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 #ifdef DEBUG_RSS/////TODO ifndef
 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
 413                         UserName = NewStrBuf();
 414                         EmailAddress = NewStrBuf();
 415                         EncBuf = NewStrBuf();
 416 ////TODO!
 417                         From = html_to_ascii(ChrPtr(ri->author_or_creator),
 418                                              StrLength(ri->author_or_creator),
 419                                              512, 0);
 420                         FromAt = strchr(From, '@') != NULL;
 421                         if (!FromAt && StrLength (ri->author_email) > 0)
 422                         {
 423                                 Encoded = NewStrBuf();
 424                                 if (!IsEmptyStr(From))
 425                                 {
 426                                         StrBufPrintf(Encoded,
 427                                                      "\"%s\" <%s>",
 428                                                      From,
 429                                                      ChrPtr(ri->author_email));
 430                                 }
 431                                 else
 432                                 {
 433                                         StrBufPrintf(Encoded,
 434                                                      "<%s>",
 435                                                      ChrPtr(ri->author_email));
 436                                 }
 437                         }
 438                         else
 439                         {
 440                                 if (FromAt)
 441                                         Encoded = NewStrBufPlain(From, -1);
 442                                 else
 443                                 {
 444                                         Encoded = NewStrBuf();
 445                                         StrBufPrintf(Encoded,
 446                                                      "\"%s\" <%s>",
 447                                                      From,
 448                                                      "rss@localhost"); /// TODO: get hostname?
 449                                 }
 450                         }
 451                         free(From);
 452                         StrBufTrim(Encoded);
 453                         QPEncoded = StrBufSanitizeEmailRecipientVector(Encoded, UserName, EmailAddress, EncBuf);
 454                         msg->cm_fields['A'] = SmashStrBuf(&QPEncoded);
 455
 456                         FreeStrBuf(&Encoded);
 457                         FreeStrBuf(&UserName);
 458                         FreeStrBuf(&EmailAddress);
 459                         FreeStrBuf(&EncBuf);
 460
 461                 }
 462                 else {
 463                         msg->cm_fields['A'] = strdup("rss");
 464                 }
 465
 466                 msg->cm_fields['N'] = strdup(NODENAME);
 467                 if (ri->title != NULL) {
 468                         long len;
 469                         char *Sbj;
 470                         StrBuf *Encoded, *QPEncoded;
 471
 472                         QPEncoded = NULL;
 473                         StrBufSpaceToBlank(ri->title);
 474                         len = StrLength(ri->title);
 475                         Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
 476                         Encoded = NewStrBufPlain(Sbj, -1);
 477                         free(Sbj);
 478
 479                         StrBufTrim(Encoded);
 480                         StrBufRFC2047encode(&QPEncoded, Encoded);
 481
 482                         msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
 483                         FreeStrBuf(&Encoded);
 484                 }
 485                 msg->cm_fields['T'] = malloc(64);
 486                 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
 487                 if (ri->channel_title != NULL) {
 488                         if (StrLength(ri->channel_title) > 0) {
 489                                 msg->cm_fields['O'] = strdup(ChrPtr(ri->channel_title));
 490                         }
 491                 }
 492                 if (ri->link == NULL)
 493                         ri->link = NewStrBufPlain(HKEY(""));
 494                 msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
 495
 496                 Message = NewStrBufPlain(NULL, StrLength(ri->description));
 497
 498                 StrBufPlain(Message, HKEY(
 499                          "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 500                          "<html><body>\n"));
 501
 502                 StrBufAppendBuf(Message, ri->description, 0);
 503                 StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
 504
 505                 AppendLink(Message, ri->link, ri->linkTitle, NULL);
 506                 AppendLink(Message, ri->reLink, ri->reLinkTitle, "Reply to this");
 507                 StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
 508
 509                 msg->cm_fields['M'] = SmashStrBuf(&Message);
 510
 511                 CtdlSubmitMsg(msg, recp, NULL, 0);
 512                 CtdlFreeMessage(msg);
 513
 514                 /* write the uidl to the use table so we don't store this item again */
 515                 strcpy(ut.ut_msgid, utmsgid);
 516                 ut.ut_timestamp = time(NULL);
 517                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 518         }
 519         free_recipients(recp);
 520 }
 521
 522
 523
 524 /*
 525  * Convert an RDF/RSS datestamp into a time_t
 526  */
 527 time_t rdf_parsedate(const char *p)
 528 {
 529         struct tm tm;
 530         time_t t = 0;
 531
 532         if (!p) return 0L;
 533         if (strlen(p) < 10) return 0L;
 534
 535         memset(&tm, 0, sizeof tm);
 536
 537         /*
 538          * If the timestamp appears to be in W3C datetime format, try to
 539          * parse it.  See also: http://www.w3.org/TR/NOTE-datetime
 540          *
 541          * This code, along with parsedate.c, is a potential candidate for
 542          * moving into libcitadel.
