1 // Bring external RSS and/or Atom feeds into rooms. This module implements a
2 // very loose parser that scrapes both kinds of feeds and is not picky about
3 // the standards compliance of the source data.
5 // Copyright (c) 2007-2023 by the citadel.org team
7 // This program is open source software. Use, duplication, or disclosure
8 // is subject to the terms of the GNU General Public License, version 3.
17 #include <sys/types.h>
20 #include <curl/curl.h>
21 #include <libcitadel.h>
22 #include "../../citadel.h"
23 #include "../../server.h"
24 #include "../../citserver.h"
25 #include "../../support.h"
26 #include "../../config.h"
27 #include "../../threads.h"
28 #include "../../ctdl_module.h"
29 #include "../../msgbase.h"
30 #include "../../parsedate.h"
31 #include "../../database.h"
32 #include "../../citadel_dirs.h"
33 #include "../../context.h"
34 #include "../../internet_addressing.h"
44 struct rssroom *rooms;
49 struct CtdlMessage *msg;
53 struct rssroom *rooms;
57 struct rssurl *rsstodo = NULL;
60 // This handler is called whenever an XML tag opens.
61 void rss_start_element(void *data, const char *el, const char **attribute) {
62 struct rssparser *r = (struct rssparser *)data;
65 if (server_shutting_down) return; // shunt the whole operation if we're exiting
68 (!strcasecmp(el, "entry"))
69 || (!strcasecmp(el, "item"))
71 // this is the start of a new item(rss) or entry(atom)
76 r->msg = malloc(sizeof(struct CtdlMessage));
77 memset(r->msg, 0, sizeof(struct CtdlMessage));
78 r->msg->cm_magic = CTDLMESSAGE_MAGIC;
79 r->msg->cm_anon_type = MES_NORMAL;
80 r->msg->cm_format_type = FMT_RFC822;
83 else if (!strcasecmp(el, "link")) { // atom feeds have the link as an attribute
84 for(i = 0; attribute[i]; i += 2) {
85 if (!strcasecmp(attribute[i], "href")) {
86 if (r->link != NULL) {
90 r->link = strdup(attribute[i+1]);
98 // This handler is called whenever an XML tag closes.
99 void rss_end_element(void *data, const char *el) {
100 struct rssparser *r = (struct rssparser *)data;
101 StrBuf *encoded_field;
103 if (server_shutting_down) return; // shunt the whole operation if we're exiting
105 if (StrLength(r->CData) > 0) { // strip leading/trailing whitespace from field
106 StrBufTrim(r->CData);
109 if ( // end of a new item(rss) or entry(atom)
110 (!strcasecmp(el, "entry"))
111 || (!strcasecmp(el, "item"))
113 if (r->msg != NULL) { // Save the message to the rooms
115 // use the link as an item id if nothing else is available
116 if ((r->item_id == NULL) && (r->link != NULL)) {
117 r->item_id = strdup(r->link);
120 // check the use table
121 StrBuf *u = NewStrBuf();
122 StrBufAppendPrintf(u, "rss/%s", r->item_id);
123 int already_seen = CheckIfAlreadySeen(u);
126 if (already_seen == 0) {
128 // Compose the message text
129 StrBuf *TheMessage = NewStrBuf();
130 StrBufAppendPrintf(TheMessage,
131 "Content-type: text/html\n\n"
133 "<html><head></head><body>"
136 if (r->description != NULL) {
137 StrBufAppendPrintf(TheMessage, "%s<br><br>\r\n", r->description);
138 free(r->description);
139 r->description = NULL;
142 if (r->link != NULL) {
143 StrBufAppendPrintf(TheMessage, "<a href=\"%s\">%s</a>\r\n", r->link, r->link);
148 StrBufAppendPrintf(TheMessage, "</body></html>\r\n");
