citadel/modules/rssclient/serv_rssclient.c

   1 /*
   2  * Bring external RSS feeds into rooms.
   3  *
   4  * Copyright (c) 2007-2010 by the citadel.org team
   5  *
   6  * This program is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License as published by
   8  * the Free Software Foundation; either version 3 of the License, or
   9  * (at your option) any later version.
  10  *
  11  * This program is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  * GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * along with this program; if not, write to the Free Software
  18  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  19  */
  20
  21 #include <stdlib.h>
  22 #include <unistd.h>
  23 #include <stdio.h>
  24
  25 #if TIME_WITH_SYS_TIME
  26 # include <sys/time.h>
  27 # include <time.h>
  28 #else
  29 # if HAVE_SYS_TIME_H
  30 #  include <sys/time.h>
  31 # else
  32 #  include <time.h>
  33 # endif
  34 #endif
  35
  36 #include <ctype.h>
  37 #include <string.h>
  38 #include <errno.h>
  39 #include <sys/types.h>
  40 #include <sys/stat.h>
  41 #include <expat.h>
  42 #include <curl/curl.h>
  43 #include <libcitadel.h>
  44 #include "citadel.h"
  45 #include "server.h"
  46 #include "citserver.h"
  47 #include "support.h"
  48 #include "config.h"
  49 #include "threads.h"
  50 #include "ctdl_module.h"
  51 #include "clientsocket.h"
  52 #include "msgbase.h"
  53 #include "parsedate.h"
  54 #include "database.h"
  55 #include "citadel_dirs.h"
  56 #include "md5.h"
  57 #include "context.h"
  58
  59 typedef struct rssnetcfg rssnetcfg;
  60 struct rssnetcfg {
  61         rssnetcfg *next;
  62         char url[256];
  63         char *rooms;
  64         time_t last_error_when;
  65         int ItemType;
  66 };
  67
  68 #define RSS_UNSET       (1<<0)
  69 #define RSS_RSS         (1<<1)
  70 #define RSS_ATOM        (1<<2)
  71 #define RSS_REQUIRE_BUF (1<<3)
  72
  73 typedef struct _rss_item {
  74         char *roomlist;
  75         int done_parsing;
  76         StrBuf *guid;
  77         StrBuf *title;
  78         StrBuf *link;
  79         StrBuf *linkTitle;
  80         StrBuf *reLink;
  81         StrBuf *reLinkTitle;
  82         StrBuf *description;
  83         time_t pubdate;
  84         StrBuf *channel_title;
  85         int item_tag_nesting;
  86         StrBuf *author_or_creator;
  87         StrBuf *author_url;
  88         StrBuf *author_email;
  89 }rss_item;
  90
  91
  92 typedef void (*rss_handler_func)(StrBuf *CData,
  93                                  rss_item *ri,
  94                                  rssnetcfg *Cfg,
  95                                  const char** Attr);
  96
  97 typedef struct __rss_xml_handler {
  98         int Flags;
  99         rss_handler_func Handler;
 100 }rss_xml_handler;
 101
 102
 103 typedef struct _rsscollection {
 104         StrBuf *CData;
 105         StrBuf *Key;
 106
 107         rss_item *Item;
 108         rssnetcfg *Cfg;
 109
 110         rss_xml_handler *Current;
 111 } rsscollection;
 112
 113 struct rssnetcfg *rnclist = NULL;
 114 HashList *StartHandlers;
 115 HashList *EndHandlers;
 116 HashList *KnownNameSpaces;
 117 void AddRSSStartHandler(rss_handler_func Handler, int Flags, const char *key, long len)
 118 {
 119         rss_xml_handler *h;
 120         h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
 121         h->Flags = Flags;
 122         h->Handler = Handler;
 123         Put(StartHandlers, key, len, h, NULL);
 124 }
 125 void AddRSSEndHandler(rss_handler_func Handler, int Flags, const char *key, long len)
 126 {
 127         rss_xml_handler *h;
 128         h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
 129         h->Flags = Flags;
 130         h->Handler = Handler;
 131         Put(EndHandlers, key, len, h, NULL);
 132 }
 133
 134 ///#if 0
 135 //#ifdef HAVE_ICONV
 136 #include <iconv.h>
 137
 138
 139 /*
 140  * dug this out of the trashcan of the midgard project, lets see if it works for us.
 141  * original code by Alexander Bokovoy <bokovoy@avilink.ne> distributed under GPL V2 or later
 142  */
 143
 144 /* Returns:
 145  >= 0 - successfull, 0 means conversion doesn't use multibyte sequences
 146    -1 - error during iconv_open call
 147    -2 - error during iconv_close call
 148    ----------------------------------
 149    This function expects that multibyte encoding in 'charset' wouldn't have
 150    characters with more than 3 bytes. It is not intended to convert UTF-8 because
 151    we'll never receive UTF-8 in our handler (it is handled by Exat itself).