 543          */
 544         if ( (p[4] == '-') && (p[7] == '-') ) {
 545                 tm.tm_year = atoi(&p[0]) - 1900;
 546                 tm.tm_mon = atoi(&p[5]) - 1;
 547                 tm.tm_mday = atoi(&p[8]);
 548                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 549                         tm.tm_hour = atoi(&p[11]);
 550                         tm.tm_min = atoi(&p[14]);
 551                 }
 552                 return mktime(&tm);
 553         }
 554
 555         /* hmm... try RFC822 date stamp format */
 556
 557         t = parsedate(p);
 558         if (t > 0) return(t);
 559
 560         /* yeesh.  ok, just return the current date and time. */
 561         return(time(NULL));
 562 }
 563
 564 void flush_rss_item(rss_item *ri)
 565 {
 566         /* Initialize the feed item data structure */
 567         FreeStrBuf(&ri->guid);
 568         FreeStrBuf(&ri->title);
 569         FreeStrBuf(&ri->link);
 570         FreeStrBuf(&ri->author_or_creator);
 571         FreeStrBuf(&ri->author_email);
 572         FreeStrBuf(&ri->author_url);
 573         FreeStrBuf(&ri->description);
 574 }
 575
 576 void rss_xml_start(void *data, const char *supplied_el, const char **attr)
 577 {
 578         rss_xml_handler *h;
 579         rsscollection   *rssc = (rsscollection*) data;
 580         rssnetcfg       *Cfg = rssc->Cfg;
 581         rss_item        *ri = rssc->Item;
 582         void            *pv;
 583         const char      *pel;
 584         char            *sep = NULL;
 585
 586         /* Axe the namespace, we don't care about it */
 587 ///     CtdlLogPrintf(0, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
 588         pel = supplied_el;
 589         while (sep = strchr(pel, ':'), sep) {
 590                 pel = sep + 1;
 591         }
 592
 593         if (pel != supplied_el)
 594         {
 595                 void *v;
 596
 597                 if (!GetHash(KnownNameSpaces,
 598                              supplied_el,
 599                              pel - supplied_el - 1,
 600                              &v))
 601                 {
 602 #ifdef DEBUG_RSS
 603                         CtdlLogPrintf(0, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
 604                                       supplied_el);
 605 #endif
 606                         return;
 607                 }
 608         }
 609
 610         StrBufPlain(rssc->Key, pel, -1);
 611         StrBufLowerCase(rssc->Key);
 612         if (GetHash(StartHandlers, SKEY(rssc->Key), &pv))
 613         {
 614                 rssc->Current = h = (rss_xml_handler*) pv;
 615
 616                 if (((h->Flags & RSS_UNSET) != 0) &&
 617                     (Cfg->ItemType == RSS_UNSET))
 618                 {
 619                         h->Handler(rssc->CData, ri, Cfg, attr);
 620                 }
 621                 else if (((h->Flags & RSS_RSS) != 0) &&
 622                     (Cfg->ItemType == RSS_RSS))
 623                 {
 624                         h->Handler(rssc->CData, ri, Cfg, attr);
 625                 }
 626                 else if (((h->Flags & RSS_ATOM) != 0) &&
 627                          (Cfg->ItemType == RSS_ATOM))
 628                 {
 629                         h->Handler(rssc->CData, ri, Cfg, attr);
 630                 }
 631 #ifdef DEBUG_RSS
 632                 else
 633                         CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
 634 #endif
 635         }
 636 #ifdef DEBUG_RSS
 637         else
 638                 CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel,  supplied_el);
 639 #endif
 640 }
 641
 642 void rss_xml_end(void *data, const char *supplied_el)
 643 {
 644         rss_xml_handler *h;
 645         rsscollection   *rssc = (rsscollection*) data;
 646         rssnetcfg       *Cfg = rssc->Cfg;
 647         rss_item        *ri = rssc->Item;
 648         const char      *pel;
 649         char            *sep = NULL;
 650         void            *pv;
 651
 652         /* Axe the namespace, we don't care about it */
 653         pel = supplied_el;
 654         while (sep = strchr(pel, ':'), sep) {
 655                 pel = sep + 1;
 656         }
 657 //      CtdlLogPrintf(0, "RSS: END %s...\n", el);
 658         if (pel != supplied_el)
 659         {
 660                 void *v;
 661
 662                 if (!GetHash(KnownNameSpaces,
 663                              supplied_el,
 664                              pel - supplied_el - 1,
 665                              &v))
 666                 {
 667 #ifdef DEBUG_RSS
 668                         CtdlLogPrintf(0, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
 669                                       supplied_el, ChrPtr(rssc->CData));
 670 #endif
 671                         FlushStrBuf(rssc->CData);
 672                         return;
 673                 }
 674         }
 675
 676         StrBufPlain(rssc->Key, pel, -1);
 677         StrBufLowerCase(rssc->Key);
 678         if (GetHash(EndHandlers, SKEY(rssc->Key), &pv))
 679         {
 680                 h = (rss_xml_handler*) pv;
 681
 682                 if (((h->Flags & RSS_UNSET) != 0) &&
 683                     (Cfg->ItemType == RSS_UNSET))
 684                 {
