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_defs.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
130 StrBuf *TheMessage = NewStrBuf();
131 StrBufAppendPrintf(TheMessage, "<html><head></head><body>");
133 if (r->description != NULL) {
134 StrBufAppendPrintf(TheMessage, "%s<br><br>\r\n", r->description);
135 free(r->description);
136 r->description = NULL;
139 if (r->link != NULL) {
140 StrBufAppendPrintf(TheMessage, "<a href=\"%s\">%s</a>\r\n", r->link, r->link);
145 StrBufAppendPrintf(TheMessage, "</body></html>\r\n");
147 // Quoted-Printable encode the HTML message, because RSS and Atom make no guarantee of line length limits.
148 StrBuf *TheMessage_Encoded = StrBufQuotedPrintableEncode(TheMessage);
150 // Now we reuse TheMessage -- this time it will contain the MIME headers concatenated with the encoded message.
151 FlushStrBuf(TheMessage);
152 StrBufAppendBufPlain(TheMessage, HKEY(
153 "Content-type: text/html; charset=UTF-8\r\n"
154 "Content-Transfer-Encoding: quoted-printable\r\n"
158 StrBufAppendBuf(TheMessage, TheMessage_Encoded, 0);
159 FreeStrBuf(&TheMessage_Encoded);
161 CM_SetField(r->msg, eMesageText, ChrPtr(TheMessage), StrLength(TheMessage));
162 FreeStrBuf(&TheMessage);
164 if (CM_IsEmpty(r->msg, eAuthor)) {
165 CM_SetField(r->msg, eAuthor, HKEY("rss"));
168 if (CM_IsEmpty(r->msg, eTimestamp)) {
169 CM_SetFieldLONG(r->msg, eTimestamp, time(NULL));
172 // Save it to the room(s)
173 struct rssroom *rr = NULL;
175 for (rr=r->rooms; rr!=NULL; rr=rr->next) {
176 if (rr == r->rooms) {
177 msgnum = CtdlSubmitMsg(r->msg, NULL, rr->room); // in first room, save msg
180 CtdlSaveMsgPointerInRoom(rr->room, msgnum, 0, NULL); // elsewhere, save a pointer
182 syslog(LOG_DEBUG, "rssclient: saved message %ld to %s", msgnum, rr->room);
186 syslog(LOG_DEBUG, "rssclient: already seen %s", r->item_id);
193 if (r->item_id != NULL) {
199 else if (!strcasecmp(el, "title")) { // item subject (rss and atom)
200 if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eMsgSubject))) {
201 encoded_field = NewStrBuf();
202 StrBufRFC2047encode(&encoded_field, r->CData);
203 CM_SetAsFieldSB(r->msg, eMsgSubject, &encoded_field);
207 else if (!strcasecmp(el, "creator")) { // <creator> can be used if <author> is not present
208 if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eAuthor))) {
209 encoded_field = NewStrBuf();
210 StrBufRFC2047encode(&encoded_field, r->CData);
211 CM_SetAsFieldSB(r->msg, eAuthor, &encoded_field);
215 else if (!strcasecmp(el, "author")) { // <author> supercedes <creator> if both are present
216 if (r->msg != NULL) {
217 encoded_field = NewStrBuf();
218 StrBufRFC2047encode(&encoded_field, r->CData);
219 CM_SetAsFieldSB(r->msg, eAuthor, &encoded_field);
223 else if (!strcasecmp(el, "pubdate")) { // date/time stamp (rss) Sat, 25 Feb 2017 14:28:01 EST
224 if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
225 CM_SetFieldLONG(r->msg, eTimestamp, parsedate(ChrPtr(r->CData)));
229 else if (!strcasecmp(el, "updated")) { // date/time stamp (atom) 2003-12-13T18:30:02Z
230 if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
233 memset(&t, 0, sizeof t);
234 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);
237 CM_SetFieldLONG(r->msg, eTimestamp, mktime(&t));
241 else if (!strcasecmp(el, "link")) { // link to story (rss)
242 if (r->link != NULL) {
246 r->link = strdup(ChrPtr(r->CData));
250 (!strcasecmp(el, "guid")) // unique item id (rss)
251 || (!strcasecmp(el, "id")) // unique item id (atom)
253 if (r->item_id != NULL) {
257 r->item_id = strdup(ChrPtr(r->CData));
261 (!strcasecmp(el, "description")) // message text (rss)
262 || (!strcasecmp(el, "summary")) // message text (atom)
263 || (!strcasecmp(el, "content")) // message text (atom)
265 if (r->description != NULL) {
266 free(r->description);
267 r->description = NULL;
269 r->description = strdup(ChrPtr(r->CData));
272 if (r->CData != NULL) {
273 FreeStrBuf(&r->CData);
279 // This handler is called whenever data appears between opening and closing tags.
