X-Git-Url: https://code.citadel.org/?a=blobdiff_plain;f=citadel%2Fmodules%2Frssclient%2Fserv_rssclient.c;h=11ffb905f138de66eb3b8525ab89b543155867a6;hb=c71df26b9d35c07eb9c1322a7e83caa86337ad82;hp=1b4c27ed5a47022e93ce257ceaef8ad32f138a68;hpb=f7aeebfe212b8f88e154f12a642d60efe4bc35a9;p=citadel.git diff --git a/citadel/modules/rssclient/serv_rssclient.c b/citadel/modules/rssclient/serv_rssclient.c index 1b4c27ed5..11ffb905f 100644 --- a/citadel/modules/rssclient/serv_rssclient.c +++ b/citadel/modules/rssclient/serv_rssclient.c @@ -1,7 +1,9 @@ /* - * Bring external RSS feeds into rooms. + * Bring external RSS and/or Atom feeds into rooms. This module implements a + * very loose parser that scrapes both kinds of feeds and is not picky about + * the standards compliance of the source data. * - * Copyright (c) 2007-2017 by the citadel.org team + * Copyright (c) 2007-2018 by the citadel.org team * * This program is open source software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 3. @@ -71,7 +73,6 @@ struct rssparser { }; time_t last_run = 0L; -struct CitContext rss_CC; struct rssurl *rsstodo = NULL; @@ -134,7 +135,7 @@ void rss_end_element(void *data, const char *el) // check the use table StrBuf *u = NewStrBuf(); StrBufAppendPrintf(u, "rss/%s", r->item_id); - time_t already_seen = CheckIfAlreadySeen(u, time(NULL), 0, eUpdate); + int already_seen = CheckIfAlreadySeen(u); FreeStrBuf(&u); if (already_seen == 0) { @@ -181,11 +182,11 @@ void rss_end_element(void *data, const char *el) else { CtdlSaveMsgPointerInRoom(rr->room, msgnum, 0, NULL); } - syslog(LOG_DEBUG, "Saved message %ld to %s", msgnum, rr->room); + syslog(LOG_DEBUG, "rssclient: saved message %ld to %s", msgnum, rr->room); } } else { - syslog(LOG_DEBUG, "%s was already seen %ld seconds ago", r->item_id, already_seen); + syslog(LOG_DEBUG, "rssclient: already seen %s", r->item_id); } CM_Free(r->msg); @@ -205,13 +206,20 @@ void rss_end_element(void *data, const char *el) } } - else if (!strcasecmp(el, "author")) { // author of item (rss and maybe atom) + else if (!strcasecmp(el, "creator")) { // can be used if is not present if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eAuthor))) { CM_SetField(r->msg, eAuthor, ChrPtr(r->CData), StrLength(r->CData)); striplt(r->msg->cm_fields[eAuthor]); } } + else if (!strcasecmp(el, "author")) { // supercedes if both are present + if (r->msg != NULL) { + CM_SetField(r->msg, eAuthor, ChrPtr(r->CData), StrLength(r->CData)); // CM_SetField will free() the previous value + striplt(r->msg->cm_fields[eAuthor]); + } + } + else if (!strcasecmp(el, "pubdate")) { // date/time stamp (rss) Sat, 25 Feb 2017 14:28:01 EST if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) { CM_SetFieldLONG(r->msg, eTimestamp, parsedate(ChrPtr(r->CData))); @@ -310,7 +318,7 @@ void rssclient_push_todo(char *rssurl, char *roomname) struct rssurl *thisone = NULL; struct rssroom *newroom = NULL; - syslog(LOG_DEBUG, "rssclient_push_todo(%s, %s)", rssurl, roomname); + syslog(LOG_DEBUG, "rssclient: will fetch %s to %s", rssurl, roomname); for (r=rsstodo; r!=NULL; r=r->next) { if (!strcasecmp(r->url, rssurl)) { @@ -332,17 +340,6 @@ void rssclient_push_todo(char *rssurl, char *roomname) } -// Callback function for curl -// -size_t rss_pof_write_data(void *buffer, size_t size, size_t nmemb, void *userp) -{ - StrBuf *Downloaded = (StrBuf *)userp; - size_t bytes = size * nmemb; - StrBufAppendBufPlain(Downloaded, buffer, bytes, 0); - return(bytes); -} - - // pull one feed (possibly multiple rooms) // void rss_pull_one_feed(struct rssurl *url) @@ -351,7 +348,7 @@ void rss_pull_one_feed(struct rssurl *url) CURLcode res; StrBuf *Downloaded = NULL; - syslog(LOG_DEBUG, "rss_pull_one_feed(%s)", url->url); + syslog(LOG_DEBUG, "rssclient: fetching %s", url->url); curl = curl_easy_init(); if (!curl) { @@ -361,13 +358,15 @@ void rss_pull_one_feed(struct rssurl *url) Downloaded = NewStrBuf(); curl_easy_setopt(curl, CURLOPT_URL, url->url); + curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0L); + curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0L); curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1L); // Follow redirects - curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_pof_write_data); // What to do with downloaded data + curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback); // What to do with downloaded data curl_easy_setopt(curl, CURLOPT_WRITEDATA, Downloaded); // Give it our StrBuf to work with curl_easy_setopt(curl, CURLOPT_TIMEOUT, 20L); // Time out after 20 seconds res = curl_easy_perform(curl); // Perform the request if (res != CURLE_OK) { - syslog(LOG_WARNING, "Failed to load feed: %s", curl_easy_strerror(res)); + syslog(LOG_WARNING, "rssclient: failed to load feed: %s", curl_easy_strerror(res)); } curl_easy_cleanup(curl); @@ -439,18 +438,17 @@ void rssclient_scan(void) { /* Run no more than once every 15 minutes. */ if ((now - last_run) < 900) { syslog(LOG_DEBUG, - "Client: polling interval not yet reached; last run was %ldm%lds ago", + "rssclient: polling interval not yet reached; last run was %ldm%lds ago", ((now - last_run) / 60), ((now - last_run) % 60) ); return; } - become_session(&rss_CC); - syslog(LOG_DEBUG, "rssclient started"); + syslog(LOG_DEBUG, "rssclient: started"); CtdlForEachRoom(rssclient_scan_room, NULL); rss_pull_feeds(); - syslog(LOG_DEBUG, "rssclient ended"); + syslog(LOG_DEBUG, "rssclient: ended"); last_run = time(NULL); return; } @@ -460,13 +458,9 @@ CTDL_MODULE_INIT(rssclient) { if (!threading) { - syslog(LOG_INFO, "%s", curl_version()); + syslog(LOG_INFO, "rssclient: using %s", curl_version()); CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300); } - else - { - CtdlFillSystemContext(&rss_CC, "rssclient"); - } return "rssclient"; }