#include <limits.h>
#include "citadel.h"
#include "server.h"
-#include "sysdep_decls.h"
#include "citserver.h"
#include "support.h"
#include "config.h"
-#include "serv_extensions.h"
#include "database.h"
#include "msgbase.h"
#include "control.h"
#include "ft_wordbreaker.h"
+#include "ctdl_module.h"
+
+
+
long ft_newhighest = 0L;
long *ft_newmsgs = NULL;
int ft_num_msgs = 0;
int ft_num_alloc = 0;
+int ftc_num_msgs[65536];
+long *ftc_msgs[65536];
+
+
/*
* Compare function
*/
return(0);
}
-
+/*
+ * Flush our index cache out to disk.
+ */
+void ft_flush_cache(void) {
+ int i;
+ time_t last_update = 0;
+
+ for (i=0; i<65536; ++i) {
+ if ((time(NULL) - last_update) >= 10) {
+ lprintf(CTDL_INFO,
+ "Flushing index cache to disk (%d%% complete)\n",
+ (i * 100 / 65536)
+ );
+ last_update = time(NULL);
+ }
+ if (ftc_msgs[i] != NULL) {
+ cdb_store(CDB_FULLTEXT, &i, sizeof(int), ftc_msgs[i],
+ (ftc_num_msgs[i] * sizeof(long)));
+ ftc_num_msgs[i] = 0;
+ free(ftc_msgs[i]);
+ ftc_msgs[i] = NULL;
+ }
+ }
+ lprintf(CTDL_INFO, "Flushed index cache to disk (100%% complete)\n");
+}
/*
* 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, j;
struct cdbdata *cdb_bucket;
- int num_msgs;
- long *msgs;
+ char *msgtext;
+ int tok;
lprintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
(op ? "adding" : "removing") , msgnum
);
- msg = CtdlFetchMessage(msgnum, 1);
- if (msg == NULL) return;
-
- if (msg->cm_fields['M'] != NULL) {
- wordbreaker(msg->cm_fields['M'], &num_tokens, &tokens);
- }
- CtdlFreeMessage(msg);
-
+ /* 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);
+ 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 [%d tokens]\n", msgnum, num_tokens);
if (num_tokens > 0) {
for (i=0; i<num_tokens; ++i) {
/* Add the message to the relevant token bucket */
- /* FIXME lock the file */
- 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 = NULL;
- num_msgs = 0;
- }
- else {
- num_msgs = cdb_bucket->len / sizeof(long);
- }
-
- 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 */
- /* FIXME do this */
- }
-
- /* sort and purge dups */
- if (num_msgs > 1) {
- qsort(msgs, num_msgs, sizeof(long), longcmp);
- for (j=0; j<(num_msgs-1); ++j) {
- if (msgs[j] == msgs[j+1]) {
- memmove(&msgs[j], &msgs[j+1],
- ((num_msgs - j)*sizeof(long)));
- --num_msgs;
+ /* search for tokens[i] */
+ tok = tokens[i];
+
+ if ( (tok >= 0) && (tok <= 65535) ) {
+ /* fetch the bucket, Liza */
+ if (ftc_msgs[tok] == NULL) {
+ cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
+ if (cdb_bucket != NULL) {
+ ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
+ ftc_msgs[tok] = (long *)cdb_bucket->ptr;
+ cdb_bucket->ptr = NULL;
+ cdb_free(cdb_bucket);
+ }
+ else {
+ ftc_num_msgs[tok] = 0;
+ ftc_msgs[tok] = malloc(sizeof(long));
+ }
+ }
+
+
+ if (op == 1) { /* add to index */
+ ++ftc_num_msgs[tok];
+ ftc_msgs[tok] = realloc(ftc_msgs[tok],
+ ftc_num_msgs[tok]*sizeof(long));
+ ftc_msgs[tok][ftc_num_msgs[tok] - 1] = msgnum;
+ }
+
+ if (op == 0) { /* remove from index */
+ if (ftc_num_msgs[tok] >= 1) {
+ for (j=0; j<ftc_num_msgs[tok]; ++j) {
+ if (ftc_msgs[tok][j] == msgnum) {
+ memmove(&ftc_msgs[tok][j], &ftc_msgs[tok][j+1], ((ftc_num_msgs[tok] - j - 1)*sizeof(long)));
+ --ftc_num_msgs[tok];
+ --j;
+ }
+ }
}
}
}
-
- cdb_store(CDB_FULLTEXT, &tokens[i], sizeof(int),
- msgs, (num_msgs*sizeof(long)) );
-
- cdb_free(cdb_bucket);
-
- /* FIXME unlock the file */
+ else {
+ lprintf(CTDL_ALERT, "Invalid token %d !!\n", tok);
+ }
}
free(tokens);
void ft_index_room(struct ctdlroom *qrbuf, void *data)
{
getroom(&CC->room, qrbuf->QRname);
- CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, ft_index_msg, NULL);
+ CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, NULL, ft_index_msg, NULL);
}
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.
