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                         SaveMsg->Msg.cm_fields[eAuthor] = SmashStrBuf(&Encoded);
 311                         SaveMsg->Msg.cm_fields[eMessagePath] =
 312                                 SmashStrBuf(&SaveMsg->author_email);
 313                 }
 314                 else
 315                 {
 316                         if (FromAt)
 317                         {
 318                                 SaveMsg->Msg.cm_fields[eAuthor] =
 319                                         SmashStrBuf(&SaveMsg->author_or_creator);
 320                                 SaveMsg->Msg.cm_fields[eMessagePath] =
 321                                         strdup(SaveMsg->Msg.cm_fields[eAuthor]);
 322                         }
 323                         else
 324                         {
 325                                 StrBufRFC2047encode(&Encoded,
 326                                                     SaveMsg->author_or_creator);
 327                                 SaveMsg->Msg.cm_fields[eAuthor] =
 328                                         SmashStrBuf(&Encoded);
 329                                 SaveMsg->Msg.cm_fields[eMessagePath] =
 330                                         strdup("rss@localhost");
 331
 332                         }
 333                 }
 334         }
 335         else {
 336                 SaveMsg->Msg.cm_fields[eAuthor] = strdup("rss");
 337         }
 338
 339         SaveMsg->Msg.cm_fields[eNodeName] = strdup(NODENAME);
 340         if (SaveMsg->title != NULL) {
 341                 long len;
 342                 char *Sbj;
 343                 StrBuf *Encoded, *QPEncoded;
 344
 345                 QPEncoded = NULL;
 346                 StrBufSpaceToBlank(SaveMsg->title);
 347                 len = StrLength(SaveMsg->title);
 348                 Sbj = html_to_ascii(ChrPtr(SaveMsg->title), len, 512, 0);
 349                 len = strlen(Sbj);
 350                 if ((len > 0) && (Sbj[len - 1] == '\n'))
 351                 {
 352                         len --;
 353                         Sbj[len] = '\0';
 354                 }
 355                 Encoded = NewStrBufPlain(Sbj, len);
 356                 free(Sbj);
 357
 358                 StrBufTrim(Encoded);
 359                 StrBufRFC2047encode(&QPEncoded, Encoded);
 360
 361                 SaveMsg->Msg.cm_fields[eMsgSubject] = SmashStrBuf(&QPEncoded);
 362                 FreeStrBuf(&Encoded);
 363         }
 364         if (SaveMsg->link == NULL)
 365                 SaveMsg->link = NewStrBufPlain(HKEY(""));
 366
 367 #if 0 /* temporarily disable shorter urls. */
 368         SaveMsg->Msg.cm_fields[TMP_SHORTER_URLS] =
 369                 GetShorterUrls(SaveMsg->description);
 370 #endif
 371
 372         msglen += 1024 + StrLength(SaveMsg->link) + StrLength(SaveMsg->description) ;
 373
 374         Message = NewStrBufPlain(NULL, msglen);
 375
 376         StrBufPlain(Message, HKEY(
 377                             "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 378                             "<html><body>\n"));
 379 #if 0 /* disable shorter url for now. */
 380         SaveMsg->Msg.cm_fields[TMP_SHORTER_URL_OFFSET] = StrLength(Message);
 381 #endif
 382         StrBufAppendBuf(Message, SaveMsg->description, 0);
 383         StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
 384
 385         AppendLink(Message, SaveMsg->link, SaveMsg->linkTitle, NULL);
 386         AppendLink(Message, SaveMsg->reLink, SaveMsg->reLinkTitle, "Reply to this");
 387         StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
 388
 389         SaveMsg->Message = Message;
 390         return 1;
 391 }
 392
 393 eNextState RSSSaveMessage(AsyncIO *IO)
 394 {
 395         long len;
 396         const char *Key;
 397         rss_aggregator *RSSAggr = (rss_aggregator *) IO->Data;
 398
 399         if (rss_format_item(IO, RSSAggr->ThisMsg))
 400         {
 401                 RSSAggr->ThisMsg->Msg.cm_fields[eMesageText] =
 402                         SmashStrBuf(&RSSAggr->ThisMsg->Message);
 403
 404                 CtdlSubmitMsg(&RSSAggr->ThisMsg->Msg, &RSSAggr->recp, NULL, 0);
 405
 406                 /* write the uidl to the use table so we don't store this item again */
 407
 408                 CheckIfAlreadySeen("RSS Item Insert", RSSAggr->ThisMsg->MsgGUID, IO->Now, 0, eWrite, CCID, IO->ID);
 409         }
 410
 411         if (GetNextHashPos(RSSAggr->Messages,
 412                            RSSAggr->Pos,
 413                            &len, &Key,
 414                            (void**) &RSSAggr->ThisMsg))
 415                 return NextDBOperation(IO, RSS_FetchNetworkUsetableEntry);
 416         else
 417                 return eAbort;
 418 }
 419
 420 eNextState RSS_FetchNetworkUsetableEntry(AsyncIO *IO)
 421 {
 422 #ifndef DEBUG_RSS
 423         const char *Key;
 424         long len;
 425         rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
 426
 427         /* Find out if we've already seen this item */
 428 // todo: expiry?
