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);
305 int rss_do_fetching(rss_aggregator *RSSAggr)
307 AsyncIO *IO = &RSSAggr->IO;
313 if ((RSSAggr->next_poll != 0) && (now < RSSAggr->next_poll))
316 ri = (rss_item*) malloc(sizeof(rss_item));
317 memset(ri, 0, sizeof(rss_item));
320 if (! InitcURLIOStruct(&RSSAggr->IO,
322 "Citadel RSS Client",
323 RSSAggregator_ParseReply,
324 RSSAggregator_Terminate,
325 RSSAggregator_TerminateDB,
326 RSSAggregator_ShutdownAbort))
328 EVRSSCM_syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
332 safestrncpy(((CitContext*)RSSAggr->IO.CitContext)->cs_host,
333 ChrPtr(RSSAggr->Url),
334 sizeof(((CitContext*)RSSAggr->IO.CitContext)->cs_host));
336 EVRSSC_syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", ChrPtr(RSSAggr->Url));
337 ParseURL(&RSSAggr->IO.ConnectMe, RSSAggr->Url, 80);
338 CurlPrepareURL(RSSAggr->IO.ConnectMe);
340 QueueCurlContext(&RSSAggr->IO);
345 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
347 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
349 StrBuf *CfgData=NULL;
352 rss_room_counter *Count = NULL;
354 char filename[PATH_MAX];
357 rss_aggregator *RSSAggr = NULL;
358 rss_aggregator *use_this_RSSAggr = NULL;
360 const char *CfgPtr, *lPtr;
363 pthread_mutex_lock(&RSSQueueMutex);
364 if (GetHash(RSSQueueRooms, LKEY(qrbuf->QRnumber), &vptr))
366 EVRSSQ_syslog(LOG_DEBUG,
367 "rssclient: [%ld] %s already in progress.\n",
370 pthread_mutex_unlock(&RSSQueueMutex);
373 pthread_mutex_unlock(&RSSQueueMutex);
375 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
377 if (server_shutting_down)
380 /* Only do net processing for rooms that have netconfigs */
381 fd = open(filename, 0);
384 "rssclient: %s no config.\n",
389 if (server_shutting_down)
392 if (fstat(fd, &statbuf) == -1) {
393 EVRSSQ_syslog(LOG_DEBUG,
394 "ERROR: could not stat configfile '%s' - %s\n",
400 if (server_shutting_down)
403 CfgData = NewStrBufPlain(NULL, statbuf.st_size + 1);
405 if (StrBufReadBLOB(CfgData, &fd, 1, statbuf.st_size, &Err) < 0) {
407 FreeStrBuf(&CfgData);
408 EVRSSQ_syslog(LOG_ERR, "ERROR: reading config '%s' - %s<br>\n",
409 filename, strerror(errno));
413 if (server_shutting_down)
417 CfgType = NewStrBuf();
418 Line = NewStrBufPlain(NULL, StrLength(CfgData));
422 Done = StrBufSipLine(Line, CfgData, &CfgPtr) == 0;
423 if (StrLength(Line) > 0)
426 StrBufExtract_NextToken(CfgType, Line, &lPtr, '|');
427 if (!strcasecmp("rssclient", ChrPtr(CfgType)))
432 sizeof(rss_room_counter));
436 RSSAggr = (rss_aggregator *) malloc(
437 sizeof(rss_aggregator));
439 memset (RSSAggr, 0, sizeof(rss_aggregator));
440 RSSAggr->QRnumber = qrbuf->QRnumber;
441 RSSAggr->roomlist_parts = 1;
442 RSSAggr->Url = NewStrBuf();
444 StrBufExtract_NextToken(RSSAggr->Url,
449 pthread_mutex_lock(&RSSQueueMutex);
450 GetHash(RSSFetchUrls,
454 use_this_RSSAggr = (rss_aggregator *)vptr;
455 if (use_this_RSSAggr != NULL)
458 StrBufAppendBufPlain(
459 use_this_RSSAggr->rooms,
