2 * This module handles fulltext indexing of the message base.
3 * Copyright (c) 2005-2021 by the citadel.org team
5 * This program is open source software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as published
7 * by the Free Software Foundation; either version 3 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
28 #include <sys/types.h>
33 #include <libcitadel.h>
36 #include "citserver.h"
42 #include "serv_fulltext.h"
43 #include "ft_wordbreaker.h"
47 #include "ctdl_module.h"
49 long ft_newhighest = 0L;
50 long *ft_newmsgs = NULL;
54 int ftc_num_msgs[65536];
55 long *ftc_msgs[65536];
61 int longcmp(const void *rec1, const void *rec2) {
64 i1 = *(const long *)rec1;
65 i2 = *(const long *)rec2;
67 if (i1 > i2) return(1);
68 if (i1 < i2) return(-1);
74 * Flush our index cache out to disk.
76 void ft_flush_cache(void) {
78 time_t last_update = 0;
80 for (i=0; i<65536; ++i) {
81 if ((time(NULL) - last_update) >= 10) {
83 "fulltext: flushing index cache to disk (%d%% complete)",
86 last_update = time(NULL);
88 if (ftc_msgs[i] != NULL) {
89 cdb_store(CDB_FULLTEXT, &i, sizeof(int), ftc_msgs[i],
90 (ftc_num_msgs[i] * sizeof(long)));
96 syslog(LOG_INFO, "fulltext: flushed index cache to disk (100%% complete)");
101 * Index or de-index a message. (op == 1 to index, 0 to de-index)
103 void ft_index_message(long msgnum, int op) {
107 struct cdbdata *cdb_bucket;
111 struct CtdlMessage *msg = NULL;
113 msg = CtdlFetchMessage(msgnum, 1);
115 syslog(LOG_ERR, "fulltext: ft_index_message() could not load msg %ld", msgnum);
119 if (!CM_IsEmpty(msg, eSuppressIdx)) {
120 syslog(LOG_DEBUG, "fulltext: ft_index_message() excluded msg %ld", msgnum);
125 syslog(LOG_DEBUG, "fulltext: ft_index_message() %s msg %ld", (op ? "adding" : "removing") , msgnum);
127 /* Output the message as text before indexing it, so we don't end up
128 * indexing a bunch of encoded base64, etc.
130 CC->redirect_buffer = NewStrBufPlain(NULL, SIZ);
131 CtdlOutputPreLoadedMsg(msg, MT_CITADEL, HEADERS_ALL, 0, 1, 0);
133 msgtext = CC->redirect_buffer;
134 CC->redirect_buffer = NULL;
135 if (msgtext != NULL) {
136 syslog(LOG_DEBUG, "fulltext: wordbreaking message %ld (%d bytes)", msgnum, StrLength(msgtext));
138 txt = SmashStrBuf(&msgtext);
139 wordbreaker(txt, &num_tokens, &tokens);
142 syslog(LOG_DEBUG, "fulltext: indexing message %ld [%d tokens]", msgnum, num_tokens);
143 if (num_tokens > 0) {
144 for (i=0; i<num_tokens; ++i) {
146 /* Add the message to the relevant token bucket */
148 /* search for tokens[i] */
151 if ( (tok >= 0) && (tok <= 65535) ) {
152 /* fetch the bucket, Liza */
153 if (ftc_msgs[tok] == NULL) {
154 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
155 if (cdb_bucket != NULL) {
156 ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
157 ftc_msgs[tok] = (long *)cdb_bucket->ptr;
158 cdb_bucket->ptr = NULL;
159 cdb_free(cdb_bucket);
162 ftc_num_msgs[tok] = 0;
163 ftc_msgs[tok] = malloc(sizeof(long));
168 if (op == 1) { /* add to index */
170 ftc_msgs[tok] = realloc(ftc_msgs[tok],
171 ftc_num_msgs[tok]*sizeof(long));
172 ftc_msgs[tok][ftc_num_msgs[tok] - 1] = msgnum;
175 if (op == 0) { /* remove from index */
176 if (ftc_num_msgs[tok] >= 1) {
177 for (j=0; j<ftc_num_msgs[tok]; ++j) {
178 if (ftc_msgs[tok][j] == msgnum) {
179 memmove(&ftc_msgs[tok][j], &ftc_msgs[tok][j+1], ((ftc_num_msgs[tok] - j - 1)*sizeof(long)));
188 syslog(LOG_ALERT, "fulltext: invalid token %d !!", tok);
198 * Add a message to the list of those to be indexed.
