2 * Bring external RSS feeds into rooms.
4 * Copyright (c) 2007-2012 by the citadel.org team
6 * This program is open source software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 3.
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.See the
12 * GNU General Public License for more details.
19 #if TIME_WITH_SYS_TIME
20 # include <sys/time.h>
33 #include <sys/types.h>
36 #include <curl/curl.h>
37 #include <libcitadel.h>
40 #include "citserver.h"
44 #include "ctdl_module.h"
46 #include "parsedate.h"
48 #include "citadel_dirs.h"
51 #include "event_client.h"
52 #include "rss_atom_parser.h"
55 #define TMP_MSGDATA 0xFF
56 #define TMP_SHORTER_URL_OFFSET 0xFE
57 #define TMP_SHORTER_URLS 0xFD
61 pthread_mutex_t RSSQueueMutex; /* locks the access to the following vars: */
62 HashList *RSSQueueRooms = NULL; /* rss_room_counter */
63 HashList *RSSFetchUrls = NULL; /*->rss_aggregator;->RefCount access locked*/
65 eNextState RSSAggregator_Terminate(AsyncIO *IO);
66 eNextState RSSAggregator_TerminateDB(AsyncIO *IO);
67 eNextState RSSAggregator_ShutdownAbort(AsyncIO *IO);
68 struct CitContext rss_CC;
70 struct rssnetcfg *rnclist = NULL;
71 int RSSClientDebugEnabled = 0;
72 #define N ((rss_aggregator*)IO->Data)->QRnumber
74 #define DBGLOG(LEVEL) if ((LEVEL != LOG_DEBUG) || (RSSClientDebugEnabled != 0))
76 #define EVRSSC_syslog(LEVEL, FORMAT, ...) \
77 DBGLOG(LEVEL) syslog(LEVEL, \
78 "IO[%ld]CC[%d][%ld]RSS" FORMAT, \
79 IO->ID, CCID, N, __VA_ARGS__)
81 #define EVRSSCM_syslog(LEVEL, FORMAT) \
82 DBGLOG(LEVEL) syslog(LEVEL, \
83 "IO[%ld]CC[%d][%ld]RSS" FORMAT, \
86 #define EVRSSQ_syslog(LEVEL, FORMAT, ...) \
87 DBGLOG(LEVEL) syslog(LEVEL, "RSS" FORMAT, \
89 #define EVRSSQM_syslog(LEVEL, FORMAT) \
90 DBGLOG(LEVEL) syslog(LEVEL, "RSS" FORMAT)
92 #define EVRSSCSM_syslog(LEVEL, FORMAT) \
93 DBGLOG(LEVEL) syslog(LEVEL, "IO[%ld][%ld]RSS" FORMAT, \
96 void DeleteRoomReference(long QRnumber)
102 rss_room_counter *pRoomC;
104 At = GetNewHashPos(RSSQueueRooms, 0);
106 if (GetHashPosFromKey(RSSQueueRooms, LKEY(QRnumber), At))
108 GetHashPos(RSSQueueRooms, At, &HKLen, &HK, &vData);
111 pRoomC = (rss_room_counter *) vData;
113 if (pRoomC->count == 0)
114 DeleteEntryFromHash(RSSQueueRooms, At);
120 void UnlinkRooms(rss_aggregator *RSSAggr)
122 DeleteRoomReference(RSSAggr->QRnumber);
123 if (RSSAggr->OtherQRnumbers != NULL)
130 At = GetNewHashPos(RSSAggr->OtherQRnumbers, 0);
131 while (! server_shutting_down &&
132 GetNextHashPos(RSSAggr->OtherQRnumbers,
138 long *lData = (long*) vData;
139 DeleteRoomReference(*lData);
146 void UnlinkRSSAggregator(rss_aggregator *RSSAggr)
150 pthread_mutex_lock(&RSSQueueMutex);
151 UnlinkRooms(RSSAggr);
153 At = GetNewHashPos(RSSFetchUrls, 0);
154 if (GetHashPosFromKey(RSSFetchUrls, SKEY(RSSAggr->Url), At))
156 DeleteEntryFromHash(RSSFetchUrls, At);
159 last_run = time(NULL);
160 pthread_mutex_unlock(&RSSQueueMutex);
163 void DeleteRssCfg(void *vptr)
165 rss_aggregator *RSSAggr = (rss_aggregator *)vptr;
166 AsyncIO *IO = &RSSAggr->IO;
167 EVRSSCM_syslog(LOG_DEBUG, "RSS: destroying\n");
169 FreeStrBuf(&RSSAggr->Url);
170 FreeStrBuf(&RSSAggr->rooms);
171 FreeStrBuf(&RSSAggr->CData);
172 FreeStrBuf(&RSSAggr->Key);
