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         CtdlLogPrintf(0, "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         CtdlLogPrintf(0, "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   CtdlLogPrintf(0, "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                 CtdlLogPrintf(CTDL_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                 CtdlLogPrintf(CTDL_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                 CtdlLogPrintf(CTDL_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                 CtdlLogPrintf(CTDL_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->author_or_creator);
 760         FreeStrBuf(&ri->author_email);
 761         FreeStrBuf(&ri->author_url);
 762         FreeStrBuf(&ri->description);
 763 }
 764
 765 void rss_xml_start(void *data, const char *supplied_el, const char **attr)
 766 {
 767         rss_xml_handler *h;
 768         rsscollection   *rssc = (rsscollection*) data;
 769         rssnetcfg       *Cfg = rssc->Cfg;
 770         rss_item        *ri = rssc->Item;
 771         void            *pv;
 772         const char      *pel;
 773         char            *sep = NULL;
 774
 775         /* Axe the namespace, we don't care about it */
 776 ///     CtdlLogPrintf(0, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
 777         pel = supplied_el;
 778         while (sep = strchr(pel, ':'), sep) {
 779                 pel = sep + 1;
 780         }
 781
 782         if (pel != supplied_el)
 783         {
 784                 void *v;
 785
 786                 if (!GetHash(KnownNameSpaces,
 787                              supplied_el,
 788                              pel - supplied_el - 1,
 789                              &v))
 790                 {
 791 #ifdef DEBUG_RSS
 792                         CtdlLogPrintf(0, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
 793                                       supplied_el);
 794 #endif
 795                         return;
 796                 }
 797         }
 798
 799         StrBufPlain(rssc->Key, pel, -1);
 800         StrBufLowerCase(rssc->Key);
 801         if (GetHash(StartHandlers, SKEY(rssc->Key), &pv))
 802         {
 803                 rssc->Current = h = (rss_xml_handler*) pv;
 804
 805                 if (((h->Flags & RSS_UNSET) != 0) &&
 806                     (Cfg->ItemType == RSS_UNSET))
 807                 {
 808                         h->Handler(rssc->CData, ri, Cfg, attr);
 809                 }
 810                 else if (((h->Flags & RSS_RSS) != 0) &&
 811                     (Cfg->ItemType == RSS_RSS))
 812                 {
 813                         h->Handler(rssc->CData, ri, Cfg, attr);
 814                 }
 815                 else if (((h->Flags & RSS_ATOM) != 0) &&
 816                          (Cfg->ItemType == RSS_ATOM))
 817                 {
 818                         h->Handler(rssc->CData, ri, Cfg, attr);
 819                 }
 820 #ifdef DEBUG_RSS
 821                 else
 822                         CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
 823 #endif
 824         }
 825 #ifdef DEBUG_RSS
 826         else
 827                 CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel,  supplied_el);
 828 #endif
 829 }
 830
 831 void rss_xml_end(void *data, const char *supplied_el)
 832 {
 833         rss_xml_handler *h;
 834         rsscollection   *rssc = (rsscollection*) data;
 835         rssnetcfg       *Cfg = rssc->Cfg;
 836         rss_item        *ri = rssc->Item;
 837         const char      *pel;
 838         char            *sep = NULL;
 839         void            *pv;
 840
 841         /* Axe the namespace, we don't care about it */
 842         pel = supplied_el;
 843         while (sep = strchr(pel, ':'), sep) {
 844                 pel = sep + 1;
 845         }
 846 //      CtdlLogPrintf(0, "RSS: END %s...\n", el);
 847         if (pel != supplied_el)
 848         {
 849                 void *v;
 850
 851                 if (!GetHash(KnownNameSpaces,
 852                              supplied_el,
 853                              pel - supplied_el - 1,
 854                              &v))
 855                 {
 856 #ifdef DEBUG_RSS
 857                         CtdlLogPrintf(0, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
 858                                       supplied_el, ChrPtr(rssc->CData));
 859 #endif
 860                         FlushStrBuf(rssc->CData);
 861                         return;
 862                 }
 863         }
 864
 865         StrBufPlain(rssc->Key, pel, -1);
 866         StrBufLowerCase(rssc->Key);
 867         if (GetHash(EndHandlers, SKEY(rssc->Key), &pv))
 868         {