149 CM_SetField(r->msg, eMesageText, ChrPtr(TheMessage), StrLength(TheMessage));
150 FreeStrBuf(&TheMessage);
152 if (CM_IsEmpty(r->msg, eAuthor)) {
153 CM_SetField(r->msg, eAuthor, HKEY("rss"));
156 if (CM_IsEmpty(r->msg, eTimestamp)) {
157 CM_SetFieldLONG(r->msg, eTimestamp, time(NULL));
160 // Save it to the room(s)
161 struct rssroom *rr = NULL;
163 for (rr=r->rooms; rr!=NULL; rr=rr->next) {
164 if (rr == r->rooms) {
165 msgnum = CtdlSubmitMsg(r->msg, NULL, rr->room); // in first room, save msg
168 CtdlSaveMsgPointerInRoom(rr->room, msgnum, 0, NULL); // elsewhere, save a pointer
170 syslog(LOG_DEBUG, "rssclient: saved message %ld to %s", msgnum, rr->room);
174 syslog(LOG_DEBUG, "rssclient: already seen %s", r->item_id);
181 if (r->item_id != NULL) {
187 else if (!strcasecmp(el, "title")) { // item subject (rss and atom)
188 if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eMsgSubject))) {
189 encoded_field = NewStrBuf();
190 StrBufRFC2047encode(&encoded_field, r->CData);
191 CM_SetAsFieldSB(r->msg, eMsgSubject, &encoded_field);
195 else if (!strcasecmp(el, "creator")) { // <creator> can be used if <author> is not present
196 if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eAuthor))) {
197 encoded_field = NewStrBuf();
198 StrBufRFC2047encode(&encoded_field, r->CData);
199 CM_SetAsFieldSB(r->msg, eAuthor, &encoded_field);
203 else if (!strcasecmp(el, "author")) { // <author> supercedes <creator> if both are present
204 if (r->msg != NULL) {
205 encoded_field = NewStrBuf();
206 StrBufRFC2047encode(&encoded_field, r->CData);
207 CM_SetAsFieldSB(r->msg, eAuthor, &encoded_field);
211 else if (!strcasecmp(el, "pubdate")) { // date/time stamp (rss) Sat, 25 Feb 2017 14:28:01 EST
212 if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
213 CM_SetFieldLONG(r->msg, eTimestamp, parsedate(ChrPtr(r->CData)));
217 else if (!strcasecmp(el, "updated")) { // date/time stamp (atom) 2003-12-13T18:30:02Z
218 if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
221 memset(&t, 0, sizeof t);
222 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);
225 CM_SetFieldLONG(r->msg, eTimestamp, mktime(&t));
229 else if (!strcasecmp(el, "link")) { // link to story (rss)
230 if (r->link != NULL) {
234 r->link = strdup(ChrPtr(r->CData));
238 (!strcasecmp(el, "guid")) // unique item id (rss)
239 || (!strcasecmp(el, "id")) // unique item id (atom)
241 if (r->item_id != NULL) {
245 r->item_id = strdup(ChrPtr(r->CData));
249 (!strcasecmp(el, "description")) // message text (rss)
250 || (!strcasecmp(el, "summary")) // message text (atom)
251 || (!strcasecmp(el, "content")) // message text (atom)
253 if (r->description != NULL) {
254 free(r->description);
255 r->description = NULL;
257 r->description = strdup(ChrPtr(r->CData));
260 if (r->CData != NULL) {
261 FreeStrBuf(&r->CData);
267 // This handler is called whenever data appears between opening and closing tags.
268 void rss_handle_data(void *data, const char *content, int length) {
269 struct rssparser *r = (struct rssparser *)data;
271 if (r->CData == NULL) {
272 r->CData = NewStrBuf();
275 StrBufAppendBufPlain(r->CData, content, length, 0);
279 // Feed has been downloaded, now parse it.