 152 */
 153 static int
 154 fill_encoding_info (const char *charset, XML_Encoding * info)
 155 {
 156   iconv_t cd = (iconv_t)(-1);
 157   int flag;
 158         CtdlLogPrintf(0, "RSS: fill encoding info ...\n");
 159
 160 #if G_BYTE_ORDER == G_LITTLE_ENDIAN
 161   cd = iconv_open ("UCS-2LE", charset);
 162 #else
 163   cd = iconv_open ("UCS-2BE", charset);
 164 #endif
 165
 166   if (cd == (iconv_t) (-1))
 167     {
 168       return -1;
 169     }
 170
 171   {
 172     unsigned short out;
 173     unsigned char buf[4];
 174     unsigned int i0, i1, i2;
 175     int result;
 176     flag = 0;
 177     for (i0 = 0; i0 < 0x100; i0++)
 178       {
 179         buf[0] = i0;
 180         info->map[i0] = 0;
 181         //result = try (cd, buf, 1, &out);
 182         if (result < 0)
 183           {
 184           }
 185         else if (result > 0)
 186           {
 187             info->map[i0] = out;
 188           }
 189         else
 190           {
 191             for (i1 = 0; i1 < 0x100; i1++)
 192               {
 193                 buf[1] = i1;
 194                 ///result = try (cd, buf, 2, &out);
 195                 if (result < 0)
 196                   {
 197                   }
 198                 else if (result > 0)
 199                   {
 200                     flag++;
 201                     info->map[i0] = -2;
 202                   }
 203                 else
 204                   {
 205                     for (i2 = 0; i2 < 0x100; i2++)
 206                       {
 207                         buf[2] = i2;
 208                         ////result = try (cd, buf, 3, &out);
 209                         if (result < 0)
 210                           {
 211                           }
 212                         else if (result > 0)
 213                           {
 214                             flag++;
 215                             info->map[i0] = -3;
 216                           }
 217                       }
 218                   }
 219               }
 220           }
 221       }
 222   }
 223
 224   if (iconv_close (cd) < 0)
 225     {
 226       return -2;
 227     }
 228   return flag;
 229 }
 230
 231 static int
 232 iconv_convertor (void *data, const char *s)
 233 {
 234   XML_Encoding *info = data;
 235   int res;
 236         CtdlLogPrintf(0, "RSS: Converting ...\n");
 237
 238   if (s == NULL)
 239     return -1;
 240 /*
 241   GByteArray *result;
 242   result = g_byte_array_new ();
 243   if (process_block (info->data, (char *) s, strlen (s), result) == 0)
 244     {
 245       res = *(result->data);
 246       g_byte_array_free (result, TRUE);
 247       return res;
 248     }
 249   g_byte_array_free (result, TRUE);
 250 */
 251   return -1;
 252 }
 253
 254 static void
 255 my_release (void *data)
 256 {
 257   iconv_t cd = (iconv_t) data;
 258   if (iconv_close (cd) != 0)
 259     {
 260 /// TODO: uh no.      exit (1);
 261     }
 262 }
 263 int
 264 handle_unknown_xml_encoding (void *encodingHandleData,
 265                              const XML_Char * name,
 266                              XML_Encoding * info)
 267 {
 268   int result;
 269   CtdlLogPrintf(0, "RSS: unknown encoding ...\n");
 270   result = fill_encoding_info (name, info);
 271   if (result >= 0)
 272     {
 273       /*
 274         Special case: client asked for reverse conversion, we'll provide him with
 275         iconv descriptor which handles it. Client should release it by himself.
 276       */
 277       if(encodingHandleData != NULL)
 278             *((iconv_t *)encodingHandleData) = iconv_open(name, "UTF-8");
 279       /*
 280          Optimization: we do not need conversion function if encoding is one-to-one,
 281          info->map table will be enough
 282        */
 283       if (result == 0)
 284         {
 285           info->data = NULL;
 286           info->convert = NULL;
 287           info->release = NULL;
 288           return 1;
 289         }
 290       /*
 291          We do need conversion function because this encoding uses multibyte sequences
 292        */
 293       info->data = (void *) iconv_open ("UTF-8", name);
 294       if ((int)info->data == -1)
 295         return -1;
 296       info->convert = iconv_convertor;
 297       info->release = my_release;
 298       return 1;
 299     }
 300   if(encodingHandleData != NULL)
 301     *(iconv_t *)encodingHandleData = NULL;
 302   return 0;
 303 }
 304
 305 ///#endif
 306 //#endif
 307
 308 void AppendLink(StrBuf *Message, StrBuf *link, StrBuf *LinkTitle, const char *Title)
 309 {
 310         if (StrLength(link) > 0)
 311         {
 312                 StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
 313                 StrBufAppendBuf(Message, link, 0);
 314                 StrBufAppendBufPlain(Message, HKEY("\">"), 0);
 315                 if (StrLength(LinkTitle) > 0)
 316                         StrBufAppendBuf(Message, LinkTitle, 0);
 317                 else if ((Title != NULL) && !IsEmptyStr(Title))
 318                         StrBufAppendBufPlain(Message, Title, -1, 0);
 319                 else
 320                         StrBufAppendBuf(Message, link, 0);
 321                 StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
 322         }
 323 }
 324 /*
 325  * Commit a fetched and parsed RSS item to disk
 326  */
 327 void rss_save_item(rss_item *ri)
 328 {
 329
 330         struct MD5Context md5context;
 331         u_char rawdigest[MD5_DIGEST_LEN];
 332         int i;
 333         char utmsgid[SIZ];
 334         struct cdbdata *cdbut;
 335         struct UseTable ut;
 336         struct CtdlMessage *msg;
 337         struct recptypes *recp = NULL;
 338         int msglen = 0;
 339         StrBuf *Message;
 340
 341         recp = (struct recptypes *) malloc(sizeof(struct recptypes));
 342         if (recp == NULL) return;
 343         memset(recp, 0, sizeof(struct recptypes));
 344         memset(&ut, 0, sizeof(struct UseTable));
 345         recp->recp_room = strdup(ri->roomlist);
 346         recp->num_room = num_tokens(ri->roomlist, '|');
 347         recp->recptypes_magic = RECPTYPES_MAGIC;
 348
 349         /* Construct a GUID to use in the S_USETABLE table.
 350          * If one is not present in the item itself, make one up.