 869                 h = (rss_xml_handler*) pv;
 870
 871                 if (((h->Flags & RSS_UNSET) != 0) &&
 872                     (Cfg->ItemType == RSS_UNSET))
 873                 {
 874                         h->Handler(rssc->CData, ri, Cfg, NULL);
 875                 }
 876                 else if (((h->Flags & RSS_RSS) != 0) &&
 877                     (Cfg->ItemType == RSS_RSS))
 878                 {
 879                         h->Handler(rssc->CData, ri, Cfg, NULL);
 880                 }
 881                 else if (((h->Flags & RSS_ATOM) != 0) &&
 882                          (Cfg->ItemType == RSS_ATOM))
 883                 {
 884                         h->Handler(rssc->CData, ri, Cfg, NULL);
 885                 }
 886 #ifdef DEBUG_RSS
 887                 else
 888                         CtdlLogPrintf(0, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 889 #endif
 890         }
 891 #ifdef DEBUG_RSS
 892         else
 893                 CtdlLogPrintf(0, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 894 #endif
 895         FlushStrBuf(rssc->CData);
 896         rssc->Current = NULL;
 897 }
 898
 899
 900
 901
 902
 903 void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 904 {
 905         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RSS feed.\n");
 906         Cfg->ItemType = RSS_RSS;
 907 }
 908
 909 void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 910 {
 911         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RDF feed.\n");
 912         Cfg->ItemType = RSS_RSS;
 913 }
 914
 915 void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 916 {
 917         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an ATOM feed.\n");
 918         Cfg->ItemType = RSS_ATOM;
 919 }
 920
 921
 922 void RSS_item_item_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 923 {
 924         ri->item_tag_nesting ++;
 925         flush_rss_item(ri);
 926 }
 927
 928 void ATOM_item_entry_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 929 {
 930 /* Atom feed... */
 931         ri->item_tag_nesting ++;
 932         flush_rss_item(ri);
 933 }
 934
 935 void ATOM_item_link_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 936 {
 937         int i;
 938         const char *pHref = NULL;
 939         const char *pType = NULL;
 940         const char *pRel = NULL;
 941         const char *pTitle = NULL;
 942
 943         for (i = 0; Attr[i] != NULL; i+=2)
 944         {
 945                 if (!strcmp(Attr[i], "href"))
 946                 {
 947                         pHref = Attr[i+1];
 948                 }
 949                 else if (!strcmp(Attr[i], "rel"))
 950                 {
 951                         pRel = Attr[i+1];
 952                 }
 953                 else if (!strcmp(Attr[i], "type"))
 954                 {
 955                         pType = Attr[i+1];
 956                 }
 957                 else if (!strcmp(Attr[i], "title"))
 958                 {
 959                         pTitle = Attr[i+1];
 960                 }
 961         }
 962         if (pHref == NULL)
 963                 return; /* WHUT? Pointing... where? */
 964         if ((pType != NULL) && !strcasecmp(pType, "application/atom+xml"))
 965                 return; /* these just point to other rss resources, we're not interested in them. */
 966         if (pRel != NULL)
 967         {
 968                 if (!strcasecmp (pRel, "replies"))
 969                 {
 970                         NewStrBufDupAppendFlush(&ri->reLink, NULL, pHref, -1);
 971                         StrBufTrim(ri->link);
 972                         NewStrBufDupAppendFlush(&ri->reLinkTitle, NULL, pTitle, -1);
 973                 }
 974                 else if (!strcasecmp(pRel, "alternate")) /* Alternative representation of this Item... */
 975                 {
 976                         NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 977                         StrBufTrim(ri->link);
 978                         NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
 979
 980                 }
 981 #if 0 /* these are also defined, but dunno what to do with them.. */
 982                 else if (!strcasecmp(pRel, "related"))
 983                 {
 984                 }
 985                 else if (!strcasecmp(pRel, "self"))
 986                 {
 987                 }
 988                 else if (!strcasecmp(pRel, "enclosure"))
 989                 {/* this reference can get big, and is probably the full article... */
 990                 }
 991                 else if (!strcasecmp(pRel, "via"))
 992                 {/* this article was provided via... */
 993                 }
 994 #endif
 995         }
 996         else if (StrLength(ri->link) == 0)