173 DeleteHash(&RSSAggr->OtherQRnumbers);
175 DeleteHashPos (&RSSAggr->Pos);
176 DeleteHash (&RSSAggr->Messages);
177 if (RSSAggr->recp.recp_room != NULL)
178 free(RSSAggr->recp.recp_room);
181 if (RSSAggr->Item != NULL)
183 flush_rss_item(RSSAggr->Item);
188 FreeAsyncIOContents(&RSSAggr->IO);
189 memset(RSSAggr, 0, sizeof(rss_aggregator));
193 eNextState RSSAggregator_Terminate(AsyncIO *IO)
195 rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
197 EVRSSCM_syslog(LOG_DEBUG, "RSS: Terminating.\n");
199 StopCurlWatchers(IO);
200 UnlinkRSSAggregator(RSSAggr);
204 eNextState RSSAggregator_TerminateDB(AsyncIO *IO)
206 rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
208 EVRSSCM_syslog(LOG_DEBUG, "RSS: Terminating.\n");
211 StopDBWatchers(&RSSAggr->IO);
212 UnlinkRSSAggregator(RSSAggr);
216 eNextState RSSAggregator_ShutdownAbort(AsyncIO *IO)
219 rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
221 pUrl = IO->ConnectMe->PlainUrl;
225 EVRSSC_syslog(LOG_DEBUG, "RSS: Aborting by shutdown: %s.\n", pUrl);
227 StopCurlWatchers(IO);
228 UnlinkRSSAggregator(RSSAggr);
232 void AppendLink(StrBuf *Message,
237 if (StrLength(link) > 0)
239 StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
240 StrBufAppendBuf(Message, link, 0);
241 StrBufAppendBufPlain(Message, HKEY("\">"), 0);
242 if (StrLength(LinkTitle) > 0)
243 StrBufAppendBuf(Message, LinkTitle, 0);
244 else if ((Title != NULL) && !IsEmptyStr(Title))
245 StrBufAppendBufPlain(Message, Title, -1, 0);
247 StrBufAppendBuf(Message, link, 0);
248 StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
253 void rss_format_item(networker_save_message *SaveMsg)
258 if (SaveMsg->author_or_creator != NULL) {
261 StrBuf *Encoded = NULL;
264 From = html_to_ascii(ChrPtr(SaveMsg->author_or_creator),
265 StrLength(SaveMsg->author_or_creator),
267 StrBufPlain(SaveMsg->author_or_creator, From, -1);
268 StrBufTrim(SaveMsg->author_or_creator);
271 FromAt = strchr(ChrPtr(SaveMsg->author_or_creator), '@') != NULL;
272 if (!FromAt && StrLength (SaveMsg->author_email) > 0)
274 StrBufRFC2047encode(&Encoded, SaveMsg->author_or_creator);
275 SaveMsg->Msg.cm_fields['A'] = SmashStrBuf(&Encoded);
276 SaveMsg->Msg.cm_fields['P'] =
277 SmashStrBuf(&SaveMsg->author_email);
283 SaveMsg->Msg.cm_fields['A'] =
284 SmashStrBuf(&SaveMsg->author_or_creator);
285 SaveMsg->Msg.cm_fields['P'] =
286 strdup(SaveMsg->Msg.cm_fields['A']);
290 StrBufRFC2047encode(&Encoded,
291 SaveMsg->author_or_creator);
292 SaveMsg->Msg.cm_fields['A'] =
293 SmashStrBuf(&Encoded);
294 SaveMsg->Msg.cm_fields['P'] =
295 strdup("rss@localhost");
301 SaveMsg->Msg.cm_fields['A'] = strdup("rss");
304 SaveMsg->Msg.cm_fields['N'] = strdup(NODENAME);
305 if (SaveMsg->title != NULL) {
308 StrBuf *Encoded, *QPEncoded;
311 StrBufSpaceToBlank(SaveMsg->title);
312 len = StrLength(SaveMsg->title);
313 Sbj = html_to_ascii(ChrPtr(SaveMsg->title), len, 512, 0);
315 if ((len > 0) && (Sbj[len - 1] == '\n'))
320 Encoded = NewStrBufPlain(Sbj, len);
324 StrBufRFC2047encode(&QPEncoded, Encoded);
326 SaveMsg->Msg.cm_fields['U'] = SmashStrBuf(&QPEncoded);
327 FreeStrBuf(&Encoded);
329 if (SaveMsg->link == NULL)
330 SaveMsg->link = NewStrBufPlain(HKEY(""));
332 #if 0 /* temporarily disable shorter urls. */
