Route all access to cm_fields[] through api functions
[citadel.git] / citadel / modules / rssclient / serv_rssclient.c
1 /*
2  * Bring external RSS feeds into rooms.
3  *
4  * Copyright (c) 2007-2012 by the citadel.org team
5  *
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.
8  *
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.
13  */
14
15 #include <stdlib.h>
16 #include <unistd.h>
17 #include <stdio.h>
18
19 #if TIME_WITH_SYS_TIME
20 # include <sys/time.h>
21 # include <time.h>
22 #else
23 # if HAVE_SYS_TIME_H
24 #include <sys/time.h>
25 # else
26 #include <time.h>
27 # endif
28 #endif
29
30 #include <ctype.h>
31 #include <string.h>
32 #include <errno.h>
33 #include <sys/types.h>
34 #include <sys/stat.h>
35 #include <expat.h>
36 #include <curl/curl.h>
37 #include <libcitadel.h>
38 #include "citadel.h"
39 #include "server.h"
40 #include "citserver.h"
41 #include "support.h"
42 #include "config.h"
43 #include "threads.h"
44 #include "ctdl_module.h"
45 #include "msgbase.h"
46 #include "parsedate.h"
47 #include "database.h"
48 #include "citadel_dirs.h"
49 #include "md5.h"
50 #include "context.h"
51 #include "event_client.h"
52 #include "rss_atom_parser.h"
53
54
55 #define TMP_MSGDATA 0xFF
56 #define TMP_SHORTER_URL_OFFSET 0xFE
57 #define TMP_SHORTER_URLS 0xFD
58
59 time_t last_run = 0L;
60
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*/
64
65 eNextState RSSAggregator_Terminate(AsyncIO *IO);
66 eNextState RSSAggregator_TerminateDB(AsyncIO *IO);
67 eNextState RSSAggregator_ShutdownAbort(AsyncIO *IO);
68 struct CitContext rss_CC;
69
70 struct rssnetcfg *rnclist = NULL;
71 int RSSClientDebugEnabled = 0;
72 #define N ((rss_aggregator*)IO->Data)->Cfg.QRnumber
73
74 #define DBGLOG(LEVEL) if ((LEVEL != LOG_DEBUG) || (RSSClientDebugEnabled != 0))
75
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__)
80
81 #define EVRSSCM_syslog(LEVEL, FORMAT)                                   \
82         DBGLOG(LEVEL) syslog(LEVEL,                                     \
83                              "IO[%ld]CC[%d][%ld]RSS" FORMAT,            \
84                              IO->ID, CCID, N)
85
86 #define EVRSSQ_syslog(LEVEL, FORMAT, ...)                               \
87         DBGLOG(LEVEL) syslog(LEVEL, "RSS" FORMAT,                       \
88                              __VA_ARGS__)
89 #define EVRSSQM_syslog(LEVEL, FORMAT)                   \
90         DBGLOG(LEVEL) syslog(LEVEL, "RSS" FORMAT)
91
92 #define EVRSSCSM_syslog(LEVEL, FORMAT)                                  \
93         DBGLOG(LEVEL) syslog(LEVEL, "IO[%ld][%ld]RSS" FORMAT,           \
94                              IO->ID, N)
95
96 typedef enum _RSSState {
97         eRSSCreated,
98         eRSSFetching,
99         eRSSFailure,
100         eRSSParsing,
101         eRSSUT
102 } RSSState;
103 ConstStr RSSStates[] = {
104         {HKEY("Aggregator created")},
105         {HKEY("Fetching content")},
106         {HKEY("Failed")},
107         {HKEY("parsing content")},
108         {HKEY("checking usetable")}
109 };
110
111 static void SetRSSState(AsyncIO *IO, RSSState State)
112 {
113         CitContext* CCC = IO->CitContext;
114         if (CCC != NULL)
115                 memcpy(CCC->cs_clientname, RSSStates[State].Key, RSSStates[State].len + 1);
116 }
117
118 void DeleteRoomReference(long QRnumber)
119 {
120         HashPos *At;
121         long HKLen;
122         const char *HK;
123         void *vData = NULL;
124         rss_room_counter *pRoomC;
125
126         At = GetNewHashPos(RSSQueueRooms, 0);
127
128         if (GetHashPosFromKey(RSSQueueRooms, LKEY(QRnumber), At))
129         {
130                 GetHashPos(RSSQueueRooms, At, &HKLen, &HK, &vData);
131                 if (vData != NULL)
132                 {
133                         pRoomC = (rss_room_counter *) vData;
134                         pRoomC->count --;
135                         if (pRoomC->count == 0)
136                                 DeleteEntryFromHash(RSSQueueRooms, At);
137                 }
138         }
139         DeleteHashPos(&At);
140 }
141
142 void UnlinkRooms(rss_aggregator *RSSAggr)
143 {
144         DeleteRoomReference(RSSAggr->Cfg.QRnumber);
