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
  60 typedef struct rssnetcfg rssnetcfg;
  61 struct rssnetcfg {
  62         rssnetcfg *next;
  63         char url[256];
  64         char *rooms;
  65         time_t last_error_when;
  66         int ItemType;
  67         time_t next_poll;
  68 };
  69
  70 #define RSS_UNSET       (1<<0)
  71 #define RSS_RSS         (1<<1)
  72 #define RSS_ATOM        (1<<2)
  73 #define RSS_REQUIRE_BUF (1<<3)
  74
  75 typedef struct _rss_item {
  76         char *roomlist;
  77         int done_parsing;
  78         StrBuf *guid;
  79         StrBuf *title;
  80         StrBuf *link;
  81         StrBuf *linkTitle;
  82         StrBuf *reLink;
  83         StrBuf *reLinkTitle;
  84         StrBuf *description;
  85         time_t pubdate;
  86         StrBuf *channel_title;
  87         int item_tag_nesting;
  88         StrBuf *author_or_creator;
  89         StrBuf *author_url;
  90         StrBuf *author_email;
  91 }rss_item;
  92
  93
  94 typedef void (*rss_handler_func)(StrBuf *CData,
  95                                  rss_item *ri,
  96                                  rssnetcfg *Cfg,
  97                                  const char** Attr);
  98
  99 typedef struct __rss_xml_handler {
 100         int Flags;
 101         rss_handler_func Handler;
 102 }rss_xml_handler;
 103
 104
 105 typedef struct _rsscollection {
 106         StrBuf *CData;
 107         StrBuf *Key;
 108
 109         rss_item *Item;
 110         rssnetcfg *Cfg;
 111
 112         rss_xml_handler *Current;
 113 } rsscollection;
 114
 115 struct rssnetcfg *rnclist = NULL;
 116 HashList *StartHandlers = NULL;
 117 HashList *EndHandlers = NULL;
 118 HashList *KnownNameSpaces = NULL;
 119 HashList *UrlShorteners = NULL;
 120 void AddRSSStartHandler(rss_handler_func Handler, int Flags, const char *key, long len)
 121 {
 122         rss_xml_handler *h;
 123         h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
 124         h->Flags = Flags;
 125         h->Handler = Handler;
 126         Put(StartHandlers, key, len, h, NULL);
 127 }
 128 void AddRSSEndHandler(rss_handler_func Handler, int Flags, const char *key, long len)
 129 {
 130         rss_xml_handler *h;
 131         h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
 132         h->Flags = Flags;
 133         h->Handler = Handler;
 134         Put(EndHandlers, key, len, h, NULL);
 135 }
 136
 137 #if 0
 138 //#ifdef HAVE_ICONV
 139 #include <iconv.h>
 140
 141
 142 /*
 143  * dug this out of the trashcan of the midgard project, lets see if it works for us.
 144  * original code by Alexander Bokovoy <bokovoy@avilink.ne> distributed under GPL V2 or later
 145  */
 146
 147 /* Returns:
 148  >= 0 - successfull, 0 means conversion doesn't use multibyte sequences
 149    -1 - error during iconv_open call
 150    -2 - error during iconv_close call
 151    ----------------------------------
 152    This function expects that multibyte encoding in 'charset' wouldn't have
 153    characters with more than 3 bytes. It is not intended to convert UTF-8 because
 154    we'll never receive UTF-8 in our handler (it is handled by Exat itself).
 155 */
 156 static int
 157 fill_encoding_info (const char *charset, XML_Encoding * info)
 158 {
 159   iconv_t cd = (iconv_t)(-1);
 160   int flag;
 161         syslog(LOG_EMERG, "RSS: fill encoding info ...\n");
 162
 163 #if G_BYTE_ORDER == G_LITTLE_ENDIAN
 164   cd = iconv_open ("UCS-2LE", charset);
 165 #else
 166   cd = iconv_open ("UCS-2BE", charset);
 167 #endif
 168
 169   if (cd == (iconv_t) (-1))
 170     {
 171       return -1;
 172     }
 173
 174   {
 175     unsigned short out = 0;
 176     unsigned char buf[4];
 177     unsigned int i0, i1, i2;
 178     int result = 0;
 179     flag = 0;
 180     for (i0 = 0; i0 < 0x100; i0++)
 181       {
 182         buf[0] = i0;
 183         info->map[i0] = 0;
 184         //result = try (cd, buf, 1, &out);
 185         if (result < 0)
 186           {
 187           }
 188         else if (result > 0)
 189           {
 190             info->map[i0] = out;
 191           }
 192         else
 193           {
 194             for (i1 = 0; i1 < 0x100; i1++)
 195               {
 196                 buf[1] = i1;
 197                 ///result = try (cd, buf, 2, &out);
 198                 if (result < 0)
 199                   {
 200                   }
 201                 else if (result > 0)
 202                   {
 203                     flag++;
 204                     info->map[i0] = -2;
 205                   }
 206                 else
 207                   {
 208                     for (i2 = 0; i2 < 0x100; i2++)
 209                       {
 210                         buf[2] = i2;
 211                         ////result = try (cd, buf, 3, &out);
 212                         if (result < 0)
 213                           {
 214                           }
 215                         else if (result > 0)
 216                           {
 217                             flag++;
 218                             info->map[i0] = -3;
 219                           }
 220                       }
 221                   }
 222               }
 223           }
 224       }
 225   }
 226
 227   if (iconv_close (cd) < 0)
 228     {
 229       return -2;
 230     }
 231   return flag;
 232 }
 233
 234 static int
 235 iconv_convertor (void *data, const char *s)
 236 {
 237   XML_Encoding *info = data;
 238   int res;
 239         syslog(LOG_EMERG, "RSS: Converting ...\n");
 240
 241   if (s == NULL)
 242     return -1;
 243 /*
 244   GByteArray *result;
 245   result = g_byte_array_new ();
 246   if (process_block (info->data, (char *) s, strlen (s), result) == 0)
 247     {
 248       res = *(result->data);
 249       g_byte_array_free (result, TRUE);
 250       return res;
 251     }
 252   g_byte_array_free (result, TRUE);
 253 */
 254   return -1;
 255 }
 256
 257 static void
 258 my_release (void *data)
 259 {
 260   iconv_t cd = (iconv_t) data;
 261   if (iconv_close (cd) != 0)
 262     {
 263 /// TODO: uh no.      exit (1);
 264     }
 265 }
 266 int
 267 handle_unknown_xml_encoding (void *encodingHandleData,
 268                              const XML_Char * name,
 269                              XML_Encoding * info)
 270 {
 271   int result;
 272   syslog(LOG_EMERG, "RSS: unknown encoding ...\n");
 273   result = fill_encoding_info (name, info);
 274   if (result >= 0)
 275     {
 276       /*
 277         Special case: client asked for reverse conversion, we'll provide him with
 278         iconv descriptor which handles it. Client should release it by himself.
