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