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 msg = CtdlFetchMessage(msgnum, 1);
85 if (msg == NULL) return;
87 if (msg->cm_fields['M'] != NULL) {
88 wordbreaker(msg->cm_fields['M'], &num_tokens, &tokens);
93 for (i=0; i<num_tokens; ++i) {
95 /* Add the message to the relevant token bucket */
97 /* FIXME do "if op=1" ... */
99 /* FIXME lock the file */
100 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tokens[i], sizeof(int));
101 if (cdb_bucket == NULL) {
102 cdb_bucket = malloc(sizeof(struct cdbdata));
104 cdb_bucket->ptr = NULL;
108 num_msgs = cdb_bucket->len / sizeof(long);
112 cdb_bucket->ptr = realloc(cdb_bucket->ptr, num_msgs*sizeof(long) );
113 msgs = (long *) cdb_bucket->ptr;
114 msgs[num_msgs - 1] = msgnum;
116 /* lprintf(CTDL_DEBUG, "bucket <%5d> position <%2d> msg <%ld>\n",
117 tokens[i], num_msgs-1, msgnum); */
119 /* sort and purge dups */
121 qsort(msgs, num_msgs, sizeof(long), longcmp);
122 for (j=0; j<(num_msgs-1); ++j) {
123 if (msgs[j] == msgs[j+1]) {
124 memmove(&msgs[j], &msgs[j+1],
125 ((num_msgs - j)*sizeof(long)));
131 cdb_store(CDB_FULLTEXT, &tokens[i], sizeof(int),
132 msgs, (num_msgs*sizeof(long)) );
134 cdb_free(cdb_bucket);
136 /* FIXME unlock the file */
146 * Add a message to the list of those to be indexed.
148 void ft_index_msg(long msgnum, void *userdata) {
150 if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
152 if (ft_num_msgs > ft_num_alloc) {
153 ft_num_alloc += 1024;
154 ft_newmsgs = realloc(ft_newmsgs,
155 (ft_num_alloc * sizeof(long)));
157 ft_newmsgs[ft_num_msgs - 1] = msgnum;
163 * Scan a room for messages to index.
165 void ft_index_room(struct ctdlroom *qrbuf, void *data)
167 getroom(&CC->room, qrbuf->QRname);
168 CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, ft_index_msg, NULL);
173 * Begin the fulltext indexing process. (Called as an EVT_TIMER event)
175 void do_fulltext_indexing(void) {
177 static time_t last_index = 0L;
180 * Make sure we don't run the indexer too frequently.
181 * FIXME move the setting into config
183 if ( (time(NULL) - last_index) < 300L) {
188 * Check to see whether the fulltext index is up to date; if there
189 * are no messages to index, don't waste any more time trying.
191 lprintf(CTDL_DEBUG, "CitControl.MMhighest = %ld\n", CitControl.MMhighest);
192 lprintf(CTDL_DEBUG, "CitControl.MMfulltext = %ld\n", CitControl.MMfulltext);
193 if (CitControl.MMfulltext >= CitControl.MMhighest) {
198 lprintf(CTDL_DEBUG, "do_fulltext_indexing() started\n");
201 * If we've switched wordbreaker modules, burn the index and start
202 * over. FIXME write this...
204 if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
205 lprintf(CTDL_INFO, "(re)initializing full text index\n");
206 cdb_trunc(CDB_FULLTEXT);
207 CitControl.MMfulltext = 0L;
212 * Now go through each room and find messages to index.
214 ft_newhighest = CitControl.MMhighest;
215 ForEachRoom(ft_index_room, NULL); /* load all msg pointers */
217 if (ft_num_msgs > 0) {
218 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
219 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
220 if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
221 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
222 ((ft_num_msgs - i)*sizeof(long)));
227 /* Here it is ... do each message! */
228 for (i=0; i<ft_num_msgs; ++i) {
229 ft_index_message(ft_newmsgs[i], 1);
238 /* Save our place so we don't have to do this again */
239 CitControl.MMfulltext = ft_newhighest;
240 CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
242 last_index = time(NULL);
244 lprintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
250 * Tentative form of our search command
252 void cmd_srch(char *argbuf) {
253 char search_string[256];
257 struct cdbdata *cdb_bucket;
260 int num_all_msgs = 0;
261 long *all_msgs = NULL;
263 if (CtdlAccessCheck(ac_logged_in)) return;
264 extract_token(search_string, argbuf, 0, '|', sizeof search_string);
265 wordbreaker(search_string, &num_tokens, &tokens);
267 cprintf("%d msgs matching search words:\n", LISTING_FOLLOWS);
268 if (num_tokens > 0) {
269 for (i=0; i<num_tokens; ++i) {
271 /* search for tokens[i] */
272 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tokens[i], sizeof(int));
273 if (cdb_bucket != NULL) {
274 num_msgs = cdb_bucket->len / sizeof(long);
275 msgs = (long *)cdb_bucket->ptr;
277 num_all_msgs += num_msgs;
278 all_msgs = realloc(all_msgs, num_all_msgs*sizeof(long) );
279 memcpy(&all_msgs[num_all_msgs - num_msgs], msgs, num_msgs*sizeof(long) );
281 cdb_free(cdb_bucket);
286 qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
289 * At this point, if a message appears num_tokens times in the
290 * list, then it contains all of the search tokens.
292 if (num_all_msgs >= num_tokens) for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
293 if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
294 cprintf("%ld\n", all_msgs[j]);
304 /*****************************************************************************/
306 char *serv_fulltext_init(void)
308 CtdlRegisterSessionHook(do_fulltext_indexing, EVT_TIMER);
309 CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");