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 void rss_format_item(networker_save_message *SaveMsg)
 278 {
 279         StrBuf *Message;
 280         int msglen = 0;
 281
 282         if (SaveMsg->author_or_creator != NULL) {
 283
 284                 char *From;
 285                 StrBuf *Encoded = NULL;
 286                 int FromAt;
 287
 288                 From = html_to_ascii(ChrPtr(SaveMsg->author_or_creator),
 289                                      StrLength(SaveMsg->author_or_creator),
 290                                      512, 0);
 291                 StrBufPlain(SaveMsg->author_or_creator, From, -1);
 292                 StrBufTrim(SaveMsg->author_or_creator);
 293                 free(From);
 294
 295                 FromAt = strchr(ChrPtr(SaveMsg->author_or_creator), '@') != NULL;
 296                 if (!FromAt && StrLength (SaveMsg->author_email) > 0)
 297                 {
 298                         StrBufRFC2047encode(&Encoded, SaveMsg->author_or_creator);
 299                         SaveMsg->Msg.cm_fields['A'] = SmashStrBuf(&Encoded);
 300                         SaveMsg->Msg.cm_fields['P'] =
 301                                 SmashStrBuf(&SaveMsg->author_email);
 302                 }
 303                 else
 304                 {
 305                         if (FromAt)
 306                         {
 307                                 SaveMsg->Msg.cm_fields['A'] =
 308                                         SmashStrBuf(&SaveMsg->author_or_creator);
 309                                 SaveMsg->Msg.cm_fields['P'] =
 310                                         strdup(SaveMsg->Msg.cm_fields['A']);
 311                         }
 312                         else
 313                         {
 314                                 StrBufRFC2047encode(&Encoded,
 315                                                     SaveMsg->author_or_creator);
 316                                 SaveMsg->Msg.cm_fields['A'] =
 317                                         SmashStrBuf(&Encoded);
 318                                 SaveMsg->Msg.cm_fields['P'] =
 319                                         strdup("rss@localhost");
 320
 321                         }
 322                 }
 323         }
 324         else {
 325                 SaveMsg->Msg.cm_fields['A'] = strdup("rss");
 326         }
 327
 328         SaveMsg->Msg.cm_fields['N'] = strdup(NODENAME);
 329         if (SaveMsg->title != NULL) {
 330                 long len;
 331                 char *Sbj;
 332                 StrBuf *Encoded, *QPEncoded;
 333
 334                 QPEncoded = NULL;
 335                 StrBufSpaceToBlank(SaveMsg->title);
 336                 len = StrLength(SaveMsg->title);
 337                 Sbj = html_to_ascii(ChrPtr(SaveMsg->title), len, 512, 0);
 338                 len = strlen(Sbj);
 339                 if ((len > 0) && (Sbj[len - 1] == '\n'))
 340                 {
 341                         len --;
 342                         Sbj[len] = '\0';
 343                 }
 344                 Encoded = NewStrBufPlain(Sbj, len);
 345                 free(Sbj);
 346
 347                 StrBufTrim(Encoded);
 348                 StrBufRFC2047encode(&QPEncoded, Encoded);
 349
 350                 SaveMsg->Msg.cm_fields['U'] = SmashStrBuf(&QPEncoded);
 351                 FreeStrBuf(&Encoded);
 352         }
 353         if (SaveMsg->link == NULL)
 354                 SaveMsg->link = NewStrBufPlain(HKEY(""));
 355
 356 #if 0 /* temporarily disable shorter urls. */
 357         SaveMsg->Msg.cm_fields[TMP_SHORTER_URLS] =
 358                 GetShorterUrls(SaveMsg->description);
 359 #endif
 360
 361         msglen += 1024 + StrLength(SaveMsg->link) + StrLength(SaveMsg->description) ;
 362
 363         Message = NewStrBufPlain(NULL, msglen);
 364
 365         StrBufPlain(Message, HKEY(
 366                             "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 367                             "<html><body>\n"));
 368 #if 0 /* disable shorter url for now. */
 369         SaveMsg->Msg.cm_fields[TMP_SHORTER_URL_OFFSET] = StrLength(Message);
