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) {
79 struct cdbdata *cdb_bucket;
84 lprintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
85 (op ? "adding" : "removing") , msgnum
88 /* Output the message as text before indexing it, so we don't end up
89 * indexing a bunch of encoded base64, etc.
91 lprintf(CTDL_DEBUG, "Fetching...\n");
92 CC->redirect_buffer = malloc(SIZ);
94 CC->redirect_alloc = SIZ;
95 CtdlOutputMsg(msgnum, MT_CITADEL, HEADERS_ALL, 0, 1);
96 msgtext = CC->redirect_buffer;
97 CC->redirect_buffer = NULL;
99 CC->redirect_alloc = 0;
100 lprintf(CTDL_DEBUG, "Wordbreaking...\n");
101 wordbreaker(msgtext, &num_tokens, &tokens);
104 lprintf(CTDL_DEBUG, "Indexing...\n");
105 if (num_tokens > 0) {
106 for (i=0; i<num_tokens; ++i) {
108 /* Add the message to the relevant token bucket */
110 /* FIXME lock the file */
111 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tokens[i], sizeof(int));
112 if (cdb_bucket == NULL) {
113 cdb_bucket = malloc(sizeof(struct cdbdata));
115 cdb_bucket->ptr = NULL;
119 num_msgs = cdb_bucket->len / sizeof(long);
122 if (op == 1) { /* add to index */
124 cdb_bucket->ptr = realloc(cdb_bucket->ptr, num_msgs*sizeof(long) );
125 msgs = (long *) cdb_bucket->ptr;
126 msgs[num_msgs - 1] = msgnum;
129 if (op == 0) { /* remove from index */
131 msgs = (long *) cdb_bucket->ptr;
132 for (j=0; j<num_msgs; ++j) {
133 if (msgs[j] == msgnum) {
134 memmove(&msgs[j], &msgs[j+1], ((num_msgs - j - 1)*sizeof(long)));
141 /* sort and purge dups */
142 if ( (op == 1) && (num_msgs > 1) ) {
143 msgs = (long *) cdb_bucket->ptr;
144 qsort(msgs, num_msgs, sizeof(long), longcmp);
145 for (j=0; j<(num_msgs-1); ++j) {
146 if (msgs[j] == msgs[j+1]) {
147 memmove(&msgs[j], &msgs[j+1], ((num_msgs - j - 1)*sizeof(long)));
153 cdb_store(CDB_FULLTEXT, &tokens[i], sizeof(int),
154 msgs, (num_msgs*sizeof(long)) );
156 cdb_free(cdb_bucket);
158 /* FIXME unlock the file */
168 * Add a message to the list of those to be indexed.
170 void ft_index_msg(long msgnum, void *userdata) {
172 if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
174 if (ft_num_msgs > ft_num_alloc) {
175 ft_num_alloc += 1024;
176 ft_newmsgs = realloc(ft_newmsgs,
177 (ft_num_alloc * sizeof(long)));
179 ft_newmsgs[ft_num_msgs - 1] = msgnum;
185 * Scan a room for messages to index.
187 void ft_index_room(struct ctdlroom *qrbuf, void *data)
189 getroom(&CC->room, qrbuf->QRname);
190 CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, ft_index_msg, NULL);
195 * Begin the fulltext indexing process. (Called as an EVT_TIMER event)
197 void do_fulltext_indexing(void) {
199 static time_t last_index = 0L;
200 static time_t last_progress = 0L;
203 * Make sure we don't run the indexer too frequently.
204 * FIXME move the setting into config
206 if ( (time(NULL) - last_index) < 300L) {
211 * Check to see whether the fulltext index is up to date; if there
212 * are no messages to index, don't waste any more time trying.
214 if (CitControl.MMfulltext >= CitControl.MMhighest) {
215 return; /* nothing to do! */
218 lprintf(CTDL_DEBUG, "do_fulltext_indexing() started\n");
221 * If we've switched wordbreaker modules, burn the index and start
222 * over. FIXME write this...
224 if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
225 lprintf(CTDL_INFO, "(re)initializing full text index\n");
226 cdb_trunc(CDB_FULLTEXT);
227 CitControl.MMfulltext = 0L;
232 * Now go through each room and find messages to index.
234 ft_newhighest = CitControl.MMhighest;
235 ForEachRoom(ft_index_room, NULL); /* load all msg pointers */
237 if (ft_num_msgs > 0) {
238 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
239 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
240 if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
241 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
242 ((ft_num_msgs - i - 1)*sizeof(long)));
247 /* Here it is ... do each message! */
248 for (i=0; i<ft_num_msgs; ++i) {
249 if ((time(NULL) - last_progress) > 10) {
251 "Indexed %d of %d messages (%d%%)\n",
253 ((i*100) / ft_num_msgs)
255 last_progress = time(NULL);
257 ft_index_message(ft_newmsgs[i], 1);
266 /* Save our place so we don't have to do this again */
267 CitControl.MMfulltext = ft_newhighest;
268 CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
270 last_index = time(NULL);
272 lprintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
278 * API call to perform searches.
279 * (This one does the "all of these words" search.)
280 * Caller is responsible for freeing the message list.
282 void ft_search(int *fts_num_msgs, long **fts_msgs, char *search_string) {
286 struct cdbdata *cdb_bucket;
289 int num_all_msgs = 0;
290 long *all_msgs = NULL;
291 int num_ret_msgs = 0;
292 int num_ret_alloc = 0;
293 long *ret_msgs = NULL;
295 wordbreaker(search_string, &num_tokens, &tokens);
296 if (num_tokens > 0) {
297 for (i=0; i<num_tokens; ++i) {
299 /* search for tokens[i] */
300 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tokens[i],
302 if (cdb_bucket != NULL) {
303 num_msgs = cdb_bucket->len / sizeof(long);
304 msgs = (long *)cdb_bucket->ptr;
306 num_all_msgs += num_msgs;
307 if (num_all_msgs > 0) {
308 all_msgs = realloc(all_msgs,
309 num_all_msgs*sizeof(long) );
310 memcpy(&all_msgs[num_all_msgs-num_msgs],
311 msgs, num_msgs*sizeof(long) );
314 cdb_free(cdb_bucket);
319 qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
322 * At this point, if a message appears num_tokens times in the
323 * list, then it contains all of the search tokens.
325 if (num_all_msgs >= num_tokens)
326 for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
327 if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
330 if (num_ret_msgs > num_ret_alloc) {
332 ret_msgs = realloc(ret_msgs,
333 (num_ret_alloc*sizeof(long)) );
335 ret_msgs[num_ret_msgs - 1] = all_msgs[j];
343 *fts_num_msgs = num_ret_msgs;
344 *fts_msgs = ret_msgs;
349 * Tentative form of a search command
351 void cmd_srch(char *argbuf) {
355 char search_string[256];
357 if (CtdlAccessCheck(ac_logged_in)) return;
358 extract_token(search_string, argbuf, 0, '|', sizeof search_string);
359 ft_search(&num_msgs, &msgs, search_string);
361 cprintf("%d %d msgs match all search words:\n",
362 LISTING_FOLLOWS, num_msgs);
364 for (i=0; i<num_msgs; ++i) {
365 cprintf("%ld\n", msgs[i]);
368 if (msgs != NULL) free(msgs);
373 /*****************************************************************************/
375 char *serv_fulltext_init(void)
377 CtdlRegisterSessionHook(do_fulltext_indexing, EVT_TIMER);
378 CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");