citadel/modules/rssclient/serv_rssclient.c

   1 /*
   2  * Bring external RSS feeds into rooms.
   3  *
   4  * Copyright (c) 2007-2010 by the citadel.org team
   5  *
   6  * This program is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License as published by
   8  * the Free Software Foundation; either version 3 of the License, or
   9  * (at your option) any later version.
  10  *
  11  * This program is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  * GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * along with this program; if not, write to the Free Software
  18  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  19  */
  20
  21 #include <stdlib.h>
  22 #include <unistd.h>
  23 #include <stdio.h>
  24
  25 #if TIME_WITH_SYS_TIME
  26 # include <sys/time.h>
  27 # include <time.h>
  28 #else
  29 # if HAVE_SYS_TIME_H
  30 #  include <sys/time.h>
  31 # else
  32 #  include <time.h>
  33 # endif
  34 #endif
  35
  36 #include <ctype.h>
  37 #include <string.h>
  38 #include <errno.h>
  39 #include <sys/types.h>
  40 #include <sys/stat.h>
  41 #include <expat.h>
  42 #include <curl/curl.h>
  43 #include <libcitadel.h>
  44 #include "citadel.h"
  45 #include "server.h"
  46 #include "citserver.h"
  47 #include "support.h"
  48 #include "config.h"
  49 #include "threads.h"
  50 #include "ctdl_module.h"
  51 #include "clientsocket.h"
  52 #include "msgbase.h"
  53 #include "parsedate.h"
  54 #include "database.h"
  55 #include "citadel_dirs.h"
  56 #include "md5.h"
  57 #include "context.h"
  58
  59
  60 typedef struct rssnetcfg rssnetcfg;
  61 struct rssnetcfg {
  62         rssnetcfg *next;
  63         char url[256];
  64         char *rooms;
  65         time_t last_error_when;
  66         int ItemType;
  67         time_t next_poll;
  68 };
  69
  70 #define RSS_UNSET       (1<<0)
  71 #define RSS_RSS         (1<<1)
  72 #define RSS_ATOM        (1<<2)
  73 #define RSS_REQUIRE_BUF (1<<3)
  74
  75 typedef struct _rss_item {
  76         char *roomlist;
  77         int done_parsing;
  78         StrBuf *guid;
  79         StrBuf *title;
  80         StrBuf *link;
  81         StrBuf *linkTitle;
  82         StrBuf *reLink;
  83         StrBuf *reLinkTitle;
  84         StrBuf *description;
  85         time_t pubdate;
  86         StrBuf *channel_title;
  87         int item_tag_nesting;
  88         StrBuf *author_or_creator;
  89         StrBuf *author_url;
  90         StrBuf *author_email;
  91 }rss_item;
  92
  93
  94 typedef void (*rss_handler_func)(StrBuf *CData,
  95                                  rss_item *ri,
  96                                  rssnetcfg *Cfg,
  97                                  const char** Attr);
  98
  99 typedef struct __rss_xml_handler {
 100         int Flags;
 101         rss_handler_func Handler;
 102 }rss_xml_handler;
 103
 104
 105 typedef struct _rsscollection {
 106         StrBuf *CData;
 107         StrBuf *Key;
 108
 109         rss_item *Item;
 110         rssnetcfg *Cfg;
 111
 112         rss_xml_handler *Current;
 113 } rsscollection;
 114
 115 struct rssnetcfg *rnclist = NULL;
 116 HashList *StartHandlers = NULL;
 117 HashList *EndHandlers = NULL;
 118 HashList *KnownNameSpaces = NULL;
 119 HashList *UrlShorteners = NULL;
 120 void AddRSSStartHandler(rss_handler_func Handler, int Flags, const char *key, long len)
 121 {
 122         rss_xml_handler *h;
 123         h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
 124         h->Flags = Flags;
 125         h->Handler = Handler;
 126         Put(StartHandlers, key, len, h, NULL);
 127 }
 128 void AddRSSEndHandler(rss_handler_func Handler, int Flags, const char *key, long len)
 129 {
 130         rss_xml_handler *h;
 131         h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
 132         h->Flags = Flags;
 133         h->Handler = Handler;
 134         Put(EndHandlers, key, len, h, NULL);
 135 }
 136
 137 #if 0
 138 //#ifdef HAVE_ICONV
 139 #include <iconv.h>
 140
 141
 142 /*
 143  * dug this out of the trashcan of the midgard project, lets see if it works for us.
