4 * This module handles fulltext indexing of the message base.
17 #include <sys/types.h>
19 #if TIME_WITH_SYS_TIME
20 # include <sys/time.h>
24 # include <sys/time.h>
35 #include "sysdep_decls.h"
36 #include "citserver.h"
39 #include "serv_extensions.h"
45 #include "serv_fulltext.h"
46 #include "ft_wordbreaker.h"
49 long ft_newhighest = 0L;
50 long *ft_newmsgs = NULL;
55 int ftc_num_msgs[65536];
56 long *ftc_msgs[65536];
62 int longcmp(const void *rec1, const void *rec2) {
65 i1 = *(const long *)rec1;
66 i2 = *(const long *)rec2;
68 if (i1 > i2) return(1);
69 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 "Flushing index cache to disk (%d%% complete)\n",
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 lprintf(CTDL_INFO, "Flushed index cache to disk (100%% complete)\n");
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 lprintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
112 (op ? "adding" : "removing") , msgnum
115 /* Output the message as text before indexing it, so we don't end up
116 * indexing a bunch of encoded base64, etc.
118 CC->redirect_buffer = malloc(SIZ);
119 CC->redirect_len = 0;
120 CC->redirect_alloc = SIZ;
121 CtdlOutputMsg(msgnum, MT_CITADEL, HEADERS_ALL, 0, 1, NULL);
122 msgtext = CC->redirect_buffer;
123 CC->redirect_buffer = NULL;
124 CC->redirect_len = 0;
125 CC->redirect_alloc = 0;
126 lprintf(CTDL_DEBUG, "Wordbreaking message %ld...\n", msgnum);
127 wordbreaker(msgtext, &num_tokens, &tokens);
130 lprintf(CTDL_DEBUG, "Indexing message %ld...\n", msgnum);
131 if (num_tokens > 0) {
132 for (i=0; i<num_tokens; ++i) {
134 /* Add the message to the relevant token bucket */
136 /* search for tokens[i] */
139 /* fetch the bucket, Liza */
140 if (ftc_msgs[tok] == NULL) {
141 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
142 if (cdb_bucket != NULL) {
143 ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
144 ftc_msgs[tok] = (long *)cdb_bucket->ptr;
145 cdb_bucket->ptr = NULL;
146 cdb_free(cdb_bucket);
149 ftc_num_msgs[tok] = 0;
150 ftc_msgs[tok] = malloc(sizeof(long));
155 if (op == 1) { /* add to index */
157 ftc_msgs[tok] = realloc(ftc_msgs[tok],
158 ftc_num_msgs[tok]*sizeof(long));
159 ftc_msgs[tok][ftc_num_msgs[tok] - 1] = msgnum;
162 if (op == 0) { /* remove from index */
163 if (ftc_num_msgs[tok] >= 1) {
164 for (j=0; j<ftc_num_msgs[tok]; ++j) {
165 if (ftc_msgs[tok][j] == msgnum) {
166 memmove(&ftc_msgs[tok][j], &ftc_msgs[tok][j+1], ((ftc_num_msgs[tok] - j - 1)*sizeof(long)));
174 /* FIXME do we need to unlock the file here? */
184 * Add a message to the list of those to be indexed.
186 void ft_index_msg(long msgnum, void *userdata) {
188 if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
190 if (ft_num_msgs > ft_num_alloc) {
191 ft_num_alloc += 1024;
192 ft_newmsgs = realloc(ft_newmsgs,
193 (ft_num_alloc * sizeof(long)));
195 ft_newmsgs[ft_num_msgs - 1] = msgnum;
201 * Scan a room for messages to index.
203 void ft_index_room(struct ctdlroom *qrbuf, void *data)
205 getroom(&CC->room, qrbuf->QRname);
206 CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, ft_index_msg, NULL);
211 * Begin the fulltext indexing process. (Called as an EVT_TIMER event)
213 void do_fulltext_indexing(void) {
215 static time_t last_index = 0L;
216 static time_t last_progress = 0L;
219 * Don't do this if the site doesn't have it enabled.
221 if (!config.c_enable_fulltext) {
226 * Make sure we don't run the indexer too frequently.
227 * FIXME move the setting into config
229 if ( (time(NULL) - last_index) < 300L) {
234 * Check to see whether the fulltext index is up to date; if there
235 * are no messages to index, don't waste any more time trying.