*/
+ begin_critical_section(S_CONTROL);
+ lprintf(CTDL_DEBUG, "wb ver on disk = %d, code ver = %d\n",
+ CitControl.fulltext_wordbreaker, FT_WORDBREAKER_ID);
if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
lprintf(CTDL_INFO, "(re)initializing full text index\n");
cdb_trunc(CDB_FULLTEXT);
CitControl.MMfulltext = 0L;
put_control();
}
+ end_critical_section(S_CONTROL);
/*
* Now go through each room and find messages to index.
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) {
+ 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;
+ }
+
+ /* Check to see if we have to maybe flush to disk */
+ if (i >= FT_MAX_CACHE) {
+ lprintf(CTDL_DEBUG, "Time to flush.\n");
+ ft_newhighest = ft_newmsgs[i];
+ break;
+ }
+
}
free(ft_newmsgs);
}
/* Save our place so we don't have to do this again */
+ ft_flush_cache();
+ begin_critical_section(S_CONTROL);
CitControl.MMfulltext = ft_newhighest;
CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
put_control();
+ end_critical_section(S_CONTROL);
last_index = time(NULL);
lprintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
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);
+}
+
+
/*
* API call to perform searches.
int *tokens = NULL;
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;
+ int tok;
wordbreaker(search_string, &num_tokens, &tokens);
if (num_tokens > 0) {
for (i=0; i<num_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;
+ tok = tokens[i];
+
+ /* fetch the bucket, Liza */
+ if (ftc_msgs[tok] == NULL) {
+ cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
+ if (cdb_bucket != NULL) {
+ ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
+ ftc_msgs[tok] = (long *)cdb_bucket->ptr;
+ cdb_bucket->ptr = NULL;
+ cdb_free(cdb_bucket);
+ }
+ else {
+ ftc_num_msgs[tok] = 0;
+ ftc_msgs[tok] = malloc(sizeof(long));
+ }
+ }
- num_all_msgs += num_msgs;
+ num_all_msgs += ftc_num_msgs[tok];
+ 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);
+ memcpy(&all_msgs[num_all_msgs-ftc_num_msgs[tok]],
+ ftc_msgs[tok], ftc_num_msgs[tok]*sizeof(long) );
}
}
* 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 (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 = 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
+ * This search command is for diagnostic purposes and may be removed or replaced.
*/
void cmd_srch(char *argbuf) {
int num_msgs = 0;
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);
+ return;
+ }
+
extract_token(search_string, argbuf, 0, '|', sizeof search_string);
ft_search(&num_msgs, &msgs, search_string);
cprintf("000\n");
}
+/*
+ * Zero out our index cache.
+ */
+void initialize_ft_cache(void) {
+ memset(ftc_num_msgs, 0, (65536 * sizeof(int)));
+ memset(ftc_msgs, 0, (65536 * sizeof(long *)));
+}
+
/*****************************************************************************/
-char *serv_fulltext_init(void)
+CTDL_MODULE_INIT(fulltext)
{
- CtdlRegisterSessionHook(do_fulltext_indexing, EVT_TIMER);
- CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
+ initialize_ft_cache();
+ CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
+
+ /* return our Subversion id for the Log */
return "$Id$";
}
projects / citadel.git / blobdiff