 144  * original code by Alexander Bokovoy <bokovoy@avilink.ne> distributed under GPL V2 or later
 145  */
 146
 147 /* Returns:
 148  >= 0 - successfull, 0 means conversion doesn't use multibyte sequences
 149    -1 - error during iconv_open call
 150    -2 - error during iconv_close call
 151    ----------------------------------
 152    This function expects that multibyte encoding in 'charset' wouldn't have
 153    characters with more than 3 bytes. It is not intended to convert UTF-8 because
 154    we'll never receive UTF-8 in our handler (it is handled by Exat itself).
 155 */
 156 static int
 157 fill_encoding_info (const char *charset, XML_Encoding * info)
 158 {
 159   iconv_t cd = (iconv_t)(-1);
 160   int flag;
 161         syslog(LOG_EMERG, "RSS: fill encoding info ...\n");
 162
 163 #if G_BYTE_ORDER == G_LITTLE_ENDIAN
 164   cd = iconv_open ("UCS-2LE", charset);
 165 #else
 166   cd = iconv_open ("UCS-2BE", charset);
 167 #endif
 168
 169   if (cd == (iconv_t) (-1))
 170     {
 171       return -1;
 172     }
 173
 174   {
 175     unsigned short out = 0;
 176     unsigned char buf[4];
 177     unsigned int i0, i1, i2;
 178     int result = 0;
 179     flag = 0;
 180     for (i0 = 0; i0 < 0x100; i0++)
 181       {
 182         buf[0] = i0;
 183         info->map[i0] = 0;
 184         //result = try (cd, buf, 1, &out);
 185         if (result < 0)
 186           {
 187           }
 188         else if (result > 0)
 189           {
 190             info->map[i0] = out;
 191           }
 192         else
 193           {
 194             for (i1 = 0; i1 < 0x100; i1++)
 195               {
 196                 buf[1] = i1;
 197                 ///result = try (cd, buf, 2, &out);
 198                 if (result < 0)
 199                   {
 200                   }
 201                 else if (result > 0)
 202                   {
 203                     flag++;
 204                     info->map[i0] = -2;
 205                   }
 206                 else
 207                   {
 208                     for (i2 = 0; i2 < 0x100; i2++)
 209                       {
 210                         buf[2] = i2;
 211                         ////result = try (cd, buf, 3, &out);
 212                         if (result < 0)
 213                           {
 214                           }
 215                         else if (result > 0)
 216                           {
 217                             flag++;
 218                             info->map[i0] = -3;
 219                           }
 220                       }
 221                   }
 222               }
 223           }
 224       }
 225   }
 226
 227   if (iconv_close (cd) < 0)
 228     {
 229       return -2;
 230     }
 231   return flag;
 232 }
 233
 234 static int
 235 iconv_convertor (void *data, const char *s)
 236 {
 237   XML_Encoding *info = data;
 238   int res;
 239         syslog(LOG_EMERG, "RSS: Converting ...\n");
 240
 241   if (s == NULL)
 242     return -1;
 243 /*
 244   GByteArray *result;
 245   result = g_byte_array_new ();
 246   if (process_block (info->data, (char *) s, strlen (s), result) == 0)
 247     {
 248       res = *(result->data);
 249       g_byte_array_free (result, TRUE);
 250       return res;
 251     }
 252   g_byte_array_free (result, TRUE);
 253 */
 254   return -1;
 255 }
 256
 257 static void
 258 my_release (void *data)
 259 {
 260   iconv_t cd = (iconv_t) data;
 261   if (iconv_close (cd) != 0)
 262     {
 263 /// TODO: uh no.      exit (1);
 264     }
 265 }
 266 int
 267 handle_unknown_xml_encoding (void *encodingHandleData,
 268                              const XML_Char * name,
 269                              XML_Encoding * info)
 270 {
 271   int result;
 272   syslog(LOG_EMERG, "RSS: unknown encoding ...\n");
 273   result = fill_encoding_info (name, info);
 274   if (result >= 0)
 275     {
 276       /*
 277         Special case: client asked for reverse conversion, we'll provide him with
 278         iconv descriptor which handles it. Client should release it by himself.