 429         SetRSSState(IO, eRSSUT);
 430         if (CheckIfAlreadySeen("RSS Item Seen",
 431                                Ctx->ThisMsg->MsgGUID,
 432                                IO->Now,
 433                                IO->Now - USETABLE_ANTIEXPIRE_HIRES,
 434                                eCheckUpdate,
 435                                CCID, IO->ID)
 436             != 0)
 437         {
 438                 /* Item has already been seen */
 439                 EVRSSC_syslog(LOG_DEBUG,
 440                           "%s has already been seen\n",
 441                           ChrPtr(Ctx->ThisMsg->MsgGUID));
 442                 SetRSSState(IO, eRSSParsing);
 443
 444                 if (GetNextHashPos(Ctx->Messages,
 445                                    Ctx->Pos,
 446                                    &len, &Key,
 447                                    (void**) &Ctx->ThisMsg))
 448                         return NextDBOperation(
 449                                 IO,
 450                                 RSS_FetchNetworkUsetableEntry);
 451                 else
 452                         return eAbort;
 453         }
 454         else
 455 #endif
 456         {
 457                 SetRSSState(IO, eRSSParsing);
 458
 459                 NextDBOperation(IO, RSSSaveMessage);
 460                 return eSendMore;
 461         }
 462         return eSendMore;
 463 }
 464
 465 void UpdateLastKnownGood(pRSSConfig *pCfg, time_t now)
 466 {
 467         OneRoomNetCfg* pRNCfg;
 468         begin_critical_section(S_NETCONFIGS);
 469         pRNCfg = CtdlGetNetCfgForRoom (pCfg->QRnumber);
 470         if (pRNCfg != NULL)
 471         {
 472                 RSSCfgLine *RSSCfg = (RSSCfgLine *)pRNCfg->NetConfigs[rssclient];
 473
 474                 while (RSSCfg != NULL)
 475                 {
 476                         if (RSSCfg == pCfg->pCfg)
 477                                 break;
 478
 479                         RSSCfg = RSSCfg->next;
 480                 }
 481                 if (RSSCfg != NULL)
 482                 {
 483                         pRNCfg->changed = 1;
 484                         RSSCfg->last_known_good = now;
 485                 }
 486         }
 487
 488         end_critical_section(S_NETCONFIGS);
 489 }
 490
 491 eNextState RSSAggregator_AnalyseReply(AsyncIO *IO)
 492 {
 493         HashPos *it = NULL;
 494         long len;
 495         const char *Key;
 496         pRSSConfig *pCfg;
 497         u_char rawdigest[MD5_DIGEST_LEN];
 498         struct MD5Context md5context;
 499         StrBuf *guid;
 500         rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
 501
 502         if (IO->HttpReq.httpcode != 200)
 503         {
 504                 StrBuf *ErrMsg;
 505                 long lens[2];
 506                 const char *strs[2];
 507
 508                 SetRSSState(IO, eRSSFailure);
 509                 ErrMsg = NewStrBuf();
 510                 EVRSSC_syslog(LOG_ALERT, "need a 200, got a %ld !\n",
 511                               IO->HttpReq.httpcode);
 512
 513                 strs[0] = ChrPtr(Ctx->Url);
 514                 lens[0] = StrLength(Ctx->Url);
 515
 516                 strs[1] = ChrPtr(Ctx->rooms);
 517                 lens[1] = StrLength(Ctx->rooms);
 518
 519                 if (IO->HttpReq.CurlError == NULL)
 520                         IO->HttpReq.CurlError = "";
 521
 522                 StrBufPrintf(ErrMsg,
 523                              "Error while RSS-Aggregation Run of %s\n"
 524                              " need a 200, got a %ld !\n"
 525                              " Curl Error message: \n%s / %s\n"
 526                              " Response text was: \n"
 527                              " \n %s\n",
 528                              ChrPtr(Ctx->Url),
 529                              IO->HttpReq.httpcode,