 997         {
 998                 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 999                 StrBufTrim(ri->link);
1000                 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
1001         }
1002 }
1003
1004
1005
1006
1007 void ATOMRSS_item_title_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1008 {
1009         if ((ri->item_tag_nesting == 0) && (StrLength(CData) > 0)) {
1010                 NewStrBufDupAppendFlush(&ri->channel_title, CData, NULL, 0);
1011                 StrBufTrim(ri->channel_title);
1012         }
1013 }
1014
1015 void RSS_item_guid_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1016 {
1017         if (StrLength(CData) > 0) {
1018                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
1019         }
1020 }
1021
1022 void ATOM_item_id_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1023 {
1024         if (StrLength(CData) > 0) {
1025                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
1026         }
1027 }
1028
1029
1030 void RSS_item_link_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1031 {
1032         if (StrLength(CData) > 0) {
1033                 NewStrBufDupAppendFlush(&ri->link, CData, NULL, 0);
1034                 StrBufTrim(ri->link);
1035         }
1036 }
1037 void RSS_item_relink_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1038 {
1039         if (StrLength(CData) > 0) {
1040                 NewStrBufDupAppendFlush(&ri->reLink, CData, NULL, 0);
1041                 StrBufTrim(ri->reLink);
1042         }
1043 }
1044
1045 void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1046 {
1047         if (StrLength(CData) > 0) {
1048                 NewStrBufDupAppendFlush(&ri->title, CData, NULL, 0);
1049                 StrBufTrim(ri->title);
1050         }
1051 }
1052
1053 void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1054 {
1055         long olen = StrLength (ri->description);
1056         long clen = StrLength (CData);
1057         if (clen > 0)
1058         {
1059                 if (olen == 0) {
1060                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1061                         StrBufTrim(ri->description);
1062                 }
1063                 else if (olen < clen) {
1064                         FlushStrBuf(ri->description);
1065                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1066                         StrBufTrim(ri->description);
1067                 }
1068         }
1069 }
1070 void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1071 {
1072         /* this can contain an abstract of the article. but we don't want to verwrite a full document if we already have it. */
1073         if ((StrLength(CData) > 0) && (StrLength(ri->description) == 0))
1074         {
1075                 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1076                 StrBufTrim(ri->description);
1077         }
1078 }
1079
1080 void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1081 {
1082         long olen = StrLength (ri->description);
1083         long clen = StrLength (CData);
1084         if (clen > 0)
1085         {
1086                 if (olen == 0) {
1087                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1088                         StrBufTrim(ri->description);
1089                 }
1090                 else if (olen < clen) {
1091                         FlushStrBuf(ri->description);
1092                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1093                         StrBufTrim(ri->description);
1094                 }
1095         }
1096 }
1097
1098 void ATOM_item_published_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1099 {
1100         if (StrLength(CData) > 0) {
1101                 StrBufTrim(CData);
1102                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1103         }
1104 }
1105
1106 void ATOM_item_updated_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1107 {
1108         if (StrLength(CData) > 0) {
1109                 StrBufTrim(CData);
1110                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1111         }
1112 }
1113
1114 void RSS_item_pubdate_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1115 {
1116         if (StrLength(CData) > 0) {
1117                 StrBufTrim(CData);
1118                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1119         }
1120 }
1121
1122
1123 void RSS_item_date_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1124 {
1125         if (StrLength(CData) > 0) {
1126                 StrBufTrim(CData);
1127                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1128         }
1129 }
1130
1131
1132
