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 eNextState RSSSaveMessage(AsyncIO *IO)
236 rss_aggregator *RSSAggr = (rss_aggregator *) IO->Data;
238 RSSAggr->ThisMsg->Msg.cm_fields['M'] =
239 SmashStrBuf(&RSSAggr->ThisMsg->Message);
241 CtdlSubmitMsg(&RSSAggr->ThisMsg->Msg, &RSSAggr->recp, NULL, 0);
243 /* write the uidl to the use table so we don't store this item again */
244 cdb_store(CDB_USETABLE,
245 SKEY(RSSAggr->ThisMsg->MsgGUID),
246 &RSSAggr->ThisMsg->ut,
247 sizeof(struct UseTable) );
249 if (GetNextHashPos(RSSAggr->Messages,
252 (void**) &RSSAggr->ThisMsg))
253 return NextDBOperation(IO, RSS_FetchNetworkUsetableEntry);
258 eNextState RSS_FetchNetworkUsetableEntry(AsyncIO *IO)
262 struct cdbdata *cdbut;
263 rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
265 /* Find out if we've already seen this item */
266 strcpy(Ctx->ThisMsg->ut.ut_msgid,
267 ChrPtr(Ctx->ThisMsg->MsgGUID)); /// TODO
268 Ctx->ThisMsg->ut.ut_timestamp = time(NULL);
270 cdbut = cdb_fetch(CDB_USETABLE, SKEY(Ctx->ThisMsg->MsgGUID));
273 /* Item has already been seen */
274 EVRSSC_syslog(LOG_DEBUG,
275 "%s has already been seen\n",
276 ChrPtr(Ctx->ThisMsg->MsgGUID));
279 /* rewrite the record anyway, to update the timestamp */
280 cdb_store(CDB_USETABLE,
281 SKEY(Ctx->ThisMsg->MsgGUID),
282 &Ctx->ThisMsg->ut, sizeof(struct UseTable) );
284 if (GetNextHashPos(Ctx->Messages,
287 (void**) &Ctx->ThisMsg))
288 return NextDBOperation(
290 RSS_FetchNetworkUsetableEntry);
297 NextDBOperation(IO, RSSSaveMessage);
302 eNextState RSSAggregator_AnalyseReply(AsyncIO *IO)
305 u_char rawdigest[MD5_DIGEST_LEN];
306 struct MD5Context md5context;
308 struct cdbdata *cdbut;
309 rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
311 if (IO->HttpReq.httpcode != 200)
314 EVRSSC_syslog(LOG_ALERT, "need a 200, got a %ld !\n",
315 IO->HttpReq.httpcode);
316 // TODO: aide error message with rate limit
320 MD5Init(&md5context);
322 MD5Update(&md5context,
323 (const unsigned char*)SKEY(IO->HttpReq.ReplyData));
325 MD5Update(&md5context,
326 (const unsigned char*)SKEY(Ctx->Url));
328 MD5Final(rawdigest, &md5context);
329 guid = NewStrBufPlain(NULL,
330 MD5_DIGEST_LEN * 2 + 12 /* _rss2ctdl*/);
331 StrBufHexEscAppend(guid, NULL, rawdigest, MD5_DIGEST_LEN);
332 StrBufAppendBufPlain(guid, HKEY("_rssFM"), 0);
333 if (StrLength(guid) > 40)
334 StrBufCutAt(guid, 40, NULL);
335 /* Find out if we've already seen this item */
336 memcpy(ut.ut_msgid, SKEY(guid));
337 ut.ut_timestamp = time(NULL);
339 cdbut = cdb_fetch(CDB_USETABLE, SKEY(guid));
342 /* Item has already been seen */
343 EVRSSC_syslog(LOG_DEBUG,
344 "%s has already been seen\n",
349 /* rewrite the record anyway, to update the timestamp */
350 cdb_store(CDB_USETABLE,
352 &ut, sizeof(struct UseTable) );
354 if (cdbut != NULL) return eAbort;
356 return RSSAggregator_ParseReply(IO);
359 eNextState RSSAggregator_FinishHttp(AsyncIO *IO)
361 return QueueDBOperation(IO, RSSAggregator_AnalyseReply);