 279       */
 280       if(encodingHandleData != NULL)
 281             *((iconv_t *)encodingHandleData) = iconv_open(name, "UTF-8");
 282       /*
 283          Optimization: we do not need conversion function if encoding is one-to-one,
 284          info->map table will be enough
 285        */
 286       if (result == 0)
 287         {
 288           info->data = NULL;
 289           info->convert = NULL;
 290           info->release = NULL;
 291           return 1;
 292         }
 293       /*
 294          We do need conversion function because this encoding uses multibyte sequences
 295        */
 296       info->data = (void *) iconv_open ("UTF-8", name);
 297       if ((int)info->data == -1)
 298         return -1;
 299       info->convert = iconv_convertor;
 300       info->release = my_release;
 301       return 1;
 302     }
 303   if(encodingHandleData != NULL)
 304     *(iconv_t *)encodingHandleData = NULL;
 305   return 0;
 306 }
 307
 308 ///#endif
 309 #endif
 310 size_t GetLocationString( void *ptr, size_t size, size_t nmemb, void *userdata)
 311 {
 312 #define LOCATION "location"
 313         if (strncasecmp((char*)ptr, LOCATION, sizeof(LOCATION) - 1) == 0)
 314         {
 315                 StrBuf *pURL = (StrBuf*) userdata;
 316                 char *pch = (char*) ptr;
 317                 char *pche;
 318
 319                 pche = pch + (size * nmemb);
 320                 pch += sizeof(LOCATION);
 321
 322                 while (isspace(*pch) || (*pch == ':'))
 323                         pch ++;
 324
 325                 while (isspace(*pche) || (*pche == '\0'))
 326                         pche--;
 327
 328                 FlushStrBuf(pURL);
 329                 StrBufPlain(pURL, pch, pche - pch + 1);
 330         }
 331         return size * nmemb;
 332 }
 333
 334 int LookupUrl(StrBuf *ShorterUrlStr)
 335 {
 336         CURL *curl;
 337         char errmsg[1024] = "";
 338         StrBuf *Answer;
 339         int rc = 0;
 340
 341         curl = curl_easy_init();
 342         if (!curl) {
 343                 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
 344                 return 0;
 345         }
 346         Answer = NewStrBufPlain(NULL, SIZ);
 347
 348         curl_easy_setopt(curl, CURLOPT_URL, ChrPtr(ShorterUrlStr));
 349         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
 350         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
 351         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
 352 //      curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
 353         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
 354         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
 355         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
 356 #ifdef CURLOPT_HTTP_CONTENT_DECODING
 357         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
 358         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
 359 #endif
 360         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
 361         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
 362         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 0);
 363
 364         curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION , GetLocationString);
 365         curl_easy_setopt(curl, CURLOPT_WRITEHEADER, ShorterUrlStr);
 366
 367
 368         if (
 369                 (!IsEmptyStr(config.c_ip_addr))
 370                 && (strcmp(config.c_ip_addr, "*"))
 371                 && (strcmp(config.c_ip_addr, "::"))
 372                 && (strcmp(config.c_ip_addr, "0.0.0.0"))
 373         ) {
 374                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
 375         }
 376
 377         if (server_shutting_down)
 378                 goto shutdown ;
 379
 380         rc = curl_easy_perform(curl);
 381         if (rc) {
 382                 syslog(LOG_ALERT, "libcurl error %d: %s\n", rc, errmsg);
 383                 rc = 0;
 384         }
 385         else
 386                 rc = 1;
 387
 388 shutdown:
 389         FreeStrBuf(&Answer);
 390         curl_easy_cleanup(curl);
 391
 392         return rc;
 393
 394 }
 395
 396
 397
 398 void CrawlMessageForShorterUrls(HashList *pUrls, StrBuf *Message)
 399 {
 400         int nHits = 0;
 401         void *pv;