1133 void RSS_item_author_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1134 {
1135         if (StrLength(CData) > 0) {
1136                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1137                 StrBufTrim(ri->author_or_creator);
1138         }
1139 }
1140
1141
1142 void ATOM_item_name_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1143 {
1144         if (StrLength(CData) > 0) {
1145                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1146                 StrBufTrim(ri->author_or_creator);
1147         }
1148 }
1149
1150 void ATOM_item_email_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1151 {
1152         if (StrLength(CData) > 0) {
1153                 NewStrBufDupAppendFlush(&ri->author_email, CData, NULL, 0);
1154                 StrBufTrim(ri->author_email);
1155         }
1156 }
1157
1158 void RSS_item_creator_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1159 {
1160         if ((StrLength(CData) > 0) &&
1161             (StrLength(ri->author_or_creator) == 0))
1162         {
1163                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1164                 StrBufTrim(ri->author_or_creator);
1165         }
1166 }
1167
1168
1169 void ATOM_item_uri_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1170 {
1171         if (StrLength(CData) > 0) {
1172                 NewStrBufDupAppendFlush(&ri->author_url, CData, NULL, 0);
1173                 StrBufTrim(ri->author_url);
1174         }
1175 }
1176
1177 void RSS_item_item_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1178 {
1179         --ri->item_tag_nesting;
1180         rss_save_item(ri);
1181 }
1182
1183
1184 void ATOM_item_entry_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1185 {
1186         --ri->item_tag_nesting;
1187         rss_save_item(ri);
1188 }
1189
1190 void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1191 {
1192 //              CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
1193         ri->done_parsing = 1;
1194
1195 }
1196 void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1197 {
1198 //              CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
1199         ri->done_parsing = 1;
1200 }
1201
1202
1203 void RSSATOM_item_ignore(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1204 {
1205 }
1206
1207
1208
1209 /*
1210  * This callback stores up the data which appears in between tags.
1211  */
1212 void rss_xml_cdata_start(void *data)
1213 {
1214         rsscollection *rssc = (rsscollection*) data;
1215
1216         FlushStrBuf(rssc->CData);
1217 }
1218
1219 void rss_xml_cdata_end(void *data)
1220 {
1221 }
1222 void rss_xml_chardata(void *data, const XML_Char *s, int len)
1223 {
1224         rsscollection *rssc = (rsscollection*) data;
1225
1226         StrBufAppendBufPlain (rssc->CData, s, len, 0);
1227 }
1228
1229 /*
1230  * Callback function for passing libcurl's output to expat for parsing
1231  */
1232 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
1233 {
1234         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
1235         return (size*nmemb);
1236 }
1237
1238
1239
1240 /*
1241  * Begin a feed parse
1242  */
1243 void rss_do_fetching(rssnetcfg *Cfg) {
1244         rsscollection rssc;
1245         rss_item ri;
1246         XML_Parser xp = NULL;
1247         StrBuf *Answer;
1248
1249         CURL *curl;
1250         CURLcode res;
1251         char errmsg[1024] = "";
1252         char *ptr;
1253         const char *at;
1254         long len;
1255
1256         time_t now;
1257
1258         now = time(NULL);
1259
1260         if ((Cfg->next_poll != 0) && (now < Cfg->next_poll))
1261                 return;
1262         memset(&ri, 0, sizeof(rss_item));
1263         rssc.Item = &ri;
1264         rssc.Cfg = Cfg;
1265
1266         CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
1267
1268         curl = curl_easy_init();
1269         if (!curl) {
1270                 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
1271                 return;
1272         }
1273         Answer = NewStrBufPlain(NULL, SIZ);
1274
1275         curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
1276         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
1277         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
1278         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
1279 //      curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
1280         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
1281         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
1282         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
1283 #ifdef CURLOPT_HTTP_CONTENT_DECODING
1284         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