367 int rss_do_fetching(rss_aggregator *RSSAggr)
369 AsyncIO *IO = &RSSAggr->IO;
375 if ((RSSAggr->next_poll != 0) && (now < RSSAggr->next_poll))
378 ri = (rss_item*) malloc(sizeof(rss_item));
379 memset(ri, 0, sizeof(rss_item));
382 if (! InitcURLIOStruct(&RSSAggr->IO,
384 "Citadel RSS Client",
385 RSSAggregator_FinishHttp,
386 RSSAggregator_Terminate,
387 RSSAggregator_TerminateDB,
388 RSSAggregator_ShutdownAbort))
390 EVRSSCM_syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
394 safestrncpy(((CitContext*)RSSAggr->IO.CitContext)->cs_host,
395 ChrPtr(RSSAggr->Url),
396 sizeof(((CitContext*)RSSAggr->IO.CitContext)->cs_host));
398 EVRSSC_syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", ChrPtr(RSSAggr->Url));
399 ParseURL(&RSSAggr->IO.ConnectMe, RSSAggr->Url, 80);
400 CurlPrepareURL(RSSAggr->IO.ConnectMe);
402 QueueCurlContext(&RSSAggr->IO);
407 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
409 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
411 StrBuf *CfgData=NULL;
414 rss_room_counter *Count = NULL;
416 char filename[PATH_MAX];
419 rss_aggregator *RSSAggr = NULL;
420 rss_aggregator *use_this_RSSAggr = NULL;
422 const char *CfgPtr, *lPtr;
425 pthread_mutex_lock(&RSSQueueMutex);
426 if (GetHash(RSSQueueRooms, LKEY(qrbuf->QRnumber), &vptr))
428 EVRSSQ_syslog(LOG_DEBUG,
429 "rssclient: [%ld] %s already in progress.\n",
432 pthread_mutex_unlock(&RSSQueueMutex);
435 pthread_mutex_unlock(&RSSQueueMutex);
437 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
439 if (server_shutting_down)
442 /* Only do net processing for rooms that have netconfigs */
443 fd = open(filename, 0);
446 "rssclient: %s no config.\n",
451 if (server_shutting_down)
454 if (fstat(fd, &statbuf) == -1) {
455 EVRSSQ_syslog(LOG_DEBUG,
456 "ERROR: could not stat configfile '%s' - %s\n",
462 if (server_shutting_down)
465 CfgData = NewStrBufPlain(NULL, statbuf.st_size + 1);
467 if (StrBufReadBLOB(CfgData, &fd, 1, statbuf.st_size, &Err) < 0) {
469 FreeStrBuf(&CfgData);
470 EVRSSQ_syslog(LOG_ERR, "ERROR: reading config '%s' - %s<br>\n",
471 filename, strerror(errno));
475 if (server_shutting_down)
479 CfgType = NewStrBuf();
480 Line = NewStrBufPlain(NULL, StrLength(CfgData));
484 Done = StrBufSipLine(Line, CfgData, &CfgPtr) == 0;
485 if (StrLength(Line) > 0)
488 StrBufExtract_NextToken(CfgType, Line, &lPtr, '|');
489 if (!strcasecmp("rssclient", ChrPtr(CfgType)))
494 sizeof(rss_room_counter));
498 RSSAggr = (rss_aggregator *) malloc(
499 sizeof(rss_aggregator));
501 memset (RSSAggr, 0, sizeof(rss_aggregator));
502 RSSAggr->QRnumber = qrbuf->QRnumber;
503 RSSAggr->roomlist_parts = 1;
504 RSSAggr->Url = NewStrBuf();
506 StrBufExtract_NextToken(RSSAggr->Url,
511 pthread_mutex_lock(&RSSQueueMutex);
512 GetHash(RSSFetchUrls,
516 use_this_RSSAggr = (rss_aggregator *)vptr;
517 if (use_this_RSSAggr != NULL)
520 StrBufAppendBufPlain(
521 use_this_RSSAggr->rooms,
