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;
58 int longcmp(const void *rec1, const void *rec2) {
61 i1 = *(const long *)rec1;
62 i2 = *(const long *)rec2;
64 if (i1 > i2) return(1);
65 if (i1 < i2) return(-1);
73 * Index or de-index a message. (op == 1 to index, 0 to de-index)
75 void ft_index_message(long msgnum, int op) {
76 struct CtdlMessage *msg;
80 struct cdbdata *cdb_bucket;
84 lprintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
85 (op ? "adding" : "removing") , msgnum
88 msg = CtdlFetchMessage(msgnum, 1);
89 if (msg == NULL) return;
91 if (msg->cm_fields['M'] != NULL) {
92 wordbreaker(msg->cm_fields['M'], &num_tokens, &tokens);
97 for (i=0; i<num_tokens; ++i) {
99 /* Add the message to the relevant token bucket */
101 /* FIXME lock the file */
102 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tokens[i], sizeof(int));
103 if (cdb_bucket == NULL) {
104 cdb_bucket = malloc(sizeof(struct cdbdata));
106 cdb_bucket->ptr = NULL;
110 num_msgs = cdb_bucket->len / sizeof(long);
113 if (op == 1) { /* add to index */
115 cdb_bucket->ptr = realloc(cdb_bucket->ptr, num_msgs*sizeof(long) );
116 msgs = (long *) cdb_bucket->ptr;
117 msgs[num_msgs - 1] = msgnum;
120 if (op == 0) { /* remove from index */
122 msgs = (long *) cdb_bucket->ptr;
123 for (j=0; j<num_msgs; ++j) {
124 if (msgs[j] == msgnum) {
125 lprintf(CTDL_DEBUG, "nix0ring %ld\n", msgs[j]);
126 memmove(&msgs[j], &msgs[j+1],
127 ((num_msgs - j - 1)*sizeof(long)));
134 /* sort and purge dups */
135 if ( (op == 1) && (num_msgs > 1) ) {
136 msgs = (long *) cdb_bucket->ptr;
137 qsort(msgs, num_msgs, sizeof(long), longcmp);
138 for (j=0; j<(num_msgs-1); ++j) {
139 if (msgs[j] == msgs[j+1]) {
140 memmove(&msgs[j], &msgs[j+1],
141 ((num_msgs - j - 1)*sizeof(long)));
147 cdb_store(CDB_FULLTEXT, &tokens[i], sizeof(int),
148 msgs, (num_msgs*sizeof(long)) );
150 cdb_free(cdb_bucket);
152 /* FIXME unlock the file */
162 * Add a message to the list of those to be indexed.
164 void ft_index_msg(long msgnum, void *userdata) {
166 if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
168 if (ft_num_msgs > ft_num_alloc) {
169 ft_num_alloc += 1024;
170 ft_newmsgs = realloc(ft_newmsgs,
171 (ft_num_alloc * sizeof(long)));
173 ft_newmsgs[ft_num_msgs - 1] = msgnum;
179 * Scan a room for messages to index.
181 void ft_index_room(struct ctdlroom *qrbuf, void *data)
183 getroom(&CC->room, qrbuf->QRname);
184 CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, ft_index_msg, NULL);
189 * Begin the fulltext indexing process. (Called as an EVT_TIMER event)
191 void do_fulltext_indexing(void) {
193 static time_t last_index = 0L;
196 * Make sure we don't run the indexer too frequently.
197 * FIXME move the setting into config
199 if ( (time(NULL) - last_index) < 300L) {
204 * Check to see whether the fulltext index is up to date; if there
205 * are no messages to index, don't waste any more time trying.
207 lprintf(CTDL_DEBUG, "CitControl.MMhighest = %ld\n", CitControl.MMhighest);
208 lprintf(CTDL_DEBUG, "CitControl.MMfulltext = %ld\n", CitControl.MMfulltext);
209 if (CitControl.MMfulltext >= CitControl.MMhighest) {
214 lprintf(CTDL_DEBUG, "do_fulltext_indexing() started\n");
217 * If we've switched wordbreaker modules, burn the index and start
218 * over. FIXME write this...
220 if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
221 lprintf(CTDL_INFO, "(re)initializing full text index\n");
222 cdb_trunc(CDB_FULLTEXT);
223 CitControl.MMfulltext = 0L;
228 * Now go through each room and find messages to index.
230 ft_newhighest = CitControl.MMhighest;
231 ForEachRoom(ft_index_room, NULL); /* load all msg pointers */
233 if (ft_num_msgs > 0) {
234 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
235 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
236 if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
237 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
238 ((ft_num_msgs - i - 1)*sizeof(long)));
243 /* Here it is ... do each message! */
244 for (i=0; i<ft_num_msgs; ++i) {
245 ft_index_message(ft_newmsgs[i], 1);
254 /* Save our place so we don't have to do this again */
255 CitControl.MMfulltext = ft_newhighest;
256 CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
258 last_index = time(NULL);
260 lprintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
266 * API call to perform searches.
267 * (This one does the "all of these words" search.)
268 * Caller is responsible for freeing the message list.
270 void ft_search(int *fts_num_msgs, long **fts_msgs, char *search_string) {
274 struct cdbdata *cdb_bucket;
277 int num_all_msgs = 0;
278 long *all_msgs = NULL;
279 int num_ret_msgs = 0;
280 int num_ret_alloc = 0;
281 long *ret_msgs = NULL;
283 wordbreaker(search_string, &num_tokens, &tokens);
284 if (num_tokens > 0) {
285 for (i=0; i<num_tokens; ++i) {
287 /* search for tokens[i] */
288 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tokens[i], sizeof(int));
289 if (cdb_bucket != NULL) {
290 num_msgs = cdb_bucket->len / sizeof(long);
291 msgs = (long *)cdb_bucket->ptr;
293 num_all_msgs += num_msgs;
294 if (num_all_msgs > 0) {
295 all_msgs = realloc(all_msgs, num_all_msgs*sizeof(long) );
296 memcpy(&all_msgs[num_all_msgs - num_msgs], msgs,
297 num_msgs*sizeof(long) );
300 cdb_free(cdb_bucket);
305 qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
308 * At this point, if a message appears num_tokens times in the
309 * list, then it contains all of the search tokens.
311 if (num_all_msgs >= num_tokens) for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
312 if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
315 if (num_ret_msgs > num_ret_alloc) {
317 ret_msgs = realloc(ret_msgs, (num_ret_alloc*sizeof(long)) );
319 ret_msgs[num_ret_msgs - 1] = all_msgs[j];
327 *fts_num_msgs = num_ret_msgs;
328 *fts_msgs = ret_msgs;
333 * Tentative form of a search command
335 void cmd_srch(char *argbuf) {
339 char search_string[256];
341 if (CtdlAccessCheck(ac_logged_in)) return;
342 extract_token(search_string, argbuf, 0, '|', sizeof search_string);
343 ft_search(&num_msgs, &msgs, search_string);
345 cprintf("%d %d msgs match all search words:\n",
346 LISTING_FOLLOWS, num_msgs);
348 for (i=0; i<num_msgs; ++i) {
349 cprintf("%ld\n", msgs[i]);
352 if (msgs != NULL) free(msgs);
357 /*****************************************************************************/
359 char *serv_fulltext_init(void)
361 CtdlRegisterSessionHook(do_fulltext_indexing, EVT_TIMER);
362 CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");