serv_fulltext: don't try to index a null message

[citadel.git] / citadel / server / modules / fulltext / serv_fulltext.c

diff --git a/citadel/server/modules/fulltext/serv_fulltext.c b/citadel/server/modules/fulltext/serv_fulltext.c
index 7f7dd7959b2a984eb91eb41e1873fed92bcf45ab..959e32637e6d14faa4d2b56b9738036dc160650a 100644 (file)
--- a/citadel/server/modules/fulltext/serv_fulltext.c
+++ b/citadel/server/modules/fulltext/serv_fulltext.c
@@ -34,7 +34,6 @@
 #include "../../context.h"
 #include "../../ctdl_module.h"
 
-
 // These can be global variables because only one indexer runs at a time.
 Array *messages_to_be_indexed = NULL;
 long highest_msg_already_indexed = 0;
@@ -101,6 +100,10 @@ void ft_index_message(long msgnum, int op) {
                array_len(tokens_in_this_message)
        );
 
+       if (tokens_in_this_message == NULL) {
+               return;
+       }
+
        if (array_len(tokens_in_this_message) > 0) {
                begin_critical_section(S_INDEXER);
                cdb_begin_transaction();
@@ -189,15 +192,15 @@ void do_fulltext_indexing(void) {
        int i;
        static time_t last_progress = 0L;
        static int is_running = 0;
+
        if (is_running) return;         // Concurrency check - only one can run 
        is_running = 1;
 
        // Don't do this if the site doesn't have it enabled.
        if (!CtdlGetConfigInt("c_enable_fulltext")) {
+               is_running = 0;
                return;
        }
-       syslog(LOG_DEBUG, "fulltext: indexing started");
-
        // If we've switched wordbreaker modules, burn the index and start over.
        begin_critical_section(S_CONTROL);
        if (CtdlGetConfigInt("MM_fulltext_wordbreaker") != FT_WORDBREAKER_ID) {
@@ -212,11 +215,14 @@ void do_fulltext_indexing(void) {
 
        // Silently return if our fulltext index is up to date with new messages.
        if ((CtdlGetConfigLong("MMfulltext") >= CtdlGetConfigLong("MMhighest"))) {
+               is_running = 0;
                return;         // nothing to do!
        }
 
        highest_msg_already_indexed = CtdlGetConfigLong("MMfulltext");
        highest_msg_to_be_indexed = CtdlGetConfigLong("MMhighest");
+       syslog(LOG_DEBUG, "fulltext: indexing started. msgs %ld--%ld", highest_msg_already_indexed, highest_msg_to_be_indexed);
+
        messages_to_be_indexed = array_new(sizeof(long));
 
        // Now go through each room and find messages to index.
@@ -226,21 +232,32 @@ void do_fulltext_indexing(void) {
        // Here it is ... do each message!
        long msgnum = 0;
        long prev_msgnum = 0;
-       for (i=0; i<array_len(messages_to_be_indexed); ++i) {
+       time_t started_indexing_at = time(NULL);
+       int yielded = 0;
+       for (i=0; ((i<array_len(messages_to_be_indexed)) && (yielded == 0)); ++i) {
                memcpy(&msgnum, array_get_element_at(messages_to_be_indexed, i), sizeof(long));
 
                if (msgnum != prev_msgnum) {                            // careful to avoid dupes
                        ft_index_message(msgnum, 1);
                }
-
                prev_msgnum = msgnum;
+               
+               // If we run too long, yield the thread so the server can do other things.  We'll be back.
+               if (time(NULL) - started_indexing_at >= MAXIMUM_INDEXER_RUN_TIME) {
+                       yielded = 1;
+               }
        }
 
        array_free(messages_to_be_indexed);
-
        CtdlSetConfigInt("MM_fulltext_wordbreaker", FT_WORDBREAKER_ID);
+       syslog(LOG_DEBUG, "fulltext: indexer has run for %ld seconds; yielded=%d", time(NULL) - started_indexing_at, yielded);
+
+       // This keeps the indexer from starting up over and over if the highest message in the list was deleted
+       // before we had a chance to index it.
+       if (!yielded) {
+               CtdlSetConfigLong("MMfulltext", highest_msg_to_be_indexed);
+       }
 
-       syslog(LOG_DEBUG, "fulltext: indexing finished");
        is_running = 0;
        return;
 }
@@ -344,7 +361,7 @@ char *ctdl_module_init_fulltext(void) {
        if (!threading) {
                CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
                CtdlRegisterDeleteHook(ft_delete_remove);
-               CtdlRegisterSessionHook(do_fulltext_indexing, EVT_TIMER, PRIO_CLEANUP + 300);
+               CtdlRegisterSessionHook(do_fulltext_indexing, EVT_HOUSE, PRIO_CLEANUP + 300);
        }
        // return our module name for the log
        return "fulltext";