 530                              IO->HttpReq.errdesc,
 531                              IO->HttpReq.CurlError,
 532                              ChrPtr(IO->HttpReq.ReplyData)
 533                         );
 534
 535                 CtdlAideFPMessage(
 536                         ChrPtr(ErrMsg),
 537                         "RSS Aggregation run failure",
 538                         2, strs, (long*) &lens,
 539                         IO->Now,
 540                         IO->ID, CCID);
 541
 542                 FreeStrBuf(&ErrMsg);
 543                 EVRSSC_syslog(LOG_DEBUG,
 544                               "RSS feed returned an invalid http status code. <%s><HTTP %ld>\n",
 545                               ChrPtr(Ctx->Url),
 546                               IO->HttpReq.httpcode);
 547                 return eAbort;
 548         }
 549
 550         pCfg = &Ctx->Cfg;
 551
 552         while (pCfg != NULL)
 553         {
 554                 UpdateLastKnownGood (pCfg, IO->Now);
 555                 if ((Ctx->roomlist_parts > 1) &&
 556                     (it == NULL))
 557                 {
 558                         it = GetNewHashPos(RSSFetchUrls, 0);
 559                 }
 560                 if (it != NULL)
 561                 {
 562                         void *vptr;
 563                         if (GetNextHashPos(Ctx->OtherQRnumbers, it, &len, &Key, &vptr))
 564                                 pCfg = vptr;
 565                         else
 566                                 pCfg = NULL;
 567                 }
 568                 else
 569                         pCfg = NULL;
 570         }
 571         DeleteHashPos (&it);
 572
 573         SetRSSState(IO, eRSSUT);
 574
 575         MD5Init(&md5context);
 576
 577         MD5Update(&md5context,
 578                   (const unsigned char*)SKEY(IO->HttpReq.ReplyData));
 579
 580         MD5Update(&md5context,
 581                   (const unsigned char*)SKEY(Ctx->Url));
 582
 583         MD5Final(rawdigest, &md5context);
 584         guid = NewStrBufPlain(NULL,
 585                               MD5_DIGEST_LEN * 2 + 12 /* _rss2ctdl*/);
 586         StrBufHexEscAppend(guid, NULL, rawdigest, MD5_DIGEST_LEN);
 587         StrBufAppendBufPlain(guid, HKEY("_rssFM"), 0);
 588         if (StrLength(guid) > 40)
 589                 StrBufCutAt(guid, 40, NULL);
 590         /* Find out if we've already seen this item */
 591
 592 #ifndef DEBUG_RSS
 593
 594         if (CheckIfAlreadySeen("RSS Whole",
 595                                guid,
 596                                IO->Now,
 597                                IO->Now - USETABLE_ANTIEXPIRE,
 598                                eCheckUpdate,
 599                                CCID, IO->ID)
 600             != 0)
 601         {
 602                 FreeStrBuf(&guid);
 603
 604                 EVRSSC_syslog(LOG_DEBUG, "RSS feed already seen. <%s>\n", ChrPtr(Ctx->Url));
 605                 return eAbort;
 606         }
 607         FreeStrBuf(&guid);
 608 #endif
 609         SetRSSState(IO, eRSSParsing);
 610         return RSSAggregator_ParseReply(IO);
 611 }
 612
 613 eNextState RSSAggregator_FinishHttp(AsyncIO *IO)
 614 {
 615         return CurlQueueDBOperation(IO, RSSAggregator_AnalyseReply);
 616 }
 617
 618 /*
 619  * Begin a feed parse
 620  */
 621 int rss_do_fetching(rss_aggregator *RSSAggr)
 622 {
 623         AsyncIO         *IO = &RSSAggr->IO;
 624         rss_item *ri;
 625         time_t now;
 626
 627         now = time(NULL);
 628
 629         if ((RSSAggr->next_poll != 0) && (now < RSSAggr->next_poll))
 630                 return 0;
 631
 632         ri = (rss_item*) malloc(sizeof(rss_item));
 633         memset(ri, 0, sizeof(rss_item));
 634         RSSAggr->Item = ri;
 635
 636         if (! InitcURLIOStruct(&RSSAggr->IO,