 279       */
 280       if(encodingHandleData != NULL)
 281             *((iconv_t *)encodingHandleData) = iconv_open(name, "UTF-8");
 282       /*
 283          Optimization: we do not need conversion function if encoding is one-to-one,
 284          info->map table will be enough
 285        */
 286       if (result == 0)
 287         {
 288           info->data = NULL;
 289           info->convert = NULL;
 290           info->release = NULL;
 291           return 1;
 292         }
 293       /*
 294          We do need conversion function because this encoding uses multibyte sequences
 295        */
 296       info->data = (void *) iconv_open ("UTF-8", name);
 297       if ((int)info->data == -1)
 298         return -1;
 299       info->convert = iconv_convertor;
 300       info->release = my_release;
 301       return 1;
 302     }
 303   if(encodingHandleData != NULL)
 304     *(iconv_t *)encodingHandleData = NULL;
 305   return 0;
 306 }
 307
 308 ///#endif
 309 #endif
 310 size_t GetLocationString( void *ptr, size_t size, size_t nmemb, void *userdata)
 311 {
 312 #define LOCATION "location"
 313         if (strncasecmp((char*)ptr, LOCATION, sizeof(LOCATION) - 1) == 0)
 314         {
 315                 StrBuf *pURL = (StrBuf*) userdata;
 316                 char *pch = (char*) ptr;
 317                 char *pche;
 318
 319                 pche = pch + (size * nmemb);
 320                 pch += sizeof(LOCATION);
 321
 322                 while (isspace(*pch) || (*pch == ':'))
 323                         pch ++;
 324
 325                 while (isspace(*pche) || (*pche == '\0'))
 326                         pche--;
 327
 328                 FlushStrBuf(pURL);
 329                 StrBufPlain(pURL, pch, pche - pch + 1);
 330         }
 331         return size * nmemb;
 332 }
 333
 334 int LookupUrl(StrBuf *ShorterUrlStr)
 335 {
 336         CURL *curl;
 337         char errmsg[1024] = "";
 338         StrBuf *Answer;
 339         int rc = 0;
 340
 341         curl = curl_easy_init();
 342         if (!curl) {
 343                 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
 344                 return 0;
 345         }
 346         Answer = NewStrBufPlain(NULL, SIZ);
 347
 348         curl_easy_setopt(curl, CURLOPT_URL, ChrPtr(ShorterUrlStr));
 349         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
 350         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
 351         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
 352 //      curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
 353         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
 354         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
 355         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
 356 #ifdef CURLOPT_HTTP_CONTENT_DECODING
 357         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
 358         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
 359 #endif
 360         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
 361         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
 362         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 0);
 363
 364         curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION , GetLocationString);
 365         curl_easy_setopt(curl, CURLOPT_WRITEHEADER, ShorterUrlStr);
 366
 367
 368         if (
 369                 (!IsEmptyStr(config.c_ip_addr))
 370                 && (strcmp(config.c_ip_addr, "*"))
 371                 && (strcmp(config.c_ip_addr, "::"))
 372                 && (strcmp(config.c_ip_addr, "0.0.0.0"))
 373         ) {
 374                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
 375         }
 376
 377         if (CtdlThreadCheckStop())
 378                 goto shutdown ;
 379
 380         rc = curl_easy_perform(curl);
 381         if (rc) {
 382                 syslog(LOG_ALERT, "libcurl error %d: %s\n", rc, errmsg);
 383                 rc = 0;
 384         }
 385         else
 386                 rc = 1;
 387
 388 shutdown:
 389         curl_easy_cleanup(curl);
 390
 391         return rc;
 392
 393 }
 394
 395
 396
 397 void CrawlMessageForShorterUrls(HashList *pUrls, StrBuf *Message)
 398 {
 399         int nHits = 0;
 400         void *pv;
 401         int nShorter = 0;
 402         const char *pch;
 403         const char *pUrl;
 404         ConstStr *pCUrl;
 405
 406         while (GetHash(UrlShorteners, IKEY(nShorter), &pv))
 407         {
 408                 nShorter++;
 409                 pch = ChrPtr(Message);
 410                 pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 411                 while ((pUrl != NULL) && (nHits < 99))
 412                 {
 413                         pCUrl = malloc(sizeof(ConstStr));
 414
 415                         pCUrl->Key = pUrl;
 416                         pch = pUrl + StrLength((StrBuf*)pv);
 417                         while (isalnum(*pch)||(*pch == '-')||(*pch == '/'))
 418                                 pch++;
 419                         pCUrl->len = pch - pCUrl->Key;
 420
 421                         Put(pUrls, IKEY(nHits), pCUrl, NULL);
 422                         nHits ++;
 423                         pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 424                 }
 425         }
 426 }
 427
 428 int SortConstStrByPosition(const void *Item1, const void *Item2)
 429 {
 430         const ConstStr *p1, *p2;
 431         p1 = (const ConstStr*) Item1;
 432         p2 = (const ConstStr*) Item2;
 433         if (p1->Key == p2->Key)
 434                 return 0;
 435         if (p1->Key > p2->Key)
 436                 return 1;
 437         return -1;
 438 }
 439
 440 void ExpandShortUrls(StrBuf *Message)
 441 {
 442         StrBuf *Shadow;
 443         HashList *pUrls;
 444         ConstStr *pCUrl;
 445         const char *pch;
 446         const char *pche;
 447
 448         /* we just suspect URL shorteners to be inside of feeds from twitter
 449          * or other short content messages, so don't crawl through real blogs.