200 void ft_index_msg(long msgnum, void *userdata) {
202 if ((msgnum > CtdlGetConfigLong("MMfulltext")) && (msgnum <= ft_newhighest)) {
204 if (ft_num_msgs > ft_num_alloc) {
205 ft_num_alloc += 1024;
206 ft_newmsgs = realloc(ft_newmsgs, (ft_num_alloc * sizeof(long)));
208 ft_newmsgs[ft_num_msgs - 1] = msgnum;
215 * Scan a room for messages to index.
217 void ft_index_room(struct ctdlroom *qrbuf, void *data)
219 if (server_shutting_down)
222 CtdlGetRoom(&CC->room, qrbuf->QRname);
223 CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, NULL, ft_index_msg, NULL);
228 * Begin the fulltext indexing process.
230 void do_fulltext_indexing(void) {
232 static time_t last_progress = 0L;
233 static int is_running = 0;
234 if (is_running) return; /* Concurrency check - only one can run */
238 * Don't do this if the site doesn't have it enabled.
240 if (!CtdlGetConfigInt("c_enable_fulltext")) {
245 * If we've switched wordbreaker modules, burn the index and start over.
247 begin_critical_section(S_CONTROL);
248 if (CtdlGetConfigInt("MM_fulltext_wordbreaker") != FT_WORDBREAKER_ID) {
249 syslog(LOG_DEBUG, "fulltext: wb ver on disk = %d, code ver = %d",
250 CtdlGetConfigInt("MM_fulltext_wordbreaker"), FT_WORDBREAKER_ID
252 syslog(LOG_INFO, "fulltext: (re)initializing index");
253 cdb_trunc(CDB_FULLTEXT);
254 CtdlSetConfigLong("MMfulltext", 0);
256 end_critical_section(S_CONTROL);
259 * Silently return if our fulltext index is up to date with new messages.
261 if ((CtdlGetConfigLong("MMfulltext") >= CtdlGetConfigLong("MMhighest"))) {
262 return; /* nothing to do! */
266 * Now go through each room and find messages to index.
268 ft_newhighest = CtdlGetConfigLong("MMhighest");
269 CtdlForEachRoom(ft_index_room, NULL); /* load all msg pointers */
271 if (ft_num_msgs > 0) {
272 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
273 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
274 if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
275 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
276 ((ft_num_msgs - i - 1)*sizeof(long)));
282 /* Here it is ... do each message! */
283 for (i=0; i<ft_num_msgs; ++i) {
284 if (time(NULL) != last_progress) {
286 "fulltext: indexed %d of %d messages (%d%%)",
288 ((i*100) / ft_num_msgs)
290 last_progress = time(NULL);
292 ft_index_message(ft_newmsgs[i], 1);
294 /* Check to see if we need to quit early */
295 if (server_shutting_down) {
296 syslog(LOG_DEBUG, "fulltext: indexer quitting early");
297 ft_newhighest = ft_newmsgs[i];
301 /* Check to see if we have to maybe flush to disk */
302 if (i >= FT_MAX_CACHE) {
303 syslog(LOG_DEBUG, "fulltext: time to flush.");
304 ft_newhighest = ft_newmsgs[i];
316 if (server_shutting_down) {
321 /* Save our place so we don't have to do this again */
323 begin_critical_section(S_CONTROL);
324 CtdlSetConfigLong("MMfulltext", ft_newhighest);
325 CtdlSetConfigInt("MM_fulltext_wordbreaker", FT_WORDBREAKER_ID);
326 end_critical_section(S_CONTROL);
328 syslog(LOG_DEBUG, "fulltext: indexing finished");
335 * API call to perform searches.
