citadel/modules/rssclient/serv_rssclient.c

   1 /*
   2  * Bring external RSS feeds into rooms.
   3  *
   4  * Copyright (c) 2007-2015 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                              "%s[%ld]CC[%d][%ld]RSS" FORMAT,            \
  79                              IOSTR, IO->ID, CCID, N, __VA_ARGS__)
  80
  81 #define EVRSSCM_syslog(LEVEL, FORMAT)                                   \
  82         DBGLOG(LEVEL) syslog(LEVEL,                                     \
  83                              "%s[%ld]CC[%d][%ld]RSS" FORMAT,            \
  84                              IOSTR, 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, "%s[%ld][%ld]RSS" FORMAT,           \
  94                              IOSTR, 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
 112 static size_t GetLocationString( void *ptr, size_t size, size_t nmemb, void *userdata)
 113 {
 114 #define LOCATION "location"
 115         if (strncasecmp((char*)ptr, LOCATION, sizeof(LOCATION) - 1) == 0)
 116         {
 117                 AsyncIO *IO = (AsyncIO *) userdata;
 118                 rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
 119
 120                 char *pch = (char*) ptr;
 121                 char *pche;
 122
 123                 pche = pch + (size * nmemb);
 124                 pch += sizeof(LOCATION);
 125
 126                 while (isspace(*pch) || (*pch == ':'))
 127                         pch ++;
 128
 129                 while (isspace(*pche) || (*pche == '\0'))
 130                         pche--;
 131                 if (RSSAggr->RedirectUrl == NULL) {
 132                         RSSAggr->RedirectUrl = NewStrBufPlain(pch, pche - pch + 1);
 133                 }
 134                 else {
 135                         FlushStrBuf(RSSAggr->RedirectUrl);
 136                         StrBufPlain(RSSAggr->RedirectUrl, pch, pche - pch + 1);
 137                 }
 138         }
 139         return size * nmemb;
 140 }
 141
 142 static void SetRSSState(AsyncIO *IO, RSSState State)
 143 {
 144         CitContext* CCC = IO->CitContext;
 145         if (CCC != NULL)
 146                 memcpy(CCC->cs_clientname, RSSStates[State].Key, RSSStates[State].len + 1);
 147 }
 148
 149 void DeleteRoomReference(long QRnumber)
 150 {
 151         HashPos *At;
 152         long HKLen;
 153         const char *HK;
 154         void *vData = NULL;
 155         rss_room_counter *pRoomC;
 156
 157         At = GetNewHashPos(RSSQueueRooms, 0);
 158
 159         if (GetHashPosFromKey(RSSQueueRooms, LKEY(QRnumber), At))
 160         {
 161                 GetHashPos(RSSQueueRooms, At, &HKLen, &HK, &vData);
 162                 if (vData != NULL)
 163                 {
 164                         pRoomC = (rss_room_counter *) vData;
 165                         pRoomC->count --;
 166                         if (pRoomC->count == 0)
 167                                 DeleteEntryFromHash(RSSQueueRooms, At);
 168                 }
 169         }
 170         DeleteHashPos(&At);
 171 }
 172
 173 void UnlinkRooms(rss_aggregator *RSSAggr)
 174 {
 175         DeleteRoomReference(RSSAggr->Cfg.QRnumber);
 176         if (RSSAggr->OtherQRnumbers != NULL)
 177         {
 178                 long HKLen;
 179                 const char *HK;
 180                 HashPos *At;
 181                 void *vData;
 182
 183                 At = GetNewHashPos(RSSAggr->OtherQRnumbers, 0);
 184                 while (! server_shutting_down &&
 185                        GetNextHashPos(RSSAggr->OtherQRnumbers,
 186                                       At,
 187                                       &HKLen, &HK,
 188                                       &vData) &&
 189                        (vData != NULL))
 190                 {
 191                         pRSSConfig *Data = (pRSSConfig*) vData;
 192                         DeleteRoomReference(Data->QRnumber);
 193                 }
 194
 195                 DeleteHashPos(&At);
 196         }
 197 }
 198
 199 void UnlinkRSSAggregator(rss_aggregator *RSSAggr)
 200 {
 201         HashPos *At;
 202
 203         pthread_mutex_lock(&RSSQueueMutex);
 204         UnlinkRooms(RSSAggr);
 205
 206         At = GetNewHashPos(RSSFetchUrls, 0);
 207         if (GetHashPosFromKey(RSSFetchUrls, SKEY(RSSAggr->Url), At))
 208         {
 209                 DeleteEntryFromHash(RSSFetchUrls, At);
 210         }
 211         DeleteHashPos(&At);
 212         last_run = time(NULL);
 213         pthread_mutex_unlock(&RSSQueueMutex);
 214 }
 215
 216 void DeleteRssCfg(void *vptr)
 217 {
 218         rss_aggregator *RSSAggr = (rss_aggregator *)vptr;
