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.
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.See the
14 * GNU General Public License for more details.
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA02111-1307USA
25 #if TIME_WITH_SYS_TIME
26 # include <sys/time.h>
39 #include <sys/types.h>
42 #include <curl/curl.h>
43 #include <libcitadel.h>
46 #include "citserver.h"
50 #include "ctdl_module.h"
52 #include "parsedate.h"
54 #include "citadel_dirs.h"
57 #include "event_client.h"
58 #include "rss_atom_parser.h"
61 #define TMP_MSGDATA 0xFF
62 #define TMP_SHORTER_URL_OFFSET 0xFE
63 #define TMP_SHORTER_URLS 0xFD
67 pthread_mutex_t RSSQueueMutex; /* locks the access to the following vars: */
68 HashList *RSSQueueRooms = NULL; /* rss_room_counter */
69 HashList *RSSFetchUrls = NULL; /*->rss_aggregator;->RefCount access locked*/
71 eNextState RSSAggregator_Terminate(AsyncIO *IO);
72 eNextState RSSAggregator_TerminateDB(AsyncIO *IO);
73 eNextState RSSAggregator_ShutdownAbort(AsyncIO *IO);
74 struct CitContext rss_CC;
76 struct rssnetcfg *rnclist = NULL;
79 void DeleteRoomReference(long QRnumber)
85 rss_room_counter *pRoomC;
87 At = GetNewHashPos(RSSQueueRooms, 0);
89 if (GetHashPosFromKey(RSSQueueRooms, LKEY(QRnumber), At))
91 GetHashPos(RSSQueueRooms, At, &HKLen, &HK, &vData);
94 pRoomC = (rss_room_counter *) vData;
96 if (pRoomC->count == 0)
97 DeleteEntryFromHash(RSSQueueRooms, At);
103 void UnlinkRooms(rss_aggregator *Cfg)
105 DeleteRoomReference(Cfg->QRnumber);
106 if (Cfg->OtherQRnumbers != NULL)
113 At = GetNewHashPos(Cfg->OtherQRnumbers, 0);
114 while (! server_shutting_down &&
115 GetNextHashPos(Cfg->OtherQRnumbers,
121 long *lData = (long*) vData;
122 DeleteRoomReference(*lData);
129 void UnlinkRSSAggregator(rss_aggregator *Cfg)
135 At = GetNewHashPos(RSSFetchUrls, 0);
136 if (GetHashPosFromKey(RSSFetchUrls, SKEY(Cfg->Url), At))
138 DeleteEntryFromHash(RSSFetchUrls, At);
141 last_run = time(NULL);
144 void DeleteRssCfg(void *vptr)
146 rss_aggregator *RSSAggr = (rss_aggregator *)vptr;
147 AsyncIO *IO = &RSSAggr->IO;
148 EVM_syslog(LOG_DEBUG, "RSS: destroying\n");
150 FreeStrBuf(&RSSAggr->Url);
151 FreeStrBuf(&RSSAggr->rooms);
152 FreeStrBuf(&RSSAggr->CData);
153 FreeStrBuf(&RSSAggr->Key);
154 DeleteHash(&RSSAggr->OtherQRnumbers);
156 DeleteHashPos (&RSSAggr->Pos);
157 DeleteHash (&RSSAggr->Messages);
158 if (RSSAggr->recp.recp_room != NULL)
159 free(RSSAggr->recp.recp_room);
162 if (RSSAggr->Item != NULL)
164 flush_rss_item(RSSAggr->Item);
169 FreeAsyncIOContents(&RSSAggr->IO);
173 eNextState RSSAggregator_Terminate(AsyncIO *IO)
175 rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
177 EVM_syslog(LOG_DEBUG, "RSS: Terminating.\n");
180 UnlinkRSSAggregator(RSSAggr);
184 eNextState RSSAggregator_TerminateDB(AsyncIO *IO)
186 rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
188 EVM_syslog(LOG_DEBUG, "RSS: Terminating.\n");
191 UnlinkRSSAggregator(RSSAggr);
195 eNextState RSSAggregator_ShutdownAbort(AsyncIO *IO)
198 rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
200 pUrl = IO->ConnectMe->PlainUrl;
204 EV_syslog(LOG_DEBUG, "RSS: Aborting by shutdown: %s.\n", pUrl);
207 UnlinkRSSAggregator(RSSAggr);
212 eNextState AbortNetworkSaveMessage (AsyncIO *IO)
214 return eAbort; ///TODO