 637                                RSSAggr,
 638                                "Citadel RSS Client",
 639                                RSSAggregator_FinishHttp,
 640                                RSSAggregator_Terminate,
 641                                RSSAggregator_TerminateDB,
 642                                RSSAggregator_ShutdownAbort))
 643         {
 644                 EVRSSCM_syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
 645                 return 0;
 646         }
 647         SetRSSState(IO, eRSSCreated);
 648
 649         safestrncpy(((CitContext*)RSSAggr->IO.CitContext)->cs_host,
 650                     ChrPtr(RSSAggr->Url),
 651                     sizeof(((CitContext*)RSSAggr->IO.CitContext)->cs_host));
 652
 653         EVRSSC_syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", ChrPtr(RSSAggr->Url));
 654         ParseURL(&RSSAggr->IO.ConnectMe, RSSAggr->Url, 80);
 655         CurlPrepareURL(RSSAggr->IO.ConnectMe);
 656
 657         SetRSSState(IO, eRSSFetching);
 658         QueueCurlContext(&RSSAggr->IO);
 659         return 1;
 660 }
 661
 662 /*
 663  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
 664  */
 665 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data, OneRoomNetCfg *OneRNCFG)
 666 {
 667         const RSSCfgLine *RSSCfg = (RSSCfgLine *)OneRNCFG->NetConfigs[rssclient];
 668         rss_aggregator *RSSAggr = NULL;
 669         rss_aggregator *use_this_RSSAggr = NULL;
 670         void *vptr;
 671
 672         pthread_mutex_lock(&RSSQueueMutex);
 673         if (GetHash(RSSQueueRooms, LKEY(qrbuf->QRnumber), &vptr))
 674         {
 675                 EVRSSQ_syslog(LOG_DEBUG,
 676                               "rssclient: [%ld] %s already in progress.\n",
 677                               qrbuf->QRnumber,
 678                               qrbuf->QRname);
 679                 pthread_mutex_unlock(&RSSQueueMutex);
 680                 return;
 681         }
 682         pthread_mutex_unlock(&RSSQueueMutex);
 683
 684         if (server_shutting_down) return;
 685
 686         while (RSSCfg != NULL)
 687         {
 688                 pthread_mutex_lock(&RSSQueueMutex);
 689                 GetHash(RSSFetchUrls,
 690                         SKEY(RSSCfg->Url),
 691                         &vptr);
 692
 693                 use_this_RSSAggr = (rss_aggregator *)vptr;
 694                 if (use_this_RSSAggr != NULL)
 695                 {
 696                         pRSSConfig *pRSSCfg;
 697
 698                         StrBufAppendBufPlain(
 699                                 use_this_RSSAggr->rooms,
 700                                 qrbuf->QRname,
 701                                 -1, 0);
 702                         if (use_this_RSSAggr->roomlist_parts==1)
 703                         {
 704                                 use_this_RSSAggr->OtherQRnumbers
 705                                         = NewHash(1, lFlathash);
 706                         }
 707
 708                         pRSSCfg = (pRSSConfig *) malloc(sizeof(pRSSConfig));
 709
 710                         pRSSCfg->QRnumber = qrbuf->QRnumber;
 711                         pRSSCfg->pCfg = RSSCfg;
 712
 713                         Put(use_this_RSSAggr->OtherQRnumbers,
 714                             LKEY(qrbuf->QRnumber),
 715                             pRSSCfg,
 716                             NULL);
 717                         use_this_RSSAggr->roomlist_parts++;
 718
 719                         pthread_mutex_unlock(&RSSQueueMutex);
 720
 721                         RSSCfg = RSSCfg->next;
 722                         continue;
 723                 }
 724                 pthread_mutex_unlock(&RSSQueueMutex);
 725
 726                 RSSAggr = (rss_aggregator *) malloc(
 727                         sizeof(rss_aggregator));