145         if (RSSAggr->OtherQRnumbers != NULL)
146         {
147                 long HKLen;
148                 const char *HK;
149                 HashPos *At;
150                 void *vData;
151
152                 At = GetNewHashPos(RSSAggr->OtherQRnumbers, 0);
153                 while (! server_shutting_down &&
154                        GetNextHashPos(RSSAggr->OtherQRnumbers,
155                                       At,
156                                       &HKLen, &HK,
157                                       &vData) &&
158                        (vData != NULL))
159                 {
160                         pRSSConfig *Data = (pRSSConfig*) vData;
161                         DeleteRoomReference(Data->QRnumber);
162                 }
163
164                 DeleteHashPos(&At);
165         }
166 }
167
168 void UnlinkRSSAggregator(rss_aggregator *RSSAggr)
169 {
170         HashPos *At;
171
172         pthread_mutex_lock(&RSSQueueMutex);
173         UnlinkRooms(RSSAggr);
174
175         At = GetNewHashPos(RSSFetchUrls, 0);
176         if (GetHashPosFromKey(RSSFetchUrls, SKEY(RSSAggr->Url), At))
177         {
178                 DeleteEntryFromHash(RSSFetchUrls, At);
179         }
180         DeleteHashPos(&At);
181         last_run = time(NULL);
182         pthread_mutex_unlock(&RSSQueueMutex);
183 }
184
185 void DeleteRssCfg(void *vptr)
186 {
187         rss_aggregator *RSSAggr = (rss_aggregator *)vptr;
188         AsyncIO *IO = &RSSAggr->IO;
189
190         if (IO->CitContext != NULL)
191                 EVRSSCM_syslog(LOG_DEBUG, "RSS: destroying\n");
192
193         FreeStrBuf(&RSSAggr->Url);
194         FreeStrBuf(&RSSAggr->rooms);
195         FreeStrBuf(&RSSAggr->CData);
196         FreeStrBuf(&RSSAggr->Key);
197         DeleteHash(&RSSAggr->OtherQRnumbers);
198
199         DeleteHashPos (&RSSAggr->Pos);
200         DeleteHash (&RSSAggr->Messages);
201         if (RSSAggr->recp.recp_room != NULL)
202                 free(RSSAggr->recp.recp_room);
203
204
205         if (RSSAggr->Item != NULL)
206         {
207                 flush_rss_item(RSSAggr->Item);
208
209                 free(RSSAggr->Item);
210         }
211
212         FreeAsyncIOContents(&RSSAggr->IO);
213         memset(RSSAggr, 0, sizeof(rss_aggregator));
214         free(RSSAggr);
215 }
216
217 eNextState RSSAggregator_Terminate(AsyncIO *IO)
218 {
219         rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
220
221         EVRSSCM_syslog(LOG_DEBUG, "RSS: Terminating.\n");
222
223         StopCurlWatchers(IO);
224         UnlinkRSSAggregator(RSSAggr);
225         return eAbort;
226 }
227
228 eNextState RSSAggregator_TerminateDB(AsyncIO *IO)
229 {
230         rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
231
232         EVRSSCM_syslog(LOG_DEBUG, "RSS: Terminating.\n");
233
234
235         StopDBWatchers(&RSSAggr->IO);
236         UnlinkRSSAggregator(RSSAggr);
237         return eAbort;
238 }
239
240 eNextState RSSAggregator_ShutdownAbort(AsyncIO *IO)
241 {
242         const char *pUrl;
243         rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
244
245         pUrl = IO->ConnectMe->PlainUrl;
246         if (pUrl == NULL)
247                 pUrl = "";
248
249         EVRSSC_syslog(LOG_DEBUG, "RSS: Aborting by shutdown: %s.\n", pUrl);
250
251         StopCurlWatchers(IO);
252         UnlinkRSSAggregator(RSSAggr);
253         return eAbort;
254 }
255
256 void AppendLink(StrBuf *Message,
257                 StrBuf *link,
258                 StrBuf *LinkTitle,
259                 const char *Title)
260 {
261         if (StrLength(link) > 0)
262         {
263                 StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
264                 StrBufAppendBuf(Message, link, 0);
265                 StrBufAppendBufPlain(Message, HKEY("\">"), 0);
266                 if (StrLength(LinkTitle) > 0)
267                         StrBufAppendBuf(Message, LinkTitle, 0);
268                 else if ((Title != NULL) && !IsEmptyStr(Title))
269                         StrBufAppendBufPlain(Message, Title, -1, 0);
270                 else
271                         StrBufAppendBuf(Message, link, 0);