333 SaveMsg->Msg.cm_fields[TMP_SHORTER_URLS] =
334 GetShorterUrls(SaveMsg->description);
337 msglen += 1024 + StrLength(SaveMsg->link) + StrLength(SaveMsg->description) ;
339 Message = NewStrBufPlain(NULL, msglen);
341 StrBufPlain(Message, HKEY(
342 "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
344 #if 0 /* disable shorter url for now. */
345 SaveMsg->Msg.cm_fields[TMP_SHORTER_URL_OFFSET] = StrLength(Message);
347 StrBufAppendBuf(Message, SaveMsg->description, 0);
348 StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
350 AppendLink(Message, SaveMsg->link, SaveMsg->linkTitle, NULL);
351 AppendLink(Message, SaveMsg->reLink, SaveMsg->reLinkTitle, "Reply to this");
352 StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
355 SaveMsg->Message = Message;
358 eNextState RSSSaveMessage(AsyncIO *IO)
362 rss_aggregator *RSSAggr = (rss_aggregator *) IO->Data;
364 rss_format_item(RSSAggr->ThisMsg);
366 RSSAggr->ThisMsg->Msg.cm_fields['M'] =
367 SmashStrBuf(&RSSAggr->ThisMsg->Message);
369 CtdlSubmitMsg(&RSSAggr->ThisMsg->Msg, &RSSAggr->recp, NULL, 0);
371 /* write the uidl to the use table so we don't store this item again */
372 cdb_store(CDB_USETABLE,
373 SKEY(RSSAggr->ThisMsg->MsgGUID),
374 &RSSAggr->ThisMsg->ut,
375 sizeof(struct UseTable) );
377 if (GetNextHashPos(RSSAggr->Messages,
380 (void**) &RSSAggr->ThisMsg))
381 return NextDBOperation(IO, RSS_FetchNetworkUsetableEntry);
386 eNextState RSS_FetchNetworkUsetableEntry(AsyncIO *IO)
390 struct cdbdata *cdbut;
391 rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
393 /* Find out if we've already seen this item */
394 strcpy(Ctx->ThisMsg->ut.ut_msgid,
395 ChrPtr(Ctx->ThisMsg->MsgGUID)); /// TODO
396 Ctx->ThisMsg->ut.ut_timestamp = time(NULL);
398 cdbut = cdb_fetch(CDB_USETABLE, SKEY(Ctx->ThisMsg->MsgGUID));
401 /* Item has already been seen */
402 EVRSSC_syslog(LOG_DEBUG,
403 "%s has already been seen\n",
404 ChrPtr(Ctx->ThisMsg->MsgGUID));
407 /* rewrite the record anyway, to update the timestamp */
408 cdb_store(CDB_USETABLE,
409 SKEY(Ctx->ThisMsg->MsgGUID),
410 &Ctx->ThisMsg->ut, sizeof(struct UseTable) );
412 if (GetNextHashPos(Ctx->Messages,
415 (void**) &Ctx->ThisMsg))
416 return NextDBOperation(
418 RSS_FetchNetworkUsetableEntry);
425 NextDBOperation(IO, RSSSaveMessage);
430 eNextState RSSAggregator_AnalyseReply(AsyncIO *IO)
433 u_char rawdigest[MD5_DIGEST_LEN];
434 struct MD5Context md5context;
436 struct cdbdata *cdbut;
437 rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
439 if (IO->HttpReq.httpcode != 200)
442 EVRSSC_syslog(LOG_ALERT, "need a 200, got a %ld !\n",
443 IO->HttpReq.httpcode);
444 // TODO: aide error message with rate limit
448 MD5Init(&md5context);
450 MD5Update(&md5context,
451 (const unsigned char*)SKEY(IO->HttpReq.ReplyData));
453 MD5Update(&md5context,
454 (const unsigned char*)SKEY(Ctx->Url));
456 MD5Final(rawdigest, &md5context);
457 guid = NewStrBufPlain(NULL,
458 MD5_DIGEST_LEN * 2 + 12 /* _rss2ctdl*/);
459 StrBufHexEscAppend(guid, NULL, rawdigest, MD5_DIGEST_LEN);
460 StrBufAppendBufPlain(guid, HKEY("_rssFM"), 0);
461 if (StrLength(guid) > 40)
462 StrBufCutAt(guid, 40, NULL);
463 /* Find out if we've already seen this item */