 685                         h->Handler(rssc->CData, ri, Cfg, NULL);
 686                 }
 687                 else if (((h->Flags & RSS_RSS) != 0) &&
 688                     (Cfg->ItemType == RSS_RSS))
 689                 {
 690                         h->Handler(rssc->CData, ri, Cfg, NULL);
 691                 }
 692                 else if (((h->Flags & RSS_ATOM) != 0) &&
 693                          (Cfg->ItemType == RSS_ATOM))
 694                 {
 695                         h->Handler(rssc->CData, ri, Cfg, NULL);
 696                 }
 697 #ifdef DEBUG_RSS
 698                 else
 699                         CtdlLogPrintf(0, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 700 #endif
 701         }
 702 #ifdef DEBUG_RSS
 703         else
 704                 CtdlLogPrintf(0, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 705 #endif
 706         FlushStrBuf(rssc->CData);
 707         rssc->Current = NULL;
 708 }
 709
 710
 711
 712
 713
 714 void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 715 {
 716         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RSS feed.\n");
 717         Cfg->ItemType = RSS_RSS;
 718 }
 719
 720 void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 721 {
 722         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RDF feed.\n");
 723         Cfg->ItemType = RSS_RSS;
 724 }
 725
 726 void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 727 {
 728         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an ATOM feed.\n");
 729         Cfg->ItemType = RSS_ATOM;
 730 }
 731
 732
 733 void RSS_item_item_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 734 {
 735         ri->item_tag_nesting ++;
 736         flush_rss_item(ri);
 737 }
 738
 739 void ATOM_item_entry_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 740 {
 741 /* Atom feed... */
 742         ri->item_tag_nesting ++;
 743         flush_rss_item(ri);
 744 }
 745
 746 void ATOM_item_link_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 747 {
 748         int i;
 749         const char *pHref = NULL;
 750         const char *pType = NULL;
 751         const char *pRel = NULL;
 752         const char *pTitle = NULL;
 753
 754         for (i = 0; Attr[i] != NULL; i+=2)
 755         {
 756                 if (!strcmp(Attr[i], "href"))
 757                 {
 758                         pHref = Attr[i+1];
 759                 }
 760                 else if (!strcmp(Attr[i], "rel"))
 761                 {
 762                         pRel = Attr[i+1];
 763                 }
 764                 else if (!strcmp(Attr[i], "type"))
 765                 {
 766                         pType = Attr[i+1];
 767                 }
 768                 else if (!strcmp(Attr[i], "title"))
 769                 {
 770                         pTitle = Attr[i+1];
 771                 }
 772         }
 773         if (pHref == NULL)
 774                 return; /* WHUT? Pointing... where? */
 775         if ((pType != NULL) && !strcasecmp(pType, "application/atom+xml"))
 776                 return; /* these just point to other rss resources, we're not interested in them. */
 777         if (pRel != NULL)
 778         {
 779                 if (!strcasecmp (pRel, "replies"))
 780                 {
 781                         NewStrBufDupAppendFlush(&ri->reLink, NULL, pHref, -1);
 782                         StrBufTrim(ri->link);
 783                         NewStrBufDupAppendFlush(&ri->reLinkTitle, NULL, pTitle, -1);
 784                 }
 785                 else if (!strcasecmp(pRel, "alternate")) /* Alternative representation of this Item... */
 786                 {
 787                         NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 788                         StrBufTrim(ri->link);
 789                         NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
 790
 791                 }
 792 #if 0 /* these are also defined, but dunno what to do with them.. */
 793                 else if (!strcasecmp(pRel, "related"))
 794                 {
 795                 }
 796                 else if (!strcasecmp(pRel, "self"))
 797                 {
 798                 }
 799                 else if (!strcasecmp(pRel, "enclosure"))
 800                 {/* this reference can get big, and is probably the full article... */
 801                 }
 802                 else if (!strcasecmp(pRel, "via"))
 803                 {/* this article was provided via... */
 804                 }
 805 #endif
 806         }
 807         else if (StrLength(ri->link) == 0)
 808         {
 809                 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 810                 StrBufTrim(ri->link);
 811                 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
 812         }
 813 }
 814
 815
 816
 817
 818 void ATOMRSS_item_title_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 819 {
 820         if ((ri->item_tag_nesting == 0) && (StrLength(CData) > 0)) {
 821                 NewStrBufDupAppendFlush(&ri->channel_title, CData, NULL, 0);
 822                 StrBufTrim(ri->channel_title);
 823         }
 824 }
 825
 826 void RSS_item_guid_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 827 {
 828         if (StrLength(CData) > 0) {