272                 StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
273         }
274 }
275
276
277 int rss_format_item(AsyncIO *IO, networker_save_message *SaveMsg)
278 {
279         StrBuf *Message;
280         int msglen = 0;
281
282         if (StrLength(SaveMsg->description) + 
283             StrLength(SaveMsg->link) + 
284             StrLength(SaveMsg->linkTitle) + 
285             StrLength(SaveMsg->reLink) +
286             StrLength(SaveMsg->reLinkTitle) +
287             StrLength(SaveMsg->title) == 0)
288         {
289                 EVRSSCM_syslog(LOG_INFO, "Refusing to save empty message.");
290                 return 0;
291         }
292
293         if (SaveMsg->author_or_creator != NULL) {
294
295                 char *From;
296                 StrBuf *Encoded = NULL;
297                 int FromAt;
298
299                 From = html_to_ascii(ChrPtr(SaveMsg->author_or_creator),
300                                      StrLength(SaveMsg->author_or_creator),
301                                      512, 0);
302                 StrBufPlain(SaveMsg->author_or_creator, From, -1);
303                 StrBufTrim(SaveMsg->author_or_creator);
304                 free(From);
305
306                 FromAt = strchr(ChrPtr(SaveMsg->author_or_creator), '@') != NULL;
307                 if (!FromAt && StrLength (SaveMsg->author_email) > 0)
308                 {
309                         StrBufRFC2047encode(&Encoded, SaveMsg->author_or_creator);
310                         CM_SetAsFieldSB(&SaveMsg->Msg, eAuthor, &Encoded);
311                         CM_SetAsFieldSB(&SaveMsg->Msg, eMessagePath, &SaveMsg->author_email);
312                 }
313                 else
314                 {
315                         if (FromAt)
316                         {
317                                 CM_SetAsFieldSB(&SaveMsg->Msg, eAuthor, &SaveMsg->author_or_creator);
318                                 CM_CopyField(&SaveMsg->Msg, eMessagePath, eAuthor);
319                         }
320                         else
321                         {
322                                 StrBufRFC2047encode(&Encoded,
323                                                     SaveMsg->author_or_creator);
324                                 CM_SetAsFieldSB(&SaveMsg->Msg, eAuthor, &Encoded);
325                                 CM_SetField(&SaveMsg->Msg, eMessagePath, HKEY("rss@localhost"));
326
327                         }
328                 }
329         }
330         else {
331                 CM_SetField(&SaveMsg->Msg, eAuthor, HKEY("rss"));
332         }
333
334         CM_SetField(&SaveMsg->Msg, eNodeName, NODENAME, strlen(NODENAME));
335         if (SaveMsg->title != NULL) {
336                 long len;
337                 char *Sbj;
338                 StrBuf *Encoded, *QPEncoded;
339
340                 QPEncoded = NULL;
341                 StrBufSpaceToBlank(SaveMsg->title);
342                 len = StrLength(SaveMsg->title);
343                 Sbj = html_to_ascii(ChrPtr(SaveMsg->title), len, 512, 0);
344                 len = strlen(Sbj);
345                 if ((len > 0) && (Sbj[len - 1] == '\n'))
346                 {
347                         len --;
348                         Sbj[len] = '\0';
349                 }
350                 Encoded = NewStrBufPlain(Sbj, len);
351                 free(Sbj);
352
353                 StrBufTrim(Encoded);
354                 StrBufRFC2047encode(&QPEncoded, Encoded);
355
356                 CM_SetAsFieldSB(&SaveMsg->Msg, eMsgSubject, &QPEncoded);
357                 FreeStrBuf(&Encoded);
358         }
359         if (SaveMsg->link == NULL)
360                 SaveMsg->link = NewStrBufPlain(HKEY(""));
361
362 #if 0 /* temporarily disable shorter urls. */
363         SaveMsg->Msg.cm_fields[TMP_SHORTER_URLS] =
364                 GetShorterUrls(SaveMsg->description);
365 #endif
366
367         msglen += 1024 + StrLength(SaveMsg->link) + StrLength(SaveMsg->description) ;
368
369         Message = NewStrBufPlain(NULL, msglen);
370
371         StrBufPlain(Message, HKEY(
372                             "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