 351          */
 352         if (ri->guid != NULL) {
 353                 StrBufSpaceToBlank(ri->guid);
 354                 StrBufTrim(ri->guid);
 355                 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ChrPtr(ri->guid));
 356         }
 357         else {
 358                 MD5Init(&md5context);
 359                 if (ri->title != NULL) {
 360                         MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->title), StrLength(ri->title));
 361                 }
 362                 if (ri->link != NULL) {
 363                         MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->link), StrLength(ri->link));
 364                 }
 365                 MD5Final(rawdigest, &md5context);
 366                 for (i=0; i<MD5_DIGEST_LEN; i++) {
 367                         sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
 368                         utmsgid[i*2] = tolower(utmsgid[i*2]);
 369                         utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
 370                 }
 371                 strcat(utmsgid, "_rss2ctdl");
 372         }
 373
 374         /* Find out if we've already seen this item */
 375
 376         cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
 377 #ifndef DEBUG_RSS
 378         if (cdbut != NULL) {
 379                 /* Item has already been seen */
 380                 CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
 381                 cdb_free(cdbut);
 382
 383                 /* rewrite the record anyway, to update the timestamp */
 384                 strcpy(ut.ut_msgid, utmsgid);
 385                 ut.ut_timestamp = time(NULL);
 386                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 387         }
 388         else
 389 #endif
 390 {
 391                 /* Item has not been seen, so save it. */
 392                 CtdlLogPrintf(CTDL_DEBUG, "RSS: saving item...\n");
 393                 if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY(""));
 394                 StrBufSpaceToBlank(ri->description);
 395                 msg = malloc(sizeof(struct CtdlMessage));
 396                 memset(msg, 0, sizeof(struct CtdlMessage));
 397                 msg->cm_magic = CTDLMESSAGE_MAGIC;
 398                 msg->cm_anon_type = MES_NORMAL;
 399                 msg->cm_format_type = FMT_RFC822;
 400
 401                 if (ri->guid != NULL) {
 402                         msg->cm_fields['E'] = strdup(ChrPtr(ri->guid));
 403                 }
 404
 405                 if (ri->author_or_creator != NULL) {
 406                         char *From;
 407                         StrBuf *Encoded, *QPEncoded;
 408                         StrBuf *UserName;
 409                         StrBuf *EmailAddress;
 410                         StrBuf *EncBuf;
 411                         int FromAt;
 412                         int FromLen;
 413
 414                         UserName = NewStrBuf();
 415                         EmailAddress = NewStrBuf();
 416                         EncBuf = NewStrBuf();
 417 ////TODO!
 418                         StrBufTrim(ri->author_or_creator);
 419                         From = html_to_ascii(ChrPtr(ri->author_or_creator),
 420                                              StrLength(ri->author_or_creator),
 421                                              512, 0);
 422                         FromLen = strlen(From);
 423                         if (From[FromLen - 1] == '\n')
 424                         {
 425                                 From[FromLen - 1] = '\0';
 426                         }
 427                         FromAt = strchr(From, '@') != NULL;
 428                         if (!FromAt && StrLength (ri->author_email) > 0)
 429                         {
 430                                 Encoded = NewStrBuf();
 431                                 if (!IsEmptyStr(From))
 432                                 {
 433                                         StrBufPrintf(Encoded,
 434                                                      "\"%s\" <%s>",
 435                                                      From,
 436                                                      ChrPtr(ri->author_email));
 437                                 }
 438                                 else
 439                                 {
 440                                         StrBufPrintf(Encoded,
 441                                                      "<%s>",
 442                                                      ChrPtr(ri->author_email));
 443                                 }
 444                         }
 445                         else
 446                         {
 447                                 if (FromAt)
 448                                         Encoded = NewStrBufPlain(From, -1);
 449                                 else
 450                                 {
 451                                         Encoded = NewStrBuf();
 452                                         StrBufPrintf(Encoded,
 453                                                      "\"%s\" <%s>",
 454                                                      From,
 455                                                      "rss@localhost"); /// TODO: get hostname?
 456                                 }
 457                         }
 458                         free(From);
 459                         StrBufTrim(Encoded);
 460                         QPEncoded = StrBufSanitizeEmailRecipientVector(Encoded, UserName, EmailAddress, EncBuf);
 461                         msg->cm_fields['A'] = SmashStrBuf(&QPEncoded);
 462
 463                         FreeStrBuf(&Encoded);
 464                         FreeStrBuf(&UserName);
 465                         FreeStrBuf(&EmailAddress);
 466                         FreeStrBuf(&EncBuf);
 467
 468                 }
 469                 else {
 470                         msg->cm_fields['A'] = strdup("rss");
 471                 }
 472
 473                 msg->cm_fields['N'] = strdup(NODENAME);
 474                 if (ri->title != NULL) {
 475                         long len;
 476                         char *Sbj;
 477                         StrBuf *Encoded, *QPEncoded;
 478
 479                         QPEncoded = NULL;
 480                         StrBufSpaceToBlank(ri->title);
 481                         len = StrLength(ri->title);
 482                         Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
 483                         len = strlen(Sbj);