 219         AsyncIO *IO = &RSSAggr->IO;
 220
 221         if (IO->CitContext != NULL)
 222                 EVRSSCM_syslog(LOG_DEBUG, "RSS: destroying\n");
 223
 224         FreeStrBuf(&RSSAggr->Url);
 225         FreeStrBuf(&RSSAggr->RedirectUrl);
 226         FreeStrBuf(&RSSAggr->rooms);
 227         FreeStrBuf(&RSSAggr->CData);
 228         FreeStrBuf(&RSSAggr->Key);
 229         DeleteHash(&RSSAggr->OtherQRnumbers);
 230
 231         DeleteHashPos (&RSSAggr->Pos);
 232         DeleteHash (&RSSAggr->Messages);
 233         if (RSSAggr->recp.recp_room != NULL)
 234                 free(RSSAggr->recp.recp_room);
 235
 236
 237         if (RSSAggr->Item != NULL)
 238         {
 239                 flush_rss_item(RSSAggr->Item);
 240
 241                 free(RSSAggr->Item);
 242         }
 243
 244         FreeAsyncIOContents(&RSSAggr->IO);
 245         memset(RSSAggr, 0, sizeof(rss_aggregator));
 246         free(RSSAggr);
 247 }
 248
 249 eNextState RSSAggregator_Terminate(AsyncIO *IO)
 250 {
 251         rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
 252
 253         EVRSSCM_syslog(LOG_DEBUG, "RSS: Terminating.\n");
 254
 255         StopCurlWatchers(IO);
 256         UnlinkRSSAggregator(RSSAggr);
 257         return eAbort;
 258 }
 259
 260 eNextState RSSAggregator_TerminateDB(AsyncIO *IO)
 261 {
 262         rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
 263
 264         EVRSSCM_syslog(LOG_DEBUG, "RSS: Terminating.\n");
 265
 266
 267         StopDBWatchers(&RSSAggr->IO);
 268         UnlinkRSSAggregator(RSSAggr);
 269         return eAbort;
 270 }
 271
 272 eNextState RSSAggregator_ShutdownAbort(AsyncIO *IO)
 273 {
 274         const char *pUrl;
 275         rss_aggregator *RSSAggr = (rss_aggregator *)IO->Data;
 276
 277         pUrl = IO->ConnectMe->PlainUrl;
 278         if (pUrl == NULL)
 279                 pUrl = "";
 280
 281         EVRSSC_syslog(LOG_DEBUG, "RSS: Aborting by shutdown: %s.\n", pUrl);
 282
 283         StopCurlWatchers(IO);
 284         UnlinkRSSAggregator(RSSAggr);
 285         return eAbort;
 286 }
 287
 288 void AppendLink(StrBuf *Message,
 289                 StrBuf *link,
 290                 StrBuf *LinkTitle,
 291                 const char *Title)
 292 {
 293         if (StrLength(link) > 0)
 294         {
 295                 StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
 296                 StrBufAppendBuf(Message, link, 0);
 297                 StrBufAppendBufPlain(Message, HKEY("\">"), 0);
 298                 if (StrLength(LinkTitle) > 0)
 299                         StrBufAppendBuf(Message, LinkTitle, 0);
 300                 else if ((Title != NULL) && !IsEmptyStr(Title))
 301                         StrBufAppendBufPlain(Message, Title, -1, 0);
 302                 else
 303                         StrBufAppendBuf(Message, link, 0);
 304                 StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
 305         }
 306 }
 307
 308
 309 int rss_format_item(AsyncIO *IO, networker_save_message *SaveMsg)
 310 {
 311         StrBuf *Message;
 312         int msglen = 0;
 313
 314         if (StrLength(SaveMsg->description) +
 315             StrLength(SaveMsg->link) +
 316             StrLength(SaveMsg->linkTitle) +
 317             StrLength(SaveMsg->reLink) +
 318             StrLength(SaveMsg->reLinkTitle) +
 319             StrLength(SaveMsg->title) == 0)
 320         {
 321                 EVRSSCM_syslog(LOG_INFO, "Refusing to save empty message.");
 322                 return 0;
 323         }
 324
 325         CM_Flush(&SaveMsg->Msg);
 326
 327         if (SaveMsg->author_or_creator != NULL) {
 328
 329                 char *From;
 330                 StrBuf *Encoded = NULL;
 331                 int FromAt;
 332
 333                 From = html_to_ascii(ChrPtr(SaveMsg->author_or_creator),
 334                                      StrLength(SaveMsg->author_or_creator),
 335                                      512, 0);
 336                 StrBufPlain(SaveMsg->author_or_creator, From, -1);
 337                 StrBufTrim(SaveMsg->author_or_creator);
 338                 free(From);
 339
 340                 FromAt = strchr(ChrPtr(SaveMsg->author_or_creator), '@') != NULL;
 341                 if (!FromAt && StrLength (SaveMsg->author_email) > 0)
 342                 {
 343                         StrBufRFC2047encode(&Encoded, SaveMsg->author_or_creator);
 344                         CM_SetAsFieldSB(&SaveMsg->Msg, eAuthor, &Encoded);
 345                         CM_SetAsFieldSB(&SaveMsg->Msg, eMessagePath, &SaveMsg->author_email);
 346                 }
 347                 else
 348                 {