462 if (use_this_RSSAggr->roomlist_parts==1)
464 use_this_RSSAggr->OtherQRnumbers
465 = NewHash(1, lFlathash);
467 QRnumber = (long*)malloc(sizeof(long));
468 *QRnumber = qrbuf->QRnumber;
469 Put(use_this_RSSAggr->OtherQRnumbers,
470 LKEY(qrbuf->QRnumber),
473 use_this_RSSAggr->roomlist_parts++;
475 pthread_mutex_unlock(&RSSQueueMutex);
477 FreeStrBuf(&RSSAggr->Url);
482 pthread_mutex_unlock(&RSSQueueMutex);
484 RSSAggr->ItemType = RSS_UNSET;
486 RSSAggr->rooms = NewStrBufPlain(
489 pthread_mutex_lock(&RSSQueueMutex);
496 pthread_mutex_unlock(&RSSQueueMutex);
502 Count->QRnumber = qrbuf->QRnumber;
503 pthread_mutex_lock(&RSSQueueMutex);
504 EVRSSQ_syslog(LOG_DEBUG, "client: [%ld] %s now starting.\n",
505 qrbuf->QRnumber, qrbuf->QRname);
506 Put(RSSQueueRooms, LKEY(qrbuf->QRnumber), Count, NULL);
507 pthread_mutex_unlock(&RSSQueueMutex);
509 FreeStrBuf(&CfgData);
510 FreeStrBuf(&CfgType);
515 * Scan for rooms that have RSS client requests configured
517 void rssclient_scan(void) {
518 int RSSRoomCount, RSSCount;
519 rss_aggregator *rptr = NULL;
524 time_t now = time(NULL);
526 /* Run no more than once every 15 minutes. */
527 if ((now - last_run) < 900) {
528 EVRSSQ_syslog(LOG_DEBUG,
529 "Client: polling interval not yet reached; last run was %ldm%lds ago",
530 ((now - last_run) / 60),
531 ((now - last_run) % 60)
537 * This is a simple concurrency check to make sure only one rssclient
538 * run is done at a time.
540 pthread_mutex_lock(&RSSQueueMutex);
541 RSSCount = GetCount(RSSFetchUrls);
542 RSSRoomCount = GetCount(RSSQueueRooms);
543 pthread_mutex_unlock(&RSSQueueMutex);
545 if ((RSSRoomCount > 0) || (RSSCount > 0)) {
546 EVRSSQ_syslog(LOG_DEBUG,
547 "rssclient: concurrency check failed; %d rooms and %d url's are queued",
548 RSSRoomCount, RSSCount
553 become_session(&rss_CC);
554 EVRSSQM_syslog(LOG_DEBUG, "rssclient started\n");
555 CtdlForEachRoom(rssclient_scan_room, NULL);
557 pthread_mutex_lock(&RSSQueueMutex);
559 it = GetNewHashPos(RSSFetchUrls, 0);
560 while (!server_shutting_down &&
561 GetNextHashPos(RSSFetchUrls, it, &len, &Key, &vrptr) &&
563 rptr = (rss_aggregator *)vrptr;
564 if (!rss_do_fetching(rptr))
565 UnlinkRSSAggregator(rptr);
568 pthread_mutex_unlock(&RSSQueueMutex);
570 EVRSSQM_syslog(LOG_DEBUG, "rssclient ended\n");
574 void rss_cleanup(void)
576 /* citthread_mutex_destroy(&RSSQueueMutex); TODO */
577 DeleteHash(&RSSFetchUrls);
578 DeleteHash(&RSSQueueRooms);
581 void LogDebugEnableRSSClient(const int n)
583 RSSClientDebugEnabled = n;
586 CTDL_MODULE_INIT(rssclient)
590 CtdlFillSystemContext(&rss_CC, "rssclient");
591 pthread_mutex_init(&RSSQueueMutex, NULL);
592 RSSQueueRooms = NewHash(1, lFlathash);
593 RSSFetchUrls = NewHash(1, NULL);
594 syslog(LOG_INFO, "%s\n", curl_version());
595 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);
596 CtdlRegisterEVCleanupHook(rss_cleanup);
597 CtdlRegisterDebugFlagHook(HKEY("rssclient"), LogDebugEnableRSSClient, &RSSClientDebugEnabled);