 829                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
 830         }
 831 }
 832
 833 void ATOM_item_id_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 834 {
 835         if (StrLength(CData) > 0) {
 836                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
 837         }
 838 }
 839
 840
 841 void RSS_item_link_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 842 {
 843         if (StrLength(CData) > 0) {
 844                 NewStrBufDupAppendFlush(&ri->link, CData, NULL, 0);
 845                 StrBufTrim(ri->link);
 846         }
 847 }
 848
 849 void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 850 {
 851         if (StrLength(CData) > 0) {
 852                 NewStrBufDupAppendFlush(&ri->title, CData, NULL, 0);
 853                 StrBufTrim(ri->title);
 854         }
 855 }
 856
 857 void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 858 {
 859         if (StrLength(CData) > 0) {
 860                 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 861                 StrBufTrim(ri->description);
 862         }
 863 }
 864 void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 865 {
 866         /* this can contain an abstract of the article. but we don't want to verwrite a full document if we already have it. */
 867         if ((StrLength(CData) > 0) && (StrLength(ri->description) == 0))
 868         {
 869                 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 870                 StrBufTrim(ri->description);
 871         }
 872 }
 873
 874 void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 875 {
 876         if (StrLength(CData) > 0) {
 877                 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 878         }
 879 }
 880
 881 void ATOM_item_published_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 882 {
 883         if (StrLength(CData) > 0) {
 884                 StrBufTrim(CData);
 885                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 886         }
 887 }
 888
 889 void ATOM_item_updated_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 890 {
 891         if (StrLength(CData) > 0) {
 892                 StrBufTrim(CData);
 893                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 894         }
 895 }
 896
 897 void RSS_item_pubdate_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 898 {
 899         if (StrLength(CData) > 0) {
 900                 StrBufTrim(CData);
 901                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 902         }
 903 }
 904
 905
 906 void RSS_item_date_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 907 {
 908         if (StrLength(CData) > 0) {
 909                 StrBufTrim(CData);
 910                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 911         }
 912 }
 913
 914
 915
 916 void RSS_item_author_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 917 {
 918         if (StrLength(CData) > 0) {
 919                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
 920                 StrBufTrim(ri->author_or_creator);
 921         }
 922 }
 923
 924
 925 void ATOM_item_name_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 926 {
 927         if (StrLength(CData) > 0) {
 928                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
 929                 StrBufTrim(ri->author_or_creator);
 930         }
 931 }
 932
 933 void ATOM_item_email_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 934 {
 935         if (StrLength(CData) > 0) {
 936                 NewStrBufDupAppendFlush(&ri->author_email, CData, NULL, 0);
 937                 StrBufTrim(ri->author_email);
 938         }
 939 }
 940
 941 void RSS_item_creator_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 942 {
 943         if ((StrLength(CData) > 0) &&
 944             (StrLength(ri->author_or_creator) == 0))
 945         {
 946                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
 947                 StrBufTrim(ri->author_or_creator);
 948         }
 949 }
 950
 951
 952 void ATOM_item_uri_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 953 {
 954         if (StrLength(CData) > 0) {
 955                 NewStrBufDupAppendFlush(&ri->author_url, CData, NULL, 0);
 956                 StrBufTrim(ri->author_url);
 957         }
 958 }
 959
 960 void RSS_item_item_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 961 {
 962         --ri->item_tag_nesting;
 963         rss_save_item(ri);
 964 }
 965
 966
 967 void ATOM_item_entry_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 968 {
 969         --ri->item_tag_nesting;
 970         rss_save_item(ri);
 971 }
 972
 973 void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 974 {
 975 //              CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
 976         ri->done_parsing = 1;
 977
 978 }
 979 void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 980 {
 981 //              CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
 982         ri->done_parsing = 1;
 983 }
 984
 985
 986 void RSSATOM_item_ignore(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 987 {
 988 }
 989
 990
 991
 992 /*
 993  * This callback stores up the data which appears in between tags.