 484                         if (Sbj[len - 1] == '\n')
 485                         {
 486                                 len --;
 487                                 Sbj[len] = '\0';
 488                         }
 489                         Encoded = NewStrBufPlain(Sbj, len);
 490                         free(Sbj);
 491
 492                         StrBufTrim(Encoded);
 493                         StrBufRFC2047encode(&QPEncoded, Encoded);
 494
 495                         msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
 496                         FreeStrBuf(&Encoded);
 497                 }
 498                 msg->cm_fields['T'] = malloc(64);
 499                 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
 500                 if (ri->channel_title != NULL) {
 501                         if (StrLength(ri->channel_title) > 0) {
 502                                 msg->cm_fields['O'] = strdup(ChrPtr(ri->channel_title));
 503                         }
 504                 }
 505                 if (ri->link == NULL)
 506                         ri->link = NewStrBufPlain(HKEY(""));
 507                 msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
 508
 509                 Message = NewStrBufPlain(NULL, StrLength(ri->description));
 510
 511                 StrBufPlain(Message, HKEY(
 512                          "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 513                          "<html><body>\n"));
 514
 515                 StrBufAppendBuf(Message, ri->description, 0);
 516                 StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
 517
 518                 AppendLink(Message, ri->link, ri->linkTitle, NULL);
 519                 AppendLink(Message, ri->reLink, ri->reLinkTitle, "Reply to this");
 520                 StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
 521
 522                 msg->cm_fields['M'] = SmashStrBuf(&Message);
 523
 524                 CtdlSubmitMsg(msg, recp, NULL, 0);
 525                 CtdlFreeMessage(msg);
 526
 527                 /* write the uidl to the use table so we don't store this item again */
 528                 strcpy(ut.ut_msgid, utmsgid);
 529                 ut.ut_timestamp = time(NULL);
 530                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 531         }
 532         free_recipients(recp);
 533 }
 534
 535
 536
 537 /*
 538  * Convert an RDF/RSS datestamp into a time_t
 539  */
 540 time_t rdf_parsedate(const char *p)
 541 {
 542         struct tm tm;
 543         time_t t = 0;
 544
 545         if (!p) return 0L;
 546         if (strlen(p) < 10) return 0L;
 547
 548         memset(&tm, 0, sizeof tm);
 549
 550         /*
 551          * If the timestamp appears to be in W3C datetime format, try to
 552          * parse it.  See also: http://www.w3.org/TR/NOTE-datetime
 553          *
 554          * This code, along with parsedate.c, is a potential candidate for
 555          * moving into libcitadel.
 556          */
 557         if ( (p[4] == '-') && (p[7] == '-') ) {
 558                 tm.tm_year = atoi(&p[0]) - 1900;
 559                 tm.tm_mon = atoi(&p[5]) - 1;
 560                 tm.tm_mday = atoi(&p[8]);
 561                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 562                         tm.tm_hour = atoi(&p[11]);
 563                         tm.tm_min = atoi(&p[14]);
 564                 }
 565                 return mktime(&tm);
 566         }
 567
 568         /* hmm... try RFC822 date stamp format */
 569
 570         t = parsedate(p);
 571         if (t > 0) return(t);
 572
 573         /* yeesh.  ok, just return the current date and time. */
 574         return(time(NULL));
 575 }
 576
 577 void flush_rss_item(rss_item *ri)
 578 {
 579         /* Initialize the feed item data structure */
 580         FreeStrBuf(&ri->guid);
 581         FreeStrBuf(&ri->title);
 582         FreeStrBuf(&ri->link);
 583         FreeStrBuf(&ri->author_or_creator);
 584         FreeStrBuf(&ri->author_email);
 585         FreeStrBuf(&ri->author_url);
 586         FreeStrBuf(&ri->description);
 587 }
 588
 589 void rss_xml_start(void *data, const char *supplied_el, const char **attr)
 590 {
 591         rss_xml_handler *h;
 592         rsscollection   *rssc = (rsscollection*) data;
 593         rssnetcfg       *Cfg = rssc->Cfg;
 594         rss_item        *ri = rssc->Item;
 595         void            *pv;
 596         const char      *pel;
 597         char            *sep = NULL;
 598
 599         /* Axe the namespace, we don't care about it */
 600 ///     CtdlLogPrintf(0, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
 601         pel = supplied_el;
 602         while (sep = strchr(pel, ':'), sep) {
 603                 pel = sep + 1;
 604         }
 605
 606         if (pel != supplied_el)
 607         {
 608                 void *v;
 609
 610                 if (!GetHash(KnownNameSpaces,
 611                              supplied_el,
 612                              pel - supplied_el - 1,
 613                              &v))
 614                 {
 615 #ifdef DEBUG_RSS
 616                         CtdlLogPrintf(0, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
 617                                       supplied_el);
 618 #endif
 619                         return;
 620                 }
 621         }
 622
 623         StrBufPlain(rssc->Key, pel, -1);
 624         StrBufLowerCase(rssc->Key);
 625         if (GetHash(StartHandlers, SKEY(rssc->Key), &pv))
 626         {
 627                 rssc->Current = h = (rss_xml_handler*) pv;
 628
 629                 if (((h->Flags & RSS_UNSET) != 0) &&
 630                     (Cfg->ItemType == RSS_UNSET))
 631                 {
 632                         h->Handler(rssc->CData, ri, Cfg, attr);
 633                 }
 634                 else if (((h->Flags & RSS_RSS) != 0) &&
 635                     (Cfg->ItemType == RSS_RSS))
 636                 {
 637                         h->Handler(rssc->CData, ri, Cfg, attr);
 638                 }
 639                 else if (((h->Flags & RSS_ATOM) != 0) &&
 640                          (Cfg->ItemType == RSS_ATOM))
 641                 {
 642                         h->Handler(rssc->CData, ri, Cfg, attr);
 643                 }
 644 #ifdef DEBUG_RSS
 645                 else
 646                         CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
 647 #endif
 648         }
 649 #ifdef DEBUG_RSS
 650         else
 651                 CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel,  supplied_el);