373                             "<html><body>\n"));
374 #if 0 /* disable shorter url for now. */
375         SaveMsg->Msg.cm_fields[TMP_SHORTER_URL_OFFSET] = StrLength(Message);
376 #endif
377         StrBufAppendBuf(Message, SaveMsg->description, 0);
378         StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
379
380         AppendLink(Message, SaveMsg->link, SaveMsg->linkTitle, NULL);
381         AppendLink(Message, SaveMsg->reLink, SaveMsg->reLinkTitle, "Reply to this");
382         StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
383
384         SaveMsg->Message = Message;
385         return 1;
386 }
387
388 eNextState RSSSaveMessage(AsyncIO *IO)
389 {
390         long len;
391         const char *Key;
392         rss_aggregator *RSSAggr = (rss_aggregator *) IO->Data;
393
394         if (rss_format_item(IO, RSSAggr->ThisMsg))
395         {
396                 CM_SetAsFieldSB(&RSSAggr->ThisMsg->Msg, eMesageText,
397                                        &RSSAggr->ThisMsg->Message);
398
399                 CtdlSubmitMsg(&RSSAggr->ThisMsg->Msg, &RSSAggr->recp, NULL, 0);
400                 
401                 /* write the uidl to the use table so we don't store this item again */
402                 
403                 CheckIfAlreadySeen("RSS Item Insert", RSSAggr->ThisMsg->MsgGUID, IO->Now, 0, eWrite, CCID, IO->ID);
404         }
405
406         if (GetNextHashPos(RSSAggr->Messages,
407                            RSSAggr->Pos,
408                            &len, &Key,
409                            (void**) &RSSAggr->ThisMsg))
410                 return NextDBOperation(IO, RSS_FetchNetworkUsetableEntry);
411         else
412                 return eAbort;
413 }
414
415 eNextState RSS_FetchNetworkUsetableEntry(AsyncIO *IO)
416 {
417 #ifndef DEBUG_RSS
418         const char *Key;
419         long len;
420         rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
421
422         /* Find out if we've already seen this item */
423 // todo: expiry?
424         SetRSSState(IO, eRSSUT);
425         if (CheckIfAlreadySeen("RSS Item Seen",
426                                Ctx->ThisMsg->MsgGUID,
427                                IO->Now,
428                                IO->Now - USETABLE_ANTIEXPIRE_HIRES,
429                                eCheckUpdate,
430                                CCID, IO->ID)
431             != 0)
432         {
433                 /* Item has already been seen */
434                 EVRSSC_syslog(LOG_DEBUG,
435                           "%s has already been seen\n",
436                           ChrPtr(Ctx->ThisMsg->MsgGUID));
437                 SetRSSState(IO, eRSSParsing);
438
439                 if (GetNextHashPos(Ctx->Messages,
440                                    Ctx->Pos,
441                                    &len, &Key,
442                                    (void**) &Ctx->ThisMsg))
443                         return NextDBOperation(
444                                 IO,
445                                 RSS_FetchNetworkUsetableEntry);
446                 else
447                         return eAbort;
448         }
449         else
450 #endif
451         {
452                 SetRSSState(IO, eRSSParsing);
453
454                 NextDBOperation(IO, RSSSaveMessage);
455                 return eSendMore;
456         }
457         return eSendMore;
458 }
459
460 void UpdateLastKnownGood(pRSSConfig *pCfg, time_t now)
461 {
462         OneRoomNetCfg* pRNCfg;
463         begin_critical_section(S_NETCONFIGS);
464         pRNCfg = CtdlGetNetCfgForRoom (pCfg->QRnumber);
465         if (pRNCfg != NULL)
466         {
467                 RSSCfgLine *RSSCfg = (RSSCfgLine *)pRNCfg->NetConfigs[rssclient];
468
469                 while (RSSCfg != NULL)
470                 {
471                         if (RSSCfg == pCfg->pCfg)
472                                 break;
473
474                         RSSCfg = RSSCfg->next;
475                 }
476                 if (RSSCfg != NULL)
477                 {
478                         pRNCfg->changed = 1;
479                         RSSCfg->last_known_good = now;