 728
 729                 memset (RSSAggr, 0, sizeof(rss_aggregator));
 730                 RSSAggr->Cfg.QRnumber = qrbuf->QRnumber;
 731                 RSSAggr->Cfg.pCfg = RSSCfg;
 732                 RSSAggr->roomlist_parts = 1;
 733                 RSSAggr->Url = NewStrBufDup(RSSCfg->Url);
 734
 735                 RSSAggr->ItemType = RSS_UNSET;
 736
 737                 RSSAggr->rooms = NewStrBufPlain(
 738                         qrbuf->QRname, -1);
 739
 740                 pthread_mutex_lock(&RSSQueueMutex);
 741
 742                 Put(RSSFetchUrls,
 743                     SKEY(RSSAggr->Url),
 744                     RSSAggr,
 745                     DeleteRssCfg);
 746
 747                 pthread_mutex_unlock(&RSSQueueMutex);
 748                 RSSCfg = RSSCfg->next;
 749         }
 750 }
 751
 752 /*
 753  * Scan for rooms that have RSS client requests configured
 754  */
 755 void rssclient_scan(void) {
 756         int RSSRoomCount, RSSCount;
 757         rss_aggregator *rptr = NULL;
 758         void *vrptr = NULL;
 759         HashPos *it;
 760         long len;
 761         const char *Key;
 762         time_t now = time(NULL);
 763
 764         /* Run no more than once every 15 minutes. */
 765         if ((now - last_run) < 900) {
 766                 EVRSSQ_syslog(LOG_DEBUG,
 767                               "Client: polling interval not yet reached; last run was %ldm%lds ago",
 768                               ((now - last_run) / 60),
 769                               ((now - last_run) % 60)
 770                 );
 771                 return;
 772         }
 773
 774         /*
 775          * This is a simple concurrency check to make sure only one rssclient
 776          * run is done at a time.
 777          */
 778         pthread_mutex_lock(&RSSQueueMutex);
 779         RSSCount = GetCount(RSSFetchUrls);
 780         RSSRoomCount = GetCount(RSSQueueRooms);
 781         pthread_mutex_unlock(&RSSQueueMutex);
 782
 783         if ((RSSRoomCount > 0) || (RSSCount > 0)) {
 784                 EVRSSQ_syslog(LOG_DEBUG,
 785                               "rssclient: concurrency check failed; %d rooms and %d url's are queued",
 786                               RSSRoomCount, RSSCount
 787                         );
 788                 return;
 789         }
 790
 791         become_session(&rss_CC);
 792         EVRSSQM_syslog(LOG_DEBUG, "rssclient started");
 793         CtdlForEachNetCfgRoom(rssclient_scan_room, NULL, rssclient);
 794
 795         if (GetCount(RSSFetchUrls) > 0)
 796         {
 797                 pthread_mutex_lock(&RSSQueueMutex);
 798                 EVRSSQ_syslog(LOG_DEBUG,
 799                                "rssclient starting %d Clients",
 800                                GetCount(RSSFetchUrls));
 801
 802                 it = GetNewHashPos(RSSFetchUrls, 0);
 803                 while (!server_shutting_down &&
 804                        GetNextHashPos(RSSFetchUrls, it, &len, &Key, &vrptr) &&
 805                        (vrptr != NULL)) {
 806                         rptr = (rss_aggregator *)vrptr;
 807                         if (!rss_do_fetching(rptr))
 808                                 UnlinkRSSAggregator(rptr);
 809                 }
 810                 DeleteHashPos(&it);
 811                 pthread_mutex_unlock(&RSSQueueMutex);
 812         }
 813         else
 814                 EVRSSQM_syslog(LOG_DEBUG, "Nothing to do.");
 815
 816         EVRSSQM_syslog(LOG_DEBUG, "rssclient ended\n");
 817         return;
 818 }
 819
 820 void rss_cleanup(void)
 821 {
 822         /* citthread_mutex_destroy(&RSSQueueMutex); TODO */
 823         DeleteHash(&RSSFetchUrls);
 824         DeleteHash(&RSSQueueRooms);
 825 }
 826
 827 void LogDebugEnableRSSClient(const int n)
 828 {
 829         RSSClientDebugEnabled = n;