 450          */
 451         if (StrLength(Message) > 500)
 452                 return;
 453
 454         pUrls = NewHash(1, Flathash);
 455         CrawlMessageForShorterUrls(pUrls, Message);
 456
 457         if (GetCount(pUrls) > 0)
 458         {
 459                 StrBuf *ShorterUrlStr;
 460                 HashPos *Pos;
 461                 const char *Key;
 462                 void *pv;
 463                 long len;
 464
 465                 Shadow = NewStrBufPlain(NULL, StrLength(Message));
 466                 SortByPayload (pUrls, SortConstStrByPosition);
 467
 468                 ShorterUrlStr = NewStrBufPlain(NULL, StrLength(Message));
 469
 470                 pch = ChrPtr(Message);
 471                 pche = pch + StrLength(Message);
 472                 Pos = GetNewHashPos(pUrls, 1);
 473                 while (GetNextHashPos(pUrls, Pos, &len, &Key, &pv))
 474                 {
 475                         pCUrl = (ConstStr*) pv;
 476
 477                         if (pch != pCUrl->Key)
 478                                 StrBufAppendBufPlain(Shadow, pch, pCUrl->Key - pch, 0);
 479
 480                         StrBufPlain(ShorterUrlStr, CKEY(*pCUrl));
 481                         if (LookupUrl(ShorterUrlStr))
 482                         {
 483                                 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 484                                 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 485                                 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 486                                 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 487                                 StrBufAppendBufPlain(Shadow, HKEY("["), 0);
 488                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 489                                 StrBufAppendBufPlain(Shadow, HKEY("]</a>"), 0);
 490                         }
 491                         else
 492                         {
 493                                 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 494                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 495                                 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 496                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 497                                 StrBufAppendBufPlain(Shadow, HKEY("</a>"), 0);
 498                         }
 499                         pch = pCUrl->Key + pCUrl->len + 1;
 500
 501                 }
 502                 if (pch < pche)
 503                         StrBufAppendBufPlain(Shadow, pch, pche - pch, 0);
 504                 FlushStrBuf(Message);
 505                 StrBufAppendBuf(Message, Shadow, 0);
 506
 507                 FreeStrBuf(&ShorterUrlStr);
 508                 FreeStrBuf(&Shadow);
 509                 DeleteHashPos(&Pos);
 510         }
 511
 512         DeleteHash(&pUrls);
 513 }
 514
 515
 516 void AppendLink(StrBuf *Message, StrBuf *link, StrBuf *LinkTitle, const char *Title)
 517 {
 518         if (StrLength(link) > 0)
 519         {
 520                 StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
 521                 StrBufAppendBuf(Message, link, 0);
 522                 StrBufAppendBufPlain(Message, HKEY("\">"), 0);
 523                 if (StrLength(LinkTitle) > 0)
 524                         StrBufAppendBuf(Message, LinkTitle, 0);
 525                 else if ((Title != NULL) && !IsEmptyStr(Title))
 526                         StrBufAppendBufPlain(Message, Title, -1, 0);
 527                 else
 528                         StrBufAppendBuf(Message, link, 0);
 529                 StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
 530         }
 531 }
 532 /*
 533  * Commit a fetched and parsed RSS item to disk
 534  */
 535 void rss_save_item(rss_item *ri)
 536 {
 537
 538         struct MD5Context md5context;
 539         u_char rawdigest[MD5_DIGEST_LEN];
 540         int i;
 541         char utmsgid[SIZ];
 542         struct cdbdata *cdbut;
 543         struct UseTable ut;
 544         struct CtdlMessage *msg;
 545         struct recptypes *recp = NULL;
 546         int msglen = 0;
 547         StrBuf *Message;
 548
 549         recp = (struct recptypes *) malloc(sizeof(struct recptypes));
 550         if (recp == NULL) return;
 551         memset(recp, 0, sizeof(struct recptypes));
 552         memset(&ut, 0, sizeof(struct UseTable));
 553         recp->recp_room = strdup(ri->roomlist);
 554         recp->num_room = num_tokens(ri->roomlist, '|');
 555         recp->recptypes_magic = RECPTYPES_MAGIC;
 556
 557         /* Construct a GUID to use in the S_USETABLE table.
 558          * If one is not present in the item itself, make one up.
 559          */
 560         if (ri->guid != NULL) {
 561                 StrBufSpaceToBlank(ri->guid);
 562                 StrBufTrim(ri->guid);
 563                 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ChrPtr(ri->guid));
 564         }
 565         else {
 566                 MD5Init(&md5context);
 567                 if (ri->title != NULL) {
 568                         MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->title), StrLength(ri->title));
 569                 }
 570                 if (ri->link != NULL) {
 571                         MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->link), StrLength(ri->link));
 572                 }
 573                 MD5Final(rawdigest, &md5context);
 574                 for (i=0; i<MD5_DIGEST_LEN; i++) {
 575                         sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
 576                         utmsgid[i*2] = tolower(utmsgid[i*2]);
 577                         utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
 578                 }
 579                 strcat(utmsgid, "_rss2ctdl");
 580         }
 581
 582         /* Find out if we've already seen this item */
 583
 584         cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
 585 #ifndef DEBUG_RSS
 586         if (cdbut != NULL) {
 587                 /* Item has already been seen */
 588                 syslog(LOG_DEBUG, "%s has already been seen\n", utmsgid);
 589                 cdb_free(cdbut);
 590
 591                 /* rewrite the record anyway, to update the timestamp */
 592                 strcpy(ut.ut_msgid, utmsgid);
 593                 ut.ut_timestamp = time(NULL);
 594                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 595         }
 596         else
 597 #endif
 598 {
 599                 /* Item has not been seen, so save it. */
 600                 syslog(LOG_DEBUG, "RSS: saving item...\n");
 601                 if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY(""));
 602                 StrBufSpaceToBlank(ri->description);
 603                 msg = malloc(sizeof(struct CtdlMessage));
 604                 memset(msg, 0, sizeof(struct CtdlMessage));
 605                 msg->cm_magic = CTDLMESSAGE_MAGIC;
 606                 msg->cm_anon_type = MES_NORMAL;
 607                 msg->cm_format_type = FMT_RFC822;
 608
 609                 if (ri->guid != NULL) {
 610                         msg->cm_fields['E'] = strdup(ChrPtr(ri->guid));
 611                 }
 612
 613                 if (ri->author_or_creator != NULL) {
 614                         char *From;
 615                         StrBuf *Encoded = NULL;
 616                         int FromAt;
 617