464 memcpy(ut.ut_msgid, SKEY(guid));
465 ut.ut_timestamp = time(NULL);
467 cdbut = cdb_fetch(CDB_USETABLE, SKEY(guid));
470 /* Item has already been seen */
471 EVRSSC_syslog(LOG_DEBUG,
472 "%s has already been seen\n",
477 /* rewrite the record anyway, to update the timestamp */
478 cdb_store(CDB_USETABLE,
480 &ut, sizeof(struct UseTable) );
482 if (cdbut != NULL) return eAbort;
484 return RSSAggregator_ParseReply(IO);
487 eNextState RSSAggregator_FinishHttp(AsyncIO *IO)
489 return QueueDBOperation(IO, RSSAggregator_AnalyseReply);
495 int rss_do_fetching(rss_aggregator *RSSAggr)
497 AsyncIO *IO = &RSSAggr->IO;
503 if ((RSSAggr->next_poll != 0) && (now < RSSAggr->next_poll))
506 ri = (rss_item*) malloc(sizeof(rss_item));
507 memset(ri, 0, sizeof(rss_item));
510 if (! InitcURLIOStruct(&RSSAggr->IO,
512 "Citadel RSS Client",
513 RSSAggregator_FinishHttp,
514 RSSAggregator_Terminate,
515 RSSAggregator_TerminateDB,
516 RSSAggregator_ShutdownAbort))
518 EVRSSCM_syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
522 safestrncpy(((CitContext*)RSSAggr->IO.CitContext)->cs_host,
523 ChrPtr(RSSAggr->Url),
524 sizeof(((CitContext*)RSSAggr->IO.CitContext)->cs_host));
526 EVRSSC_syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", ChrPtr(RSSAggr->Url));
527 ParseURL(&RSSAggr->IO.ConnectMe, RSSAggr->Url, 80);
528 CurlPrepareURL(RSSAggr->IO.ConnectMe);
530 QueueCurlContext(&RSSAggr->IO);
535 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
537 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
539 StrBuf *CfgData=NULL;
542 rss_room_counter *Count = NULL;
544 char filename[PATH_MAX];
547 rss_aggregator *RSSAggr = NULL;
548 rss_aggregator *use_this_RSSAggr = NULL;
550 const char *CfgPtr, *lPtr;
553 pthread_mutex_lock(&RSSQueueMutex);
554 if (GetHash(RSSQueueRooms, LKEY(qrbuf->QRnumber), &vptr))
556 EVRSSQ_syslog(LOG_DEBUG,
557 "rssclient: [%ld] %s already in progress.\n",
560 pthread_mutex_unlock(&RSSQueueMutex);
563 pthread_mutex_unlock(&RSSQueueMutex);
565 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
567 if (server_shutting_down)
570 /* Only do net processing for rooms that have netconfigs */
571 fd = open(filename, 0);
574 "rssclient: %s no config.\n",
579 if (server_shutting_down)
582 if (fstat(fd, &statbuf) == -1) {
583 EVRSSQ_syslog(LOG_DEBUG,
584 "ERROR: could not stat configfile '%s' - %s\n",
590 if (server_shutting_down)
593 CfgData = NewStrBufPlain(NULL, statbuf.st_size + 1);
595 if (StrBufReadBLOB(CfgData, &fd, 1, statbuf.st_size, &Err) < 0) {
597 FreeStrBuf(&CfgData);
598 EVRSSQ_syslog(LOG_ERR, "ERROR: reading config '%s' - %s<br>\n",
599 filename, strerror(errno));
603 if (server_shutting_down)
607 CfgType = NewStrBuf();
608 Line = NewStrBufPlain(NULL, StrLength(CfgData));
612 Done = StrBufSipLine(Line, CfgData, &CfgPtr) == 0;
613 if (StrLength(Line) > 0)
616 StrBufExtract_NextToken(CfgType, Line, &lPtr, '|');
617 if (!strcasecmp("rssclient", ChrPtr(CfgType)))
622 sizeof(rss_room_counter));
626 RSSAggr = (rss_aggregator *) malloc(
627 sizeof(rss_aggregator));
629 memset (RSSAggr, 0, sizeof(rss_aggregator));
630 RSSAggr->QRnumber = qrbuf->QRnumber;
631 RSSAggr->roomlist_parts = 1;
632 RSSAggr->Url = NewStrBuf();