 402         int nShorter = 0;
 403         const char *pch;
 404         const char *pUrl;
 405         ConstStr *pCUrl;
 406
 407         while (GetHash(UrlShorteners, IKEY(nShorter), &pv))
 408         {
 409                 nShorter++;
 410                 pch = ChrPtr(Message);
 411                 pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 412                 while ((pUrl != NULL) && (nHits < 99))
 413                 {
 414                         pCUrl = malloc(sizeof(ConstStr));
 415
 416                         pCUrl->Key = pUrl;
 417                         pch = pUrl + StrLength((StrBuf*)pv);
 418                         while (isalnum(*pch)||(*pch == '-')||(*pch == '/'))
 419                                 pch++;
 420                         pCUrl->len = pch - pCUrl->Key;
 421
 422                         Put(pUrls, IKEY(nHits), pCUrl, NULL);
 423                         nHits ++;
 424                         pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 425                 }
 426         }
 427 }
 428
 429 int SortConstStrByPosition(const void *Item1, const void *Item2)
 430 {
 431         const ConstStr *p1, *p2;
 432         p1 = (const ConstStr*) Item1;
 433         p2 = (const ConstStr*) Item2;
 434         if (p1->Key == p2->Key)
 435                 return 0;
 436         if (p1->Key > p2->Key)
 437                 return 1;
 438         return -1;
 439 }
 440
 441 void ExpandShortUrls(StrBuf *Message)
 442 {
 443         StrBuf *Shadow;
 444         HashList *pUrls;
 445         ConstStr *pCUrl;
 446         const char *pch;
 447         const char *pche;
 448
 449         /* we just suspect URL shorteners to be inside of feeds from twitter
 450          * or other short content messages, so don't crawl through real blogs.
 451          */
 452         if (StrLength(Message) > 500)
 453                 return;
 454
 455         pUrls = NewHash(1, Flathash);
 456         CrawlMessageForShorterUrls(pUrls, Message);
 457
 458         if (GetCount(pUrls) > 0)
 459         {
 460                 StrBuf *ShorterUrlStr;
 461                 HashPos *Pos;
 462                 const char *RetrKey;
 463                 void *pv;
 464                 long len;
 465
 466                 Shadow = NewStrBufPlain(NULL, StrLength(Message));
 467                 SortByPayload (pUrls, SortConstStrByPosition);
 468
 469                 ShorterUrlStr = NewStrBufPlain(NULL, StrLength(Message));
 470
 471                 pch = ChrPtr(Message);
 472                 pche = pch + StrLength(Message);
 473                 Pos = GetNewHashPos(pUrls, 1);
 474                 while (GetNextHashPos(pUrls, Pos, &len, &RetrKey, &pv))
 475                 {
 476                         pCUrl = (ConstStr*) pv;
 477
 478                         if (pch != pCUrl->Key)
 479                                 StrBufAppendBufPlain(Shadow, pch, pCUrl->Key - pch, 0);
 480
 481                         StrBufPlain(ShorterUrlStr, CKEY(*pCUrl));
 482                         if (LookupUrl(ShorterUrlStr))
 483                         {
 484                                 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 485                                 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 486                                 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 487                                 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 488                                 StrBufAppendBufPlain(Shadow, HKEY("["), 0);
 489                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 490                                 StrBufAppendBufPlain(Shadow, HKEY("]</a>"), 0);
 491                         }
 492                         else
 493                         {
 494                                 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 495                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 496                                 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 497                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 498                                 StrBufAppendBufPlain(Shadow, HKEY("</a>"), 0);
 499                         }
 500                         pch = pCUrl->Key + pCUrl->len + 1;