 618                         From = html_to_ascii(ChrPtr(ri->author_or_creator),
 619                                              StrLength(ri->author_or_creator),
 620                                              512, 0);
 621                         StrBufPlain(ri->author_or_creator, From, -1);
 622                         StrBufTrim(ri->author_or_creator);
 623                         free(From);
 624
 625                         FromAt = strchr(ChrPtr(ri->author_or_creator), '@') != NULL;
 626                         if (!FromAt && StrLength (ri->author_email) > 0)
 627                         {
 628                                 StrBufRFC2047encode(&Encoded, ri->author_or_creator);
 629                                 msg->cm_fields['A'] = SmashStrBuf(&Encoded);
 630                                 msg->cm_fields['P'] = SmashStrBuf(&ri->author_email);
 631                         }
 632                         else
 633                         {
 634                                 if (FromAt)
 635                                         msg->cm_fields['P'] = SmashStrBuf(&ri->author_or_creator);
 636                                 else
 637                                 {
 638                                         StrBufRFC2047encode(&Encoded, ri->author_or_creator);
 639                                         msg->cm_fields['A'] = SmashStrBuf(&Encoded);
 640                                         msg->cm_fields['P'] = strdup("rss@localhost");
 641                                 }
 642                         }
 643                 }
 644                 else {
 645                         msg->cm_fields['A'] = strdup("rss");
 646                 }
 647
 648                 msg->cm_fields['N'] = strdup(NODENAME);
 649                 if (ri->title != NULL) {
 650                         long len;
 651                         char *Sbj;
 652                         StrBuf *Encoded, *QPEncoded;
 653
 654                         QPEncoded = NULL;
 655                         StrBufSpaceToBlank(ri->title);
 656                         len = StrLength(ri->title);
 657                         Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
 658                         len = strlen(Sbj);
 659                         if (Sbj[len - 1] == '\n')
 660                         {
 661                                 len --;
 662                                 Sbj[len] = '\0';
 663                         }
 664                         Encoded = NewStrBufPlain(Sbj, len);
 665                         free(Sbj);
 666
 667                         StrBufTrim(Encoded);
 668                         StrBufRFC2047encode(&QPEncoded, Encoded);
 669
 670                         msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
 671                         FreeStrBuf(&Encoded);
 672                 }
 673                 msg->cm_fields['T'] = malloc(64);
 674                 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
 675                 if (ri->channel_title != NULL) {
 676                         if (StrLength(ri->channel_title) > 0) {
 677                                 msg->cm_fields['O'] = strdup(ChrPtr(ri->channel_title));
 678                         }
 679                 }
 680                 if (ri->link == NULL)
 681                         ri->link = NewStrBufPlain(HKEY(""));
 682                 ExpandShortUrls(ri->description);
 683                 msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
 684
 685                 Message = NewStrBufPlain(NULL, StrLength(ri->description));
 686
 687                 StrBufPlain(Message, HKEY(
 688                          "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 689                          "<html><body>\n"));
 690
 691                 StrBufAppendBuf(Message, ri->description, 0);
 692                 StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
 693
 694                 AppendLink(Message, ri->link, ri->linkTitle, NULL);
 695                 AppendLink(Message, ri->reLink, ri->reLinkTitle, "Reply to this");
 696                 StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
 697
 698                 msg->cm_fields['M'] = SmashStrBuf(&Message);
 699
 700                 CtdlSubmitMsg(msg, recp, NULL, 0);
 701                 CtdlFreeMessage(msg);
 702
 703                 /* write the uidl to the use table so we don't store this item again */
 704                 strcpy(ut.ut_msgid, utmsgid);
 705                 ut.ut_timestamp = time(NULL);
 706                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 707         }
 708         free_recipients(recp);
 709 }
 710
 711
 712
 713 /*
 714  * Convert an RDF/RSS datestamp into a time_t
 715  */
 716 time_t rdf_parsedate(const char *p)
 717 {
 718         struct tm tm;
 719         time_t t = 0;
 720
 721         if (!p) return 0L;
 722         if (strlen(p) < 10) return 0L;
 723
 724         memset(&tm, 0, sizeof tm);
 725
 726         /*
 727          * If the timestamp appears to be in W3C datetime format, try to
 728          * parse it.  See also: http://www.w3.org/TR/NOTE-datetime
 729          *
 730          * This code, along with parsedate.c, is a potential candidate for
 731          * moving into libcitadel.
 732          */
 733         if ( (p[4] == '-') && (p[7] == '-') ) {
 734                 tm.tm_year = atoi(&p[0]) - 1900;
 735                 tm.tm_mon = atoi(&p[5]) - 1;
 736                 tm.tm_mday = atoi(&p[8]);
 737                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 738                         tm.tm_hour = atoi(&p[11]);
 739                         tm.tm_min = atoi(&p[14]);
 740                 }
 741                 return mktime(&tm);
 742         }
 743
 744         /* hmm... try RFC822 date stamp format */
 745
 746         t = parsedate(p);
 747         if (t > 0) return(t);
 748
 749         /* yeesh.  ok, just return the current date and time. */
 750         return(time(NULL));
 751 }
 752
 753 void flush_rss_item(rss_item *ri)
 754 {
 755         /* Initialize the feed item data structure */
 756         FreeStrBuf(&ri->guid);
 757         FreeStrBuf(&ri->title);
 758         FreeStrBuf(&ri->link);
 759         FreeStrBuf(&ri->linkTitle);
 760         FreeStrBuf(&ri->reLink);
 761         FreeStrBuf(&ri->reLinkTitle);
 762         FreeStrBuf(&ri->description);
 763         FreeStrBuf(&ri->channel_title);
 764         FreeStrBuf(&ri->author_or_creator);
 765         FreeStrBuf(&ri->author_url);
 766         FreeStrBuf(&ri->author_email);
 767 }
 768
 769 void rss_xml_start(void *data, const char *supplied_el, const char **attr)
 770 {
 771         rss_xml_handler *h;
 772         rsscollection   *rssc = (rsscollection*) data;
 773         rssnetcfg       *Cfg = rssc->Cfg;
 774         rss_item        *ri = rssc->Item;
 775         void            *pv;
 776         const char      *pel;
 777         char            *sep = NULL;
 778
 779         /* Axe the namespace, we don't care about it */
 780 ///     syslog(LOG_EMERG, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
 781         pel = supplied_el;
 782         while (sep = strchr(pel, ':'), sep) {
 783                 pel = sep + 1;
 784         }
 785
 786         if (pel != supplied_el)
 787         {
 788                 void *v;
 789
 790                 if (!GetHash(KnownNameSpaces,
 791                              supplied_el,
 792                              pel - supplied_el - 1,