 652 #endif
 653 }
 654
 655 void rss_xml_end(void *data, const char *supplied_el)
 656 {
 657         rss_xml_handler *h;
 658         rsscollection   *rssc = (rsscollection*) data;
 659         rssnetcfg       *Cfg = rssc->Cfg;
 660         rss_item        *ri = rssc->Item;
 661         const char      *pel;
 662         char            *sep = NULL;
 663         void            *pv;
 664
 665         /* Axe the namespace, we don't care about it */
 666         pel = supplied_el;
 667         while (sep = strchr(pel, ':'), sep) {
 668                 pel = sep + 1;
 669         }
 670 //      CtdlLogPrintf(0, "RSS: END %s...\n", el);
 671         if (pel != supplied_el)
 672         {
 673                 void *v;
 674
 675                 if (!GetHash(KnownNameSpaces,
 676                              supplied_el,
 677                              pel - supplied_el - 1,
 678                              &v))
 679                 {
 680 #ifdef DEBUG_RSS
 681                         CtdlLogPrintf(0, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
 682                                       supplied_el, ChrPtr(rssc->CData));
 683 #endif
 684                         FlushStrBuf(rssc->CData);
 685                         return;
 686                 }
 687         }
 688
 689         StrBufPlain(rssc->Key, pel, -1);
 690         StrBufLowerCase(rssc->Key);
 691         if (GetHash(EndHandlers, SKEY(rssc->Key), &pv))
 692         {
 693                 h = (rss_xml_handler*) pv;
 694
 695                 if (((h->Flags & RSS_UNSET) != 0) &&
 696                     (Cfg->ItemType == RSS_UNSET))
 697                 {
 698                         h->Handler(rssc->CData, ri, Cfg, NULL);
 699                 }
 700                 else if (((h->Flags & RSS_RSS) != 0) &&
 701                     (Cfg->ItemType == RSS_RSS))
 702                 {
 703                         h->Handler(rssc->CData, ri, Cfg, NULL);
 704                 }
 705                 else if (((h->Flags & RSS_ATOM) != 0) &&
 706                          (Cfg->ItemType == RSS_ATOM))
 707                 {
 708                         h->Handler(rssc->CData, ri, Cfg, NULL);
 709                 }
 710 #ifdef DEBUG_RSS
 711                 else
 712                         CtdlLogPrintf(0, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 713 #endif
 714         }
 715 #ifdef DEBUG_RSS
 716         else
 717                 CtdlLogPrintf(0, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 718 #endif
 719         FlushStrBuf(rssc->CData);
 720         rssc->Current = NULL;
 721 }
 722
 723
 724
 725
 726
 727 void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 728 {
 729         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RSS feed.\n");
 730         Cfg->ItemType = RSS_RSS;
 731 }
 732
 733 void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 734 {
 735         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RDF feed.\n");
 736         Cfg->ItemType = RSS_RSS;
 737 }
 738
 739 void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 740 {
 741         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an ATOM feed.\n");
 742         Cfg->ItemType = RSS_ATOM;
 743 }
 744
 745
 746 void RSS_item_item_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 747 {
 748         ri->item_tag_nesting ++;
 749         flush_rss_item(ri);
 750 }
 751
 752 void ATOM_item_entry_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 753 {
 754 /* Atom feed... */
 755         ri->item_tag_nesting ++;
 756         flush_rss_item(ri);
 757 }
 758
 759 void ATOM_item_link_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 760 {
 761         int i;
 762         const char *pHref = NULL;
 763         const char *pType = NULL;
 764         const char *pRel = NULL;
 765         const char *pTitle = NULL;
 766
 767         for (i = 0; Attr[i] != NULL; i+=2)
 768         {
 769                 if (!strcmp(Attr[i], "href"))
 770                 {
 771                         pHref = Attr[i+1];
 772                 }
 773                 else if (!strcmp(Attr[i], "rel"))
 774                 {
 775                         pRel = Attr[i+1];
 776                 }
 777                 else if (!strcmp(Attr[i], "type"))
 778                 {
 779                         pType = Attr[i+1];
 780                 }
 781                 else if (!strcmp(Attr[i], "title"))
 782                 {
 783                         pTitle = Attr[i+1];
 784                 }
 785         }
 786         if (pHref == NULL)
 787                 return; /* WHUT? Pointing... where? */
 788         if ((pType != NULL) && !strcasecmp(pType, "application/atom+xml"))
 789                 return; /* these just point to other rss resources, we're not interested in them. */
 790         if (pRel != NULL)
 791         {
 792                 if (!strcasecmp (pRel, "replies"))
 793                 {
 794                         NewStrBufDupAppendFlush(&ri->reLink, NULL, pHref, -1);
 795                         StrBufTrim(ri->link);
 796                         NewStrBufDupAppendFlush(&ri->reLinkTitle, NULL, pTitle, -1);
 797                 }
 798                 else if (!strcasecmp(pRel, "alternate")) /* Alternative representation of this Item... */
 799                 {
 800                         NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 801                         StrBufTrim(ri->link);
 802                         NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
 803
 804                 }
 805 #if 0 /* these are also defined, but dunno what to do with them.. */
 806                 else if (!strcasecmp(pRel, "related"))
 807                 {
 808                 }
 809                 else if (!strcasecmp(pRel, "self"))
 810                 {
 811                 }
 812                 else if (!strcasecmp(pRel, "enclosure"))
 813                 {/* this reference can get big, and is probably the full article... */
 814                 }
 815                 else if (!strcasecmp(pRel, "via"))
 816                 {/* this article was provided via... */
 817                 }
 818 #endif
 819         }
 820         else if (StrLength(ri->link) == 0)
 821         {
 822                 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 823                 StrBufTrim(ri->link);
 824                 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
 825         }
 826 }
 827
 828
 829
 830
 831 void ATOMRSS_item_title_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 832 {