 501
 502                 }
 503                 if (pch < pche)
 504                         StrBufAppendBufPlain(Shadow, pch, pche - pch, 0);
 505                 FlushStrBuf(Message);
 506                 StrBufAppendBuf(Message, Shadow, 0);
 507
 508                 FreeStrBuf(&ShorterUrlStr);
 509                 FreeStrBuf(&Shadow);
 510                 DeleteHashPos(&Pos);
 511         }
 512
 513         DeleteHash(&pUrls);
 514 }
 515
 516
 517 void AppendLink(StrBuf *Message, StrBuf *link, StrBuf *LinkTitle, const char *Title)
 518 {
 519         if (StrLength(link) > 0)
 520         {
 521                 StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
 522                 StrBufAppendBuf(Message, link, 0);
 523                 StrBufAppendBufPlain(Message, HKEY("\">"), 0);
 524                 if (StrLength(LinkTitle) > 0)
 525                         StrBufAppendBuf(Message, LinkTitle, 0);
 526                 else if ((Title != NULL) && !IsEmptyStr(Title))
 527                         StrBufAppendBufPlain(Message, Title, -1, 0);
 528                 else
 529                         StrBufAppendBuf(Message, link, 0);
 530                 StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
 531         }
 532 }
 533 /*
 534  * Commit a fetched and parsed RSS item to disk
 535  */
 536 void rss_save_item(rss_item *ri)
 537 {
 538
 539         struct MD5Context md5context;
 540         u_char rawdigest[MD5_DIGEST_LEN];
 541         int i;
 542         char utmsgid[SIZ];
 543         struct cdbdata *cdbut;
 544         struct UseTable ut;
 545         struct CtdlMessage *msg;
 546         struct recptypes *recp = NULL;
 547         int msglen = 0;
 548         StrBuf *Message;
 549
 550         recp = (struct recptypes *) malloc(sizeof(struct recptypes));
 551         if (recp == NULL) return;
 552         memset(recp, 0, sizeof(struct recptypes));
 553         memset(&ut, 0, sizeof(struct UseTable));
 554         recp->recp_room = strdup(ri->roomlist);
 555         recp->num_room = num_tokens(ri->roomlist, '|');
 556         recp->recptypes_magic = RECPTYPES_MAGIC;
 557
 558         /* Construct a GUID to use in the S_USETABLE table.
 559          * If one is not present in the item itself, make one up.
 560          */
 561         if (ri->guid != NULL) {
 562                 StrBufSpaceToBlank(ri->guid);
 563                 StrBufTrim(ri->guid);
 564                 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ChrPtr(ri->guid));
 565         }
 566         else {
 567                 MD5Init(&md5context);
 568                 if (ri->title != NULL) {
 569                         MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->title), StrLength(ri->title));
 570                 }
 571                 if (ri->link != NULL) {
 572                         MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->link), StrLength(ri->link));
 573                 }
 574                 MD5Final(rawdigest, &md5context);
 575                 for (i=0; i<MD5_DIGEST_LEN; i++) {
 576                         sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
 577                         utmsgid[i*2] = tolower(utmsgid[i*2]);
 578                         utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
 579                 }
 580                 strcat(utmsgid, "_rss2ctdl");
 581         }
 582
 583         /* Find out if we've already seen this item */
 584
 585         cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
 586 #ifndef DEBUG_RSS
 587         if (cdbut != NULL) {
 588                 /* Item has already been seen */
 589                 syslog(LOG_DEBUG, "%s has already been seen\n", utmsgid);
 590                 cdb_free(cdbut);
 591
 592                 /* rewrite the record anyway, to update the timestamp */
 593                 strcpy(ut.ut_msgid, utmsgid);
 594                 ut.ut_timestamp = time(NULL);
 595                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 596         }
 597         else
 598 #endif
 599 {
 600                 /* Item has not been seen, so save it. */
 601                 syslog(LOG_DEBUG, "RSS: saving item...\n");
 602                 if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY(""));