480                 }
481         }
482
483         end_critical_section(S_NETCONFIGS);
484 }
485
486 eNextState RSSAggregator_AnalyseReply(AsyncIO *IO)
487 {
488         HashPos *it = NULL;
489         long len;
490         const char *Key;
491         pRSSConfig *pCfg;
492         u_char rawdigest[MD5_DIGEST_LEN];
493         struct MD5Context md5context;
494         StrBuf *guid;
495         rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
496
497         if (IO->HttpReq.httpcode != 200)
498         {
499                 StrBuf *ErrMsg;
500                 long lens[2];
501                 const char *strs[2];
502
503                 SetRSSState(IO, eRSSFailure);
504                 ErrMsg = NewStrBuf();
505                 EVRSSC_syslog(LOG_ALERT, "need a 200, got a %ld !\n",
506                               IO->HttpReq.httpcode);
507                 
508                 strs[0] = ChrPtr(Ctx->Url);
509                 lens[0] = StrLength(Ctx->Url);
510
511                 strs[1] = ChrPtr(Ctx->rooms);
512                 lens[1] = StrLength(Ctx->rooms);
513
514                 if (IO->HttpReq.CurlError == NULL)
515                         IO->HttpReq.CurlError = "";
516
517                 StrBufPrintf(ErrMsg,
518                              "Error while RSS-Aggregation Run of %s\n"
519                              " need a 200, got a %ld !\n"
520                              " Curl Error message: \n%s / %s\n"
521                              " Response text was: \n"
522                              " \n %s\n",
523                              ChrPtr(Ctx->Url),
524                              IO->HttpReq.httpcode,
525                              IO->HttpReq.errdesc,
526                              IO->HttpReq.CurlError,
527                              ChrPtr(IO->HttpReq.ReplyData)
528                         );
529
530                 CtdlAideFPMessage(
531                         ChrPtr(ErrMsg),
532                         "RSS Aggregation run failure",
533                         2, strs, (long*) &lens,
534                         IO->Now,
535                         IO->ID, CCID);
536                 
537                 FreeStrBuf(&ErrMsg);
538                 EVRSSC_syslog(LOG_DEBUG,
539                               "RSS feed returned an invalid http status code. <%s><HTTP %ld>\n",
540                               ChrPtr(Ctx->Url),
541                               IO->HttpReq.httpcode);
542                 return eAbort;
543         }
544
545         pCfg = &Ctx->Cfg;
546
547         while (pCfg != NULL)
548         {
549                 UpdateLastKnownGood (pCfg, IO->Now);
550                 if ((Ctx->roomlist_parts > 1) && 
551                     (it == NULL))
552                 {
553                         it = GetNewHashPos(RSSFetchUrls, 0);
554                 }
555                 if (it != NULL)
556                 {
557                         void *vptr;
558                         if (GetNextHashPos(Ctx->OtherQRnumbers, it, &len, &Key, &vptr))
559                                 pCfg = vptr;
560                         else
561                                 pCfg = NULL;
562                 }
563                 else 
564                         pCfg = NULL;
565         }
566         DeleteHashPos (&it);
567
568         SetRSSState(IO, eRSSUT);
569
570         MD5Init(&md5context);
571
572         MD5Update(&md5context,
573                   (const unsigned char*)SKEY(IO->HttpReq.ReplyData));
574
575         MD5Update(&md5context,
576                   (const unsigned char*)SKEY(Ctx->Url));
577
578         MD5Final(rawdigest, &md5context);
579         guid = NewStrBufPlain(NULL,
580                               MD5_DIGEST_LEN * 2 + 12 /* _rss2ctdl*/);
581         StrBufHexEscAppend(guid, NULL, rawdigest, MD5_DIGEST_LEN);
582         StrBufAppendBufPlain(guid, HKEY("_rssFM"), 0);
583         if (StrLength(guid) > 40)
584                 StrBufCutAt(guid, 40, NULL);
585         /* Find out if we've already seen this item */
586
587 #ifndef DEBUG_RSS
588
589         if (CheckIfAlreadySeen("RSS Whole",
590                                guid,
