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) {
82 lprintf(5, "Flushing index cache to disk (%d%% complete)\n",
85 last_update = time(NULL);
87 if (ftc_msgs[i] != NULL) {
88 cdb_store(CDB_FULLTEXT, &i, sizeof(int), ftc_msgs[i],
89 (ftc_num_msgs[i] * sizeof(long)));
95 lprintf(5, "Cache flush has been completed\n");
100 * Index or de-index a message. (op == 1 to index, 0 to de-index)
102 void ft_index_message(long msgnum, int op) {
106 struct cdbdata *cdb_bucket;
110 lprintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
111 (op ? "adding" : "removing") , msgnum
114 /* Output the message as text before indexing it, so we don't end up
115 * indexing a bunch of encoded base64, etc.
117 CC->redirect_buffer = malloc(SIZ);
118 CC->redirect_len = 0;
119 CC->redirect_alloc = SIZ;
120 CtdlOutputMsg(msgnum, MT_CITADEL, HEADERS_ALL, 0, 1);
121 msgtext = CC->redirect_buffer;
122 CC->redirect_buffer = NULL;
123 CC->redirect_len = 0;
124 CC->redirect_alloc = 0;
125 lprintf(CTDL_DEBUG, "Wordbreaking message %ld...\n", msgnum);
126 wordbreaker(msgtext, &num_tokens, &tokens);
129 lprintf(CTDL_DEBUG, "Indexing message %ld...\n", msgnum);
130 if (num_tokens > 0) {
131 for (i=0; i<num_tokens; ++i) {
133 /* Add the message to the relevant token bucket */
135 /* search for tokens[i] */
138 /* fetch the bucket, Liza */
139 if (ftc_msgs[tok] == NULL) {
140 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
141 if (cdb_bucket != NULL) {
142 ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
143 ftc_msgs[tok] = (long *)cdb_bucket->ptr;
144 cdb_bucket->ptr = NULL;
145 cdb_free(cdb_bucket);
148 ftc_num_msgs[tok] = 0;
149 ftc_msgs[tok] = malloc(sizeof(long));
154 if (op == 1) { /* add to index */
156 ftc_msgs[tok] = realloc(ftc_msgs[tok],
157 ftc_num_msgs[tok]*sizeof(long));
158 ftc_msgs[tok][ftc_num_msgs[tok] - 1] = msgnum;
161 if (op == 0) { /* remove from index */
162 if (ftc_num_msgs[tok] >= 1) {
163 for (j=0; j<ftc_num_msgs[tok]; ++j) {
164 if (ftc_msgs[tok][j] == msgnum) {
165 memmove(&ftc_msgs[tok][j], &ftc_msgs[tok][j+1], ((ftc_num_msgs[tok] - j - 1)*sizeof(long)));
173 /* FIXME do we need to unlock the file here? */
183 * Add a message to the list of those to be indexed.
185 void ft_index_msg(long msgnum, void *userdata) {
187 if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
189 if (ft_num_msgs > ft_num_alloc) {
190 ft_num_alloc += 1024;
191 ft_newmsgs = realloc(ft_newmsgs,
192 (ft_num_alloc * sizeof(long)));
194 ft_newmsgs[ft_num_msgs - 1] = msgnum;
200 * Scan a room for messages to index.
202 void ft_index_room(struct ctdlroom *qrbuf, void *data)
204 getroom(&CC->room, qrbuf->QRname);
205 CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, ft_index_msg, NULL);
210 * Begin the fulltext indexing process. (Called as an EVT_TIMER event)
212 void do_fulltext_indexing(void) {
214 static time_t last_index = 0L;
215 static time_t last_progress = 0L;
218 * Don't do this if the site doesn't have it enabled.
220 if (!config.c_enable_fulltext) {
225 * Make sure we don't run the indexer too frequently.