 994  */
 995 void rss_xml_cdata_start(void *data)
 996 {
 997         rsscollection *rssc = (rsscollection*) data;
 998
 999         FlushStrBuf(rssc->CData);
1000 }
1001
1002 void rss_xml_cdata_end(void *data)
1003 {
1004 }
1005 void rss_xml_chardata(void *data, const XML_Char *s, int len)
1006 {
1007         rsscollection *rssc = (rsscollection*) data;
1008
1009         StrBufAppendBufPlain (rssc->CData, s, len, 0);
1010 }
1011
1012 /*
1013  * Callback function for passing libcurl's output to expat for parsing
1014  */
1015 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
1016 {
1017         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
1018         return (size*nmemb);
1019 }
1020
1021
1022
1023 /*
1024  * Begin a feed parse
1025  */
1026 void rss_do_fetching(rssnetcfg *Cfg) {
1027         rsscollection rssc;
1028         rss_item ri;
1029         XML_Parser xp;
1030         StrBuf *Answer;
1031
1032         CURL *curl;
1033         CURLcode res;
1034         char errmsg[1024] = "";
1035         char *ptr;
1036         const char *at;
1037         long len;
1038
1039         memset(&ri, 0, sizeof(rss_item));
1040         rssc.Item = &ri;
1041         rssc.Cfg = Cfg;
1042
1043         CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
1044
1045         curl = curl_easy_init();
1046         if (!curl) {
1047                 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
1048                 return;
1049         }
1050         Answer = NewStrBufPlain(NULL, SIZ);
1051
1052         curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
1053         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
1054         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
1055         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
1056 //      curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
1057         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
1058         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
1059         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
1060 #ifdef CURLOPT_HTTP_CONTENT_DECODING
1061         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
1062         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
1063 #endif
1064         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
1065         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
1066         if (
1067                 (!IsEmptyStr(config.c_ip_addr))
1068                 && (strcmp(config.c_ip_addr, "*"))
1069                 && (strcmp(config.c_ip_addr, "::"))
1070                 && (strcmp(config.c_ip_addr, "0.0.0.0"))
1071         ) {
1072                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
1073         }
1074
1075         if (CtdlThreadCheckStop())
1076         {
1077                 curl_easy_cleanup(curl);
1078                 return;
1079         }
1080
1081         if (CtdlThreadCheckStop())
1082                 goto shutdown ;
1083
1084         res = curl_easy_perform(curl);
1085         if (res) {
1086                 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
1087         }
1088
1089         if (CtdlThreadCheckStop())
1090                 goto shutdown ;
1091
1092
1093
1094
1095         memset(&ri, 0, sizeof(rss_item));
1096         ri.roomlist = Cfg->rooms;
1097         rssc.CData = NewStrBufPlain(NULL, SIZ);
1098         rssc.Key = NewStrBuf();
1099         at = NULL;
1100         StrBufSipLine(rssc.Key, Answer, &at);
1101         ptr = NULL;
1102
1103 #define encoding "encoding=\""
1104         ptr = strstr(ChrPtr(rssc.Key), encoding);
1105         if (ptr != NULL)
1106         {
1107                 char *pche;
1108
1109                 ptr += sizeof (encoding) - 1;
1110                 pche = strchr(ptr, '"');
1111                 if (pche != NULL)
1112                         StrBufCutAt(rssc.Key, -1, pche);
1113                 else
1114                         ptr = "UTF-8";
1115         }
1116         else
1117                 ptr = "UTF-8";
1118
1119
1120         xp = XML_ParserCreateNS(ptr, ':');
1121         if (!xp) {
1122                 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
1123                 goto shutdown;
1124         }
1125         FlushStrBuf(rssc.Key);
1126 //#ifdef HAVE_ICONV
1127 #if 0
1128         XML_SetUnknownEncodingHandler(xp,
1129                                       handle_unknown_xml_encoding,
1130                                       &rssc);
1131 #endif
1132 //#endif
1133         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
1134         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
1135         XML_SetUserData(xp, &rssc);
1136         XML_SetCdataSectionHandler(xp,
1137                                    rss_xml_cdata_start,
1138                                    rss_xml_cdata_end);
1139
1140
1141         len = StrLength(Answer);