591                                IO->Now,
592                                IO->Now - USETABLE_ANTIEXPIRE,
593                                eCheckUpdate,
594                                CCID, IO->ID)
595             != 0)
596         {
597                 FreeStrBuf(&guid);
598
599                 EVRSSC_syslog(LOG_DEBUG, "RSS feed already seen. <%s>\n", ChrPtr(Ctx->Url));
600                 return eAbort;
601         }
602         FreeStrBuf(&guid);
603 #endif
604         SetRSSState(IO, eRSSParsing);
605         return RSSAggregator_ParseReply(IO);
606 }
607
608 eNextState RSSAggregator_FinishHttp(AsyncIO *IO)
609 {
610         return CurlQueueDBOperation(IO, RSSAggregator_AnalyseReply);
611 }
612
613 /*
614  * Begin a feed parse
615  */
616 int rss_do_fetching(rss_aggregator *RSSAggr)
617 {
618         AsyncIO         *IO = &RSSAggr->IO;
619         rss_item *ri;
620         time_t now;
621
622         now = time(NULL);
623
624         if ((RSSAggr->next_poll != 0) && (now < RSSAggr->next_poll))
625                 return 0;
626
627         ri = (rss_item*) malloc(sizeof(rss_item));
628         memset(ri, 0, sizeof(rss_item));
629         RSSAggr->Item = ri;
630
631         if (! InitcURLIOStruct(&RSSAggr->IO,
632                                RSSAggr,
633                                "Citadel RSS Client",
634                                RSSAggregator_FinishHttp,
635                                RSSAggregator_Terminate,
636                                RSSAggregator_TerminateDB,
637                                RSSAggregator_ShutdownAbort))
638         {
639                 EVRSSCM_syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
640                 return 0;
641         }
642         SetRSSState(IO, eRSSCreated);
643
644         safestrncpy(((CitContext*)RSSAggr->IO.CitContext)->cs_host,
645                     ChrPtr(RSSAggr->Url),
646                     sizeof(((CitContext*)RSSAggr->IO.CitContext)->cs_host));
647
648         EVRSSC_syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", ChrPtr(RSSAggr->Url));
649         ParseURL(&RSSAggr->IO.ConnectMe, RSSAggr->Url, 80);
650         CurlPrepareURL(RSSAggr->IO.ConnectMe);
651
652         SetRSSState(IO, eRSSFetching);
653         QueueCurlContext(&RSSAggr->IO);
654         return 1;
655 }
656
657 /*
658  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
659  */
660 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data, OneRoomNetCfg *OneRNCFG)
661 {
662         const RSSCfgLine *RSSCfg = (RSSCfgLine *)OneRNCFG->NetConfigs[rssclient];
663         rss_aggregator *RSSAggr = NULL;
664         rss_aggregator *use_this_RSSAggr = NULL;
665         void *vptr;
666
667         pthread_mutex_lock(&RSSQueueMutex);
668         if (GetHash(RSSQueueRooms, LKEY(qrbuf->QRnumber), &vptr))
669         {
670                 EVRSSQ_syslog(LOG_DEBUG,
671                               "rssclient: [%ld] %s already in progress.\n",
672                               qrbuf->QRnumber,
673                               qrbuf->QRname);
674                 pthread_mutex_unlock(&RSSQueueMutex);
675                 return;
676         }
677         pthread_mutex_unlock(&RSSQueueMutex);
678
679         if (server_shutting_down) return;
680
681         while (RSSCfg != NULL)
682         {
683                 pthread_mutex_lock(&RSSQueueMutex);
684                 GetHash(RSSFetchUrls,
685                         SKEY(RSSCfg->Url),
686                         &vptr);
687
688                 use_this_RSSAggr = (rss_aggregator *)vptr;
689                 if (use_this_RSSAggr != NULL)
690                 {
691                         pRSSConfig *pRSSCfg;
692
693                         StrBufAppendBufPlain(
694                                 use_this_RSSAggr->rooms,
695                                 qrbuf->QRname,
696                                 -1, 0);
697                         if (use_this_RSSAggr->roomlist_parts==1)
698                         {
699                                 use_this_RSSAggr->OtherQRnumbers
700                                         = NewHash(1, lFlathash);
701                         }
702
