Hunting a bug in scheduled threads that caused a segflt. Got it and

[citadel.git] / citadel / modules / rssclient / serv_rssclient.c

diff --git a/citadel/modules/rssclient/serv_rssclient.c b/citadel/modules/rssclient/serv_rssclient.c
index 80fe3908c9f4e9501d2973d5fd79aaeda5d4600b..e27f888d41a53153b5707bc6ac581a4cfe6eeff2 100644 (file)
--- a/citadel/modules/rssclient/serv_rssclient.c
+++ b/citadel/modules/rssclient/serv_rssclient.c
@@ -25,12 +25,12 @@
 #include <errno.h>
 #include <sys/types.h>
 #include <sys/stat.h>
+#include <libcitadel.h>
 #include "citadel.h"
 #include "server.h"
 #include "citserver.h"
 #include "support.h"
 #include "config.h"
-#include "tools.h"
 #include "room_ops.h"
 #include "ctdl_module.h"
 #include "clientsocket.h"
@@ -59,6 +59,8 @@ struct rss_item {
        char *link;
        char *description;
        time_t pubdate;
+       char channel_title[256];
+       int item_tag_nesting;
 };
 
 struct rssnetcfg *rnclist = NULL;
@@ -124,6 +126,7 @@ void rss_save_item(struct rss_item *ri) {
        else {
                /* Item has not been seen, so save it. */
 
+               if (ri->description == NULL) ri->description = strdup("");
                for (i=strlen(ri->description); i>=0; --i) {
                        if (isspace(ri->description[i])) {
                                ri->description[i] = ' ';
@@ -140,6 +143,9 @@ void rss_save_item(struct rss_item *ri) {
                msg->cm_fields['U'] = strdup(ri->title);
                msg->cm_fields['T'] = malloc(64);
                snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
+               if (!IsEmptyStr(ri->channel_title)) {
+                       msg->cm_fields['O'] = strdup(ri->channel_title);
+               }
 
                msglen = 1024 + strlen(ri->link) + strlen(ri->description) ;
                msg->cm_fields['M'] = malloc(msglen);
@@ -156,13 +162,13 @@ void rss_save_item(struct rss_item *ri) {
 
                CtdlSubmitMsg(msg, recp, NULL);
                CtdlFreeMessage(msg);
-               free_recipients(recp);
 
                /* write the uidl to the use table so we don't store this item again */
                strcpy(ut.ut_msgid, utmsgid);
                ut.ut_timestamp = time(NULL);
                cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
        }
+       free_recipients(recp);
 }
 
 
@@ -200,10 +206,19 @@ time_t rdf_parsedate(char *p)
 
 
 
-void rss_xml_start(void *data, const char *el, const char **attr) {
+void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
        struct rss_item *ri = (struct rss_item *) data;
+       char el[256];
+       char *sep = NULL;
+
+       /* Axe the namespace, we don't care about it */
+       safestrncpy(el, supplied_el, sizeof el);
+       while (sep = strchr(el, ':'), sep) {
+               strcpy(el, ++sep);
+       }
 
        if (!strcasecmp(el, "item")) {
+               ++ri->item_tag_nesting;
 
                /* Initialize the feed item data structure */
                if (ri->guid != NULL) free(ri->guid);
@@ -232,43 +247,47 @@ void rss_xml_end(void *data, const char *supplied_el) {
        char el[256];
        char *sep = NULL;
 
-
        /* Axe the namespace, we don't care about it */
-
        safestrncpy(el, supplied_el, sizeof el);
        while (sep = strchr(el, ':'), sep) {
                strcpy(el, ++sep);
        }
 
-       if (!strcasecmp(el, "guid")) {
+       if ( (!strcasecmp(el, "title")) && (ri->item_tag_nesting == 0) && (ri->chardata != NULL) ) {
+               safestrncpy(ri->channel_title, ri->chardata, sizeof ri->channel_title);
+               striplt(ri->channel_title);
+       }
+
+       if ( (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) {
                if (ri->guid != NULL) free(ri->guid);
                striplt(ri->chardata);
                ri->guid = strdup(ri->chardata);
        }
 