 833         if ((ri->item_tag_nesting == 0) && (StrLength(CData) > 0)) {
 834                 NewStrBufDupAppendFlush(&ri->channel_title, CData, NULL, 0);
 835                 StrBufTrim(ri->channel_title);
 836         }
 837 }
 838
 839 void RSS_item_guid_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 840 {
 841         if (StrLength(CData) > 0) {
 842                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
 843         }
 844 }
 845
 846 void ATOM_item_id_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 847 {
 848         if (StrLength(CData) > 0) {
 849                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
 850         }
 851 }
 852
 853
 854 void RSS_item_link_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 855 {
 856         if (StrLength(CData) > 0) {
 857                 NewStrBufDupAppendFlush(&ri->link, CData, NULL, 0);
 858                 StrBufTrim(ri->link);
 859         }
 860 }
 861
 862 void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 863 {
 864         if (StrLength(CData) > 0) {
 865                 NewStrBufDupAppendFlush(&ri->title, CData, NULL, 0);
 866                 StrBufTrim(ri->title);
 867         }
 868 }
 869
 870 void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 871 {
 872         long olen = StrLength (ri->description);
 873         long clen = StrLength (CData);
 874         if (clen > 0)
 875         {
 876                 if (olen == 0) {
 877                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 878                         StrBufTrim(ri->description);
 879                 }
 880                 else if (olen < clen) {
 881                         FlushStrBuf(ri->description);
 882                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 883                         StrBufTrim(ri->description);
 884                 }
 885         }
 886 }
 887 void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 888 {
 889         /* this can contain an abstract of the article. but we don't want to verwrite a full document if we already have it. */
 890         if ((StrLength(CData) > 0) && (StrLength(ri->description) == 0))
 891         {
 892                 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 893                 StrBufTrim(ri->description);
 894         }
 895 }
 896
 897 void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 898 {
 899         long olen = StrLength (ri->description);
 900         long clen = StrLength (CData);
 901         if (clen > 0)
 902         {
 903                 if (olen == 0) {
 904                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 905                         StrBufTrim(ri->description);
 906                 }
 907                 else if (olen < clen) {
 908                         FlushStrBuf(ri->description);
 909                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 910                         StrBufTrim(ri->description);
 911                 }
 912         }
 913 }
 914
 915 void ATOM_item_published_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 916 {
 917         if (StrLength(CData) > 0) {
 918                 StrBufTrim(CData);
 919                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 920         }
 921 }
 922
 923 void ATOM_item_updated_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 924 {
 925         if (StrLength(CData) > 0) {
 926                 StrBufTrim(CData);
 927                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 928         }
 929 }
 930
 931 void RSS_item_pubdate_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 932 {
 933         if (StrLength(CData) > 0) {
 934                 StrBufTrim(CData);
 935                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 936         }
 937 }
 938
 939
 940 void RSS_item_date_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 941 {
 942         if (StrLength(CData) > 0) {
 943                 StrBufTrim(CData);
 944                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 945         }
 946 }
 947
 948
 949
 950 void RSS_item_author_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 951 {
 952         if (StrLength(CData) > 0) {
 953                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
 954                 StrBufTrim(ri->author_or_creator);
 955         }
 956 }
 957
 958
 959 void ATOM_item_name_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 960 {
 961         if (StrLength(CData) > 0) {
 962                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
 963                 StrBufTrim(ri->author_or_creator);
 964         }
 965 }
 966
 967 void ATOM_item_email_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 968 {
 969         if (StrLength(CData) > 0) {
 970                 NewStrBufDupAppendFlush(&ri->author_email, CData, NULL, 0);
 971                 StrBufTrim(ri->author_email);
 972         }
 973 }
 974
 975 void RSS_item_creator_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 976 {
 977         if ((StrLength(CData) > 0) &&
 978             (StrLength(ri->author_or_creator) == 0))
 979         {
 980                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
 981                 StrBufTrim(ri->author_or_creator);
 982         }
 983 }
 984
 985
 986 void ATOM_item_uri_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 987 {
 988         if (StrLength(CData) > 0) {
 989                 NewStrBufDupAppendFlush(&ri->author_url, CData, NULL, 0);
 990                 StrBufTrim(ri->author_url);
 991         }
 992 }
 993
 994 void RSS_item_item_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 995 {
 996         --ri->item_tag_nesting;
 997         rss_save_item(ri);
 998 }
 999
1000
1001 void ATOM_item_entry_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1002 {
1003         --ri->item_tag_nesting;
1004         rss_save_item(ri);
1005 }
1006
1007 void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1008 {
1009 //              CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
1010         ri->done_parsing = 1;
1011
1012 }
1013 void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1014 {
1015 //              CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
1016         ri->done_parsing = 1;
1017 }
1018
1019
1020 void RSSATOM_item_ignore(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1021 {
1022 }
1023
1024
1025
1026 /*
1027  * This callback stores up the data which appears in between tags.