 349                         if (FromAt)
 350                         {
 351                                 CM_SetAsFieldSB(&SaveMsg->Msg, eAuthor, &SaveMsg->author_or_creator);
 352                                 CM_CopyField(&SaveMsg->Msg, eMessagePath, eAuthor);
 353                         }
 354                         else
 355                         {
 356                                 StrBufRFC2047encode(&Encoded,
 357                                                     SaveMsg->author_or_creator);
 358                                 CM_SetAsFieldSB(&SaveMsg->Msg, eAuthor, &Encoded);
 359                                 CM_SetField(&SaveMsg->Msg, eMessagePath, HKEY("rss@localhost"));
 360
 361                         }
 362                 }
 363         }
 364         else {
 365                 CM_SetField(&SaveMsg->Msg, eAuthor, HKEY("rss"));
 366         }
 367
 368         CM_SetField(&SaveMsg->Msg, eNodeName, CtdlGetConfigStr("c_nodename"), strlen(CtdlGetConfigStr("c_nodename")));
 369         if (SaveMsg->title != NULL) {
 370                 long len;
 371                 char *Sbj;
 372                 StrBuf *Encoded, *QPEncoded;
 373
 374                 QPEncoded = NULL;
 375                 StrBufSpaceToBlank(SaveMsg->title);
 376                 len = StrLength(SaveMsg->title);
 377                 Sbj = html_to_ascii(ChrPtr(SaveMsg->title), len, 512, 0);
 378                 len = strlen(Sbj);
 379                 if ((len > 0) && (Sbj[len - 1] == '\n'))
 380                 {
 381                         len --;
 382                         Sbj[len] = '\0';
 383                 }
 384                 Encoded = NewStrBufPlain(Sbj, len);
 385                 free(Sbj);
 386
 387                 StrBufTrim(Encoded);
 388                 StrBufRFC2047encode(&QPEncoded, Encoded);
 389
 390                 CM_SetAsFieldSB(&SaveMsg->Msg, eMsgSubject, &QPEncoded);
 391                 FreeStrBuf(&Encoded);
 392         }
 393         if (SaveMsg->link == NULL)
 394                 SaveMsg->link = NewStrBufPlain(HKEY(""));
 395
 396 #if 0 /* temporarily disable shorter urls. */
 397         SaveMsg->Msg.cm_fields[TMP_SHORTER_URLS] =
 398                 GetShorterUrls(SaveMsg->description);
 399 #endif
 400
 401         msglen += 1024 + StrLength(SaveMsg->link) + StrLength(SaveMsg->description) ;
 402
 403         Message = NewStrBufPlain(NULL, msglen);
 404
 405         StrBufPlain(Message, HKEY(
 406                             "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 407                             "<html><body>\n"));
 408 #if 0 /* disable shorter url for now. */
 409         SaveMsg->Msg.cm_fields[TMP_SHORTER_URL_OFFSET] = StrLength(Message);
 410 #endif
 411         StrBufAppendBuf(Message, SaveMsg->description, 0);
 412         StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
 413
 414         AppendLink(Message, SaveMsg->link, SaveMsg->linkTitle, NULL);
 415         AppendLink(Message, SaveMsg->reLink, SaveMsg->reLinkTitle, "Reply to this");
 416         StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
 417
 418         SaveMsg->Message = Message;
 419         return 1;
 420 }
 421
 422 eNextState RSSSaveMessage(AsyncIO *IO)
 423 {
 424         long len;
 425         const char *Key;
 426         rss_aggregator *RSSAggr = (rss_aggregator *) IO->Data;
 427
 428         if (rss_format_item(IO, RSSAggr->ThisMsg))
 429         {
 430                 CM_SetAsFieldSB(&RSSAggr->ThisMsg->Msg, eMesageText,
 431                                        &RSSAggr->ThisMsg->Message);
 432
 433                 CtdlSubmitMsg(&RSSAggr->ThisMsg->Msg, &RSSAggr->recp, NULL, 0);
 434
 435                 /* write the uidl to the use table so we don't store this item again */
 436
 437                 CheckIfAlreadySeen("RSS Item Insert", RSSAggr->ThisMsg->MsgGUID, EvGetNow(IO), 0, eWrite, CCID, IO->ID);
 438         }
 439
 440         if (GetNextHashPos(RSSAggr->Messages,
 441                            RSSAggr->Pos,
 442                            &len, &Key,
 443                            (void**) &RSSAggr->ThisMsg))
 444                 return NextDBOperation(IO, RSS_FetchNetworkUsetableEntry);
 445         else
 446                 return eAbort;
 447 }
 448
 449 eNextState RSS_FetchNetworkUsetableEntry(AsyncIO *IO)
 450 {
 451         static const time_t antiExpire = USETABLE_ANTIEXPIRE_HIRES;
 452 #ifndef DEBUG_RSS
 453         time_t seenstamp = 0;
 454         const char *Key;
 455         long len;
 456         rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
 457
 458         /* Find out if we've already seen this item */
 459 // todo: expiry?