 370 #endif
 371         StrBufAppendBuf(Message, SaveMsg->description, 0);
 372         StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
 373
 374         AppendLink(Message, SaveMsg->link, SaveMsg->linkTitle, NULL);
 375         AppendLink(Message, SaveMsg->reLink, SaveMsg->reLinkTitle, "Reply to this");
 376         StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
 377
 378
 379         SaveMsg->Message = Message;
 380 }
 381
 382 eNextState RSSSaveMessage(AsyncIO *IO)
 383 {
 384         long len;
 385         const char *Key;
 386         rss_aggregator *RSSAggr = (rss_aggregator *) IO->Data;
 387
 388         rss_format_item(RSSAggr->ThisMsg);
 389
 390         RSSAggr->ThisMsg->Msg.cm_fields['M'] =
 391                 SmashStrBuf(&RSSAggr->ThisMsg->Message);
 392
 393         CtdlSubmitMsg(&RSSAggr->ThisMsg->Msg, &RSSAggr->recp, NULL, 0);
 394
 395         /* write the uidl to the use table so we don't store this item again */
 396
 397         CheckIfAlreadySeen("RSS Item Insert", RSSAggr->ThisMsg->MsgGUID, IO->Now, 0, eWrite, IO->ID, CCID);
 398
 399         if (GetNextHashPos(RSSAggr->Messages,
 400                            RSSAggr->Pos,
 401                            &len, &Key,
 402                            (void**) &RSSAggr->ThisMsg))
 403                 return NextDBOperation(IO, RSS_FetchNetworkUsetableEntry);
 404         else
 405                 return eAbort;
 406 }
 407
 408 eNextState RSS_FetchNetworkUsetableEntry(AsyncIO *IO)
 409 {
 410         const char *Key;
 411         long len;
 412         rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
 413
 414         /* Find out if we've already seen this item */
 415 // todo: expiry?
 416 #ifndef DEBUG_RSS
 417         SetRSSState(IO, eRSSUT);
 418         if (CheckIfAlreadySeen("RSS Item Seen",
 419                                Ctx->ThisMsg->MsgGUID,
 420                                IO->Now,
 421                                IO->Now - USETABLE_ANTIEXPIRE,
 422                                eCheckUpdate,
 423                                IO->ID, CCID)
 424             != 0)
 425         {
 426                 /* Item has already been seen */
 427                 EVRSSC_syslog(LOG_DEBUG,
 428                           "%s has already been seen\n",
 429                           ChrPtr(Ctx->ThisMsg->MsgGUID));
 430                 SetRSSState(IO, eRSSParsing);
 431
 432                 if (GetNextHashPos(Ctx->Messages,
 433                                    Ctx->Pos,
 434                                    &len, &Key,
 435                                    (void**) &Ctx->ThisMsg))
 436                         return NextDBOperation(
 437                                 IO,
 438                                 RSS_FetchNetworkUsetableEntry);
 439                 else
 440                         return eAbort;
 441         }
 442         else
 443 #endif
 444         {
 445                 SetRSSState(IO, eRSSParsing);
 446
 447                 NextDBOperation(IO, RSSSaveMessage);
 448                 return eSendMore;
 449         }
 450         return eSendMore;
 451 }
 452
 453 void UpdateLastKnownGood(pRSSConfig *pCfg, time_t now)
 454 {
 455         OneRoomNetCfg* pRNCfg;
 456         begin_critical_section(S_NETCONFIGS);
 457         pRNCfg = CtdlGetNetCfgForRoom (pCfg->QRnumber);
 458         if (pRNCfg != NULL)
 459         {
 460                 RSSCfgLine *RSSCfg = (RSSCfgLine *)pRNCfg->NetConfigs[rssclient];
 461
 462                 while (RSSCfg != NULL)
 463                 {
 464                         if (RSSCfg == pCfg->pCfg)
 465                                 break;
 466
 467                         RSSCfg = RSSCfg->next;
 468                 }
 469                 if (RSSCfg != NULL)
 470                 {
 471                         pRNCfg->changed = 1;
 472                         RSSCfg->last_known_good = now;
 473                 }
 474         }
 475
 476         end_critical_section(S_NETCONFIGS);
 477 }
 478
 479 eNextState RSSAggregator_AnalyseReply(AsyncIO *IO)
 480 {
 481         HashPos *it = NULL;