 603                 StrBufSpaceToBlank(ri->description);
 604                 msg = malloc(sizeof(struct CtdlMessage));
 605                 memset(msg, 0, sizeof(struct CtdlMessage));
 606                 msg->cm_magic = CTDLMESSAGE_MAGIC;
 607                 msg->cm_anon_type = MES_NORMAL;
 608                 msg->cm_format_type = FMT_RFC822;
 609
 610                 if (ri->guid != NULL) {
 611                         msg->cm_fields['E'] = strdup(ChrPtr(ri->guid));
 612                 }
 613
 614                 if (ri->author_or_creator != NULL) {
 615                         char *From;
 616                         StrBuf *Encoded = NULL;
 617                         int FromAt;
 618
 619                         From = html_to_ascii(ChrPtr(ri->author_or_creator),
 620                                              StrLength(ri->author_or_creator),
 621                                              512, 0);
 622                         StrBufPlain(ri->author_or_creator, From, -1);
 623                         StrBufTrim(ri->author_or_creator);
 624                         free(From);
 625
 626                         FromAt = strchr(ChrPtr(ri->author_or_creator), '@') != NULL;
 627                         if (!FromAt && StrLength (ri->author_email) > 0)
 628                         {
 629                                 StrBufRFC2047encode(&Encoded, ri->author_or_creator);
 630                                 msg->cm_fields['A'] = SmashStrBuf(&Encoded);
 631                                 msg->cm_fields['P'] = SmashStrBuf(&ri->author_email);
 632                         }
 633                         else
 634                         {
 635                                 if (FromAt)
 636                                 {
 637                                         msg->cm_fields['A'] = SmashStrBuf(&ri->author_or_creator);
 638                                         msg->cm_fields['P'] = strdup(msg->cm_fields['A']);
 639                                 }
 640                                 else
 641                                 {
 642                                         StrBufRFC2047encode(&Encoded, ri->author_or_creator);
 643                                         msg->cm_fields['A'] = SmashStrBuf(&Encoded);
 644                                         msg->cm_fields['P'] = strdup("rss@localhost");
 645                                 }
 646                         }
 647                 }
 648                 else {
 649                         msg->cm_fields['A'] = strdup("rss");
 650                 }
 651
 652                 msg->cm_fields['N'] = strdup(NODENAME);
 653                 if (ri->title != NULL) {
 654                         long len;
 655                         char *Sbj;
 656                         StrBuf *Encoded, *QPEncoded;
 657
 658                         QPEncoded = NULL;
 659                         StrBufSpaceToBlank(ri->title);
 660                         len = StrLength(ri->title);
 661                         Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
 662                         len = strlen(Sbj);
 663                         if (Sbj[len - 1] == '\n')
 664                         {
 665                                 len --;
 666                                 Sbj[len] = '\0';
 667                         }
 668                         Encoded = NewStrBufPlain(Sbj, len);
 669                         free(Sbj);
 670
 671                         StrBufTrim(Encoded);
 672                         StrBufRFC2047encode(&QPEncoded, Encoded);
 673
 674                         msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
 675                         FreeStrBuf(&Encoded);
 676                 }
 677
 678                 if (ri->pubdate <= 0) {
 679                         ri->pubdate = time(NULL);
 680                 }
 681                 msg->cm_fields['T'] = malloc(64);
 682                 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
 683
 684                 if (ri->channel_title != NULL) {
 685                         if (StrLength(ri->channel_title) > 0) {
 686                                 msg->cm_fields['O'] = strdup(ChrPtr(ri->channel_title));
 687                         }
 688                 }
 689                 if (ri->link == NULL)