1028  */
1029 void rss_xml_cdata_start(void *data)
1030 {
1031         rsscollection *rssc = (rsscollection*) data;
1032
1033         FlushStrBuf(rssc->CData);
1034 }
1035
1036 void rss_xml_cdata_end(void *data)
1037 {
1038 }
1039 void rss_xml_chardata(void *data, const XML_Char *s, int len)
1040 {
1041         rsscollection *rssc = (rsscollection*) data;
1042
1043         StrBufAppendBufPlain (rssc->CData, s, len, 0);
1044 }
1045
1046 /*
1047  * Callback function for passing libcurl's output to expat for parsing
1048  */
1049 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
1050 {
1051         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
1052         return (size*nmemb);
1053 }
1054
1055
1056
1057 /*
1058  * Begin a feed parse
1059  */
1060 void rss_do_fetching(rssnetcfg *Cfg) {
1061         rsscollection rssc;
1062         rss_item ri;
1063         XML_Parser xp;
1064         StrBuf *Answer;
1065
1066         CURL *curl;
1067         CURLcode res;
1068         char errmsg[1024] = "";
1069         char *ptr;
1070         const char *at;
1071         long len;
1072
1073         memset(&ri, 0, sizeof(rss_item));
1074         rssc.Item = &ri;
1075         rssc.Cfg = Cfg;
1076
1077         CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
1078
1079         curl = curl_easy_init();
1080         if (!curl) {
1081                 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
1082                 return;
1083         }
1084         Answer = NewStrBufPlain(NULL, SIZ);
1085
1086         curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
1087         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
1088         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
1089         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
1090 //      curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
1091         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
1092         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
1093         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
1094 #ifdef CURLOPT_HTTP_CONTENT_DECODING
1095         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
1096         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
1097 #endif
1098         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
1099         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
1100         if (
1101                 (!IsEmptyStr(config.c_ip_addr))
1102                 && (strcmp(config.c_ip_addr, "*"))
1103                 && (strcmp(config.c_ip_addr, "::"))
1104                 && (strcmp(config.c_ip_addr, "0.0.0.0"))
1105         ) {
1106                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
1107         }
1108
1109         if (CtdlThreadCheckStop())
1110         {
1111                 curl_easy_cleanup(curl);
1112                 return;
1113         }
1114
1115         if (CtdlThreadCheckStop())
1116                 goto shutdown ;
1117
1118         res = curl_easy_perform(curl);
1119         if (res) {
1120                 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
1121         }
1122
1123         if (CtdlThreadCheckStop())
1124                 goto shutdown ;
1125
1126
1127
1128
1129         memset(&ri, 0, sizeof(rss_item));
1130         ri.roomlist = Cfg->rooms;
1131         rssc.CData = NewStrBufPlain(NULL, SIZ);
1132         rssc.Key = NewStrBuf();
1133         at = NULL;
1134         StrBufSipLine(rssc.Key, Answer, &at);
1135         ptr = NULL;
1136
1137 #define encoding "encoding=\""
1138         ptr = strstr(ChrPtr(rssc.Key), encoding);
1139         if (ptr != NULL)
1140         {
1141                 char *pche;
1142
1143                 ptr += sizeof (encoding) - 1;
1144                 pche = strchr(ptr, '"');
1145                 if (pche != NULL)
1146                         StrBufCutAt(rssc.Key, -1, pche);
1147                 else
1148                         ptr = "UTF-8";
1149         }
1150         else
1151                 ptr = "UTF-8";
1152
1153
1154         xp = XML_ParserCreateNS(ptr, ':');
1155         if (!xp) {
1156                 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
1157                 goto shutdown;
1158         }
1159         FlushStrBuf(rssc.Key);
1160 //#ifdef HAVE_ICONV
1161 #if 0
1162         XML_SetUnknownEncodingHandler(xp,
1163                                       handle_unknown_xml_encoding,
1164                                       &rssc);
1165 #endif
1166 //#endif
1167         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
1168         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
1169         XML_SetUserData(xp, &rssc);
1170         XML_SetCdataSectionHandler(xp,
1171                                    rss_xml_cdata_start,
1172                                    rss_xml_cdata_end);
1173
1174
1175         len = StrLength(Answer);
1176         ptr = SmashStrBuf(&Answer);
1177         XML_Parse(xp, ptr, len, 0);
1178         free (ptr);
1179         if (ri.done_parsing == 0)
1180                 XML_Parse(xp, "", 0, 1);
1181
1182
1183         CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n",
1184                       XML_ErrorString(
1185                               XML_GetErrorCode(xp)));
1186
1187 shutdown:
1188         curl_easy_cleanup(curl);
1189         XML_ParserFree(xp);
1190
1191         flush_rss_item(&ri);
1192         FreeStrBuf(&rssc.CData);
1193         FreeStrBuf(&rssc.Key);
1194 }
1195
1196
1197 /*
1198  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
1199  */
1200 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
1201 {
1202         char filename[PATH_MAX];
1203         char buf[1024];
1204         char instr[32];
1205         FILE *fp;
1206         char feedurl[256];
1207         rssnetcfg *rncptr = NULL;
1208         rssnetcfg *use_this_rncptr = NULL;
1209         int len = 0;
1210         char *ptr = NULL;
1211
1212         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
1213
1214         if (CtdlThreadCheckStop())
1215                 return;
1216
1217         /* Only do net processing for rooms that have netconfigs */
1218         fp = fopen(filename, "r");
1219         if (fp == NULL) {
1220                 return;
1221         }
1222
1223         while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
1224                 buf[strlen(buf)-1] = 0;
1225
1226                 extract_token(instr, buf, 0, '|', sizeof instr);
1227                 if (!strcasecmp(instr, "rssclient")) {
1228
1229                         use_this_rncptr = NULL;
1230
1231                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
1232
1233                         /* If any other rooms have requested the same feed, then we will just add this
1234                          * room to the target list for that client request.