 482         long len;
 483         const char *Key;
 484         pRSSConfig *pCfg;
 485         u_char rawdigest[MD5_DIGEST_LEN];
 486         struct MD5Context md5context;
 487         StrBuf *guid;
 488         rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
 489
 490         if (IO->HttpReq.httpcode != 200)
 491         {
 492                 StrBuf *ErrMsg;
 493                 long lens[2];
 494                 const char *strs[2];
 495
 496                 SetRSSState(IO, eRSSFailure);
 497                 ErrMsg = NewStrBuf();
 498                 EVRSSC_syslog(LOG_ALERT, "need a 200, got a %ld !\n",
 499                               IO->HttpReq.httpcode);
 500
 501                 strs[0] = ChrPtr(Ctx->Url);
 502                 lens[0] = StrLength(Ctx->Url);
 503
 504                 strs[1] = ChrPtr(Ctx->rooms);
 505                 lens[1] = StrLength(Ctx->rooms);
 506                 StrBufPrintf(ErrMsg,
 507                              "Error while RSS-Aggregation Run of %s\n"
 508                              " need a 200, got a %ld !\n"
 509                              " Response text was: \n"
 510                              " \n %s\n",
 511                              ChrPtr(Ctx->Url),
 512                              IO->HttpReq.httpcode,
 513                              ChrPtr(IO->HttpReq.ReplyData));
 514                 CtdlAideFPMessage(
 515                         ChrPtr(ErrMsg),
 516                         "RSS Aggregation run failure",
 517                         2, strs, (long*) &lens,
 518                         IO->Now,
 519                         IO->ID, CCID);
 520
 521                 FreeStrBuf(&ErrMsg);
 522                 EVRSSC_syslog(LOG_DEBUG,
 523                               "RSS feed returned an invalid http status code. <%s><HTTP %ld>\n",
 524                               ChrPtr(Ctx->Url),
 525                               IO->HttpReq.httpcode);
 526                 return eAbort;
 527         }
 528
 529         pCfg = &Ctx->Cfg;
 530
 531         while (pCfg != NULL)
 532         {
 533                 UpdateLastKnownGood (pCfg, IO->Now);
 534                 if ((Ctx->roomlist_parts > 1) &&
 535                     (it == NULL))
 536                 {
 537                         it = GetNewHashPos(RSSFetchUrls, 0);
 538                 }
 539                 if (it != NULL)
 540                 {
 541                         void *vptr;
 542                         if (GetNextHashPos(Ctx->OtherQRnumbers, it, &len, &Key, &vptr))
 543                                 pCfg = vptr;
 544                         else
 545                                 pCfg = NULL;
 546                 }
 547                 else
 548                         pCfg = NULL;
 549         }
 550         DeleteHashPos (&it);
 551
 552         SetRSSState(IO, eRSSUT);
 553
 554         MD5Init(&md5context);
 555
 556         MD5Update(&md5context,
 557                   (const unsigned char*)SKEY(IO->HttpReq.ReplyData));
 558
 559         MD5Update(&md5context,
 560                   (const unsigned char*)SKEY(Ctx->Url));
 561
 562         MD5Final(rawdigest, &md5context);
 563         guid = NewStrBufPlain(NULL,
 564                               MD5_DIGEST_LEN * 2 + 12 /* _rss2ctdl*/);
 565         StrBufHexEscAppend(guid, NULL, rawdigest, MD5_DIGEST_LEN);
 566         StrBufAppendBufPlain(guid, HKEY("_rssFM"), 0);
 567         if (StrLength(guid) > 40)
 568                 StrBufCutAt(guid, 40, NULL);
 569         /* Find out if we've already seen this item */
 570
 571 #ifndef DEBUG_RSS
 572
 573         if (CheckIfAlreadySeen("RSS Whole",
 574                                guid,
 575                                IO->Now,
 576                                IO->Now - USETABLE_ANTIEXPIRE,
 577                                eCheckUpdate,
 578                                IO->ID, CCID)
 579             != 0)
 580         {
 581                 FreeStrBuf(&guid);
 582
 583                 EVRSSC_syslog(LOG_DEBUG, "RSS feed already seen. <%s>\n", ChrPtr(Ctx->Url));
 584                 return eAbort;
 585         }