1285         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
1286 #endif
1287         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
1288         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
1289         if (
1290                 (!IsEmptyStr(config.c_ip_addr))
1291                 && (strcmp(config.c_ip_addr, "*"))
1292                 && (strcmp(config.c_ip_addr, "::"))
1293                 && (strcmp(config.c_ip_addr, "0.0.0.0"))
1294         ) {
1295                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
1296         }
1297
1298         if (CtdlThreadCheckStop())
1299         {
1300                 curl_easy_cleanup(curl);
1301                 return;
1302         }
1303
1304         if (CtdlThreadCheckStop())
1305                 goto shutdown ;
1306
1307         res = curl_easy_perform(curl);
1308         if (res) {
1309                 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
1310         }
1311
1312         if (CtdlThreadCheckStop())
1313                 goto shutdown ;
1314
1315
1316
1317
1318         memset(&ri, 0, sizeof(rss_item));
1319         ri.roomlist = Cfg->rooms;
1320         rssc.CData = NewStrBufPlain(NULL, SIZ);
1321         rssc.Key = NewStrBuf();
1322         at = NULL;
1323         StrBufSipLine(rssc.Key, Answer, &at);
1324         ptr = NULL;
1325
1326 #define encoding "encoding=\""
1327         ptr = strstr(ChrPtr(rssc.Key), encoding);
1328         if (ptr != NULL)
1329         {
1330                 char *pche;
1331
1332                 ptr += sizeof (encoding) - 1;
1333                 pche = strchr(ptr, '"');
1334                 if (pche != NULL)
1335                         StrBufCutAt(rssc.Key, -1, pche);
1336                 else
1337                         ptr = "UTF-8";
1338         }
1339         else
1340                 ptr = "UTF-8";
1341
1342
1343         xp = XML_ParserCreateNS(ptr, ':');
1344         if (!xp) {
1345                 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
1346                 goto shutdown;
1347         }
1348         FlushStrBuf(rssc.Key);
1349 //#ifdef HAVE_ICONV
1350 #if 0
1351         XML_SetUnknownEncodingHandler(xp,
1352                                       handle_unknown_xml_encoding,
1353                                       &rssc);
1354 #endif
1355 //#endif
1356         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
1357         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
1358         XML_SetUserData(xp, &rssc);
1359         XML_SetCdataSectionHandler(xp,
1360                                    rss_xml_cdata_start,
1361                                    rss_xml_cdata_end);
1362
1363
1364         len = StrLength(Answer);
1365         ptr = SmashStrBuf(&Answer);
1366         XML_Parse(xp, ptr, len, 0);
1367         free (ptr);
1368         if (ri.done_parsing == 0)
1369                 XML_Parse(xp, "", 0, 1);
1370
1371
1372         CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n",
1373                       XML_ErrorString(
1374                               XML_GetErrorCode(xp)));
1375
1376 shutdown:
1377         curl_easy_cleanup(curl);
1378         XML_ParserFree(xp);
1379
1380         flush_rss_item(&ri);
1381         FreeStrBuf(&rssc.CData);
1382         FreeStrBuf(&rssc.Key);
1383
1384         Cfg->next_poll = time(NULL) + config.c_net_freq;
1385 }
1386
1387
1388 /*
1389  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
1390  */
1391 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
1392 {
1393         char filename[PATH_MAX];
1394         char buf[1024];
1395         char instr[32];
1396         FILE *fp;
1397         char feedurl[256];
1398         rssnetcfg *rncptr = NULL;
1399         rssnetcfg *use_this_rncptr = NULL;
1400         int len = 0;
1401         char *ptr = NULL;
1402
1403         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
1404
1405         if (CtdlThreadCheckStop())
1406                 return;
1407
1408         /* Only do net processing for rooms that have netconfigs */
1409         fp = fopen(filename, "r");
1410         if (fp == NULL) {
1411                 return;
1412         }
1413
1414         while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
1415                 buf[strlen(buf)-1] = 0;
1416
1417                 extract_token(instr, buf, 0, '|', sizeof instr);
1418                 if (!strcasecmp(instr, "rssclient")) {
1419
1420                         use_this_rncptr = NULL;
1421
1422                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
1423
1424                         /* If any other rooms have requested the same feed, then we will just add this
1425                          * room to the target list for that client request.