1142         ptr = SmashStrBuf(&Answer);
1143         XML_Parse(xp, ptr, len, 0);
1144         free (ptr);
1145         if (ri.done_parsing == 0)
1146                 XML_Parse(xp, "", 0, 1);
1147
1148
1149         CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n",
1150                       XML_ErrorString(
1151                               XML_GetErrorCode(xp)));
1152
1153 shutdown:
1154         curl_easy_cleanup(curl);
1155         XML_ParserFree(xp);
1156
1157         flush_rss_item(&ri);
1158         FreeStrBuf(&rssc.CData);
1159         FreeStrBuf(&rssc.Key);
1160 }
1161
1162
1163 /*
1164  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
1165  */
1166 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
1167 {
1168         char filename[PATH_MAX];
1169         char buf[1024];
1170         char instr[32];
1171         FILE *fp;
1172         char feedurl[256];
1173         rssnetcfg *rncptr = NULL;
1174         rssnetcfg *use_this_rncptr = NULL;
1175         int len = 0;
1176         char *ptr = NULL;
1177
1178         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
1179
1180         if (CtdlThreadCheckStop())
1181                 return;
1182
1183         /* Only do net processing for rooms that have netconfigs */
1184         fp = fopen(filename, "r");
1185         if (fp == NULL) {
1186                 return;
1187         }
1188
1189         while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
1190                 buf[strlen(buf)-1] = 0;
1191
1192                 extract_token(instr, buf, 0, '|', sizeof instr);
1193                 if (!strcasecmp(instr, "rssclient")) {
1194
1195                         use_this_rncptr = NULL;
1196
1197                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
1198
1199                         /* If any other rooms have requested the same feed, then we will just add this
1200                          * room to the target list for that client request.
1201                          */
1202                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
1203                                 if (!strcmp(rncptr->url, feedurl)) {
1204                                         use_this_rncptr = rncptr;
1205                                 }
1206                         }
1207
1208                         /* Otherwise create a new client request */
1209                         if (use_this_rncptr == NULL) {
1210                                 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
1211                                 rncptr->ItemType = RSS_UNSET;
1212                                 if (rncptr != NULL) {
1213                                         rncptr->next = rnclist;
1214                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
1215                                         rncptr->rooms = NULL;
1216                                         rnclist = rncptr;
1217                                         use_this_rncptr = rncptr;
1218                                 }
1219                         }
1220
1221                         /* Add the room name to the request */
1222                         if (use_this_rncptr != NULL) {
1223                                 if (use_this_rncptr->rooms == NULL) {
1224                                         rncptr->rooms = strdup(qrbuf->QRname);
1225                                 }
1226                                 else {
1227                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
1228                                         ptr = realloc(use_this_rncptr->rooms, len);
1229                                         if (ptr != NULL) {
1230                                                 strcat(ptr, "|");
1231                                                 strcat(ptr, qrbuf->QRname);
1232                                                 use_this_rncptr->rooms = ptr;
1233                                         }
1234                                 }
1235                         }
1236                 }
1237
1238         }
1239
1240         fclose(fp);
1241
1242 }
1243
1244 /*
1245  * Scan for rooms that have RSS client requests configured
1246  */
1247 void *rssclient_scan(void *args) {
1248         static time_t last_run = 0L;
1249         static int doing_rssclient = 0;
1250         rssnetcfg *rptr = NULL;
1251         CitContext rssclientCC;
1252
1253         /* Give this thread its own private CitContext */
1254         CtdlFillSystemContext(&rssclientCC, "rssclient");
1255         citthread_setspecific(MyConKey, (void *)&rssclientCC );
1256
1257         /*
1258          * This is a simple concurrency check to make sure only one rssclient run
1259          * is done at a time.  We could do this with a mutex, but since we
1260          * don't really require extremely fine granularity here, we'll do it
1261          * with a static variable instead.