 830 }
 831
 832
 833 typedef struct __RSSVetoInfo {
 834         StrBuf *ErrMsg;
 835         time_t Now;
 836         int Veto;
 837 }RSSVetoInfo;
 838
 839 void rssclient_veto_scan_room(struct ctdlroom *qrbuf, void *data, OneRoomNetCfg *OneRNCFG)
 840 {
 841         RSSVetoInfo *Info = (RSSVetoInfo *) data;
 842         const RSSCfgLine *RSSCfg = (RSSCfgLine *)OneRNCFG->NetConfigs[rssclient];
 843
 844         while (RSSCfg != NULL)
 845         {
 846                 if ((RSSCfg->last_known_good != 0) &&
 847                     (RSSCfg->last_known_good + USETABLE_ANTIEXPIRE < Info->Now))
 848                 {
 849                         StrBufAppendPrintf(Info->ErrMsg,
 850                                            "RSS feed not seen for a %d days:: <",
 851                                            (Info->Now - RSSCfg->last_known_good) / (24 * 60 * 60));
 852
 853                         StrBufAppendBuf(Info->ErrMsg, RSSCfg->Url, 0);
 854                         StrBufAppendBufPlain(Info->ErrMsg, HKEY(">\n"), 0);
 855                 }
 856                 RSSCfg = RSSCfg->next;
 857         }
 858 }
 859
 860 int RSSCheckUsetableVeto(StrBuf *ErrMsg)
 861 {
 862         RSSVetoInfo Info;
 863
 864         Info.ErrMsg = ErrMsg;
 865         Info.Now = time (NULL);
 866         Info.Veto = 0;
 867
 868         CtdlForEachNetCfgRoom(rssclient_veto_scan_room, &Info, rssclient);
 869
 870         return Info.Veto;;
 871 }
 872
 873
 874
 875
 876 void ParseRSSClientCfgLine(const CfgLineType *ThisOne, StrBuf *Line, const char *LinePos, OneRoomNetCfg *OneRNCFG)
 877 {
 878         RSSCfgLine *RSSCfg;
 879
 880         RSSCfg = (RSSCfgLine *) malloc (sizeof(RSSCfgLine));
 881         RSSCfg->Url = NewStrBufPlain (NULL, StrLength (Line));
 882
 883
 884         StrBufExtract_NextToken(RSSCfg->Url, Line, &LinePos, '|');
 885         RSSCfg->last_known_good = StrBufExtractNext_long(Line, &LinePos, '|');
 886
 887
 888         RSSCfg->next = (RSSCfgLine *)OneRNCFG->NetConfigs[ThisOne->C];
 889         OneRNCFG->NetConfigs[ThisOne->C] = (RoomNetCfgLine*) RSSCfg;
 890 }
 891
 892 void SerializeRSSClientCfgLine(const CfgLineType *ThisOne, StrBuf *OutputBuffer, OneRoomNetCfg *RNCfg, RoomNetCfgLine *data)
 893 {
 894         RSSCfgLine *RSSCfg = (RSSCfgLine*) data;
 895
 896         StrBufAppendBufPlain(OutputBuffer, CKEY(ThisOne->Str), 0);
 897         StrBufAppendBufPlain(OutputBuffer, HKEY("|"), 0);
 898         StrBufAppendBufPlain(OutputBuffer, SKEY(RSSCfg->Url), 0);
 899         StrBufAppendPrintf(OutputBuffer, "|%ld\n", RSSCfg->last_known_good);
 900 }
 901
 902 void DeleteRSSClientCfgLine(const CfgLineType *ThisOne, RoomNetCfgLine **data)
 903 {
 904         RSSCfgLine *RSSCfg = (RSSCfgLine*) *data;
 905
 906         FreeStrBuf(&RSSCfg->Url);
 907         free(*data);
 908         *data = NULL;
 909 }
 910
 911
 912 CTDL_MODULE_INIT(rssclient)
 913 {
 914         if (!threading)
 915         {
 916                 CtdlRegisterTDAPVetoHook (RSSCheckUsetableVeto, CDB_USETABLE, 0);
 917
 918                 CtdlREGISTERRoomCfgType(rssclient, ParseRSSClientCfgLine, 0, 1, SerializeRSSClientCfgLine, DeleteRSSClientCfgLine);
 919                 pthread_mutex_init(&RSSQueueMutex, NULL);
 920                 RSSQueueRooms = NewHash(1, lFlathash);
 921                 RSSFetchUrls = NewHash(1, NULL);
 922                 syslog(LOG_INFO, "%s\n", curl_version());
 923                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);
 924                 CtdlRegisterEVCleanupHook(rss_cleanup);
 925                 CtdlRegisterDebugFlagHook(HKEY("rssclient"), LogDebugEnableRSSClient, &RSSClientDebugEnabled);
 926         }
 927         else
 928         {
 929                 CtdlFillSystemContext(&rss_CC, "rssclient");
 930         }
 931         return "rssclient";
 932 }