280 void rss_handle_data(void *data, const char *content, int length) {
281 struct rssparser *r = (struct rssparser *)data;
283 if (r->CData == NULL) {
284 r->CData = NewStrBuf();
287 StrBufAppendBufPlain(r->CData, content, length, 0);
291 // Feed has been downloaded, now parse it.
292 void rss_parse_feed(StrBuf *Feed, struct rssroom *rooms) {
295 memset(&r, 0, sizeof r);
297 XML_Parser p = XML_ParserCreate("UTF-8");
298 XML_SetElementHandler(p, rss_start_element, rss_end_element);
299 XML_SetCharacterDataHandler(p, rss_handle_data);
300 XML_SetUserData(p, (void *)&r);
301 XML_Parse(p, ChrPtr(Feed), StrLength(Feed), XML_TRUE);
306 // Add a feed/room pair into the todo list
307 void rssclient_push_todo(char *rssurl, char *roomname) {
308 struct rssurl *r = NULL;
309 struct rssurl *thisone = NULL;
310 struct rssroom *newroom = NULL;
312 syslog(LOG_DEBUG, "rssclient: will fetch %s to %s", rssurl, roomname);
314 for (r=rsstodo; r!=NULL; r=r->next) {
315 if (!strcasecmp(r->url, rssurl)) {
320 if (thisone == NULL) {
321 thisone = malloc(sizeof(struct rssurl));
322 thisone->url = strdup(rssurl);
323 thisone->rooms = NULL;
324 thisone->next = rsstodo;
328 newroom = malloc(sizeof(struct rssroom));
329 newroom->room = strdup(roomname);
330 newroom->next = thisone->rooms;
331 thisone->rooms = newroom;
335 // pull one feed (possibly multiple rooms)
336 void rss_pull_one_feed(struct rssurl *url) {
339 StrBuf *Downloaded = NULL;
341 syslog(LOG_DEBUG, "rssclient: fetching %s", url->url);
343 curl = curl_easy_init();
348 Downloaded = NewStrBuf();
350 curl_easy_setopt(curl, CURLOPT_URL, url->url);
351 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0L);
352 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0L);
353 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1L); // Follow redirects
354 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback); // What to do with downloaded data
355 curl_easy_setopt(curl, CURLOPT_WRITEDATA, Downloaded); // Give it our StrBuf to work with
356 curl_easy_setopt(curl, CURLOPT_TIMEOUT, 20L); // Time out after 20 seconds
357 res = curl_easy_perform(curl); // Perform the request
358 if (res != CURLE_OK) {
359 syslog(LOG_WARNING, "rssclient: failed to load feed: %s", curl_easy_strerror(res));
361 curl_easy_cleanup(curl);
363 rss_parse_feed(Downloaded, url->rooms); // parse the feed
364 FreeStrBuf(&Downloaded); // free the downloaded feed data
368 // We have a list, now download the feeds
369 void rss_pull_feeds(void) {
373 while ((rsstodo != NULL) && (!server_shutting_down)) {
374 rss_pull_one_feed(rsstodo);
376 rsstodo = rsstodo->next;
377 while (r->rooms != NULL) {
379 r->rooms = r->rooms->next;
389 // Scan a room's netconfig looking for RSS feed parsing requests
390 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data) {
391 char *serialized_config = NULL;
396 if (server_shutting_down) return;
398 serialized_config = LoadRoomNetConfigFile(qrbuf->QRnumber);
399 if (!serialized_config) {
403 num_configs = num_tokens(serialized_config, '\n');
404 for (i=0; i<num_configs; ++i) {
405 extract_token(cfgline, serialized_config, i, '\n', sizeof cfgline);
406 if (!strncasecmp(cfgline, HKEY("rssclient|"))) {
407 strcpy(cfgline, &cfgline[10]);
408 char *vbar = strchr(cfgline, '|');
412 rssclient_push_todo(cfgline, qrbuf->QRname);
416 free(serialized_config);
420 // Scan for rooms that have RSS client requests configured
421 void rssclient_scan(void) {
422 time_t now = time(NULL);
424 // Run no more than once every 15 minutes.
425 if ((now - last_run) < 900) {
427 "rssclient: polling interval not yet reached; last run was %ldm%lds ago",
428 ((now - last_run) / 60),
429 ((now - last_run) % 60)
434 syslog(LOG_DEBUG, "rssclient: started");
435 CtdlForEachRoom(rssclient_scan_room, NULL);
437 syslog(LOG_DEBUG, "rssclient: ended");
438 last_run = time(NULL);
443 // Initialization function, called from modules_init.c
444 char *ctdl_module_init_rssclient(void) {
446 syslog(LOG_INFO, "rssclient: using %s", curl_version());
447 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);