703                         pRSSCfg = (pRSSConfig *) malloc(sizeof(pRSSConfig));
704
705                         pRSSCfg->QRnumber = qrbuf->QRnumber;
706                         pRSSCfg->pCfg = RSSCfg;
707
708                         Put(use_this_RSSAggr->OtherQRnumbers,
709                             LKEY(qrbuf->QRnumber),
710                             pRSSCfg,
711                             NULL);
712                         use_this_RSSAggr->roomlist_parts++;
713
714                         pthread_mutex_unlock(&RSSQueueMutex);
715
716                         RSSCfg = RSSCfg->next;
717                         continue;
718                 }
719                 pthread_mutex_unlock(&RSSQueueMutex);
720
721                 RSSAggr = (rss_aggregator *) malloc(
722                         sizeof(rss_aggregator));
723
724                 memset (RSSAggr, 0, sizeof(rss_aggregator));
725                 RSSAggr->Cfg.QRnumber = qrbuf->QRnumber;
726                 RSSAggr->Cfg.pCfg = RSSCfg;
727                 RSSAggr->roomlist_parts = 1;
728                 RSSAggr->Url = NewStrBufDup(RSSCfg->Url);
729
730                 RSSAggr->ItemType = RSS_UNSET;
731
732                 RSSAggr->rooms = NewStrBufPlain(
733                         qrbuf->QRname, -1);
734
735                 pthread_mutex_lock(&RSSQueueMutex);
736
737                 Put(RSSFetchUrls,
738                     SKEY(RSSAggr->Url),
739                     RSSAggr,
740                     DeleteRssCfg);
741
742                 pthread_mutex_unlock(&RSSQueueMutex);
743                 RSSCfg = RSSCfg->next;
744         }
745 }
746
747 /*
748  * Scan for rooms that have RSS client requests configured
749  */
750 void rssclient_scan(void) {
751         int RSSRoomCount, RSSCount;
752         rss_aggregator *rptr = NULL;
753         void *vrptr = NULL;
754         HashPos *it;
755         long len;
756         const char *Key;
757         time_t now = time(NULL);
758
759         /* Run no more than once every 15 minutes. */
760         if ((now - last_run) < 900) {
761                 EVRSSQ_syslog(LOG_DEBUG,
762                               "Client: polling interval not yet reached; last run was %ldm%lds ago",
763                               ((now - last_run) / 60),
764                               ((now - last_run) % 60)
765                 );
766                 return;
767         }
768
769         /*
770          * This is a simple concurrency check to make sure only one rssclient
771          * run is done at a time.
772          */
773         pthread_mutex_lock(&RSSQueueMutex);
774         RSSCount = GetCount(RSSFetchUrls);
775         RSSRoomCount = GetCount(RSSQueueRooms);
776         pthread_mutex_unlock(&RSSQueueMutex);
777
778         if ((RSSRoomCount > 0) || (RSSCount > 0)) {
779                 EVRSSQ_syslog(LOG_DEBUG,
780                               "rssclient: concurrency check failed; %d rooms and %d url's are queued",
781                               RSSRoomCount, RSSCount
782                         );
783                 return;
784         }
785
786         become_session(&rss_CC);
787         EVRSSQM_syslog(LOG_DEBUG, "rssclient started");
788         CtdlForEachNetCfgRoom(rssclient_scan_room, NULL, rssclient);
789
790         if (GetCount(RSSFetchUrls) > 0)
791         {
792                 pthread_mutex_lock(&RSSQueueMutex);
793                 EVRSSQ_syslog(LOG_DEBUG,
794                                "rssclient starting %d Clients",
795                                GetCount(RSSFetchUrls));
796                 
797                 it = GetNewHashPos(RSSFetchUrls, 0);
798                 while (!server_shutting_down &&
799                        GetNextHashPos(RSSFetchUrls, it, &len, &Key, &vrptr) &&
800                        (vrptr != NULL)) {
801                         rptr = (rss_aggregator *)vrptr;
802                         if (!rss_do_fetching(rptr))
803                                 UnlinkRSSAggregator(rptr);
804                 }
805                 DeleteHashPos(&it);
806                 pthread_mutex_unlock(&RSSQueueMutex);
807         }
808         else