 793                              &v))
 794                 {
 795 #ifdef DEBUG_RSS
 796                         syslog(LOG_EMERG, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
 797                                       supplied_el);
 798 #endif
 799                         return;
 800                 }
 801         }
 802
 803         StrBufPlain(rssc->Key, pel, -1);
 804         StrBufLowerCase(rssc->Key);
 805         if (GetHash(StartHandlers, SKEY(rssc->Key), &pv))
 806         {
 807                 rssc->Current = h = (rss_xml_handler*) pv;
 808
 809                 if (((h->Flags & RSS_UNSET) != 0) &&
 810                     (Cfg->ItemType == RSS_UNSET))
 811                 {
 812                         h->Handler(rssc->CData, ri, Cfg, attr);
 813                 }
 814                 else if (((h->Flags & RSS_RSS) != 0) &&
 815                     (Cfg->ItemType == RSS_RSS))
 816                 {
 817                         h->Handler(rssc->CData, ri, Cfg, attr);
 818                 }
 819                 else if (((h->Flags & RSS_ATOM) != 0) &&
 820                          (Cfg->ItemType == RSS_ATOM))
 821                 {
 822                         h->Handler(rssc->CData, ri, Cfg, attr);
 823                 }
 824 #ifdef DEBUG_RSS
 825                 else
 826                         syslog(LOG_EMERG, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
 827 #endif
 828         }
 829 #ifdef DEBUG_RSS
 830         else
 831                 syslog(LOG_EMERG, "RSS: START unhandled: [%s] [%s]...\n", pel,  supplied_el);
 832 #endif
 833 }
 834
 835 void rss_xml_end(void *data, const char *supplied_el)
 836 {
 837         rss_xml_handler *h;
 838         rsscollection   *rssc = (rsscollection*) data;
 839         rssnetcfg       *Cfg = rssc->Cfg;
 840         rss_item        *ri = rssc->Item;
 841         const char      *pel;
 842         char            *sep = NULL;
 843         void            *pv;
 844
 845         /* Axe the namespace, we don't care about it */
 846         pel = supplied_el;
 847         while (sep = strchr(pel, ':'), sep) {
 848                 pel = sep + 1;
 849         }
 850 //      syslog(LOG_EMERG, "RSS: END %s...\n", el);
 851         if (pel != supplied_el)
 852         {
 853                 void *v;
 854
 855                 if (!GetHash(KnownNameSpaces,
 856                              supplied_el,
 857                              pel - supplied_el - 1,
 858                              &v))
 859                 {
 860 #ifdef DEBUG_RSS
 861                         syslog(LOG_EMERG, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
 862                                       supplied_el, ChrPtr(rssc->CData));
 863 #endif
 864                         FlushStrBuf(rssc->CData);
 865                         return;
 866                 }
 867         }
 868
 869         StrBufPlain(rssc->Key, pel, -1);
 870         StrBufLowerCase(rssc->Key);
 871         if (GetHash(EndHandlers, SKEY(rssc->Key), &pv))
 872         {
 873                 h = (rss_xml_handler*) pv;
 874
 875                 if (((h->Flags & RSS_UNSET) != 0) &&
 876                     (Cfg->ItemType == RSS_UNSET))
 877                 {
 878                         h->Handler(rssc->CData, ri, Cfg, NULL);
 879                 }
 880                 else if (((h->Flags & RSS_RSS) != 0) &&
 881                     (Cfg->ItemType == RSS_RSS))
 882                 {
 883                         h->Handler(rssc->CData, ri, Cfg, NULL);
 884                 }
 885                 else if (((h->Flags & RSS_ATOM) != 0) &&
 886                          (Cfg->ItemType == RSS_ATOM))
 887                 {
 888                         h->Handler(rssc->CData, ri, Cfg, NULL);
 889                 }
 890 #ifdef DEBUG_RSS
 891                 else
 892                         syslog(LOG_EMERG, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 893 #endif
 894         }
 895 #ifdef DEBUG_RSS
 896         else
 897                 syslog(LOG_EMERG, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 898 #endif
 899         FlushStrBuf(rssc->CData);
 900         rssc->Current = NULL;
 901 }
 902
 903
 904
 905
 906
 907 void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 908 {
 909         syslog(LOG_DEBUG, "RSS: This is an RSS feed.\n");
 910         Cfg->ItemType = RSS_RSS;
 911 }
 912
 913 void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 914 {
 915         syslog(LOG_DEBUG, "RSS: This is an RDF feed.\n");
 916         Cfg->ItemType = RSS_RSS;
 917 }
 918
 919 void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 920 {
 921         syslog(LOG_DEBUG, "RSS: This is an ATOM feed.\n");
 922         Cfg->ItemType = RSS_ATOM;
 923 }
 924
 925
 926 void RSS_item_item_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 927 {
 928         ri->item_tag_nesting ++;
 929         flush_rss_item(ri);
 930 }
 931
 932 void ATOM_item_entry_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 933 {
 934 /* Atom feed... */
 935         ri->item_tag_nesting ++;
 936         flush_rss_item(ri);
 937 }
 938
 939 void ATOM_item_link_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 940 {
 941         int i;
 942         const char *pHref = NULL;
 943         const char *pType = NULL;
 944         const char *pRel = NULL;
 945         const char *pTitle = NULL;
 946
 947         for (i = 0; Attr[i] != NULL; i+=2)
 948         {
 949                 if (!strcmp(Attr[i], "href"))
 950                 {
 951                         pHref = Attr[i+1];
 952                 }
 953                 else if (!strcmp(Attr[i], "rel"))
 954                 {
 955                         pRel = Attr[i+1];
 956                 }
 957                 else if (!strcmp(Attr[i], "type"))
 958                 {
 959                         pType = Attr[i+1];
 960                 }
 961                 else if (!strcmp(Attr[i], "title"))
 962                 {
 963                         pTitle = Attr[i+1];
 964                 }
 965         }
 966         if (pHref == NULL)
 967                 return; /* WHUT? Pointing... where? */
 968         if ((pType != NULL) && !strcasecmp(pType, "application/atom+xml"))
 969                 return; /* these just point to other rss resources, we're not interested in them. */
 970         if (pRel != NULL)
 971         {
 972                 if (!strcasecmp (pRel, "replies"))
 973                 {
 974                         NewStrBufDupAppendFlush(&ri->reLink, NULL, pHref, -1);
 975                         StrBufTrim(ri->link);
 976                         NewStrBufDupAppendFlush(&ri->reLinkTitle, NULL, pTitle, -1);
 977                 }
 978                 else if (!strcasecmp(pRel, "alternate")) /* Alternative representation of this Item... */
 979                 {
 980                         NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 981                         StrBufTrim(ri->link);
 982                         NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
 983
 984                 }
 985 #if 0 /* these are also defined, but dunno what to do with them.. */
 986                 else if (!strcasecmp(pRel, "related"))
 987                 {
 988                 }
 989                 else if (!strcasecmp(pRel, "self"))
 990                 {
 991                 }
 992                 else if (!strcasecmp(pRel, "enclosure"))