 586         FreeStrBuf(&guid);
 587 #endif
 588         SetRSSState(IO, eRSSParsing);
 589         return RSSAggregator_ParseReply(IO);
 590 }
 591
 592 eNextState RSSAggregator_FinishHttp(AsyncIO *IO)
 593 {
 594         StopCurlWatchers(IO);
 595         return QueueDBOperation(IO, RSSAggregator_AnalyseReply);
 596 }
 597
 598 /*
 599  * Begin a feed parse
 600  */
 601 int rss_do_fetching(rss_aggregator *RSSAggr)
 602 {
 603         AsyncIO         *IO = &RSSAggr->IO;
 604         rss_item *ri;
 605         time_t now;
 606
 607         now = time(NULL);
 608
 609         if ((RSSAggr->next_poll != 0) && (now < RSSAggr->next_poll))
 610                 return 0;
 611
 612         ri = (rss_item*) malloc(sizeof(rss_item));
 613         memset(ri, 0, sizeof(rss_item));
 614         RSSAggr->Item = ri;
 615
 616         if (! InitcURLIOStruct(&RSSAggr->IO,
 617                                RSSAggr,
 618                                "Citadel RSS Client",
 619                                RSSAggregator_FinishHttp,
 620                                RSSAggregator_Terminate,
 621                                RSSAggregator_TerminateDB,
 622                                RSSAggregator_ShutdownAbort))
 623         {
 624                 EVRSSCM_syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
 625                 return 0;
 626         }
 627         SetRSSState(IO, eRSSCreated);
 628
 629         safestrncpy(((CitContext*)RSSAggr->IO.CitContext)->cs_host,
 630                     ChrPtr(RSSAggr->Url),
 631                     sizeof(((CitContext*)RSSAggr->IO.CitContext)->cs_host));
 632
 633         EVRSSC_syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", ChrPtr(RSSAggr->Url));
 634         ParseURL(&RSSAggr->IO.ConnectMe, RSSAggr->Url, 80);
 635         CurlPrepareURL(RSSAggr->IO.ConnectMe);
 636
 637         SetRSSState(IO, eRSSFetching);
 638         QueueCurlContext(&RSSAggr->IO);
 639         return 1;
 640 }
 641
 642 /*
 643  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
 644  */
 645 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data, OneRoomNetCfg *OneRNCFG)
 646 {
 647         const RSSCfgLine *RSSCfg = (RSSCfgLine *)OneRNCFG->NetConfigs[rssclient];
 648         rss_aggregator *RSSAggr = NULL;
 649         rss_aggregator *use_this_RSSAggr = NULL;
 650         void *vptr;
 651
 652         pthread_mutex_lock(&RSSQueueMutex);
 653         if (GetHash(RSSQueueRooms, LKEY(qrbuf->QRnumber), &vptr))
 654         {
 655                 EVRSSQ_syslog(LOG_DEBUG,
 656                               "rssclient: [%ld] %s already in progress.\n",
 657                               qrbuf->QRnumber,
 658                               qrbuf->QRname);
 659                 pthread_mutex_unlock(&RSSQueueMutex);
 660                 return;
 661         }
 662         pthread_mutex_unlock(&RSSQueueMutex);
 663
 664         if (server_shutting_down) return;
 665
 666         while (RSSCfg != NULL)
 667         {
 668                 pthread_mutex_lock(&RSSQueueMutex);
 669                 GetHash(RSSFetchUrls,
 670                         SKEY(RSSCfg->Url),
 671                         &vptr);
 672
 673                 use_this_RSSAggr = (rss_aggregator *)vptr;
 674                 if (use_this_RSSAggr != NULL)
 675                 {
 676                         pRSSConfig *pRSSCfg;
 677
 678                         StrBufAppendBufPlain(
 679                                 use_this_RSSAggr->rooms,
 680                                 qrbuf->QRname,
 681                                 -1, 0);
 682                         if (use_this_RSSAggr->roomlist_parts==1)
 683                         {
 684                                 use_this_RSSAggr->OtherQRnumbers
 685                                         = NewHash(1, lFlathash);
 686                         }
 687
 688                         pRSSCfg = (pRSSConfig *) malloc(sizeof(pRSSConfig));
 689
 690                         pRSSCfg->QRnumber = qrbuf->QRnumber;
 691                         pRSSCfg->pCfg = RSSCfg;
 692
 693                         Put(use_this_RSSAggr->OtherQRnumbers,