809                 EVRSSQM_syslog(LOG_DEBUG, "Nothing to do.");
810
811         EVRSSQM_syslog(LOG_DEBUG, "rssclient ended\n");
812         return;
813 }
814
815 void rss_cleanup(void)
816 {
817         /* citthread_mutex_destroy(&RSSQueueMutex); TODO */
818         DeleteHash(&RSSFetchUrls);
819         DeleteHash(&RSSQueueRooms);
820 }
821
822 void LogDebugEnableRSSClient(const int n)
823 {
824         RSSClientDebugEnabled = n;
825 }
826
827
828 typedef struct __RSSVetoInfo {
829         StrBuf *ErrMsg;
830         time_t Now;
831         int Veto;
832 }RSSVetoInfo;
833
834 void rssclient_veto_scan_room(struct ctdlroom *qrbuf, void *data, OneRoomNetCfg *OneRNCFG)
835 {
836         RSSVetoInfo *Info = (RSSVetoInfo *) data;
837         const RSSCfgLine *RSSCfg = (RSSCfgLine *)OneRNCFG->NetConfigs[rssclient];
838
839         while (RSSCfg != NULL)
840         {
841                 if ((RSSCfg->last_known_good != 0) &&
842                     (RSSCfg->last_known_good + USETABLE_ANTIEXPIRE < Info->Now))
843                 {
844                         StrBufAppendPrintf(Info->ErrMsg,
845                                            "RSS feed not seen for a %d days:: <",
846                                            (Info->Now - RSSCfg->last_known_good) / (24 * 60 * 60));
847
848                         StrBufAppendBuf(Info->ErrMsg, RSSCfg->Url, 0);
849                         StrBufAppendBufPlain(Info->ErrMsg, HKEY(">\n"), 0);
850                 }
851                 RSSCfg = RSSCfg->next;
852         }
853 }
854
855 int RSSCheckUsetableVeto(StrBuf *ErrMsg)
856 {
857         RSSVetoInfo Info;
858
859         Info.ErrMsg = ErrMsg;
860         Info.Now = time (NULL);
861         Info.Veto = 0;
862
863         CtdlForEachNetCfgRoom(rssclient_veto_scan_room, &Info, rssclient);
864
865         return Info.Veto;;
866 }
867
868
869
870
871 void ParseRSSClientCfgLine(const CfgLineType *ThisOne, StrBuf *Line, const char *LinePos, OneRoomNetCfg *OneRNCFG)
872 {
873         RSSCfgLine *RSSCfg;
874
875         RSSCfg = (RSSCfgLine *) malloc (sizeof(RSSCfgLine));
876         RSSCfg->Url = NewStrBufPlain (NULL, StrLength (Line));
877         
878
879         StrBufExtract_NextToken(RSSCfg->Url, Line, &LinePos, '|');
880         RSSCfg->last_known_good = StrBufExtractNext_long(Line, &LinePos, '|');
881
882
883         RSSCfg->next = (RSSCfgLine *)OneRNCFG->NetConfigs[ThisOne->C];
884         OneRNCFG->NetConfigs[ThisOne->C] = (RoomNetCfgLine*) RSSCfg;
885 }
886
887 void SerializeRSSClientCfgLine(const CfgLineType *ThisOne, StrBuf *OutputBuffer, OneRoomNetCfg *RNCfg, RoomNetCfgLine *data)
888 {
889         RSSCfgLine *RSSCfg = (RSSCfgLine*) data;
890
891         StrBufAppendBufPlain(OutputBuffer, CKEY(ThisOne->Str), 0);
892         StrBufAppendBufPlain(OutputBuffer, HKEY("|"), 0);
893         StrBufAppendBufPlain(OutputBuffer, SKEY(RSSCfg->Url), 0);
894         StrBufAppendPrintf(OutputBuffer, "|%ld\n", RSSCfg->last_known_good);
895 }
896
897 void DeleteRSSClientCfgLine(const CfgLineType *ThisOne, RoomNetCfgLine **data)
898 {
899         RSSCfgLine *RSSCfg = (RSSCfgLine*) *data;
900
901         FreeStrBuf(&RSSCfg->Url);
902         free(*data);
903         *data = NULL;
904 }
905
906
907 CTDL_MODULE_INIT(rssclient)
908 {
909         if (!threading)
910         {
911                 CtdlRegisterTDAPVetoHook (RSSCheckUsetableVeto, CDB_USETABLE, 0);
912
913                 CtdlREGISTERRoomCfgType(rssclient, ParseRSSClientCfgLine, 0, 1, SerializeRSSClientCfgLine, DeleteRSSClientCfgLine);
914                 pthread_mutex_init(&RSSQueueMutex, NULL);
915                 RSSQueueRooms = NewHash(1, lFlathash);
916                 RSSFetchUrls = NewHash(1, NULL);
917                 syslog(LOG_INFO, "%s\n", curl_version());
918                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);
919                 CtdlRegisterEVCleanupHook(rss_cleanup);
920                 CtdlRegisterDebugFlagHook(HKEY("rssclient"), LogDebugEnableRSSClient, &RSSClientDebugEnabled);
921         }
922         else
923         {
924                 CtdlFillSystemContext(&rss_CC, "rssclient");
925         }
926         return "rssclient";
927 }