1426                          */
1427                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
1428                                 if (!strcmp(rncptr->url, feedurl)) {
1429                                         use_this_rncptr = rncptr;
1430                                 }
1431                         }
1432
1433                         /* Otherwise create a new client request */
1434                         if (use_this_rncptr == NULL) {
1435                                 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
1436                                 memset(rncptr, 0, sizeof(rssnetcfg));
1437                                 rncptr->ItemType = RSS_UNSET;
1438                                 if (rncptr != NULL) {
1439                                         rncptr->next = rnclist;
1440                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
1441                                         rncptr->rooms = NULL;
1442                                         rnclist = rncptr;
1443                                         use_this_rncptr = rncptr;
1444                                 }
1445                         }
1446
1447                         /* Add the room name to the request */
1448                         if (use_this_rncptr != NULL) {
1449                                 if (use_this_rncptr->rooms == NULL) {
1450                                         rncptr->rooms = strdup(qrbuf->QRname);
1451                                 }
1452                                 else {
1453                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
1454                                         ptr = realloc(use_this_rncptr->rooms, len);
1455                                         if (ptr != NULL) {
1456                                                 strcat(ptr, "|");
1457                                                 strcat(ptr, qrbuf->QRname);
1458                                                 use_this_rncptr->rooms = ptr;
1459                                         }
1460                                 }
1461                         }
1462                 }
1463
1464         }
1465
1466         fclose(fp);
1467
1468 }
1469
1470 /*
1471  * Scan for rooms that have RSS client requests configured
1472  */
1473 void rssclient_scan(void) {
1474         static time_t last_run = 0L;
1475         static int doing_rssclient = 0;
1476         rssnetcfg *rptr = NULL;
1477
1478         /*
1479          * This is a simple concurrency check to make sure only one rssclient run
1480          * is done at a time.  We could do this with a mutex, but since we
1481          * don't really require extremely fine granularity here, we'll do it
1482          * with a static variable instead.
1483          */
1484         if (doing_rssclient) return;
1485         doing_rssclient = 1;
1486
1487         CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
1488         CtdlForEachRoom(rssclient_scan_room, NULL);
1489
1490         while (rnclist != NULL && !CtdlThreadCheckStop()) {
1491                 rss_do_fetching(rnclist);
1492                 rptr = rnclist;
1493                 rnclist = rnclist->next;
1494                 if (rptr->rooms != NULL) free(rptr->rooms);
1495                 free(rptr);
1496         }
1497
1498         CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
1499         last_run = time(NULL);
1500         doing_rssclient = 0;
1501         return;
1502 }
1503
1504 void LoadUrlShorteners(void)
1505 {
1506         int i = 0;
1507         int fd;
1508         const char *POS = NULL;
1509         const char *Err = NULL;
1510         StrBuf *Content, *Line;
1511
1512
1513         UrlShorteners = NewHash(0, Flathash);
1514
1515         fd = open(file_citadel_urlshorteners, 0);
1516
1517         if (fd != 0)
1518         {
1519                 Content = NewStrBufPlain(NULL, SIZ);
1520                 Line = NewStrBuf();
1521                 while (POS != StrBufNOTNULL)
1522                 {
1523                         StrBufTCP_read_buffered_line_fast (Line, Content, &POS, &fd, 1, 1, &Err);
1524                         StrBufTrim(Line);
1525                         if ((*ChrPtr(Line) != '#') && (StrLength(Line) > 0))
1526                         {
1527                                 Put(UrlShorteners, IKEY(i), Line, HFreeStrBuf);
1528                                 i++;
1529                                 Line = NewStrBuf();
1530                         }
1531                         else
1532                                 FlushStrBuf(Line);
1533                         if (POS == NULL)
1534                                 POS = StrBufNOTNULL;
1535                 }
1536                 FreeStrBuf(&Line);
1537                 FreeStrBuf(&Content);
1538         }
1539         close(fd);
1540 }
1541
1542 CTDL_MODULE_INIT(rssclient)
1543 {
1544         if (threading)
1545         {
1546                 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
1547                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
1548         }
1549         else
1550         {
1551                 LoadUrlShorteners ();
1552
1553                 StartHandlers = NewHash(1, NULL);
1554                 EndHandlers = NewHash(1, NULL);
1555
1556                 AddRSSStartHandler(RSS_item_rss_start,     RSS_UNSET, HKEY("rss"));
1557                 AddRSSStartHandler(RSS_item_rdf_start,     RSS_UNSET, HKEY("rdf"));
1558                 AddRSSStartHandler(ATOM_item_feed_start,    RSS_UNSET, HKEY("feed"));
1559                 AddRSSStartHandler(RSS_item_item_start,    RSS_RSS, HKEY("item"));