280 void rss_parse_feed(StrBuf *Feed, struct rssroom *rooms) {
283 memset(&r, 0, sizeof r);
285 XML_Parser p = XML_ParserCreate("UTF-8");
286 XML_SetElementHandler(p, rss_start_element, rss_end_element);
287 XML_SetCharacterDataHandler(p, rss_handle_data);
288 XML_SetUserData(p, (void *)&r);
289 XML_Parse(p, ChrPtr(Feed), StrLength(Feed), XML_TRUE);
294 // Add a feed/room pair into the todo list
295 void rssclient_push_todo(char *rssurl, char *roomname) {
296 struct rssurl *r = NULL;
297 struct rssurl *thisone = NULL;
298 struct rssroom *newroom = NULL;
300 syslog(LOG_DEBUG, "rssclient: will fetch %s to %s", rssurl, roomname);
302 for (r=rsstodo; r!=NULL; r=r->next) {
303 if (!strcasecmp(r->url, rssurl)) {
308 if (thisone == NULL) {
309 thisone = malloc(sizeof(struct rssurl));
310 thisone->url = strdup(rssurl);
311 thisone->rooms = NULL;
312 thisone->next = rsstodo;
316 newroom = malloc(sizeof(struct rssroom));
317 newroom->room = strdup(roomname);
318 newroom->next = thisone->rooms;
319 thisone->rooms = newroom;
323 // pull one feed (possibly multiple rooms)
324 void rss_pull_one_feed(struct rssurl *url) {
327 StrBuf *Downloaded = NULL;
329 syslog(LOG_DEBUG, "rssclient: fetching %s", url->url);
331 curl = curl_easy_init();
336 Downloaded = NewStrBuf();
338 curl_easy_setopt(curl, CURLOPT_URL, url->url);
339 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0L);
340 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0L);
341 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1L); // Follow redirects
342 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback); // What to do with downloaded data
343 curl_easy_setopt(curl, CURLOPT_WRITEDATA, Downloaded); // Give it our StrBuf to work with
344 curl_easy_setopt(curl, CURLOPT_TIMEOUT, 20L); // Time out after 20 seconds
345 res = curl_easy_perform(curl); // Perform the request
346 if (res != CURLE_OK) {
347 syslog(LOG_WARNING, "rssclient: failed to load feed: %s", curl_easy_strerror(res));
349 curl_easy_cleanup(curl);
351 rss_parse_feed(Downloaded, url->rooms); // parse the feed
352 FreeStrBuf(&Downloaded); // free the downloaded feed data
356 // We have a list, now download the feeds
357 void rss_pull_feeds(void) {
361 while ((rsstodo != NULL) && (!server_shutting_down)) {
362 rss_pull_one_feed(rsstodo);
364 rsstodo = rsstodo->next;
365 while (r->rooms != NULL) {
367 r->rooms = r->rooms->next;
377 // Scan a room's netconfig looking for RSS feed parsing requests
378 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data) {
379 char *serialized_config = NULL;
384 if (server_shutting_down) return;
386 serialized_config = LoadRoomNetConfigFile(qrbuf->QRnumber);
387 if (!serialized_config) {
391 num_configs = num_tokens(serialized_config, '\n');
392 for (i=0; i<num_configs; ++i) {
393 extract_token(cfgline, serialized_config, i, '\n', sizeof cfgline);
394 if (!strncasecmp(cfgline, HKEY("rssclient|"))) {
395 strcpy(cfgline, &cfgline[10]);
396 char *vbar = strchr(cfgline, '|');
400 rssclient_push_todo(cfgline, qrbuf->QRname);
404 free(serialized_config);
408 // Scan for rooms that have RSS client requests configured
409 void rssclient_scan(void) {
410 time_t now = time(NULL);
412 // Run no more than once every 15 minutes.
413 if ((now - last_run) < 900) {
415 "rssclient: polling interval not yet reached; last run was %ldm%lds ago",
416 ((now - last_run) / 60),
417 ((now - last_run) % 60)
422 syslog(LOG_DEBUG, "rssclient: started");
423 CtdlForEachRoom(rssclient_scan_room, NULL);
425 syslog(LOG_DEBUG, "rssclient: ended");
426 last_run = time(NULL);
431 // Initialization function, called from modules_init.c
432 char *ctdl_module_init_rssclient(void) {
434 syslog(LOG_INFO, "rssclient: using %s", curl_version());
435 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);