 690                         ri->link = NewStrBufPlain(HKEY(""));
 691                 ExpandShortUrls(ri->description);
 692                 msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
 693
 694                 Message = NewStrBufPlain(NULL, StrLength(ri->description));
 695
 696                 StrBufPlain(Message, HKEY(
 697                          "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 698                          "<html><body>\n"));
 699
 700                 StrBufAppendBuf(Message, ri->description, 0);
 701                 StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
 702
 703                 AppendLink(Message, ri->link, ri->linkTitle, NULL);
 704                 AppendLink(Message, ri->reLink, ri->reLinkTitle, "Reply to this");
 705                 StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
 706
 707                 msg->cm_fields['M'] = SmashStrBuf(&Message);
 708
 709                 CtdlSubmitMsg(msg, recp, NULL, 0);
 710                 CtdlFreeMessage(msg);
 711
 712                 /* write the uidl to the use table so we don't store this item again */
 713                 strcpy(ut.ut_msgid, utmsgid);
 714                 ut.ut_timestamp = time(NULL);
 715                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 716         }
 717         free_recipients(recp);
 718 }
 719
 720
 721
 722 /*
 723  * Convert an RDF/RSS datestamp into a time_t
 724  */
 725 time_t rdf_parsedate(const char *p)
 726 {
 727         struct tm tm;
 728         time_t t = 0;
 729
 730         if (!p) return 0L;
 731         if (strlen(p) < 10) return 0L;
 732
 733         memset(&tm, 0, sizeof tm);
 734
 735         /*
 736          * If the timestamp appears to be in W3C datetime format, try to
 737          * parse it.  See also: http://www.w3.org/TR/NOTE-datetime
 738          *
 739          * This code, along with parsedate.c, is a potential candidate for
 740          * moving into libcitadel.
 741          */
 742         if ( (p[4] == '-') && (p[7] == '-') ) {
 743                 tm.tm_year = atoi(&p[0]) - 1900;
 744                 tm.tm_mon = atoi(&p[5]) - 1;
 745                 tm.tm_mday = atoi(&p[8]);
 746                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 747                         tm.tm_hour = atoi(&p[11]);
 748                         tm.tm_min = atoi(&p[14]);
 749                 }
 750                 return mktime(&tm);
 751         }
 752
 753         /* hmm... try RFC822 date stamp format */
 754
 755         t = parsedate(p);
 756         if (t > 0) return(t);
 757
 758         /* yeesh.  ok, just return the current date and time. */
 759         return(time(NULL));
 760 }
 761
 762 void flush_rss_item(rss_item *ri)
 763 {
 764         /* Initialize the feed item data structure */
 765         FreeStrBuf(&ri->guid);
 766         FreeStrBuf(&ri->title);
 767         FreeStrBuf(&ri->link);
 768         FreeStrBuf(&ri->linkTitle);
 769         FreeStrBuf(&ri->reLink);
 770         FreeStrBuf(&ri->reLinkTitle);
 771         FreeStrBuf(&ri->description);
 772         FreeStrBuf(&ri->channel_title);
 773         FreeStrBuf(&ri->author_or_creator);
 774         FreeStrBuf(&ri->author_url);
 775         FreeStrBuf(&ri->author_email);
 776 }
 777
 778 void rss_xml_start(void *data, const char *supplied_el, const char **attr)
 779 {
 780         rss_xml_handler *h;
 781         rsscollection   *rssc = (rsscollection*) data;
 782         rssnetcfg       *Cfg = rssc->Cfg;
 783         rss_item        *ri = rssc->Item;
 784         void            *pv;
 785         const char      *pel;
 786         char            *sep = NULL;
 787
 788         /* Axe the namespace, we don't care about it */
 789 ///     syslog(LOG_EMERG, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
 790         pel = supplied_el;
 791         while (sep = strchr(pel, ':'), sep) {
 792                 pel = sep + 1;
 793         }
 794
 795         if (pel != supplied_el)
 796         {
 797                 void *v;
 798
 799                 if (!GetHash(KnownNameSpaces,
 800                              supplied_el,
 801                              pel - supplied_el - 1,
 802                              &v))
 803                 {
 804 #ifdef DEBUG_RSS