 993                 {/* this reference can get big, and is probably the full article... */
 994                 }
 995                 else if (!strcasecmp(pRel, "via"))
 996                 {/* this article was provided via... */
 997                 }
 998 #endif
 999         }
1000         else if (StrLength(ri->link) == 0)
1001         {
1002                 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
1003                 StrBufTrim(ri->link);
1004                 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
1005         }
1006 }
1007
1008
1009
1010
1011 void ATOMRSS_item_title_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1012 {
1013         if ((ri->item_tag_nesting == 0) && (StrLength(CData) > 0)) {
1014                 NewStrBufDupAppendFlush(&ri->channel_title, CData, NULL, 0);
1015                 StrBufTrim(ri->channel_title);
1016         }
1017 }
1018
1019 void RSS_item_guid_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1020 {
1021         if (StrLength(CData) > 0) {
1022                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
1023         }
1024 }
1025
1026 void ATOM_item_id_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1027 {
1028         if (StrLength(CData) > 0) {
1029                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
1030         }
1031 }
1032
1033
1034 void RSS_item_link_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1035 {
1036         if (StrLength(CData) > 0) {
1037                 NewStrBufDupAppendFlush(&ri->link, CData, NULL, 0);
1038                 StrBufTrim(ri->link);
1039         }
1040 }
1041 void RSS_item_relink_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1042 {
1043         if (StrLength(CData) > 0) {
1044                 NewStrBufDupAppendFlush(&ri->reLink, CData, NULL, 0);
1045                 StrBufTrim(ri->reLink);
1046         }
1047 }
1048
1049 void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1050 {
1051         if (StrLength(CData) > 0) {
1052                 NewStrBufDupAppendFlush(&ri->title, CData, NULL, 0);
1053                 StrBufTrim(ri->title);
1054         }
1055 }
1056
1057 void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1058 {
1059         long olen = StrLength (ri->description);
1060         long clen = StrLength (CData);
1061         if (clen > 0)
1062         {
1063                 if (olen == 0) {
1064                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1065                         StrBufTrim(ri->description);
1066                 }
1067                 else if (olen < clen) {
1068                         FlushStrBuf(ri->description);
1069                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1070                         StrBufTrim(ri->description);
1071                 }
1072         }
1073 }
1074 void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1075 {
1076         /* this can contain an abstract of the article. but we don't want to verwrite a full document if we already have it. */
1077         if ((StrLength(CData) > 0) && (StrLength(ri->description) == 0))
1078         {
1079                 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1080                 StrBufTrim(ri->description);
1081         }
1082 }
1083
1084 void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1085 {
1086         long olen = StrLength (ri->description);
1087         long clen = StrLength (CData);
1088         if (clen > 0)
1089         {
1090                 if (olen == 0) {
1091                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1092                         StrBufTrim(ri->description);
1093                 }
1094                 else if (olen < clen) {
1095                         FlushStrBuf(ri->description);
1096                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1097                         StrBufTrim(ri->description);
1098                 }
1099         }
1100 }
1101
1102 void ATOM_item_published_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1103 {
1104         if (StrLength(CData) > 0) {
1105                 StrBufTrim(CData);
1106                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1107         }
1108 }
1109
1110 void ATOM_item_updated_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1111 {
1112         if (StrLength(CData) > 0) {
1113                 StrBufTrim(CData);
1114                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1115         }
1116 }
1117
1118 void RSS_item_pubdate_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1119 {
1120         if (StrLength(CData) > 0) {
1121                 StrBufTrim(CData);
1122                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1123         }
1124 }
1125
1126
1127 void RSS_item_date_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1128 {
1129         if (StrLength(CData) > 0) {
1130                 StrBufTrim(CData);
1131                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1132         }
1133 }
1134
1135
1136
1137 void RSS_item_author_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1138 {
1139         if (StrLength(CData) > 0) {
1140                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1141                 StrBufTrim(ri->author_or_creator);
1142         }
1143 }
1144
1145
1146 void ATOM_item_name_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1147 {
1148         if (StrLength(CData) > 0) {
1149                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1150                 StrBufTrim(ri->author_or_creator);
1151         }
1152 }
1153
1154 void ATOM_item_email_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1155 {
1156         if (StrLength(CData) > 0) {
1157                 NewStrBufDupAppendFlush(&ri->author_email, CData, NULL, 0);
1158                 StrBufTrim(ri->author_email);
1159         }
1160 }
1161
1162 void RSS_item_creator_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1163 {
1164         if ((StrLength(CData) > 0) &&
1165             (StrLength(ri->author_or_creator) == 0))
1166         {
1167                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1168                 StrBufTrim(ri->author_or_creator);
1169         }
1170 }
1171
1172
1173 void ATOM_item_uri_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1174 {
1175         if (StrLength(CData) > 0) {
1176                 NewStrBufDupAppendFlush(&ri->author_url, CData, NULL, 0);
1177                 StrBufTrim(ri->author_url);
1178         }
1179 }
1180
1181 void RSS_item_item_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1182 {
1183         --ri->item_tag_nesting;
1184         rss_save_item(ri);
1185 }
1186
1187
1188 void ATOM_item_entry_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1189 {
1190         --ri->item_tag_nesting;
1191         rss_save_item(ri);
1192 }
1193
1194 void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1195 {
1196 //              syslog(LOG_DEBUG, "End of feed detected.  Closing parser.\n");
1197         ri->done_parsing = 1;
1198
1199 }
1200 void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1201 {
1202 //              syslog(LOG_DEBUG, "End of feed detected.  Closing parser.\n");
1203         ri->done_parsing = 1;
1204 }
1205
1206
1207 void RSSATOM_item_ignore(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1208 {
1209 }
1210
1211
1212
1213 /*
1214  * This callback stores up the data which appears in between tags.