1262          */
1263         if (doing_rssclient) return NULL;
1264         doing_rssclient = 1;
1265
1266         CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
1267         CtdlForEachRoom(rssclient_scan_room, NULL);
1268
1269         while (rnclist != NULL && !CtdlThreadCheckStop()) {
1270                 rss_do_fetching(rnclist);
1271                 rptr = rnclist;
1272                 rnclist = rnclist->next;
1273                 if (rptr->rooms != NULL) free(rptr->rooms);
1274                 free(rptr);
1275         }
1276
1277         CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
1278         last_run = time(NULL);
1279         doing_rssclient = 0;
1280         if (!CtdlThreadCheckStop())
1281                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
1282         else
1283                 CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
1284         CtdlClearSystemContext();
1285         return NULL;
1286 }
1287
1288
1289 CTDL_MODULE_INIT(rssclient)
1290 {
1291         if (threading)
1292         {
1293                 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
1294                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
1295         }
1296
1297         StartHandlers = NewHash(1, NULL);
1298         EndHandlers = NewHash(1, NULL);
1299
1300         AddRSSStartHandler(RSS_item_rss_start,     RSS_UNSET, HKEY("rss"));
1301         AddRSSStartHandler(RSS_item_rdf_start,     RSS_UNSET, HKEY("rdf"));
1302         AddRSSStartHandler(ATOM_item_feed_start,    RSS_UNSET, HKEY("feed"));
1303         AddRSSStartHandler(RSS_item_item_start,    RSS_RSS, HKEY("item"));
1304         AddRSSStartHandler(ATOM_item_entry_start,  RSS_ATOM, HKEY("entry"));
1305         AddRSSStartHandler(ATOM_item_link_start,   RSS_ATOM, HKEY("link"));
1306
1307         AddRSSEndHandler(ATOMRSS_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1308         AddRSSEndHandler(RSS_item_guid_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
1309         AddRSSEndHandler(ATOM_item_id_end,         RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
1310         AddRSSEndHandler(RSS_item_link_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
1311         AddRSSEndHandler(RSSATOM_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1312         AddRSSEndHandler(ATOM_item_content_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
1313         AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("encoded"));
1314         AddRSSEndHandler(ATOM_item_summary_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
1315         AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
1316         AddRSSEndHandler(ATOM_item_published_end,  RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
1317         AddRSSEndHandler(ATOM_item_updated_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
1318         AddRSSEndHandler(RSS_item_pubdate_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
1319         AddRSSEndHandler(RSS_item_date_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
1320         AddRSSEndHandler(RSS_item_author_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
1321         AddRSSEndHandler(RSS_item_creator_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
1322 /* <author> */
1323         AddRSSEndHandler(ATOM_item_email_end,      RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
1324         AddRSSEndHandler(ATOM_item_name_end,       RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
1325         AddRSSEndHandler(ATOM_item_uri_end,        RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
1326 /* </author> */
1327         AddRSSEndHandler(RSS_item_item_end,        RSS_RSS, HKEY("item"));
1328         AddRSSEndHandler(RSS_item_rss_end,         RSS_RSS, HKEY("rss"));
1329         AddRSSEndHandler(RSS_item_rdf_end,         RSS_RSS, HKEY("rdf"));
1330         AddRSSEndHandler(ATOM_item_entry_end,      RSS_ATOM, HKEY("entry"));
1331
1332
1333 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
1334         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1335         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1336         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1337         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1338
1339 /* links to other feed generators... */
1340         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1341         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1342         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1343         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1344
1345         KnownNameSpaces = NewHash(1, NULL);
1346         Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
1347         Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
1348         Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
1349         Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
1350         Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
1351         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1352         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
1353         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
1354         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
1355         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1356         Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
1357         Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
1358         Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
1359         Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
1360         Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
1361         Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
1362         Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
1363         Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
1364         Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1365         Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1366         Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
1367         Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
1368         Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
1369 #if 0
1370         /* we don't like these namespaces because of they shadow our usefull parameters. */
1371         Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
1372 #endif
1373         return "rssclient";
1374 }