634 StrBufExtract_NextToken(RSSAggr->Url,
639 pthread_mutex_lock(&RSSQueueMutex);
640 GetHash(RSSFetchUrls,
644 use_this_RSSAggr = (rss_aggregator *)vptr;
645 if (use_this_RSSAggr != NULL)
648 StrBufAppendBufPlain(
649 use_this_RSSAggr->rooms,
652 if (use_this_RSSAggr->roomlist_parts==1)
654 use_this_RSSAggr->OtherQRnumbers
655 = NewHash(1, lFlathash);
657 QRnumber = (long*)malloc(sizeof(long));
658 *QRnumber = qrbuf->QRnumber;
659 Put(use_this_RSSAggr->OtherQRnumbers,
660 LKEY(qrbuf->QRnumber),
663 use_this_RSSAggr->roomlist_parts++;
665 pthread_mutex_unlock(&RSSQueueMutex);
667 FreeStrBuf(&RSSAggr->Url);
672 pthread_mutex_unlock(&RSSQueueMutex);
674 RSSAggr->ItemType = RSS_UNSET;
676 RSSAggr->rooms = NewStrBufPlain(
679 pthread_mutex_lock(&RSSQueueMutex);
686 pthread_mutex_unlock(&RSSQueueMutex);
692 Count->QRnumber = qrbuf->QRnumber;
693 pthread_mutex_lock(&RSSQueueMutex);
694 EVRSSQ_syslog(LOG_DEBUG, "client: [%ld] %s now starting.\n",
695 qrbuf->QRnumber, qrbuf->QRname);
696 Put(RSSQueueRooms, LKEY(qrbuf->QRnumber), Count, NULL);
697 pthread_mutex_unlock(&RSSQueueMutex);
699 FreeStrBuf(&CfgData);
700 FreeStrBuf(&CfgType);
705 * Scan for rooms that have RSS client requests configured
707 void rssclient_scan(void) {
708 int RSSRoomCount, RSSCount;
709 rss_aggregator *rptr = NULL;
714 time_t now = time(NULL);
716 /* Run no more than once every 15 minutes. */
717 if ((now - last_run) < 900) {
718 EVRSSQ_syslog(LOG_DEBUG,
719 "Client: polling interval not yet reached; last run was %ldm%lds ago",
720 ((now - last_run) / 60),
721 ((now - last_run) % 60)
727 * This is a simple concurrency check to make sure only one rssclient
728 * run is done at a time.
730 pthread_mutex_lock(&RSSQueueMutex);
731 RSSCount = GetCount(RSSFetchUrls);
732 RSSRoomCount = GetCount(RSSQueueRooms);
733 pthread_mutex_unlock(&RSSQueueMutex);
735 if ((RSSRoomCount > 0) || (RSSCount > 0)) {
736 EVRSSQ_syslog(LOG_DEBUG,
737 "rssclient: concurrency check failed; %d rooms and %d url's are queued",
738 RSSRoomCount, RSSCount
743 become_session(&rss_CC);
744 EVRSSQM_syslog(LOG_DEBUG, "rssclient started\n");
745 CtdlForEachRoom(rssclient_scan_room, NULL);
747 pthread_mutex_lock(&RSSQueueMutex);
749 it = GetNewHashPos(RSSFetchUrls, 0);
750 while (!server_shutting_down &&
751 GetNextHashPos(RSSFetchUrls, it, &len, &Key, &vrptr) &&
753 rptr = (rss_aggregator *)vrptr;
754 if (!rss_do_fetching(rptr))
755 UnlinkRSSAggregator(rptr);
758 pthread_mutex_unlock(&RSSQueueMutex);
760 EVRSSQM_syslog(LOG_DEBUG, "rssclient ended\n");
764 void rss_cleanup(void)
766 /* citthread_mutex_destroy(&RSSQueueMutex); TODO */
767 DeleteHash(&RSSFetchUrls);
768 DeleteHash(&RSSQueueRooms);
771 void LogDebugEnableRSSClient(const int n)
773 RSSClientDebugEnabled = n;
776 CTDL_MODULE_INIT(rssclient)
780 CtdlFillSystemContext(&rss_CC, "rssclient");
781 pthread_mutex_init(&RSSQueueMutex, NULL);
782 RSSQueueRooms = NewHash(1, lFlathash);
783 RSSFetchUrls = NewHash(1, NULL);
784 syslog(LOG_INFO, "%s\n", curl_version());
785 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);
786 CtdlRegisterEVCleanupHook(rss_cleanup);
787 CtdlRegisterDebugFlagHook(HKEY("rssclient"), LogDebugEnableRSSClient, &RSSClientDebugEnabled);