1215  */
1216 void rss_xml_cdata_start(void *data)
1217 {
1218         rsscollection *rssc = (rsscollection*) data;
1219
1220         FlushStrBuf(rssc->CData);
1221 }
1222
1223 void rss_xml_cdata_end(void *data)
1224 {
1225 }
1226 void rss_xml_chardata(void *data, const XML_Char *s, int len)
1227 {
1228         rsscollection *rssc = (rsscollection*) data;
1229
1230         StrBufAppendBufPlain (rssc->CData, s, len, 0);
1231 }
1232
1233 /*
1234  * Callback function for passing libcurl's output to expat for parsing
1235  */
1236 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
1237 {
1238         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
1239         return (size*nmemb);
1240 }
1241
1242
1243
1244 /*
1245  * Begin a feed parse
1246  */
1247 void rss_do_fetching(rssnetcfg *Cfg) {
1248         rsscollection rssc;
1249         rss_item ri;
1250         XML_Parser xp = NULL;
1251         StrBuf *Answer;
1252
1253         CURL *curl;
1254         CURLcode res;
1255         char errmsg[1024] = "";
1256         char *ptr;
1257         const char *at;
1258         long len;
1259
1260         time_t now;
1261
1262         now = time(NULL);
1263
1264         if ((Cfg->next_poll != 0) && (now < Cfg->next_poll))
1265                 return;
1266         memset(&ri, 0, sizeof(rss_item));
1267         rssc.Item = &ri;
1268         rssc.Cfg = Cfg;
1269
1270         syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
1271
1272         curl = curl_easy_init();
1273         if (!curl) {
1274                 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
1275                 return;
1276         }
1277         Answer = NewStrBufPlain(NULL, SIZ);
1278
1279         curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
1280         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
1281         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
1282         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
1283 //      curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
1284         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
1285         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
1286         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
1287 #ifdef CURLOPT_HTTP_CONTENT_DECODING
1288         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
1289         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
1290 #endif
1291         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
1292         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
1293         if (
1294                 (!IsEmptyStr(config.c_ip_addr))
1295                 && (strcmp(config.c_ip_addr, "*"))
1296                 && (strcmp(config.c_ip_addr, "::"))
1297                 && (strcmp(config.c_ip_addr, "0.0.0.0"))
1298         ) {
1299                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
1300         }
1301
1302         if (CtdlThreadCheckStop())
1303         {
1304                 curl_easy_cleanup(curl);
1305                 return;
1306         }
1307
1308         if (CtdlThreadCheckStop())
1309                 goto shutdown ;
1310
1311         res = curl_easy_perform(curl);
1312         if (res) {
1313                 syslog(LOG_ALERT, "libcurl error %d: %s\n", res, errmsg);
1314         }
1315
1316         if (CtdlThreadCheckStop())
1317                 goto shutdown ;
1318
1319
1320
1321
1322         memset(&ri, 0, sizeof(rss_item));
1323         ri.roomlist = Cfg->rooms;
1324         rssc.CData = NewStrBufPlain(NULL, SIZ);
1325         rssc.Key = NewStrBuf();
1326         at = NULL;
1327         StrBufSipLine(rssc.Key, Answer, &at);
1328         ptr = NULL;
1329
1330 #define encoding "encoding=\""
1331         ptr = strstr(ChrPtr(rssc.Key), encoding);
1332         if (ptr != NULL)
1333         {
1334                 char *pche;
1335
1336                 ptr += sizeof (encoding) - 1;
1337                 pche = strchr(ptr, '"');
1338                 if (pche != NULL)
1339                         StrBufCutAt(rssc.Key, -1, pche);
1340                 else
1341                         ptr = "UTF-8";
1342         }
1343         else
1344                 ptr = "UTF-8";
1345
1346
1347         xp = XML_ParserCreateNS(ptr, ':');
1348         if (!xp) {
1349                 syslog(LOG_ALERT, "Cannot create XML parser!\n");
1350                 goto shutdown;
1351         }
1352         FlushStrBuf(rssc.Key);
1353 //#ifdef HAVE_ICONV
1354 #if 0
1355         XML_SetUnknownEncodingHandler(xp,
1356                                       handle_unknown_xml_encoding,
1357                                       &rssc);
1358 #endif
1359 //#endif
1360         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
1361         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
1362         XML_SetUserData(xp, &rssc);
1363         XML_SetCdataSectionHandler(xp,
1364                                    rss_xml_cdata_start,
1365                                    rss_xml_cdata_end);
1366
1367
1368         len = StrLength(Answer);
1369         ptr = SmashStrBuf(&Answer);
1370         XML_Parse(xp, ptr, len, 0);
1371         free (ptr);
1372         if (ri.done_parsing == 0)
1373                 XML_Parse(xp, "", 0, 1);
1374
1375
1376         syslog(LOG_ALERT, "RSS: XML Status [%s] \n",
1377                       XML_ErrorString(
1378                               XML_GetErrorCode(xp)));
1379
1380 shutdown:
1381         FreeStrBuf(&Answer);
1382         curl_easy_cleanup(curl);
1383         XML_ParserFree(xp);
1384
1385         flush_rss_item(&ri);
1386         FreeStrBuf(&rssc.CData);
1387         FreeStrBuf(&rssc.Key);
1388
1389         Cfg->next_poll = time(NULL) + config.c_net_freq;
1390 }
1391
1392
1393 /*
1394  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
1395  */
1396 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
1397 {
1398         char filename[PATH_MAX];
1399         char buf[1024];
1400         char instr[32];
1401         FILE *fp;
1402         char feedurl[256];
1403         rssnetcfg *rncptr = NULL;
1404         rssnetcfg *use_this_rncptr = NULL;
1405         int len = 0;
1406         char *ptr = NULL;
1407
1408         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
1409
1410         if (CtdlThreadCheckStop())
1411                 return;
1412
1413         /* Only do net processing for rooms that have netconfigs */
1414         fp = fopen(filename, "r");
1415         if (fp == NULL) {
1416                 return;
1417         }
1418
1419         while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
1420                 buf[strlen(buf)-1] = 0;
1421
1422                 extract_token(instr, buf, 0, '|', sizeof instr);
1423                 if (!strcasecmp(instr, "rssclient")) {
1424
1425                         use_this_rncptr = NULL;
1426
1427                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
1428
1429                         /* If any other rooms have requested the same feed, then we will just add this
1430                          * room to the target list for that client request.
1431                          */
1432                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
1433                                 if (!strcmp(rncptr->url, feedurl)) {
1434                                         use_this_rncptr = rncptr;
1435                                 }
1436                         }
1437
1438                         /* Otherwise create a new client request */
1439                         if (use_this_rncptr == NULL) {
1440                                 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
1441                                 memset(rncptr, 0, sizeof(rssnetcfg));
1442                                 rncptr->ItemType = RSS_UNSET;
1443                                 if (rncptr != NULL) {
1444                                         rncptr->next = rnclist;
1445                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
1446                                         rncptr->rooms = NULL;
1447                                         rnclist = rncptr;
1448                                         use_this_rncptr = rncptr;
1449                                 }
1450                         }
1451
1452                         /* Add the room name to the request */
1453                         if (use_this_rncptr != NULL) {
1454                                 if (use_this_rncptr->rooms == NULL) {
1455                                         rncptr->rooms = strdup(qrbuf->QRname);
1456                                 }
1457                                 else {
1458                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
1459                                         ptr = realloc(use_this_rncptr->rooms, len);
1460                                         if (ptr != NULL) {
1461                                                 strcat(ptr, "|");
1462                                                 strcat(ptr, qrbuf->QRname);
1463                                                 use_this_rncptr->rooms = ptr;
1464                                         }
1465                                 }
1466                         }
1467                 }
1468
1469         }
1470
1471         fclose(fp);
1472
1473 }
1474
1475 /*
1476  * Scan for rooms that have RSS client requests configured
1477  */
1478 void rssclient_scan(void) {
1479         static time_t last_run = 0L;
1480         static int doing_rssclient = 0;
1481         rssnetcfg *rptr = NULL;
1482
1483         /*
1484          * This is a simple concurrency check to make sure only one rssclient run
1485          * is done at a time.  We could do this with a mutex, but since we
1486          * don't really require extremely fine granularity here, we'll do it
1487          * with a static variable instead.