-       if (!strcasecmp(el, "title")) {
+       if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) {
                if (ri->title != NULL) free(ri->title);
                striplt(ri->chardata);
                ri->title = strdup(ri->chardata);
        }
 
-       if (!strcasecmp(el, "link")) {
+       if ( (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) {
                if (ri->link != NULL) free(ri->link);
                striplt(ri->chardata);
                ri->link = strdup(ri->chardata);
        }
 
-       if (!strcasecmp(el, "description")) {
+       if ( (!strcasecmp(el, "description")) && (ri->chardata != NULL) ) {
                if (ri->description != NULL) free(ri->description);
                ri->description = strdup(ri->chardata);
        }
 
-       if ( (!strcasecmp(el, "pubdate")) || (!strcasecmp(el, "date")) ) {
+       if ( ((!strcasecmp(el, "pubdate")) || (!strcasecmp(el, "date"))) && (ri->chardata != NULL) ) {
                striplt(ri->chardata);
                ri->pubdate = rdf_parsedate(ri->chardata);
        }
 
        if (!strcasecmp(el, "item")) {
+               --ri->item_tag_nesting;
                rss_save_item(ri);
        }
 
@@ -408,6 +427,14 @@ retry:     lprintf(CTDL_NOTICE, "Connecting to <%s>\n", rsshost);
                lprintf(CTDL_DEBUG, "<%s\n", buf);
                sock_puts(sock, buf);
 
+               snprintf(buf, sizeof buf, "User-Agent: %s", CITADEL);
+               lprintf(CTDL_DEBUG, "<%s\n", buf);
+               sock_puts(sock, buf);
+
+               snprintf(buf, sizeof buf, "Accept: */*");
+               lprintf(CTDL_DEBUG, "<%s\n", buf);
+               sock_puts(sock, buf);
+
                sock_puts(sock, "");
 
                if (sock_getln(sock, buf, sizeof buf) >= 0) {
@@ -501,6 +528,9 @@ void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
 
                extract_token(instr, buf, 0, '|', sizeof instr);
                if (!strcasecmp(instr, "rssclient")) {
+
+                       use_this_rncptr = NULL;
+
                        extract_token(feedurl, buf, 1, '|', sizeof feedurl);
 
                        /* If any other rooms have requested the same feed, then we will just add this
@@ -550,17 +580,18 @@ void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
 /*
  * Scan for rooms that have RSS client requests configured
  */
-void rssclient_scan(void) {
+void *rssclient_scan(void *args) {
        static time_t last_run = 0L;
        static int doing_rssclient = 0;
        struct rssnetcfg *rptr = NULL;
 
+       CtdlThreadAllocTSD();
        /*
         * Run RSS client no more frequently than once every n seconds
         */
-       if ( (time(NULL) - last_run) < config.c_net_freq ) {
-               return;
-       }
+//     if ( (time(NULL) - last_run) < config.c_net_freq ) {
+//             return;
+//     }
 
        /*
         * This is a simple concurrency check to make sure only one rssclient run
@@ -568,7 +599,7 @@ void rssclient_scan(void) {
         * don't really require extremely fine granularity here, we'll do it
         * with a static variable instead.
         */
-       if (doing_rssclient) return;
+       if (doing_rssclient) return NULL;
        doing_rssclient = 1;
 
        lprintf(CTDL_DEBUG, "rssclient started\n");
@@ -585,6 +616,8 @@ void rssclient_scan(void) {
        lprintf(CTDL_DEBUG, "rssclient ended\n");
        last_run = time(NULL);
        doing_rssclient = 0;
+       CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
+       return NULL;
 }
 
 
@@ -592,11 +625,16 @@ void rssclient_scan(void) {
 
 CTDL_MODULE_INIT(rssclient)
 {
+       if (threading)
+       {
 #ifdef HAVE_EXPAT
-       CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
+//             CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
+               CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
 #else
-        lprintf(CTDL_INFO, "This server is missing the Expat XML parser.  RSS client will be disabled.\n");
+               lprintf(CTDL_INFO, "This server is missing the Expat XML parser.  RSS client will be disabled.\n");
 #endif
+       }
+       
        /* return our Subversion id for the Log */
         return "$Id: serv_rssclient.c 5652 2007-10-29 20:14:48Z ajc $";
 }