int ft_num_alloc = 0;
+/*
+ * Compare function
+ */
+int longcmp(const void *rec1, const void *rec2) {
+ long i1, i2;
+
+ i1 = *(const long *)rec1;
+ i2 = *(const long *)rec2;
+
+ if (i1 > i2) return(1);
+ if (i1 < i2) return(-1);
+ return(0);
+}
+
+
+
+
/*
* Index or de-index a message. (op == 1 to index, 0 to de-index)
*/
void ft_index_message(long msgnum, int op) {
- struct CtdlMessage *msg;
int num_tokens = 0;
int *tokens = NULL;
- int i;
+ int i, j;
struct cdbdata *cdb_bucket;
int num_msgs;
long *msgs;
+ char *msgtext;
- msg = CtdlFetchMessage(msgnum, 1);
- if (msg == NULL) return;
-
- if (msg->cm_fields['M'] != NULL) {
- wordbreaker(msg->cm_fields['M'], &num_tokens, &tokens);
- }
- CtdlFreeMessage(msg);
+ lprintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
+ (op ? "adding" : "removing") , msgnum
+ );
+ /* Output the message as text before indexing it, so we don't end up
+ * indexing a bunch of encoded base64, etc.
+ */
+ CC->redirect_buffer = malloc(SIZ);
+ CC->redirect_len = 0;
+ CC->redirect_alloc = SIZ;
+ CtdlOutputMsg(msgnum, MT_CITADEL, HEADERS_ALL, 0, 1);
+ msgtext = CC->redirect_buffer;
+ CC->redirect_buffer = NULL;
+ CC->redirect_len = 0;
+ CC->redirect_alloc = 0;
+ lprintf(CTDL_DEBUG, "Wordbreaking message %ld...\n", msgnum);
+ wordbreaker(msgtext, &num_tokens, &tokens);
+ free(msgtext);
+
+ lprintf(CTDL_DEBUG, "Indexing message %ld...\n", msgnum);
if (num_tokens > 0) {
for (i=0; i<num_tokens; ++i) {
/* Add the message to the relevant token bucket */
- /* FIXME do "if op=1" ... */
-
- /* FIXME lock the file */
- cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tokens[i], sizeof(int));
+ /* FIXME do we need to lock the file here? */
+ cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tokens[i],
+ sizeof(int));
if (cdb_bucket == NULL) {
cdb_bucket = malloc(sizeof(struct cdbdata));
cdb_bucket->len = 0;
- cdb_bucket->ptr = malloc(sizeof(long));
+ cdb_bucket->ptr = NULL;
+ num_msgs = 0;
+ }
+ else {
+ num_msgs = cdb_bucket->len / sizeof(long);
}
- num_msgs = cdb_bucket->len / sizeof(long);
- ++num_msgs;
- cdb_bucket->ptr = realloc(cdb_bucket->ptr, num_msgs*sizeof(long) );
- msgs = (long *) cdb_bucket->ptr;
- msgs[num_msgs - 1] = msgnum;
+ if (op == 1) { /* add to index */
+ ++num_msgs;
+ cdb_bucket->ptr = realloc(cdb_bucket->ptr,
+ num_msgs*sizeof(long));
+ msgs = (long *) cdb_bucket->ptr;
+ msgs[num_msgs - 1] = msgnum;
+ }
+
+ if (op == 0) { /* remove from index */
+ if (num_msgs >= 1) {
+ msgs = (long *) cdb_bucket->ptr;
+ for (j=0; j<num_msgs; ++j) {
+ if (msgs[j] == msgnum) {
+ memmove(&msgs[j], &msgs[j+1], ((num_msgs - j - 1)*sizeof(long)));
+ --num_msgs;
+ --j;
+ }
+ }
+ }
+ }
cdb_store(CDB_FULLTEXT, &tokens[i], sizeof(int),
- cdb_bucket->ptr, num_msgs*sizeof(long) );
+ msgs, (num_msgs*sizeof(long)) );
cdb_free(cdb_bucket);
- /* FIXME unlock the file */
+ /* FIXME do we need to unlock the file here? */
}
free(tokens);
}
-/*
- * Compare function
- */
-int longcmp(const void *rec1, const void *rec2) {
- long i1, i2;
-
- i1 = *(const long *)rec1;
- i2 = *(const long *)rec2;
-
- if (i1 > i2) return(1);
- if (i1 < i2) return(-1);
- return(0);
-}
-
-
/*
* Begin the fulltext indexing process. (Called as an EVT_TIMER event)
*/
void do_fulltext_indexing(void) {
int i;
static time_t last_index = 0L;
+ static time_t last_progress = 0L;
+
+ /*
+ * Don't do this if the site doesn't have it enabled.
+ */
+ if (!config.c_enable_fulltext) {
+ return;
+ }
/*
* Make sure we don't run the indexer too frequently.
* Check to see whether the fulltext index is up to date; if there
* are no messages to index, don't waste any more time trying.
*/
- lprintf(CTDL_DEBUG, "CitControl.MMhighest = %ld\n",
- CitControl.MMhighest);
- lprintf(CTDL_DEBUG, "CitControl.MMfulltext = %ld\n",
- CitControl.MMfulltext);
if (CitControl.MMfulltext >= CitControl.MMhighest) {
- /* nothing to do! */
- return;
+ return; /* nothing to do! */
}
lprintf(CTDL_DEBUG, "do_fulltext_indexing() started\n");
/*
* If we've switched wordbreaker modules, burn the index and start
- * over. FIXME write this...
+ * over.