 460         SetRSSState(IO, eRSSUT);
 461         seenstamp = CheckIfAlreadySeen("RSS Item Seen",
 462                                        Ctx->ThisMsg->MsgGUID,
 463                                        EvGetNow(IO),
 464                                        antiExpire,
 465                                        eCheckUpdate,
 466                                        CCID, IO->ID);
 467         if (seenstamp != 0)
 468         {
 469                 /* Item has already been seen */
 470                 EVRSSC_syslog(LOG_DEBUG,
 471                               "%s has already been seen - %ld < %ld",
 472                               ChrPtr(Ctx->ThisMsg->MsgGUID),
 473                               seenstamp, antiExpire);
 474
 475                 SetRSSState(IO, eRSSParsing);
 476
 477                 if (GetNextHashPos(Ctx->Messages,
 478                                    Ctx->Pos,
 479                                    &len, &Key,
 480                                    (void**) &Ctx->ThisMsg))
 481                         return NextDBOperation(
 482                                 IO,
 483                                 RSS_FetchNetworkUsetableEntry);
 484                 else
 485                         return eAbort;
 486         }
 487         else
 488 #endif
 489         {
 490                 /* Item has already been seen */
 491                 EVRSSC_syslog(LOG_DEBUG,
 492                               "%s Parsing - %ld >= %ld",
 493                               ChrPtr(Ctx->ThisMsg->MsgGUID),
 494                               seenstamp, antiExpire);
 495                 SetRSSState(IO, eRSSParsing);
 496
 497                 NextDBOperation(IO, RSSSaveMessage);
 498                 return eSendMore;
 499         }
 500         return eSendMore;
 501 }
 502
 503 void UpdateLastKnownGood(pRSSConfig *pCfg, time_t now)
 504 {
 505         OneRoomNetCfg* pRNCfg;
 506         begin_critical_section(S_NETCONFIGS);
 507         pRNCfg = CtdlGetNetCfgForRoom (pCfg->QRnumber);
 508         if (pRNCfg != NULL)
 509         {
 510                 RSSCfgLine *RSSCfg = (RSSCfgLine *)pRNCfg->NetConfigs[rssclient];
 511
 512                 while (RSSCfg != NULL)
 513                 {
 514                         if (RSSCfg == pCfg->pCfg)
 515                                 break;
 516
 517                         RSSCfg = RSSCfg->next;
 518                 }
 519                 if (RSSCfg != NULL)
 520                 {
 521                         pRNCfg->changed = 1;
 522                         RSSCfg->last_known_good = now;
 523                 }
 524         }
 525
 526         end_critical_section(S_NETCONFIGS);
 527 }
 528
 529 eNextState RSSAggregator_AnalyseReply(AsyncIO *IO)
 530 {
 531         HashPos *it = NULL;
 532         long len;
 533         const char *Key;
 534         pRSSConfig *pCfg;
 535         u_char rawdigest[MD5_DIGEST_LEN];
 536         struct MD5Context md5context;
 537         StrBuf *guid;
 538         rss_aggregator *Ctx = (rss_aggregator *) IO->Data;
 539
 540
 541         if ((IO->HttpReq.httpcode >= 300) &&
 542             (IO->HttpReq.httpcode < 400)  &&
 543             (Ctx->RedirectUrl != NULL)) {
 544
 545                 StrBuf *ErrMsg;
 546                 long lens[2];
 547                 const char *strs[2];
 548
 549                 SetRSSState(IO, eRSSFailure);
 550                 ErrMsg = NewStrBuf();
 551                 if (IO) EVRSSC_syslog(LOG_ALERT, "need a 200, got a %ld !\n",
 552                               IO->HttpReq.httpcode);
 553                 strs[0] = ChrPtr(Ctx->Url);
 554                 lens[0] = StrLength(Ctx->Url);
 555
 556                 strs[1] = ChrPtr(Ctx->rooms);
 557                 lens[1] = StrLength(Ctx->rooms);
 558
 559                 if (IO->HttpReq.CurlError == NULL)
 560                         IO->HttpReq.CurlError = "";
 561
 562                 StrBufPrintf(ErrMsg,
 563                              "Error while RSS-Aggregation Run of %s\n"
 564                              " need a 200, got a %ld !\n"
 565                              " Curl Error message: \n%s / %s\n"
 566                              " Redirect header points to: %s\n"
 567                              " Response text was: \n"
 568                              " \n %s\n",
 569                              ChrPtr(Ctx->Url),
 570                              IO->HttpReq.httpcode,