226 * FIXME move the setting into config
228 if ( (time(NULL) - last_index) < 300L) {
233 * Check to see whether the fulltext index is up to date; if there
234 * are no messages to index, don't waste any more time trying.
236 if (CitControl.MMfulltext >= CitControl.MMhighest) {
237 return; /* nothing to do! */
240 lprintf(CTDL_DEBUG, "do_fulltext_indexing() started\n");
243 * If we've switched wordbreaker modules, burn the index and start
246 if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
247 lprintf(CTDL_INFO, "(re)initializing full text index\n");
248 cdb_trunc(CDB_FULLTEXT);
249 CitControl.MMfulltext = 0L;
254 * Now go through each room and find messages to index.
256 ft_newhighest = CitControl.MMhighest;
257 ForEachRoom(ft_index_room, NULL); /* load all msg pointers */
259 if (ft_num_msgs > 0) {
260 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
261 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
262 if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
263 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
264 ((ft_num_msgs - i - 1)*sizeof(long)));
270 /* Here it is ... do each message! */
271 for (i=0; i<ft_num_msgs; ++i) {
272 if ((time(NULL) - last_progress) > 10) {
274 "Indexed %d of %d messages (%d%%)\n",
276 ((i*100) / ft_num_msgs)
278 last_progress = time(NULL);
280 ft_index_message(ft_newmsgs[i], 1);
282 /* Check to see if we need to quit early */
284 lprintf(CTDL_DEBUG, "Indexer quitting early\n");
285 ft_newhighest = ft_newmsgs[i];
297 /* Save our place so we don't have to do this again */
299 CitControl.MMfulltext = ft_newhighest;
300 CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
302 last_index = time(NULL);
304 lprintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
309 * Main loop for the indexer thread.
311 void *indexer_thread(void *arg) {
312 struct CitContext indexerCC;
314 lprintf(CTDL_DEBUG, "indexer_thread() initializing\n");
316 memset(&indexerCC, 0, sizeof(struct CitContext));
317 indexerCC.internal_pgm = 1;
318 indexerCC.cs_pid = 0;
319 pthread_setspecific(MyConKey, (void *)&indexerCC );
323 while (!time_to_die) {
324 do_fulltext_indexing();
328 lprintf(CTDL_DEBUG, "indexer_thread() exiting\n");
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, 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 qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
385 * At this point, if a message appears num_tokens times in the
386 * list, then it contains all of the search tokens.
388 if (num_all_msgs >= num_tokens)
389 for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
390 if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
393 if (num_ret_msgs > num_ret_alloc) {
395 ret_msgs = realloc(ret_msgs,
396 (num_ret_alloc*sizeof(long)) );
398 ret_msgs[num_ret_msgs - 1] = all_msgs[j];
406 *fts_num_msgs = num_ret_msgs;
407 *fts_msgs = ret_msgs;
412 * Tentative form of a search command
414 void cmd_srch(char *argbuf) {
418 char search_string[256];
420 if (CtdlAccessCheck(ac_logged_in)) return;
422 if (!config.c_enable_fulltext) {
423 cprintf("%d Full text index is not enabled on this server.\n",
424 ERROR + CMD_NOT_SUPPORTED);
427 extract_token(search_string, argbuf, 0, '|', sizeof search_string);
428 ft_search(&num_msgs, &msgs, search_string);
430 cprintf("%d %d msgs match all search words:\n",
431 LISTING_FOLLOWS, num_msgs);
433 for (i=0; i<num_msgs; ++i) {
434 cprintf("%ld\n", msgs[i]);
437 if (msgs != NULL) free(msgs);
442 * Zero out our index cache.
444 void initialize_ft_cache(void) {
445 memset(ftc_num_msgs, 0, (65536 * sizeof(int)));
446 memset(ftc_msgs, 0, (65536 * sizeof(long *)));
450 /*****************************************************************************/
452 char *serv_fulltext_init(void)
454 initialize_ft_cache();
455 CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");