217 eNextState RSSSaveMessage(AsyncIO *IO)
221 rss_aggregator *RSSAggr = (rss_aggregator *) IO->Data;
223 RSSAggr->ThisMsg->Msg.cm_fields['M'] =
224 SmashStrBuf(&RSSAggr->ThisMsg->Message);
226 CtdlSubmitMsg(&RSSAggr->ThisMsg->Msg, &RSSAggr->recp, NULL, 0);
228 /* write the uidl to the use table so we don't store this item again */
229 cdb_store(CDB_USETABLE,
230 SKEY(RSSAggr->ThisMsg->MsgGUID),
231 &RSSAggr->ThisMsg->ut,
232 sizeof(struct UseTable) );
234 if (GetNextHashPos(RSSAggr->Messages,
237 (void**) &RSSAggr->ThisMsg))
238 return NextDBOperation(IO, RSS_FetchNetworkUsetableEntry);
243 eNextState RSS_FetchNetworkUsetableEntry(AsyncIO *IO)
247 struct cdbdata *cdbut;
248 rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
250 /* Find out if we've already seen this item */
251 strcpy(Ctx->ThisMsg->ut.ut_msgid,
252 ChrPtr(Ctx->ThisMsg->MsgGUID)); /// TODO
253 Ctx->ThisMsg->ut.ut_timestamp = time(NULL);
255 cdbut = cdb_fetch(CDB_USETABLE, SKEY(Ctx->ThisMsg->MsgGUID));
258 /* Item has already been seen */
260 "%s has already been seen\n",
261 ChrPtr(Ctx->ThisMsg->MsgGUID));
264 /* rewrite the record anyway, to update the timestamp */
265 cdb_store(CDB_USETABLE,
266 SKEY(Ctx->ThisMsg->MsgGUID),
267 &Ctx->ThisMsg->ut, sizeof(struct UseTable) );
269 if (GetNextHashPos(Ctx->Messages,
272 (void**) &Ctx->ThisMsg))
273 return NextDBOperation(
275 RSS_FetchNetworkUsetableEntry);
282 NextDBOperation(IO, RSSSaveMessage);
290 int rss_do_fetching(rss_aggregator *Cfg)
297 if ((Cfg->next_poll != 0) && (now < Cfg->next_poll))
300 ri = (rss_item*) malloc(sizeof(rss_item));
301 memset(ri, 0, sizeof(rss_item));
304 if (! InitcURLIOStruct(&Cfg->IO,
306 "Citadel RSS Client",
307 RSSAggregator_ParseReply,
308 RSSAggregator_Terminate,
309 RSSAggregator_TerminateDB,
310 RSSAggregator_ShutdownAbort))
312 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
316 safestrncpy(((CitContext*)Cfg->IO.CitContext)->cs_host,
318 sizeof(((CitContext*)Cfg->IO.CitContext)->cs_host));
320 syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", ChrPtr(Cfg->Url));
321 ParseURL(&Cfg->IO.ConnectMe, Cfg->Url, 80);
322 CurlPrepareURL(Cfg->IO.ConnectMe);
324 QueueCurlContext(&Cfg->IO);
329 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
331 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
333 StrBuf *CfgData=NULL;
336 rss_room_counter *Count = NULL;
338 char filename[PATH_MAX];
341 rss_aggregator *RSSAggr = NULL;
342 rss_aggregator *use_this_RSSAggr = NULL;
344 const char *CfgPtr, *lPtr;
347 pthread_mutex_lock(&RSSQueueMutex);
348 if (GetHash(RSSQueueRooms, LKEY(qrbuf->QRnumber), &vptr))
351 "rssclient: [%ld] %s already in progress.\n",
354 pthread_mutex_unlock(&RSSQueueMutex);
357 pthread_mutex_unlock(&RSSQueueMutex);
359 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
361 if (server_shutting_down)
364 /* Only do net processing for rooms that have netconfigs */
365 fd = open(filename, 0);
368 "rssclient: %s no config.\n",
373 if (server_shutting_down)
376 if (fstat(fd, &statbuf) == -1) {
378 "ERROR: could not stat configfile '%s' - %s\n",
384 if (server_shutting_down)
387 CfgData = NewStrBufPlain(NULL, statbuf.st_size + 1);
389 if (StrBufReadBLOB(CfgData, &fd, 1, statbuf.st_size, &Err) < 0) {
391 FreeStrBuf(&CfgData);
392 syslog(LOG_DEBUG, "ERROR: reading config '%s' - %s<br>\n",