 571                              IO->HttpReq.errdesc,
 572                              IO->HttpReq.CurlError,
 573                              ChrPtr(Ctx->RedirectUrl),
 574                              ChrPtr(IO->HttpReq.ReplyData)
 575                         );
 576
 577                 CtdlAideFPMessage(
 578                         ChrPtr(ErrMsg),
 579                         "RSS Aggregation run failure",
 580                         2, strs, (long*) &lens,
 581                         CCID,
 582                         IO->ID,
 583                         EvGetNow(IO));
 584
 585                 FreeStrBuf(&ErrMsg);
 586                 EVRSSC_syslog(LOG_DEBUG,
 587                               "RSS feed returned an invalid http status code. <%s><HTTP %ld>\n",
 588                               ChrPtr(Ctx->Url),
 589                               IO->HttpReq.httpcode);
 590                 return eAbort;
 591         }
 592         else if (IO->HttpReq.httpcode != 200)
 593         {
 594                 StrBuf *ErrMsg;
 595                 long lens[2];
 596                 const char *strs[2];
 597
 598                 SetRSSState(IO, eRSSFailure);
 599                 ErrMsg = NewStrBuf();
 600                 if (IO) EVRSSC_syslog(LOG_ALERT, "need a 200, got a %ld !\n",
 601                               IO->HttpReq.httpcode);
 602                 strs[0] = ChrPtr(Ctx->Url);
 603                 lens[0] = StrLength(Ctx->Url);
 604
 605                 strs[1] = ChrPtr(Ctx->rooms);
 606                 lens[1] = StrLength(Ctx->rooms);
 607
 608                 if (IO->HttpReq.CurlError == NULL)
 609                         IO->HttpReq.CurlError = "";
 610
 611                 StrBufPrintf(ErrMsg,
 612                              "Error while RSS-Aggregation Run of %s\n"
 613                              " need a 200, got a %ld !\n"
 614                              " Curl Error message: \n%s / %s\n"
 615                              " Response text was: \n"
 616                              " \n %s\n",
 617                              ChrPtr(Ctx->Url),
 618                              IO->HttpReq.httpcode,
 619                              IO->HttpReq.errdesc,
 620                              IO->HttpReq.CurlError,
 621                              ChrPtr(IO->HttpReq.ReplyData)
 622                         );
 623
 624                 CtdlAideFPMessage(
 625                         ChrPtr(ErrMsg),
 626                         "RSS Aggregation run failure",
 627                         2, strs, (long*) &lens,
 628                         CCID,
 629                         IO->ID,
 630                         EvGetNow(IO));
 631
 632                 FreeStrBuf(&ErrMsg);
 633                 EVRSSC_syslog(LOG_DEBUG,
 634                               "RSS feed returned an invalid http status code. <%s><HTTP %ld>\n",
 635                               ChrPtr(Ctx->Url),
 636                               IO->HttpReq.httpcode);
 637                 return eAbort;
 638         }
 639
 640         pCfg = &Ctx->Cfg;
 641
 642         while (pCfg != NULL)
 643         {
 644                 UpdateLastKnownGood (pCfg, EvGetNow(IO));
 645                 if ((Ctx->roomlist_parts > 1) &&
 646                     (it == NULL))
 647                 {
 648                         it = GetNewHashPos(RSSFetchUrls, 0);
 649                 }
 650                 if (it != NULL)
 651                 {
 652                         void *vptr;
 653                         if (GetNextHashPos(Ctx->OtherQRnumbers, it, &len, &Key, &vptr))
 654                                 pCfg = vptr;
 655                         else
 656                                 pCfg = NULL;
 657                 }
 658                 else
 659                         pCfg = NULL;
 660         }
 661         DeleteHashPos (&it);
 662
 663         SetRSSState(IO, eRSSUT);
 664
 665         MD5Init(&md5context);
 666
 667         MD5Update(&md5context,
 668                   (const unsigned char*)SKEY(IO->HttpReq.ReplyData));
 669
 670         MD5Update(&md5context,
 671                   (const unsigned char*)SKEY(Ctx->Url));
 672
 673         MD5Final(rawdigest, &md5context);
 674         guid = NewStrBufPlain(NULL,
 675                               MD5_DIGEST_LEN * 2 + 12 /* _rss2ctdl*/);
 676         StrBufHexEscAppend(guid, NULL, rawdigest, MD5_DIGEST_LEN);