1235                          */
1236                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
1237                                 if (!strcmp(rncptr->url, feedurl)) {
1238                                         use_this_rncptr = rncptr;
1239                                 }
1240                         }
1241
1242                         /* Otherwise create a new client request */
1243                         if (use_this_rncptr == NULL) {
1244                                 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
1245                                 rncptr->ItemType = RSS_UNSET;
1246                                 if (rncptr != NULL) {
1247                                         rncptr->next = rnclist;
1248                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
1249                                         rncptr->rooms = NULL;
1250                                         rnclist = rncptr;
1251                                         use_this_rncptr = rncptr;
1252                                 }
1253                         }
1254
1255                         /* Add the room name to the request */
1256                         if (use_this_rncptr != NULL) {
1257                                 if (use_this_rncptr->rooms == NULL) {
1258                                         rncptr->rooms = strdup(qrbuf->QRname);
1259                                 }
1260                                 else {
1261                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
1262                                         ptr = realloc(use_this_rncptr->rooms, len);
1263                                         if (ptr != NULL) {
1264                                                 strcat(ptr, "|");
1265                                                 strcat(ptr, qrbuf->QRname);
1266                                                 use_this_rncptr->rooms = ptr;
1267                                         }
1268                                 }
1269                         }
1270                 }
1271
1272         }
1273
1274         fclose(fp);
1275
1276 }
1277
1278 /*
1279  * Scan for rooms that have RSS client requests configured
1280  */
1281 void *rssclient_scan(void *args) {
1282         static time_t last_run = 0L;
1283         static int doing_rssclient = 0;
1284         rssnetcfg *rptr = NULL;
1285         CitContext rssclientCC;
1286
1287         /* Give this thread its own private CitContext */
1288         CtdlFillSystemContext(&rssclientCC, "rssclient");
1289         citthread_setspecific(MyConKey, (void *)&rssclientCC );
1290
1291         /*
1292          * This is a simple concurrency check to make sure only one rssclient run
1293          * is done at a time.  We could do this with a mutex, but since we
1294          * don't really require extremely fine granularity here, we'll do it
1295          * with a static variable instead.
1296          */
1297         if (doing_rssclient) return NULL;
1298         doing_rssclient = 1;
1299
1300         CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
1301         CtdlForEachRoom(rssclient_scan_room, NULL);
1302
1303         while (rnclist != NULL && !CtdlThreadCheckStop()) {
1304                 rss_do_fetching(rnclist);
1305                 rptr = rnclist;
1306                 rnclist = rnclist->next;
1307                 if (rptr->rooms != NULL) free(rptr->rooms);
1308                 free(rptr);
1309         }
1310
1311         CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
1312         last_run = time(NULL);
1313         doing_rssclient = 0;
1314         if (!CtdlThreadCheckStop())
1315                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
1316         else
1317                 CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
1318         CtdlClearSystemContext();
1319         return NULL;
1320 }
1321
1322
1323 CTDL_MODULE_INIT(rssclient)
1324 {
1325         if (threading)
1326         {
1327                 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
1328                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
1329         }
1330
1331         StartHandlers = NewHash(1, NULL);
1332         EndHandlers = NewHash(1, NULL);
1333
1334         AddRSSStartHandler(RSS_item_rss_start,     RSS_UNSET, HKEY("rss"));
1335         AddRSSStartHandler(RSS_item_rdf_start,     RSS_UNSET, HKEY("rdf"));
1336         AddRSSStartHandler(ATOM_item_feed_start,    RSS_UNSET, HKEY("feed"));
1337         AddRSSStartHandler(RSS_item_item_start,    RSS_RSS, HKEY("item"));
1338         AddRSSStartHandler(ATOM_item_entry_start,  RSS_ATOM, HKEY("entry"));
1339         AddRSSStartHandler(ATOM_item_link_start,   RSS_ATOM, HKEY("link"));
1340
1341         AddRSSEndHandler(ATOMRSS_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1342         AddRSSEndHandler(RSS_item_guid_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
1343         AddRSSEndHandler(ATOM_item_id_end,         RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
1344         AddRSSEndHandler(RSS_item_link_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
1345         AddRSSEndHandler(RSSATOM_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1346         AddRSSEndHandler(ATOM_item_content_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
1347         AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_ATOM|RSS_REQUIRE_BUF, HKEY("encoded"));
1348         AddRSSEndHandler(ATOM_item_summary_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
1349         AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
1350         AddRSSEndHandler(ATOM_item_published_end,  RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
1351         AddRSSEndHandler(ATOM_item_updated_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
1352         AddRSSEndHandler(RSS_item_pubdate_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
1353         AddRSSEndHandler(RSS_item_date_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
1354         AddRSSEndHandler(RSS_item_author_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
1355         AddRSSEndHandler(RSS_item_creator_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
1356 /* <author> */
1357         AddRSSEndHandler(ATOM_item_email_end,      RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
1358         AddRSSEndHandler(ATOM_item_name_end,       RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
1359         AddRSSEndHandler(ATOM_item_uri_end,        RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
1360 /* </author> */
1361         AddRSSEndHandler(RSS_item_item_end,        RSS_RSS, HKEY("item"));
1362         AddRSSEndHandler(RSS_item_rss_end,         RSS_RSS, HKEY("rss"));
1363         AddRSSEndHandler(RSS_item_rdf_end,         RSS_RSS, HKEY("rdf"));
1364         AddRSSEndHandler(ATOM_item_entry_end,      RSS_ATOM, HKEY("entry"));
1365
1366
1367 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
1368         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1369         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1370         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1371         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1372
1373 /* links to other feed generators... */
1374         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1375         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1376         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1377         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1378
1379         KnownNameSpaces = NewHash(1, NULL);
1380         Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
1381         Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
1382         Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
1383         Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
1384         Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
1385         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1386         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
1387         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
1388         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
1389         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1390         Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
1391         Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
1392         Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
1393         Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
1394         Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
1395         Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
1396         Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
1397         Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
1398         Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1399         Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1400         Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
1401         Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
1402         Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
1403 #if 0
1404         /* we don't like these namespaces because of they shadow our usefull parameters. */
1405         Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
1406 #endif
1407         return "rssclient";
1408 }