 694                             LKEY(qrbuf->QRnumber),
 695                             pRSSCfg,
 696                             NULL);
 697                         use_this_RSSAggr->roomlist_parts++;
 698
 699                         pthread_mutex_unlock(&RSSQueueMutex);
 700
 701                         RSSCfg = RSSCfg->next;
 702                         continue;
 703                 }
 704                 pthread_mutex_unlock(&RSSQueueMutex);
 705
 706                 RSSAggr = (rss_aggregator *) malloc(
 707                         sizeof(rss_aggregator));
 708
 709                 memset (RSSAggr, 0, sizeof(rss_aggregator));
 710                 RSSAggr->Cfg.QRnumber = qrbuf->QRnumber;
 711                 RSSAggr->Cfg.pCfg = RSSCfg;
 712                 RSSAggr->roomlist_parts = 1;
 713                 RSSAggr->Url = NewStrBufDup(RSSCfg->Url);
 714
 715                 RSSAggr->ItemType = RSS_UNSET;
 716
 717                 RSSAggr->rooms = NewStrBufPlain(
 718                         qrbuf->QRname, -1);
 719
 720                 pthread_mutex_lock(&RSSQueueMutex);
 721
 722                 Put(RSSFetchUrls,
 723                     SKEY(RSSAggr->Url),
 724                     RSSAggr,
 725                     DeleteRssCfg);
 726
 727                 pthread_mutex_unlock(&RSSQueueMutex);
 728                 RSSCfg = RSSCfg->next;
 729         }
 730 }
 731
 732 /*
 733  * Scan for rooms that have RSS client requests configured
 734  */
 735 void rssclient_scan(void) {
 736         int RSSRoomCount, RSSCount;
 737         rss_aggregator *rptr = NULL;
 738         void *vrptr = NULL;
 739         HashPos *it;
 740         long len;
 741         const char *Key;
 742         time_t now = time(NULL);
 743
 744         /* Run no more than once every 15 minutes. */
 745         if ((now - last_run) < 900) {
 746                 EVRSSQ_syslog(LOG_DEBUG,
 747                               "Client: polling interval not yet reached; last run was %ldm%lds ago",
 748                               ((now - last_run) / 60),
 749                               ((now - last_run) % 60)
 750                 );
 751                 return;
 752         }
 753
 754         /*
 755          * This is a simple concurrency check to make sure only one rssclient
 756          * run is done at a time.
 757          */
 758         pthread_mutex_lock(&RSSQueueMutex);
 759         RSSCount = GetCount(RSSFetchUrls);
 760         RSSRoomCount = GetCount(RSSQueueRooms);
 761         pthread_mutex_unlock(&RSSQueueMutex);
 762
 763         if ((RSSRoomCount > 0) || (RSSCount > 0)) {
 764                 EVRSSQ_syslog(LOG_DEBUG,
 765                               "rssclient: concurrency check failed; %d rooms and %d url's are queued",
 766                               RSSRoomCount, RSSCount
 767                         );
 768                 return;
 769         }
 770
 771         become_session(&rss_CC);
 772         EVRSSQM_syslog(LOG_DEBUG, "rssclient started");
 773         CtdlForEachNetCfgRoom(rssclient_scan_room, NULL, rssclient);
 774
 775         if (GetCount(RSSFetchUrls) > 0)
 776         {
 777                 pthread_mutex_lock(&RSSQueueMutex);
 778                 EVRSSQ_syslog(LOG_DEBUG,
 779                                "rssclient starting %d Clients",
 780                                GetCount(RSSFetchUrls));
 781
 782                 it = GetNewHashPos(RSSFetchUrls, 0);
 783                 while (!server_shutting_down &&
 784                        GetNextHashPos(RSSFetchUrls, it, &len, &Key, &vrptr) &&
 785                        (vrptr != NULL)) {
 786                         rptr = (rss_aggregator *)vrptr;
 787                         if (!rss_do_fetching(rptr))
 788                                 UnlinkRSSAggregator(rptr);
 789                 }
 790                 DeleteHashPos(&it);
 791                 pthread_mutex_unlock(&RSSQueueMutex);
 792         }
 793         else
 794                 EVRSSQM_syslog(LOG_DEBUG, "Nothing to do.");
 795
 796         EVRSSQM_syslog(LOG_DEBUG, "rssclient ended\n");
 797         return;
 798 }