 677         StrBufAppendBufPlain(guid, HKEY("_rssFM"), 0);
 678         if (StrLength(guid) > 40)
 679                 StrBufCutAt(guid, 40, NULL);
 680         /* Find out if we've already seen this item */
 681
 682 #ifndef DEBUG_RSS
 683
 684         if (CheckIfAlreadySeen("RSS Whole",
 685                                guid,
 686                                EvGetNow(IO),
 687                                EvGetNow(IO) - USETABLE_ANTIEXPIRE,
 688                                eUpdate,
 689                                CCID, IO->ID)
 690             != 0)
 691         {
 692                 FreeStrBuf(&guid);
 693
 694                 EVRSSC_syslog(LOG_DEBUG, "RSS feed already seen. <%s>\n", ChrPtr(Ctx->Url));
 695                 return eAbort;
 696         }
 697         FreeStrBuf(&guid);
 698 #endif
 699         SetRSSState(IO, eRSSParsing);
 700         return RSSAggregator_ParseReply(IO);
 701 }
 702
 703 eNextState RSSAggregator_FinishHttp(AsyncIO *IO)
 704 {
 705         return CurlQueueDBOperation(IO, RSSAggregator_AnalyseReply);
 706 }
 707
 708 /*
 709  * Begin a feed parse
 710  */
 711 int rss_do_fetching(rss_aggregator *RSSAggr)
 712 {
 713         AsyncIO         *IO = &RSSAggr->IO;
 714         rss_item *ri;
 715         time_t now;
 716         CURLcode sta;
 717         CURL *chnd;
 718
 719
 720         now = time(NULL);
 721
 722         if ((RSSAggr->next_poll != 0) && (now < RSSAggr->next_poll))
 723                 return 0;
 724
 725         ri = (rss_item*) malloc(sizeof(rss_item));
 726         memset(ri, 0, sizeof(rss_item));
 727         RSSAggr->Item = ri;
 728
 729         if (! InitcURLIOStruct(&RSSAggr->IO,
 730                                RSSAggr,
 731                                "Citadel RSS Client",
 732                                RSSAggregator_FinishHttp,
 733                                RSSAggregator_Terminate,
 734                                RSSAggregator_TerminateDB,
 735                                RSSAggregator_ShutdownAbort))
 736         {
 737                 EVRSSCM_syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
 738                 return 0;
 739         }
 740         chnd = IO->HttpReq.chnd;
 741         OPT(HEADERDATA, IO);
 742         OPT(HEADERFUNCTION, GetLocationString);
 743         SetRSSState(IO, eRSSCreated);
 744
 745         safestrncpy(((CitContext*)RSSAggr->IO.CitContext)->cs_host,
 746                     ChrPtr(RSSAggr->Url),
 747                     sizeof(((CitContext*)RSSAggr->IO.CitContext)->cs_host));
 748
 749         EVRSSC_syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", ChrPtr(RSSAggr->Url));
 750         ParseURL(&RSSAggr->IO.ConnectMe, RSSAggr->Url, 80);
 751         CurlPrepareURL(RSSAggr->IO.ConnectMe);
 752
 753         SetRSSState(IO, eRSSFetching);
 754         QueueCurlContext(&RSSAggr->IO);
 755         return 1;
 756 }
 757
 758 /*
 759  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
 760  */
 761 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data, OneRoomNetCfg *OneRNCFG)
 762 {
 763         const RSSCfgLine *RSSCfg = (RSSCfgLine *)OneRNCFG->NetConfigs[rssclient];
 764         rss_aggregator *RSSAggr = NULL;
 765         rss_aggregator *use_this_RSSAggr = NULL;
 766         void *vptr;
 767
 768         pthread_mutex_lock(&RSSQueueMutex);
 769         if (GetHash(RSSQueueRooms, LKEY(qrbuf->QRnumber), &vptr))
 770         {
 771                 EVRSSQ_syslog(LOG_DEBUG,
 772                               "rssclient: [%ld] %s already in progress.\n",
 773                               qrbuf->QRnumber,
 774                               qrbuf->QRname);
 775                 pthread_mutex_unlock(&RSSQueueMutex);
 776                 return;
 777         }
 778         pthread_mutex_unlock(&RSSQueueMutex);
 779
 780         if (server_shutting_down) return;
 781
 782         while (RSSCfg != NULL)
 783         {
 784                 pthread_mutex_lock(&RSSQueueMutex);
 785                 GetHash(RSSFetchUrls,
 786                         SKEY(RSSCfg->Url),
 787                         &vptr);
 788
 789                 use_this_RSSAggr = (rss_aggregator *)vptr;
 790                 if (use_this_RSSAggr != NULL)
 791                 {
 792                         pRSSConfig *pRSSCfg;
 793
 794                         StrBufAppendBufPlain(
 795                                 use_this_RSSAggr->rooms,
 796                                 qrbuf->QRname,
