/*
- * $Id$
- *
* This module handles fulltext indexing of the message base.
- *
- *
* Copyright (c) 2005-2009 by the citadel.org team
*
* This program is free software; you can redistribute it and/or modify
for (i=0; i<65536; ++i) {
if ((time(NULL) - last_update) >= 10) {
- CtdlLogPrintf(CTDL_INFO,
+ syslog(LOG_INFO,
"Flushing index cache to disk (%d%% complete)\n",
(i * 100 / 65536)
);
ftc_msgs[i] = NULL;
}
}
- CtdlLogPrintf(CTDL_INFO, "Flushed index cache to disk (100%% complete)\n");
+ syslog(LOG_INFO, "Flushed index cache to disk (100%% complete)\n");
}
int *tokens = NULL;
int i, j;
struct cdbdata *cdb_bucket;
- char *msgtext;
+ StrBuf *msgtext;
+ char *txt;
int tok;
+ struct CtdlMessage *msg = NULL;
+
+ msg = CtdlFetchMessage(msgnum, 1);
+ if (msg == NULL) {
+ syslog(LOG_ERR, "ft_index_message() could not load msg %ld\n", msgnum);
+ return;
+ }
+
+ if (msg->cm_fields['1'] != NULL) {
+ syslog(LOG_DEBUG, "ft_index_message() excluded msg %ld\n", msgnum);
+ CtdlFreeMessage(msg);
+ return;
+ }
- CtdlLogPrintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
+ syslog(LOG_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, NULL, 0);
+ CC->redirect_buffer = NewStrBufPlain(NULL, SIZ);
+ CtdlOutputPreLoadedMsg(msg, MT_CITADEL, HEADERS_ALL, 0, 1, 0);
+ CtdlFreeMessage(msg);
msgtext = CC->redirect_buffer;
CC->redirect_buffer = NULL;
- CC->redirect_len = 0;
- CC->redirect_alloc = 0;
- CtdlLogPrintf(CTDL_DEBUG, "Wordbreaking message %ld...\n", msgnum);
- wordbreaker(msgtext, &num_tokens, &tokens);
- free(msgtext);
+ syslog(LOG_DEBUG, "Wordbreaking message %ld...\n", msgnum);
+ txt = SmashStrBuf(&msgtext);
+ wordbreaker(txt, &num_tokens, &tokens);
+ free(txt);
- CtdlLogPrintf(CTDL_DEBUG, "Indexing message %ld [%d tokens]\n", msgnum, num_tokens);
+ syslog(LOG_DEBUG, "Indexing message %ld [%d tokens]\n", msgnum, num_tokens);
if (num_tokens > 0) {
for (i=0; i<num_tokens; ++i) {
}
}
else {
- CtdlLogPrintf(CTDL_ALERT, "Invalid token %d !!\n", tok);
+ syslog(LOG_ALERT, "Invalid token %d !!\n", tok);
}
}
}
run_time = time(NULL);
- CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() started (%ld)\n", run_time);
+ syslog(LOG_DEBUG, "do_fulltext_indexing() started (%ld)\n", run_time);
/*
* If we've switched wordbreaker modules, burn the index and start
*/
begin_critical_section(S_CONTROL);
if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
- CtdlLogPrintf(CTDL_DEBUG, "wb ver on disk = %d, code ver = %d\n",
+ syslog(LOG_DEBUG, "wb ver on disk = %d, code ver = %d\n",
CitControl.fulltext_wordbreaker, FT_WORDBREAKER_ID);
- CtdlLogPrintf(CTDL_INFO, "(re)initializing full text index\n");
+ syslog(LOG_INFO, "(re)initializing full text index\n");
cdb_trunc(CDB_FULLTEXT);
CitControl.MMfulltext = 0L;
put_control();
/* Here it is ... do each message! */
for (i=0; i<ft_num_msgs; ++i) {
if (time(NULL) != last_progress) {
- CtdlLogPrintf(CTDL_DEBUG,
+ syslog(LOG_DEBUG,
"Indexed %d of %d messages (%d%%)\n",
i, ft_num_msgs,
((i*100) / ft_num_msgs)
/* Check to see if we need to quit early */
if (CtdlThreadCheckStop()) {
- CtdlLogPrintf(CTDL_DEBUG, "Indexer quitting early\n");
+ syslog(LOG_DEBUG, "Indexer quitting early\n");
ft_newhighest = ft_newmsgs[i];
break;
}
/* Check to see if we have to maybe flush to disk */
if (i >= FT_MAX_CACHE) {
- CtdlLogPrintf(CTDL_DEBUG, "Time to flush.\n");
+ syslog(LOG_DEBUG, "Time to flush.\n");
ft_newhighest = ft_newmsgs[i];
break;
}
if (CtdlThreadCheckStop())
return;
- CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() duration (%ld)\n", end_time - run_time);
+ syslog(LOG_DEBUG, "do_fulltext_indexing() duration (%ld)\n", end_time - run_time);
/* Save our place so we don't have to do this again */
ft_flush_cache();
end_critical_section(S_CONTROL);
last_index = time(NULL);
- CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
+ syslog(LOG_DEBUG, "do_fulltext_indexing() finished\n");
return;
}
void *indexer_thread(void *arg) {
struct CitContext indexerCC;
- CtdlLogPrintf(CTDL_DEBUG, "indexer_thread() initializing\n");
CtdlFillSystemContext(&indexerCC, "indexer");
citthread_setspecific(MyConKey, (void *)&indexerCC );
+ syslog(LOG_DEBUG, "indexer_thread() initializing\n");
while (!CtdlThreadCheckStop()) {
do_fulltext_indexing();
CtdlThreadSleep(300);
}
- CtdlLogPrintf(CTDL_DEBUG, "indexer_thread() exiting\n");
+ syslog(LOG_DEBUG, "indexer_thread() exiting\n");
+ CtdlClearSystemContext();
return NULL;
}
* (This one does the "all of these words" search.)
* Caller is responsible for freeing the message list.
*/
-void ft_search(int *fts_num_msgs, long **fts_msgs, char *search_string) {
+void ft_search(int *fts_num_msgs, long **fts_msgs, const char *search_string) {
int num_tokens = 0;
int *tokens = NULL;
int i, j;
CtdlThreadCreate("Indexer", CTDLTHREAD_BIGSTACK, indexer_thread, NULL);
}
/* return our Subversion id for the Log */
- return "$Id$";
+ return "fulltext";
}