1488          */
1489         if (doing_rssclient) return;
1490         doing_rssclient = 1;
1491
1492         syslog(LOG_DEBUG, "rssclient started\n");
1493         CtdlForEachRoom(rssclient_scan_room, NULL);
1494
1495         while (rnclist != NULL && !CtdlThreadCheckStop()) {
1496                 rss_do_fetching(rnclist);
1497                 rptr = rnclist;
1498                 rnclist = rnclist->next;
1499                 if (rptr->rooms != NULL) free(rptr->rooms);
1500                 free(rptr);
1501         }
1502
1503         syslog(LOG_DEBUG, "rssclient ended\n");
1504         last_run = time(NULL);
1505         doing_rssclient = 0;
1506         return;
1507 }
1508
1509 void LoadUrlShorteners(void)
1510 {
1511         int i = 0;
1512         int fd;
1513         const char *POS = NULL;
1514         const char *Err = NULL;
1515         StrBuf *Content, *Line;
1516
1517
1518         UrlShorteners = NewHash(0, Flathash);
1519
1520         fd = open(file_citadel_urlshorteners, 0);
1521
1522         if (fd != 0)
1523         {
1524                 Content = NewStrBufPlain(NULL, SIZ);
1525                 Line = NewStrBuf();
1526                 while (POS != StrBufNOTNULL)
1527                 {
1528                         StrBufTCP_read_buffered_line_fast (Line, Content, &POS, &fd, 1, 1, &Err);
1529                         StrBufTrim(Line);
1530                         if ((*ChrPtr(Line) != '#') && (StrLength(Line) > 0))
1531                         {
1532                                 Put(UrlShorteners, IKEY(i), Line, HFreeStrBuf);
1533                                 i++;
1534                                 Line = NewStrBuf();
1535                         }
1536                         else
1537                                 FlushStrBuf(Line);
1538                         if (POS == NULL)
1539                                 POS = StrBufNOTNULL;
1540                 }
1541                 FreeStrBuf(&Line);
1542                 FreeStrBuf(&Content);
1543         }
1544         close(fd);
1545 }
1546
1547 void rss_cleanup(void)
1548 {
1549         DeleteHash(&StartHandlers);
1550         DeleteHash(&EndHandlers);
1551         DeleteHash(&UrlShorteners);
1552         DeleteHash(&KnownNameSpaces);
1553 }
1554
1555 CTDL_MODULE_INIT(rssclient)
1556 {
1557         if (threading)
1558         {
1559                 syslog(LOG_INFO, "%s\n", curl_version());
1560                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
1561         }
1562         else
1563         {
1564                 LoadUrlShorteners ();
1565
1566                 StartHandlers = NewHash(1, NULL);
1567                 EndHandlers = NewHash(1, NULL);
1568
1569                 AddRSSStartHandler(RSS_item_rss_start,     RSS_UNSET, HKEY("rss"));
1570                 AddRSSStartHandler(RSS_item_rdf_start,     RSS_UNSET, HKEY("rdf"));
1571                 AddRSSStartHandler(ATOM_item_feed_start,    RSS_UNSET, HKEY("feed"));
1572                 AddRSSStartHandler(RSS_item_item_start,    RSS_RSS, HKEY("item"));
1573                 AddRSSStartHandler(ATOM_item_entry_start,  RSS_ATOM, HKEY("entry"));
1574                 AddRSSStartHandler(ATOM_item_link_start,   RSS_ATOM, HKEY("link"));
1575
1576                 AddRSSEndHandler(ATOMRSS_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1577                 AddRSSEndHandler(RSS_item_guid_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
1578                 AddRSSEndHandler(ATOM_item_id_end,         RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
1579                 AddRSSEndHandler(RSS_item_link_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
1580 #if 0
1581 // hm, rss to the comments of that blog, might be interesting in future, but...
1582                 AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("commentrss"));
1583 // comment count...
1584                 AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("comments"));
1585 #endif
1586                 AddRSSEndHandler(RSSATOM_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1587                 AddRSSEndHandler(ATOM_item_content_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
1588                 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_ATOM|RSS_REQUIRE_BUF, HKEY("encoded"));
1589                 AddRSSEndHandler(ATOM_item_summary_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
1590                 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
1591                 AddRSSEndHandler(ATOM_item_published_end,  RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
1592                 AddRSSEndHandler(ATOM_item_updated_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
1593                 AddRSSEndHandler(RSS_item_pubdate_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
1594                 AddRSSEndHandler(RSS_item_date_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
1595                 AddRSSEndHandler(RSS_item_author_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
1596                 AddRSSEndHandler(RSS_item_creator_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
1597 /* <author> */
1598                 AddRSSEndHandler(ATOM_item_email_end,      RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
1599                 AddRSSEndHandler(ATOM_item_name_end,       RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
1600                 AddRSSEndHandler(ATOM_item_uri_end,        RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
1601 /* </author> */
1602                 AddRSSEndHandler(RSS_item_item_end,        RSS_RSS, HKEY("item"));
1603                 AddRSSEndHandler(RSS_item_rss_end,         RSS_RSS, HKEY("rss"));
1604                 AddRSSEndHandler(RSS_item_rdf_end,         RSS_RSS, HKEY("rdf"));
1605                 AddRSSEndHandler(ATOM_item_entry_end,      RSS_ATOM, HKEY("entry"));
1606
1607
1608 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
1609                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1610                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1611                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1612                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1613
1614 /* links to other feed generators... */
1615                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1616                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1617                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1618                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1619
1620                 KnownNameSpaces = NewHash(1, NULL);
1621                 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
1622                 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
1623                 Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
1624                 Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
1625                 Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
1626                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1627                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
1628                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
1629                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
1630                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1631                 Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
1632                 Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
1633                 Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
1634                 Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
1635                 Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
1636                 Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
1637                 Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
1638                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
1639                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1640                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1641                 Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
1642                 Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
1643                 Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
1644 #if 0
1645                 /* we don't like these namespaces because of they shadow our usefull parameters. */
1646                 Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
1647 #endif
1648                 CtdlRegisterCleanupHook(rss_cleanup);
1649         }
1650         return "rssclient";
1651 }