 797                                 -1, 0);
 798                         if (use_this_RSSAggr->roomlist_parts==1)
 799                         {
 800                                 use_this_RSSAggr->OtherQRnumbers
 801                                         = NewHash(1, lFlathash);
 802                         }
 803
 804                         pRSSCfg = (pRSSConfig *) malloc(sizeof(pRSSConfig));
 805
 806                         pRSSCfg->QRnumber = qrbuf->QRnumber;
 807                         pRSSCfg->pCfg = RSSCfg;
 808
 809                         Put(use_this_RSSAggr->OtherQRnumbers,
 810                             LKEY(qrbuf->QRnumber),
 811                             pRSSCfg,
 812                             NULL);
 813                         use_this_RSSAggr->roomlist_parts++;
 814
 815                         pthread_mutex_unlock(&RSSQueueMutex);
 816
 817                         RSSCfg = RSSCfg->next;
 818                         continue;
 819                 }
 820                 pthread_mutex_unlock(&RSSQueueMutex);
 821
 822                 RSSAggr = (rss_aggregator *) malloc(
 823                         sizeof(rss_aggregator));
 824
 825                 memset (RSSAggr, 0, sizeof(rss_aggregator));
 826                 RSSAggr->Cfg.QRnumber = qrbuf->QRnumber;
 827                 RSSAggr->Cfg.pCfg = RSSCfg;
 828                 RSSAggr->roomlist_parts = 1;
 829                 RSSAggr->Url = NewStrBufDup(RSSCfg->Url);
 830
 831                 RSSAggr->ItemType = RSS_UNSET;
 832
 833                 RSSAggr->rooms = NewStrBufPlain(
 834                         qrbuf->QRname, -1);
 835
 836                 pthread_mutex_lock(&RSSQueueMutex);
 837
 838                 Put(RSSFetchUrls,
 839                     SKEY(RSSAggr->Url),
 840                     RSSAggr,
 841                     DeleteRssCfg);
 842
 843                 pthread_mutex_unlock(&RSSQueueMutex);
 844                 RSSCfg = RSSCfg->next;
 845         }
 846 }
 847
 848 /*
 849  * Scan for rooms that have RSS client requests configured
 850  */
 851 void rssclient_scan(void) {
 852         int RSSRoomCount, RSSCount;
 853         rss_aggregator *rptr = NULL;
 854         void *vrptr = NULL;
 855         HashPos *it;
 856         long len;
 857         const char *Key;
 858         time_t now = time(NULL);
 859
 860         /* Run no more than once every 15 minutes. */
 861         if ((now - last_run) < 900) {
 862                 EVRSSQ_syslog(LOG_DEBUG,
 863                               "Client: polling interval not yet reached; last run was %ldm%lds ago",
 864                               ((now - last_run) / 60),
 865                               ((now - last_run) % 60)
 866                 );
 867                 return;
 868         }
 869
 870         /*
 871          * This is a simple concurrency check to make sure only one rssclient
 872          * run is done at a time.
 873          */
 874         pthread_mutex_lock(&RSSQueueMutex);
 875         RSSCount = GetCount(RSSFetchUrls);
 876         RSSRoomCount = GetCount(RSSQueueRooms);
 877         pthread_mutex_unlock(&RSSQueueMutex);
 878
 879         if ((RSSRoomCount > 0) || (RSSCount > 0)) {
 880                 EVRSSQ_syslog(LOG_DEBUG,
 881                               "rssclient: concurrency check failed; %d rooms and %d url's are queued",
 882                               RSSRoomCount, RSSCount
 883                         );
 884                 return;
 885         }
 886
 887         become_session(&rss_CC);
 888         EVRSSQM_syslog(LOG_DEBUG, "rssclient started");
 889         CtdlForEachNetCfgRoom(rssclient_scan_room, NULL, rssclient);
 890
 891         if (GetCount(RSSFetchUrls) > 0)
 892         {
 893                 pthread_mutex_lock(&RSSQueueMutex);
 894                 EVRSSQ_syslog(LOG_DEBUG,
 895                                "rssclient starting %d Clients",
 896                                GetCount(RSSFetchUrls));
 897
 898                 it = GetNewHashPos(RSSFetchUrls, 0);
 899                 while (!server_shutting_down &&
 900                        GetNextHashPos(RSSFetchUrls, it, &len, &Key, &vrptr) &&
 901                        (vrptr != NULL)) {
 902                         rptr = (rss_aggregator *)vrptr;
 903                         if (!rss_do_fetching(rptr))
 904                                 UnlinkRSSAggregator(rptr);
 905                 }