524 if (use_this_RSSAggr->roomlist_parts==1)
526 use_this_RSSAggr->OtherQRnumbers
527 = NewHash(1, lFlathash);
529 QRnumber = (long*)malloc(sizeof(long));
530 *QRnumber = qrbuf->QRnumber;
531 Put(use_this_RSSAggr->OtherQRnumbers,
532 LKEY(qrbuf->QRnumber),
535 use_this_RSSAggr->roomlist_parts++;
537 pthread_mutex_unlock(&RSSQueueMutex);
539 FreeStrBuf(&RSSAggr->Url);
544 pthread_mutex_unlock(&RSSQueueMutex);
546 RSSAggr->ItemType = RSS_UNSET;
548 RSSAggr->rooms = NewStrBufPlain(
551 pthread_mutex_lock(&RSSQueueMutex);
558 pthread_mutex_unlock(&RSSQueueMutex);
564 Count->QRnumber = qrbuf->QRnumber;
565 pthread_mutex_lock(&RSSQueueMutex);
566 EVRSSQ_syslog(LOG_DEBUG, "client: [%ld] %s now starting.\n",
567 qrbuf->QRnumber, qrbuf->QRname);
568 Put(RSSQueueRooms, LKEY(qrbuf->QRnumber), Count, NULL);
569 pthread_mutex_unlock(&RSSQueueMutex);
571 FreeStrBuf(&CfgData);
572 FreeStrBuf(&CfgType);
577 * Scan for rooms that have RSS client requests configured
579 void rssclient_scan(void) {
580 int RSSRoomCount, RSSCount;
581 rss_aggregator *rptr = NULL;
586 time_t now = time(NULL);
588 /* Run no more than once every 15 minutes. */
589 if ((now - last_run) < 900) {
590 EVRSSQ_syslog(LOG_DEBUG,
591 "Client: polling interval not yet reached; last run was %ldm%lds ago",
592 ((now - last_run) / 60),
593 ((now - last_run) % 60)
599 * This is a simple concurrency check to make sure only one rssclient
600 * run is done at a time.
602 pthread_mutex_lock(&RSSQueueMutex);
603 RSSCount = GetCount(RSSFetchUrls);
604 RSSRoomCount = GetCount(RSSQueueRooms);
605 pthread_mutex_unlock(&RSSQueueMutex);
607 if ((RSSRoomCount > 0) || (RSSCount > 0)) {
608 EVRSSQ_syslog(LOG_DEBUG,
609 "rssclient: concurrency check failed; %d rooms and %d url's are queued",
610 RSSRoomCount, RSSCount
615 become_session(&rss_CC);
616 EVRSSQM_syslog(LOG_DEBUG, "rssclient started\n");
617 CtdlForEachRoom(rssclient_scan_room, NULL);
619 pthread_mutex_lock(&RSSQueueMutex);
621 it = GetNewHashPos(RSSFetchUrls, 0);
622 while (!server_shutting_down &&
623 GetNextHashPos(RSSFetchUrls, it, &len, &Key, &vrptr) &&
625 rptr = (rss_aggregator *)vrptr;
626 if (!rss_do_fetching(rptr))
627 UnlinkRSSAggregator(rptr);
630 pthread_mutex_unlock(&RSSQueueMutex);
632 EVRSSQM_syslog(LOG_DEBUG, "rssclient ended\n");
636 void rss_cleanup(void)
638 /* citthread_mutex_destroy(&RSSQueueMutex); TODO */
639 DeleteHash(&RSSFetchUrls);
640 DeleteHash(&RSSQueueRooms);
643 void LogDebugEnableRSSClient(const int n)
645 RSSClientDebugEnabled = n;
648 CTDL_MODULE_INIT(rssclient)
652 CtdlFillSystemContext(&rss_CC, "rssclient");
653 pthread_mutex_init(&RSSQueueMutex, NULL);
654 RSSQueueRooms = NewHash(1, lFlathash);
655 RSSFetchUrls = NewHash(1, NULL);
656 syslog(LOG_INFO, "%s\n", curl_version());
657 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);
658 CtdlRegisterEVCleanupHook(rss_cleanup);
659 CtdlRegisterDebugFlagHook(HKEY("rssclient"), LogDebugEnableRSSClient, &RSSClientDebugEnabled);