintf(CTDL_INFO, "Flushed index cache to disk (100%% complete)\n");
118 * Index or de-index a message. (op == 1 to index, 0 to de-index)
120 void ft_index_message(long msgnum, int op) {
124 struct cdbdata *cdb_bucket;
128 CtdlLogPrintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
129 (op ? "adding" : "removing") , msgnum
132 /* Output the message as text before indexing it, so we don't end up
133 * indexing a bunch of encoded base64, etc.
135 CC->redirect_buffer = malloc(SIZ);
136 CC->redirect_len = 0;
137 CC->redirect_alloc = SIZ;
138 CtdlOutputMsg(msgnum, MT_CITADEL, HEADERS_ALL, 0, 1, NULL, 0);
139 msgtext = CC->redirect_buffer;
140 CC->redirect_buffer = NULL;
141 CC->redirect_len = 0;
142 CC->redirect_alloc = 0;
143 CtdlLogPrintf(CTDL_DEBUG, "Wordbreaking message %ld...\n", msgnum);
144 wordbreaker(msgtext, &num_tokens, &tokens);
147 CtdlLogPrintf(CTDL_DEBUG, "Indexing message %ld [%d tokens]\n", msgnum, num_tokens);
148 if (num_tokens > 0) {
149 for (i=0; i<num_tokens; ++i) {
151 /* Add the message to the relevant token bucket */
153 /* search for tokens[i] */
156 if ( (tok >= 0) && (tok <= 65535) ) {
157 /* fetch the bucket, Liza */
158 if (ftc_msgs[tok] == NULL) {
159 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
160 if (cdb_bucket != NULL) {
161 ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
162 ftc_msgs[tok] = (long *)cdb_bucket->ptr;
163 cdb_bucket->ptr = NULL;
164 cdb_free(cdb_bucket);
167 ftc_num_msgs[tok] = 0;
168 ftc_msgs[tok] = malloc(sizeof(long));
173 if (op == 1) { /* add to index */
175 ftc_msgs[tok] = realloc(ftc_msgs[tok],
176 ftc_num_msgs[tok]*sizeof(long));
177 ftc_msgs[tok][ftc_num_msgs[tok] - 1] = msgnum;
180 if (op == 0) { /* remove from index */
181 if (ftc_num_msgs[tok] >= 1) {
182 for (j=0; j<ftc_num_msgs[tok]; ++j) {
183 if (ftc_msgs[tok][j] == msgnum) {
184 memmove(&ftc_msgs[tok][j], &ftc_msgs[tok][j+1], ((ftc_num_msgs[tok] - j - 1)*sizeof(long)));
193 CtdlLogPrintf(CTDL_ALERT, "Invalid token %d !!\n", tok);
204 * Add a message to the list of those to be indexed.
206 void ft_index_msg(long msgnum, void *userdata) {
208 if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
210 if (ft_num_msgs > ft_num_alloc) {
211 ft_num_alloc += 1024;
212 ft_newmsgs = realloc(ft_newmsgs,
213 (ft_num_alloc * sizeof(long)));
215 ft_newmsgs[ft_num_msgs - 1] = msgnum;
221 * Scan a room for messages to index.
223 void ft_index_room(struct ctdlroom *qrbuf, void *data)
225 if (CtdlThreadCheckStop())
228 CtdlGetRoom(&CC->room, qrbuf->QRname);
229 CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, NULL, ft_index_msg, NULL);
234 * Begin the fulltext indexing process. (Called as an EVT_TIMER event)
236 void do_fulltext_indexing(void) {
238 static time_t last_index = 0L;
239 static time_t last_progress = 0L;
240 time_t run_time = 0L;
241 time_t end_time = 0L;
244 * Don't do this if the site doesn't have it enabled.
246 if (!config.c_enable_fulltext) {
251 * Make sure we don't run the indexer too frequently.
252 * FIXME move the setting into config
255 * The thread sleeps for 300 seconds so we don't need this here any more
257 if ( (time(NULL) - last_index) < 300L) {
263 * Check to see whether the fulltext index is up to date; if there
264 * are no messages to index, don't waste any more time trying.
266 if ((CitControl.MMfulltext >= CitControl.MMhighest) && (CitControl.fulltext_wordbreaker == FT_WORDBREAKER_ID)) {
267 return; /* nothing to do! */
270 run_time = time(NULL);
271 CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() started (%ld)\n", run_time);
274 * If we've switched wordbreaker modules, burn the index and start
277 begin_critical_section(S_CONTROL);
278 if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
279 CtdlLogPrintf(CTDL_DEBUG, "wb ver on disk = %d, code ver = %d\n",
280 CitControl.fulltext_wordbreaker, FT_WORDBREAKER_ID);
281 CtdlLogPrintf(CTDL_INFO, "(re)initializing full text index\n");
282 cdb_trunc(CDB_FULLTEXT);
283 CitControl.MMfulltext = 0L;
286 end_critical_section(S_CONTROL);
289 * Now go through each room and find messages to index.