 906                 DeleteHashPos(&it);
 907                 pthread_mutex_unlock(&RSSQueueMutex);
 908         }
 909         else
 910                 EVRSSQM_syslog(LOG_DEBUG, "Nothing to do.");
 911
 912         EVRSSQM_syslog(LOG_DEBUG, "rssclient ended\n");
 913         return;
 914 }
 915
 916 void rss_cleanup(void)
 917 {
 918         /* citthread_mutex_destroy(&RSSQueueMutex); TODO */
 919         DeleteHash(&RSSFetchUrls);
 920         DeleteHash(&RSSQueueRooms);
 921 }
 922
 923 void LogDebugEnableRSSClient(const int n)
 924 {
 925         RSSClientDebugEnabled = n;
 926 }
 927
 928
 929 typedef struct __RSSVetoInfo {
 930         StrBuf *ErrMsg;
 931         time_t Now;
 932         int Veto;
 933 }RSSVetoInfo;
 934
 935 void rssclient_veto_scan_room(struct ctdlroom *qrbuf, void *data, OneRoomNetCfg *OneRNCFG)
 936 {
 937         RSSVetoInfo *Info = (RSSVetoInfo *) data;
 938         const RSSCfgLine *RSSCfg = (RSSCfgLine *)OneRNCFG->NetConfigs[rssclient];
 939
 940         while (RSSCfg != NULL)
 941         {
 942                 if ((RSSCfg->last_known_good != 0) &&
 943                     (RSSCfg->last_known_good + USETABLE_ANTIEXPIRE < Info->Now))
 944                 {
 945                         StrBufAppendPrintf(Info->ErrMsg,
 946                                            "RSS feed not seen for a %d days:: <",
 947                                            (Info->Now - RSSCfg->last_known_good) / (24 * 60 * 60));
 948
 949                         StrBufAppendBuf(Info->ErrMsg, RSSCfg->Url, 0);
 950                         StrBufAppendBufPlain(Info->ErrMsg, HKEY(">\n"), 0);
 951                 }
 952                 RSSCfg = RSSCfg->next;
 953         }
 954 }
 955
 956 int RSSCheckUsetableVeto(StrBuf *ErrMsg)
 957 {
 958         RSSVetoInfo Info;
 959
 960         Info.ErrMsg = ErrMsg;
 961         Info.Now = time (NULL);
 962         Info.Veto = 0;
 963
 964         CtdlForEachNetCfgRoom(rssclient_veto_scan_room, &Info, rssclient);
 965
 966         return Info.Veto;;
 967 }
 968
 969
 970
 971
 972 void ParseRSSClientCfgLine(const CfgLineType *ThisOne, StrBuf *Line, const char *LinePos, OneRoomNetCfg *OneRNCFG)
 973 {
 974         RSSCfgLine *RSSCfg;
 975
 976         RSSCfg = (RSSCfgLine *) malloc (sizeof(RSSCfgLine));
 977         RSSCfg->Url = NewStrBufPlain (NULL, StrLength (Line));
 978
 979
 980         StrBufExtract_NextToken(RSSCfg->Url, Line, &LinePos, '|');
 981         RSSCfg->last_known_good = StrBufExtractNext_long(Line, &LinePos, '|');
 982
 983
 984         RSSCfg->next = (RSSCfgLine *)OneRNCFG->NetConfigs[ThisOne->C];
 985         OneRNCFG->NetConfigs[ThisOne->C] = (RoomNetCfgLine*) RSSCfg;
 986 }
 987
 988 void SerializeRSSClientCfgLine(const CfgLineType *ThisOne, StrBuf *OutputBuffer, OneRoomNetCfg *RNCfg, RoomNetCfgLine *data)
 989 {
 990         RSSCfgLine *RSSCfg = (RSSCfgLine*) data;
 991
 992         StrBufAppendBufPlain(OutputBuffer, CKEY(ThisOne->Str), 0);
 993         StrBufAppendBufPlain(OutputBuffer, HKEY("|"), 0);
 994         StrBufAppendBufPlain(OutputBuffer, SKEY(RSSCfg->Url), 0);
 995         StrBufAppendPrintf(OutputBuffer, "|%ld\n", RSSCfg->last_known_good);
 996 }
 997
 998 void DeleteRSSClientCfgLine(const CfgLineType *ThisOne, RoomNetCfgLine **data)
 999 {
1000         RSSCfgLine *RSSCfg = (RSSCfgLine*) *data;
1001
1002         FreeStrBuf(&RSSCfg->Url);
1003         free(*data);
1004         *data = NULL;
1005 }
1006
1007
1008 CTDL_MODULE_INIT(rssclient)
1009 {
1010         if (!threading)
1011         {
1012                 CtdlRegisterTDAPVetoHook (RSSCheckUsetableVeto, CDB_USETABLE, 0);
1013
1014                 CtdlREGISTERRoomCfgType(rssclient, ParseRSSClientCfgLine, 0, 1, SerializeRSSClientCfgLine, DeleteRSSClientCfgLine);
1015                 pthread_mutex_init(&RSSQueueMutex, NULL);
1016                 RSSQueueRooms = NewHash(1, lFlathash);
1017                 RSSFetchUrls = NewHash(1, NULL);
1018                 syslog(LOG_INFO, "%s\n", curl_version());
1019                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);
1020                 CtdlRegisterEVCleanupHook(rss_cleanup);
1021                 CtdlRegisterDebugFlagHook(HKEY("rssclient"), LogDebugEnableRSSClient, &RSSClientDebugEnabled);
1022         }
1023         else
1024         {
1025                 CtdlFillSystemContext(&rss_CC, "rssclient");
1026         }
1027         return "rssclient";
1028 }