336 * (This one does the "all of these words" search.)
337 * Caller is responsible for freeing the message list.
339 void ft_search(int *fts_num_msgs, long **fts_msgs, const char *search_string) {
343 struct cdbdata *cdb_bucket;
344 int num_all_msgs = 0;
345 long *all_msgs = NULL;
346 int num_ret_msgs = 0;
347 int num_ret_alloc = 0;
348 long *ret_msgs = NULL;
351 wordbreaker(search_string, &num_tokens, &tokens);
352 if (num_tokens > 0) {
353 for (i=0; i<num_tokens; ++i) {
355 /* search for tokens[i] */
358 /* fetch the bucket, Liza */
359 if (ftc_msgs[tok] == NULL) {
360 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
361 if (cdb_bucket != NULL) {
362 ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
363 ftc_msgs[tok] = (long *)cdb_bucket->ptr;
364 cdb_bucket->ptr = NULL;
365 cdb_free(cdb_bucket);
368 ftc_num_msgs[tok] = 0;
369 ftc_msgs[tok] = malloc(sizeof(long));
373 num_all_msgs += ftc_num_msgs[tok];
374 if (num_all_msgs > 0) {
375 all_msgs = realloc(all_msgs, num_all_msgs*sizeof(long) );
376 memcpy(&all_msgs[num_all_msgs-ftc_num_msgs[tok]],
377 ftc_msgs[tok], ftc_num_msgs[tok]*sizeof(long) );
382 if (all_msgs != NULL) {
383 qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
386 * At this point, if a message appears num_tokens times in the
387 * list, then it contains all of the search tokens.
389 if (num_all_msgs >= num_tokens)
390 for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
391 if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
394 if (num_ret_msgs > num_ret_alloc) {
396 ret_msgs = realloc(ret_msgs,
397 (num_ret_alloc*sizeof(long)) );
399 ret_msgs[num_ret_msgs - 1] = all_msgs[j];
407 *fts_num_msgs = num_ret_msgs;
408 *fts_msgs = ret_msgs;
413 * This search command is for diagnostic purposes and may be removed or replaced.
415 void cmd_srch(char *argbuf) {
419 char search_string[256];
421 if (CtdlAccessCheck(ac_logged_in)) return;
423 if (!CtdlGetConfigInt("c_enable_fulltext")) {
424 cprintf("%d Full text index is not enabled on this server.\n",
425 ERROR + CMD_NOT_SUPPORTED);
429 extract_token(search_string, argbuf, 0, '|', sizeof search_string);
430 ft_search(&num_msgs, &msgs, search_string);
432 cprintf("%d %d msgs match all search words:\n",
433 LISTING_FOLLOWS, num_msgs);
435 for (i=0; i<num_msgs; ++i) {
436 cprintf("%ld\n", msgs[i]);
439 if (msgs != NULL) free(msgs);
445 * Zero out our index cache.
447 void initialize_ft_cache(void) {
448 memset(ftc_num_msgs, 0, (65536 * sizeof(int)));
449 memset(ftc_msgs, 0, (65536 * sizeof(long *)));
453 void ft_delete_remove(char *room, long msgnum)
457 /* Remove from fulltext index */
458 if (CtdlGetConfigInt("c_enable_fulltext")) {
459 ft_index_message(msgnum, 0);
464 /*****************************************************************************/
466 CTDL_MODULE_INIT(fulltext)
470 initialize_ft_cache();
471 CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
472 CtdlRegisterDeleteHook(ft_delete_remove);
473 CtdlRegisterSearchFuncHook(ft_search, "fulltext");
474 CtdlRegisterSessionHook(do_fulltext_indexing, EVT_TIMER, PRIO_CLEANUP + 300);
476 /* return our module name for the log */