291 ft_newhighest = CitControl.MMhighest;
292 CtdlForEachRoom(ft_index_room, NULL); /* load all msg pointers */
294 if (ft_num_msgs > 0) {
295 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
296 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
297 if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
298 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
299 ((ft_num_msgs - i - 1)*sizeof(long)));
305 /* Here it is ... do each message! */
306 for (i=0; i<ft_num_msgs; ++i) {
307 if (time(NULL) != last_progress) {
308 CtdlLogPrintf(CTDL_DEBUG,
309 "Indexed %d of %d messages (%d%%)\n",
311 ((i*100) / ft_num_msgs)
313 last_progress = time(NULL);
315 ft_index_message(ft_newmsgs[i], 1);
317 /* Check to see if we need to quit early */
318 if (CtdlThreadCheckStop()) {
319 CtdlLogPrintf(CTDL_DEBUG, "Indexer quitting early\n");
320 ft_newhighest = ft_newmsgs[i];
324 /* Check to see if we have to maybe flush to disk */
325 if (i >= FT_MAX_CACHE) {
326 CtdlLogPrintf(CTDL_DEBUG, "Time to flush.\n");
327 ft_newhighest = ft_newmsgs[i];
338 end_time = time(NULL);
340 if (CtdlThreadCheckStop())
343 CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() duration (%ld)\n", end_time - run_time);
345 /* Save our place so we don't have to do this again */
347 begin_critical_section(S_CONTROL);
348 CitControl.MMfulltext = ft_newhighest;
349 CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
351 end_critical_section(S_CONTROL);
352 last_index = time(NULL);
354 CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
359 * Main loop for the indexer thread.
361 void *indexer_thread(void *arg) {
362 struct CitContext indexerCC;
364 CtdlLogPrintf(CTDL_DEBUG, "indexer_thread() initializing\n");
366 CtdlFillSystemContext(&indexerCC, "indexer");
367 citthread_setspecific(MyConKey, (void *)&indexerCC );
369 while (!CtdlThreadCheckStop()) {
370 do_fulltext_indexing();
371 CtdlThreadSleep(300);
374 CtdlLogPrintf(CTDL_DEBUG, "indexer_thread() exiting\n");
381 * API call to perform searches.
382 * (This one does the "all of these words" search.)
383 * Caller is responsible for freeing the message list.
385 void ft_search(int *fts_num_msgs, long **fts_msgs, char *search_string) {
389 struct cdbdata *cdb_bucket;
390 int num_all_msgs = 0;
391 long *all_msgs = NULL;
392 int num_ret_msgs = 0;
393 int num_ret_alloc = 0;
394 long *ret_msgs = NULL;
397 wordbreaker(search_string, &num_tokens, &tokens);
398 if (num_tokens > 0) {
399 for (i=0; i<num_tokens; ++i) {
401 /* search for tokens[i] */
404 /* fetch the bucket, Liza */
405 if (ftc_msgs[tok] == NULL) {
406 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
407 if (cdb_bucket != NULL) {
408 ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
409 ftc_msgs[tok] = (long *)cdb_bucket->ptr;
410 cdb_bucket->ptr = NULL;
411 cdb_free(cdb_bucket);
414 ftc_num_msgs[tok] = 0;
415 ftc_msgs[tok] = malloc(sizeof(long));
419 num_all_msgs += ftc_num_msgs[tok];
420 if (num_all_msgs > 0) {
421 all_msgs = realloc(all_msgs, num_all_msgs*sizeof(long) );
422 memcpy(&all_msgs[num_all_msgs-ftc_num_msgs[tok]],
423 ftc_msgs[tok], ftc_num_msgs[tok]*sizeof(long) );
428 qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
431 * At this point, if a message appears num_tokens times in the
432 * list, then it contains all of the search tokens.
434 if (num_all_msgs >= num_tokens)
435 for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
436 if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
439 if (num_ret_msgs > num_ret_alloc) {
441 ret_msgs = realloc(ret_msgs,
442 (num_ret_alloc*sizeof(long)) );
444 ret_msgs[num_ret_msgs - 1] = all_msgs[j];
452 *fts_num_msgs = num_ret_msgs;
453 *fts_msgs = ret_msgs;
458 * This search command is for diagnostic purposes and may be removed or replaced.
460 void cmd_srch(char *argbuf) {
464 char search_string[256];
466 if (CtdlAccessCheck(ac_logged_in)) return;
468 if (!config.c_enable_fulltext) {
469 cprintf("%d Full text index is not enabled on this server.\n",
470 ERROR + CMD_NOT_SUPPORTED);
474 extract_token(search_string, argbuf, 0, '|', sizeof search_string);
475 ft_search(&num_msgs, &msgs, search_string);
477 cprintf("%d %d msgs match all search words:\n",
478 LISTING_FOLLOWS, num_msgs);
480 for (i=0; i<num_msgs; ++i) {
481 cprintf("%ld\n", msgs[i]);
484 if (msgs != NULL) free(msgs);
489 * Zero out our index cache.
491 void initialize_ft_cache(void) {
492 memset(ftc_num_msgs, 0, (65536 * sizeof(int)));
493 memset(ftc_msgs, 0, (65536 * sizeof(long *)));
497 void ft_delete_remove(char *room, long msgnum)
501 /* Remove from fulltext index */
502 if (config.c_enable_fulltext) {
503 ft_index_message(msgnum, 0);
507 /*****************************************************************************/
509 CTDL_MODULE_INIT(fulltext)
513 initialize_ft_cache();
514 initialize_noise_words();
515 CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
516 CtdlRegisterDeleteHook(ft_delete_remove);
517 CtdlRegisterSearchFuncHook(ft_search, "fulltext");
518 CtdlRegisterCleanupHook(noise_word_cleanup);
522 CtdlThreadCreate("Indexer", CTDLTHREAD_BIGSTACK, indexer_thread, NULL);
524 /* return our Subversion id for the Log */