237 if (CitControl.MMfulltext >= CitControl.MMhighest) {
238 return; /* nothing to do! */
241 lprintf(CTDL_DEBUG, "do_fulltext_indexing() started\n");
244 * If we've switched wordbreaker modules, burn the index and start
247 if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
248 lprintf(CTDL_INFO, "(re)initializing full text index\n");
249 cdb_trunc(CDB_FULLTEXT);
250 CitControl.MMfulltext = 0L;
255 * Now go through each room and find messages to index.
257 ft_newhighest = CitControl.MMhighest;
258 ForEachRoom(ft_index_room, NULL); /* load all msg pointers */
260 if (ft_num_msgs > 0) {
261 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
262 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
263 if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
264 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
265 ((ft_num_msgs - i - 1)*sizeof(long)));
271 /* Here it is ... do each message! */
272 for (i=0; i<ft_num_msgs; ++i) {
273 if ((time(NULL) - last_progress) > 10) {
275 "Indexed %d of %d messages (%d%%)\n",
277 ((i*100) / ft_num_msgs)
279 last_progress = time(NULL);
281 ft_index_message(ft_newmsgs[i], 1);
283 /* Check to see if we need to quit early */
285 lprintf(CTDL_DEBUG, "Indexer quitting early\n");
286 ft_newhighest = ft_newmsgs[i];
298 /* Save our place so we don't have to do this again */
300 CitControl.MMfulltext = ft_newhighest;
301 CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
303 last_index = time(NULL);
305 lprintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
310 * Main loop for the indexer thread.
312 void *indexer_thread(void *arg) {
313 struct CitContext indexerCC;
315 lprintf(CTDL_DEBUG, "indexer_thread() initializing\n");
317 memset(&indexerCC, 0, sizeof(struct CitContext));
318 indexerCC.internal_pgm = 1;
319 indexerCC.cs_pid = 0;
320 pthread_setspecific(MyConKey, (void *)&indexerCC );
324 while (!time_to_die) {
325 do_fulltext_indexing();
329 lprintf(CTDL_DEBUG, "indexer_thread() exiting\n");
336 * API call to perform searches.
337 * (This one does the "all of these words" search.)
338 * Caller is responsible for freeing the message list.
340 void ft_search(int *fts_num_msgs, long **fts_msgs, char *search_string) {
344 struct cdbdata *cdb_bucket;
345 int num_all_msgs = 0;
346 long *all_msgs = NULL;
347 int num_ret_msgs = 0;
348 int num_ret_alloc = 0;
349 long *ret_msgs = NULL;
352 wordbreaker(search_string, &num_tokens, &tokens);
353 if (num_tokens > 0) {
354 for (i=0; i<num_tokens; ++i) {
356 /* search for tokens[i] */
359 /* fetch the bucket, Liza */
360 if (ftc_msgs[tok] == NULL) {
361 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
362 if (cdb_bucket != NULL) {
363 ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
364 ftc_msgs[tok] = (long *)cdb_bucket->ptr;
365 cdb_bucket->ptr = NULL;
366 cdb_free(cdb_bucket);
369 ftc_num_msgs[tok] = 0;
370 ftc_msgs[tok] = malloc(sizeof(long));
374 num_all_msgs += ftc_num_msgs[tok];
375 if (num_all_msgs > 0) {
376 all_msgs = realloc(all_msgs, num_all_msgs*sizeof(long) );
377 memcpy(&all_msgs[num_all_msgs-ftc_num_msgs[tok]],
378 ftc_msgs[tok], ftc_num_msgs[tok]*sizeof(long) );
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 * Tentative form of a search command
415 void cmd_srch(char *argbuf) {
419 char search_string[256];
421 if (CtdlAccessCheck(ac_logged_in)) return;
423 if (!config.c_enable_fulltext) {
424 cprintf("%d Full text index is not enabled on this server.\n",
425 ERROR + CMD_NOT_SUPPORTED);
428 extract_token(search_string, argbuf, 0, '|', sizeof search_string);
429 ft_search(&num_msgs, &msgs, search_string);
431 cprintf("%d %d msgs match all search words:\n",
432 LISTING_FOLLOWS, num_msgs);
434 for (i=0; i<num_msgs; ++i) {
435 cprintf("%ld\n", msgs[i]);
438 if (msgs != NULL) free(msgs);
443 * Zero out our index cache.
445 void initialize_ft_cache(void) {
446 memset(ftc_num_msgs, 0, (65536 * sizeof(int)));
447 memset(ftc_msgs, 0, (65536 * sizeof(long *)));
451 /*****************************************************************************/
453 char *serv_fulltext_init(void)
455 initialize_ft_cache();
456 CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");