/*
- * 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) 2007-2020 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.
+ * 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.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.See the
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*/
#include "parsedate.h"
#include "database.h"
#include "citadel_dirs.h"
-#include "md5.h"
#include "context.h"
+#include "internet_addressing.h"
struct rssroom {
struct rssroom *next;
struct rssparser {
StrBuf *CData;
struct CtdlMessage *msg;
+ char *link;
+ char *description;
+ char *item_id;
+ struct rssroom *rooms;
};
time_t last_run = 0L;
-struct CitContext rss_CC;
struct rssurl *rsstodo = NULL;
-// This is what RSS probably looks like
-//
-// <item>
-// <title><![CDATA[Felicity flexes action chops]]></title>
-// <link>http://video.foxnews.com/v/5336254459001/</link>
-// <author>foxnewsonline@foxnews.com (Fox News Online)</author>
-// <description />
-// <pubDate>Sat, 25 Feb 2017 14:28:01 EST</pubDate>
-// </item>
-
-
// This handler is called whenever an XML tag opens.
//
void rss_start_element(void *data, const char *el, const char **attribute)
{
struct rssparser *r = (struct rssparser *)data;
+ int i;
if (
(!strcasecmp(el, "entry"))
r->msg->cm_anon_type = MES_NORMAL;
r->msg->cm_format_type = FMT_RFC822;
}
+
+ else if (!strcasecmp(el, "link")) { // atom feeds have the link as an attribute
+ for(i = 0; attribute[i]; i += 2) {
+ if (!strcasecmp(attribute[i], "href")) {
+ if (r->link != NULL) {
+ free(r->link);
+ r->link = NULL;
+ }
+ r->link = strdup(attribute[i+1]);
+ striplt(r->link);
+ }
+ }
+ }
}
void rss_end_element(void *data, const char *el)
{
struct rssparser *r = (struct rssparser *)data;
+ StrBuf *encoded_field;
+
+ if (StrLength(r->CData) > 0) { // strip leading/trailing whitespace from field
+ StrBufTrim(r->CData);
+ }
if ( // end of a new item(rss) or entry(atom)
(!strcasecmp(el, "entry"))
|| (!strcasecmp(el, "item"))
) {
+ if (r->msg != NULL) { // Save the message to the rooms
- // FIXME check the use table
+ // use the link as an item id if nothing else is available
+ if ((r->item_id == NULL) && (r->link != NULL)) {
+ r->item_id = strdup(r->link);
+ }
- if (r->msg != NULL) {
- // FIXME WRITE IT TO THE ROOMS HERE, DUMMEH
+ // check the use table
+ StrBuf *u = NewStrBuf();
+ StrBufAppendPrintf(u, "rss/%s", r->item_id);
+ int already_seen = CheckIfAlreadySeen(u);
+ FreeStrBuf(&u);
+
+ if (already_seen == 0) {
+
+ // Compose the message text
+ StrBuf *TheMessage = NewStrBuf();
+ StrBufAppendPrintf(TheMessage,
+ "Content-type: text/html\n\n"
+ "\n\n"
+ "<html><head></head><body>"
+ );
+
+ if (r->description != NULL) {
+ StrBufAppendPrintf(TheMessage, "%s<br><br>\r\n", r->description);
+ free(r->description);
+ r->description = NULL;
+ }
+
+ if (r->link != NULL) {
+ StrBufAppendPrintf(TheMessage, "<a href=\"%s\">%s</a>\r\n", r->link, r->link);
+ free(r->link);
+ r->link = NULL;
+ }
+
+ StrBufAppendPrintf(TheMessage, "</body></html>\r\n");
+ CM_SetField(r->msg, eMesageText, ChrPtr(TheMessage), StrLength(TheMessage));
+ FreeStrBuf(&TheMessage);
+
+ if (CM_IsEmpty(r->msg, eAuthor)) {
+ CM_SetField(r->msg, eAuthor, HKEY("rss"));
+ }
+
+ if (CM_IsEmpty(r->msg, eTimestamp)) {
+ CM_SetFieldLONG(r->msg, eTimestamp, time(NULL));
+ }
+
+ // Save it to the room(s)
+ struct rssroom *rr = NULL;
+ long msgnum = (-1);
+ for (rr=r->rooms; rr!=NULL; rr=rr->next) {
+ if (rr == r->rooms) {
+ msgnum = CtdlSubmitMsg(r->msg, NULL, rr->room); // in first room, save msg
+ }
+ else {
+ CtdlSaveMsgPointerInRoom(rr->room, msgnum, 0, NULL); // elsewhere, save a pointer
+ }
+ syslog(LOG_DEBUG, "rssclient: saved message %ld to %s", msgnum, rr->room);
+ }
+ }
+ else {
+ syslog(LOG_DEBUG, "rssclient: already seen %s", r->item_id);
+ }
+
CM_Free(r->msg);
r->msg = NULL;
}
+
+ if (r->item_id != NULL) {
+ free(r->item_id);
+ r->item_id = NULL;
+ }
}
else if (!strcasecmp(el, "title")) { // item subject (rss and atom)