1560                 AddRSSStartHandler(ATOM_item_entry_start,  RSS_ATOM, HKEY("entry"));
1561                 AddRSSStartHandler(ATOM_item_link_start,   RSS_ATOM, HKEY("link"));
1562
1563                 AddRSSEndHandler(ATOMRSS_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1564                 AddRSSEndHandler(RSS_item_guid_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
1565                 AddRSSEndHandler(ATOM_item_id_end,         RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
1566                 AddRSSEndHandler(RSS_item_link_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
1567 #if 0
1568 // hm, rss to the comments of that blog, might be interesting in future, but...
1569                 AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("commentrss"));
1570 // comment count...
1571                 AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("comments"));
1572 #endif
1573                 AddRSSEndHandler(RSSATOM_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1574                 AddRSSEndHandler(ATOM_item_content_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
1575                 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_ATOM|RSS_REQUIRE_BUF, HKEY("encoded"));
1576                 AddRSSEndHandler(ATOM_item_summary_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
1577                 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
1578                 AddRSSEndHandler(ATOM_item_published_end,  RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
1579                 AddRSSEndHandler(ATOM_item_updated_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
1580                 AddRSSEndHandler(RSS_item_pubdate_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
1581                 AddRSSEndHandler(RSS_item_date_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
1582                 AddRSSEndHandler(RSS_item_author_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
1583                 AddRSSEndHandler(RSS_item_creator_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
1584 /* <author> */
1585                 AddRSSEndHandler(ATOM_item_email_end,      RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
1586                 AddRSSEndHandler(ATOM_item_name_end,       RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
1587                 AddRSSEndHandler(ATOM_item_uri_end,        RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
1588 /* </author> */
1589                 AddRSSEndHandler(RSS_item_item_end,        RSS_RSS, HKEY("item"));
1590                 AddRSSEndHandler(RSS_item_rss_end,         RSS_RSS, HKEY("rss"));
1591                 AddRSSEndHandler(RSS_item_rdf_end,         RSS_RSS, HKEY("rdf"));
1592                 AddRSSEndHandler(ATOM_item_entry_end,      RSS_ATOM, HKEY("entry"));
1593
1594
1595 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
1596                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1597                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1598                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1599                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1600
1601 /* links to other feed generators... */
1602                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1603                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1604                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1605                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1606
1607                 KnownNameSpaces = NewHash(1, NULL);
1608                 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
1609                 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
1610                 Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
1611                 Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
1612                 Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
1613                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1614                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
1615                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
1616                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
1617                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1618                 Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
1619                 Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
1620                 Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
1621                 Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
1622                 Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
1623                 Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
1624                 Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
1625                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
1626                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1627                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1628                 Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
1629                 Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
1630                 Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
1631 #if 0
1632                 /* we don't like these namespaces because of they shadow our usefull parameters. */
1633                 Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
1634 #endif
1635         }
1636         return "rssclient";
1637 }