 799
 800 void rss_cleanup(void)
 801 {
 802         /* citthread_mutex_destroy(&RSSQueueMutex); TODO */
 803         DeleteHash(&RSSFetchUrls);
 804         DeleteHash(&RSSQueueRooms);
 805 }
 806
 807 void LogDebugEnableRSSClient(const int n)
 808 {
 809         RSSClientDebugEnabled = n;
 810 }
 811
 812
 813 typedef struct __RSSVetoInfo {
 814         StrBuf *ErrMsg;
 815         time_t Now;
 816         int Veto;
 817 }RSSVetoInfo;
 818
 819 void rssclient_veto_scan_room(struct ctdlroom *qrbuf, void *data, OneRoomNetCfg *OneRNCFG)
 820 {
 821         RSSVetoInfo *Info = (RSSVetoInfo *) data;
 822         const RSSCfgLine *RSSCfg = (RSSCfgLine *)OneRNCFG->NetConfigs[rssclient];
 823
 824         while (RSSCfg != NULL)
 825         {
 826                 if ((RSSCfg->last_known_good != 0) &&
 827                     (RSSCfg->last_known_good + USETABLE_ANTIEXPIRE < Info->Now))
 828                 {
 829                         StrBufAppendPrintf(Info->ErrMsg,
 830                                            "RSS feed not seen for a %d days:: <",
 831                                            (Info->Now - RSSCfg->last_known_good) / (24 * 60 * 60));
 832
 833                         StrBufAppendBuf(Info->ErrMsg, RSSCfg->Url, 0);
 834                         StrBufAppendBufPlain(Info->ErrMsg, HKEY(">\n"), 0);
 835                 }
 836                 RSSCfg = RSSCfg->next;
 837         }
 838 }
 839
 840 int RSSCheckUsetableVeto(StrBuf *ErrMsg)
 841 {
 842         RSSVetoInfo Info;
 843
 844         Info.ErrMsg = ErrMsg;
 845         Info.Now = time (NULL);
 846         Info.Veto = 0;
 847
 848         CtdlForEachNetCfgRoom(rssclient_veto_scan_room, &Info, rssclient);
 849
 850         return Info.Veto;;
 851 }
 852
 853
 854
 855
 856 void ParseRSSClientCfgLine(const CfgLineType *ThisOne, StrBuf *Line, const char *LinePos, OneRoomNetCfg *OneRNCFG)
 857 {
 858         RSSCfgLine *RSSCfg;
 859
 860         RSSCfg = (RSSCfgLine *) malloc (sizeof(RSSCfgLine));
 861         RSSCfg->Url = NewStrBufPlain (NULL, StrLength (Line));
 862
 863
 864         StrBufExtract_NextToken(RSSCfg->Url, Line, &LinePos, '|');
 865         RSSCfg->last_known_good = StrBufExtractNext_long(Line, &LinePos, '|');
 866
 867
 868         RSSCfg->next = (RSSCfgLine *)OneRNCFG->NetConfigs[ThisOne->C];
 869         OneRNCFG->NetConfigs[ThisOne->C] = (RoomNetCfgLine*) RSSCfg;
 870 }
 871
 872 void SerializeRSSClientCfgLine(const CfgLineType *ThisOne, StrBuf *OutputBuffer, OneRoomNetCfg *RNCfg, RoomNetCfgLine *data)
 873 {
 874         RSSCfgLine *RSSCfg = (RSSCfgLine*) data;
 875
 876         StrBufAppendBufPlain(OutputBuffer, CKEY(ThisOne->Str), 0);
 877         StrBufAppendBufPlain(OutputBuffer, HKEY("|"), 0);
 878         StrBufAppendBufPlain(OutputBuffer, SKEY(RSSCfg->Url), 0);
 879         StrBufAppendPrintf(OutputBuffer, "|%ld\n", RSSCfg->last_known_good);
 880 }
 881
 882 void DeleteRSSClientCfgLine(const CfgLineType *ThisOne, RoomNetCfgLine **data)
 883 {
 884         RSSCfgLine *RSSCfg = (RSSCfgLine*) *data;
 885
 886         FreeStrBuf(&RSSCfg->Url);
 887         free(*data);
 888         *data = NULL;
 889 }
 890
 891
 892 CTDL_MODULE_INIT(rssclient)
 893 {
 894         if (!threading)
 895         {
 896                 CtdlRegisterTDAPVetoHook (RSSCheckUsetableVeto, CDB_USETABLE, 0);
 897
 898                 CtdlREGISTERRoomCfgType(rssclient, ParseRSSClientCfgLine, 0, 1, SerializeRSSClientCfgLine, DeleteRSSClientCfgLine);
 899                 pthread_mutex_init(&RSSQueueMutex, NULL);
 900                 RSSQueueRooms = NewHash(1, lFlathash);
 901                 RSSFetchUrls = NewHash(1, NULL);
 902                 syslog(LOG_INFO, "%s\n", curl_version());
 903                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);
 904                 CtdlRegisterEVCleanupHook(rss_cleanup);
 905                 CtdlRegisterDebugFlagHook(HKEY("rssclient"), LogDebugEnableRSSClient, &RSSClientDebugEnabled);
 906         }
 907         else
 908         {
 909                 CtdlFillSystemContext(&rss_CC, "rssclient");
 910         }
 911         return "rssclient";
 912 }