- if ((r->msg != NULL) && (r->msg->cm_fields[eMsgSubject] == NULL)) {
- r->msg->cm_fields[eMsgSubject] = strdup(ChrPtr(r->CData));
+ if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eMsgSubject))) {
+ encoded_field = NewStrBuf();
+ StrBufRFC2047encode(&encoded_field, r->CData);
+ CM_SetAsFieldSB(r->msg, eMsgSubject, &encoded_field);
}
}
- else if (!strcasecmp(el, "author")) { // author of item (rss and maybe atom)
- if ((r->msg != NULL) && (r->msg->cm_fields[eAuthor] == NULL)) {
- r->msg->cm_fields[eAuthor] = strdup(ChrPtr(r->CData));
+ else if (!strcasecmp(el, "creator")) { // <creator> can be used if <author> is not present
+ if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eAuthor))) {
+ encoded_field = NewStrBuf();
+ StrBufRFC2047encode(&encoded_field, r->CData);
+ CM_SetAsFieldSB(r->msg, eAuthor, &encoded_field);
+ }
+ }
+
+ else if (!strcasecmp(el, "author")) { // <author> supercedes <creator> if both are present
+ if (r->msg != NULL) {
+ encoded_field = NewStrBuf();
+ StrBufRFC2047encode(&encoded_field, r->CData);
+ CM_SetAsFieldSB(r->msg, eAuthor, &encoded_field);
}
}
else if (!strcasecmp(el, "pubdate")) { // date/time stamp (rss) Sat, 25 Feb 2017 14:28:01 EST
- // FIXME parse it
+ if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
+ CM_SetFieldLONG(r->msg, eTimestamp, parsedate(ChrPtr(r->CData)));
+ }
}
else if (!strcasecmp(el, "updated")) { // date/time stamp (atom) 2003-12-13T18:30:02Z
- // FIXME parse it
+ if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
+ struct tm t;
+ char zulu;
+ memset(&t, 0, sizeof t);
+ sscanf(ChrPtr(r->CData), "%d-%d-%dT%d:%d:%d%c", &t.tm_year, &t.tm_mon, &t.tm_mday, &t.tm_hour, &t.tm_min, &t.tm_sec, &zulu);
+ t.tm_year -= 1900;
+ t.tm_mon -= 1;
+ CM_SetFieldLONG(r->msg, eTimestamp, mktime(&t));
+ }
+ }
+
+ else if (!strcasecmp(el, "link")) { // link to story (rss)
+ if (r->link != NULL) {
+ free(r->link);
+ r->link = NULL;
+ }
+ r->link = strdup(ChrPtr(r->CData));
+ }
+
+ else if (
+ (!strcasecmp(el, "guid")) // unique item id (rss)
+ || (!strcasecmp(el, "id")) // unique item id (atom)
+ ) {
+ if (r->item_id != NULL) {
+ free(r->item_id);
+ r->item_id = NULL;
+ }
+ r->item_id = strdup(ChrPtr(r->CData));
+ }
+
+ else if (
+ (!strcasecmp(el, "description")) // message text (rss)
+ || (!strcasecmp(el, "summary")) // message text (atom)
+ || (!strcasecmp(el, "content")) // message text (atom)
+ ) {
+ if (r->description != NULL) {
+ free(r->description);
+ r->description = NULL;
+ }
+ r->description = strdup(ChrPtr(r->CData));
}
if (r->CData != NULL) {
// Feed has been downloaded, now parse it.
//
-void rss_parse_feed(StrBuf *Feed)
+void rss_parse_feed(StrBuf *Feed, struct rssroom *rooms)
{
struct rssparser r;
memset(&r, 0, sizeof r);
- XML_Parser p = XML_ParserCreateNS("UTF-8", ':');
+ r.rooms = rooms;
+ XML_Parser p = XML_ParserCreate("UTF-8");
XML_SetElementHandler(p, rss_start_element, rss_end_element);
XML_SetCharacterDataHandler(p, rss_handle_data);
XML_SetUserData(p, (void *)&r);
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)) {
thisone = r;
}
}
+
if (thisone == NULL) {
thisone = malloc(sizeof(struct rssurl));
thisone->url = strdup(rssurl);
}
-// 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)
{
- struct rssroom *r;
CURL *curl;
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) {
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);
- rss_parse_feed(Downloaded); // parse the feed
-
- for (r=url->rooms; r!=NULL; r=r->next) { // we might move this somewhere else
- syslog(LOG_DEBUG, "Saving item to %s", r->room);
- // FIXME save to rooms
- }
-
- FreeStrBuf(&Downloaded);
+ rss_parse_feed(Downloaded, url->rooms); // parse the feed
+ FreeStrBuf(&Downloaded); // free the downloaded feed data
}
/* 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",
- ((now - last_run) / 60),
- ((now - last_run) % 60)
+ "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;
}
{
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";
}