*/
if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
lprintf(CTDL_INFO, "(re)initializing full text index\n");
if (ft_num_msgs > 0) {
qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
- if (i>1) for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
+ for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
- ((ft_num_msgs - i)*sizeof(long)));
+ ((ft_num_msgs - i - 1)*sizeof(long)));
--ft_num_msgs;
+ --i;
}
}
/* Here it is ... do each message! */
for (i=0; i<ft_num_msgs; ++i) {
+ if ((time(NULL) - last_progress) > 10) {
+ lprintf(CTDL_DEBUG,
+ "Indexed %d of %d messages (%d%%)\n",
+ i, ft_num_msgs,
+ ((i*100) / ft_num_msgs)
+ );
+ last_progress = time(NULL);
+ }
ft_index_message(ft_newmsgs[i], 1);
+
+ /* Check to see if we need to quit early */
+ if (time_to_die) {
+ lprintf(CTDL_DEBUG, "Indexer quitting early\n");
+ ft_newhighest = ft_newmsgs[i];
+ break;
+ }
+
}
free(ft_newmsgs);
return;
}
+/*
+ * Main loop for the indexer thread.
+ */
+void *indexer_thread(void *arg) {
+ struct CitContext indexerCC;
+
+ lprintf(CTDL_DEBUG, "indexer_thread() initializing\n");
+
+ memset(&indexerCC, 0, sizeof(struct CitContext));
+ indexerCC.internal_pgm = 1;
+ indexerCC.cs_pid = 0;
+ pthread_setspecific(MyConKey, (void *)&indexerCC );
+
+ cdb_allocate_tsd();
+
+ while (!time_to_die) {
+ do_fulltext_indexing();
+ sleep(1);
+ }
+
+ lprintf(CTDL_DEBUG, "indexer_thread() exiting\n");
+ pthread_exit(NULL);
+}
+
+
/*
- * Tentative form of our search command
+ * API call to perform searches.
+ * (This one does the "all of these words" search.)
+ * Caller is responsible for freeing the message list.
*/
-void cmd_srch(char *argbuf) {
- char search_string[256];
+void ft_search(int *fts_num_msgs, long **fts_msgs, char *search_string) {
int num_tokens = 0;
int *tokens = NULL;
- int i;
+ int i, j;
+ struct cdbdata *cdb_bucket;
+ int num_msgs;
+ long *msgs;
+ int num_all_msgs = 0;
+ long *all_msgs = NULL;
+ int num_ret_msgs = 0;
+ int num_ret_alloc = 0;
+ long *ret_msgs = NULL;
- if (CtdlAccessCheck(ac_logged_in)) return;
- extract_token(search_string, argbuf, 0, '|', sizeof search_string);
wordbreaker(search_string, &num_tokens, &tokens);
-
- cprintf("%d msgs matching search words:\n", LISTING_FOLLOWS);
if (num_tokens > 0) {
for (i=0; i<num_tokens; ++i) {
- cprintf("FIXME search for token %d\n", tokens[i]);
+ /* search for tokens[i] */
+ cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tokens[i],
+ sizeof(int));
+ if (cdb_bucket != NULL) {
+ num_msgs = cdb_bucket->len / sizeof(long);
+ msgs = (long *)cdb_bucket->ptr;
+
+ num_all_msgs += num_msgs;
+ if (num_all_msgs > 0) {
+ all_msgs = realloc(all_msgs,
+ num_all_msgs*sizeof(long) );
+ memcpy(&all_msgs[num_all_msgs-num_msgs],
+ msgs, num_msgs*sizeof(long) );
+ }
+
+ cdb_free(cdb_bucket);
+ }
}
free(tokens);
+ qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
+
+ /*
+ * At this point, if a message appears num_tokens times in the
+ * list, then it contains all of the search tokens.
+ */
+ if (num_all_msgs >= num_tokens)
+ for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
+ if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
+
+ ++num_ret_msgs;
+ if (num_ret_msgs > num_ret_alloc) {
+ num_ret_alloc += 64;
+ ret_msgs = realloc(ret_msgs,
+ (num_ret_alloc*sizeof(long)) );
+ }
+ ret_msgs[num_ret_msgs - 1] = all_msgs[j];
+
+ }
+ }
+
+ free(all_msgs);
+ }
+
+ *fts_num_msgs = num_ret_msgs;
+ *fts_msgs = ret_msgs;
+}
+
+
+/*
+ * Tentative form of a search command
+ */
+void cmd_srch(char *argbuf) {
+ int num_msgs = 0;
+ long *msgs = NULL;
+ int i;
+ char search_string[256];
+
+ if (CtdlAccessCheck(ac_logged_in)) return;
+
+ if (!config.c_enable_fulltext) {
+ cprintf("%d Full text index is not enabled on this server.\n",
+ ERROR + CMD_NOT_SUPPORTED);
+ }
+
+ extract_token(search_string, argbuf, 0, '|', sizeof search_string);
+ ft_search(&num_msgs, &msgs, search_string);
+
+ cprintf("%d %d msgs match all search words:\n",
+ LISTING_FOLLOWS, num_msgs);
+ if (num_msgs > 0) {
+ for (i=0; i<num_msgs; ++i) {
+ cprintf("%ld\n", msgs[i]);
+ }
}
+ if (msgs != NULL) free(msgs);
cprintf("000\n");
}
char *serv_fulltext_init(void)
{
- CtdlRegisterSessionHook(do_fulltext_indexing, EVT_TIMER);
- CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
+ CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
return "$Id$";
}