393 filename, strerror(errno));
397 if (server_shutting_down)
401 CfgType = NewStrBuf();
402 Line = NewStrBufPlain(NULL, StrLength(CfgData));
406 Done = StrBufSipLine(Line, CfgData, &CfgPtr) == 0;
407 if (StrLength(Line) > 0)
410 StrBufExtract_NextToken(CfgType, Line, &lPtr, '|');
411 if (!strcasecmp("rssclient", ChrPtr(CfgType)))
416 sizeof(rss_room_counter));
420 RSSAggr = (rss_aggregator *) malloc(
421 sizeof(rss_aggregator));
423 memset (RSSAggr, 0, sizeof(rss_aggregator));
424 RSSAggr->QRnumber = qrbuf->QRnumber;
425 RSSAggr->roomlist_parts = 1;
426 RSSAggr->Url = NewStrBuf();
428 StrBufExtract_NextToken(RSSAggr->Url,
433 pthread_mutex_lock(&RSSQueueMutex);
434 GetHash(RSSFetchUrls,
438 use_this_RSSAggr = (rss_aggregator *)vptr;
439 if (use_this_RSSAggr != NULL)
442 StrBufAppendBufPlain(
443 use_this_RSSAggr->rooms,
446 if (use_this_RSSAggr->roomlist_parts==1)
448 use_this_RSSAggr->OtherQRnumbers
449 = NewHash(1, lFlathash);
451 QRnumber = (long*)malloc(sizeof(long));
452 *QRnumber = qrbuf->QRnumber;
453 Put(use_this_RSSAggr->OtherQRnumbers,
454 LKEY(qrbuf->QRnumber),
457 use_this_RSSAggr->roomlist_parts++;
459 pthread_mutex_unlock(&RSSQueueMutex);
461 FreeStrBuf(&RSSAggr->Url);
466 pthread_mutex_unlock(&RSSQueueMutex);
468 RSSAggr->ItemType = RSS_UNSET;
470 RSSAggr->rooms = NewStrBufPlain(
473 pthread_mutex_lock(&RSSQueueMutex);
480 pthread_mutex_unlock(&RSSQueueMutex);
486 Count->QRnumber = qrbuf->QRnumber;
487 pthread_mutex_lock(&RSSQueueMutex);
488 syslog(LOG_DEBUG, "rssclient: [%ld] %s now starting.\n",
489 qrbuf->QRnumber, qrbuf->QRname);
490 Put(RSSQueueRooms, LKEY(qrbuf->QRnumber), Count, NULL);
491 pthread_mutex_unlock(&RSSQueueMutex);
493 FreeStrBuf(&CfgData);
494 FreeStrBuf(&CfgType);
499 * Scan for rooms that have RSS client requests configured
501 void rssclient_scan(void) {
502 rss_aggregator *rptr = NULL;
507 time_t now = time(NULL);
509 /* Run no more than once every 15 minutes. */
510 if ((now - last_run) < 900) {
515 * This is a simple concurrency check to make sure only one rssclient
516 * run is done at a time.We could do this with a mutex, but since we
517 * don't really require extremely fine granularity here, we'll do it
518 * with a static variable instead.
521 if ((GetCount(RSSQueueRooms) > 0) || (GetCount(RSSFetchUrls) > 0))
524 become_session(&rss_CC);
525 syslog(LOG_DEBUG, "rssclient started\n");
526 CtdlForEachRoom(rssclient_scan_room, NULL);
528 pthread_mutex_lock(&RSSQueueMutex);
530 it = GetNewHashPos(RSSFetchUrls, 0);
531 while (!server_shutting_down &&
532 GetNextHashPos(RSSFetchUrls, it, &len, &Key, &vrptr) &&
534 rptr = (rss_aggregator *)vrptr;
535 if (!rss_do_fetching(rptr))
536 UnlinkRSSAggregator(rptr);
539 pthread_mutex_unlock(&RSSQueueMutex);
541 syslog(LOG_DEBUG, "rssclient ended\n");
545 void rss_cleanup(void)
547 /* citthread_mutex_destroy(&RSSQueueMutex); TODO */
548 DeleteHash(&RSSFetchUrls);
549 DeleteHash(&RSSQueueRooms);
553 CTDL_MODULE_INIT(rssclient)
557 CtdlFillSystemContext(&rss_CC, "rssclient");
558 pthread_mutex_init(&RSSQueueMutex, NULL);
559 RSSQueueRooms = NewHash(1, lFlathash);
560 RSSFetchUrls = NewHash(1, NULL);
561 syslog(LOG_INFO, "%s\n", curl_version());
562 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
563 CtdlRegisterEVCleanupHook(rss_cleanup);