 805                         syslog(LOG_EMERG, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
 806                                       supplied_el);
 807 #endif
 808                         return;
 809                 }
 810         }
 811
 812         StrBufPlain(rssc->Key, pel, -1);
 813         StrBufLowerCase(rssc->Key);
 814         if (GetHash(StartHandlers, SKEY(rssc->Key), &pv))
 815         {
 816                 rssc->Current = h = (rss_xml_handler*) pv;
 817
 818                 if (((h->Flags & RSS_UNSET) != 0) &&
 819                     (Cfg->ItemType == RSS_UNSET))
 820                 {
 821                         h->Handler(rssc->CData, ri, Cfg, attr);
 822                 }
 823                 else if (((h->Flags & RSS_RSS) != 0) &&
 824                     (Cfg->ItemType == RSS_RSS))
 825                 {
 826                         h->Handler(rssc->CData, ri, Cfg, attr);
 827                 }
 828                 else if (((h->Flags & RSS_ATOM) != 0) &&
 829                          (Cfg->ItemType == RSS_ATOM))
 830                 {
 831                         h->Handler(rssc->CData, ri, Cfg, attr);
 832                 }
 833 #ifdef DEBUG_RSS
 834                 else
 835                         syslog(LOG_EMERG, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
 836 #endif
 837         }
 838 #ifdef DEBUG_RSS
 839         else
 840                 syslog(LOG_EMERG, "RSS: START unhandled: [%s] [%s]...\n", pel,  supplied_el);
 841 #endif
 842 }
 843
 844 void rss_xml_end(void *data, const char *supplied_el)
 845 {
 846         rss_xml_handler *h;
 847         rsscollection   *rssc = (rsscollection*) data;
 848         rssnetcfg       *Cfg = rssc->Cfg;
 849         rss_item        *ri = rssc->Item;
 850         const char      *pel;
 851         char            *sep = NULL;
 852         void            *pv;
 853
 854         /* Axe the namespace, we don't care about it */
 855         pel = supplied_el;
 856         while (sep = strchr(pel, ':'), sep) {
 857                 pel = sep + 1;
 858         }
 859 //      syslog(LOG_EMERG, "RSS: END %s...\n", el);
 860         if (pel != supplied_el)
 861         {
 862                 void *v;
 863
 864                 if (!GetHash(KnownNameSpaces,
 865                              supplied_el,
 866                              pel - supplied_el - 1,
 867                              &v))
 868                 {
 869 #ifdef DEBUG_RSS
 870                         syslog(LOG_EMERG, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
 871                                       supplied_el, ChrPtr(rssc->CData));
 872 #endif
 873                         FlushStrBuf(rssc->CData);
 874                         return;
 875                 }
 876         }
 877
 878         StrBufPlain(rssc->Key, pel, -1);
 879         StrBufLowerCase(rssc->Key);
 880         if (GetHash(EndHandlers, SKEY(rssc->Key), &pv))
 881         {
 882                 h = (rss_xml_handler*) pv;
 883
 884                 if (((h->Flags & RSS_UNSET) != 0) &&
 885                     (Cfg->ItemType == RSS_UNSET))
 886                 {
 887                         h->Handler(rssc->CData, ri, Cfg, NULL);
 888                 }
 889                 else if (((h->Flags & RSS_RSS) != 0) &&
 890                     (Cfg->ItemType == RSS_RSS))
 891                 {
 892                         h->Handler(rssc->CData, ri, Cfg, NULL);
 893                 }
 894                 else if (((h->Flags & RSS_ATOM) != 0) &&
 895                          (Cfg->ItemType == RSS_ATOM))
 896                 {
 897                         h->Handler(rssc->CData, ri, Cfg, NULL);
 898                 }
 899 #ifdef DEBUG_RSS
 900                 else
 901                         syslog(LOG_EMERG, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 902 #endif
 903         }
 904 #ifdef DEBUG_RSS
 905         else
 906                 syslog(LOG_EMERG, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 907 #endif
 908         FlushStrBuf(rssc->CData);
 909         rssc->Current = NULL;
 910 }
 911
 912
 913
 914
 915
 916 void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 917 {
 918         syslog(LOG_DEBUG, "RSS: This is an RSS feed.\n");
 919         Cfg->ItemType = RSS_RSS;
 920 }
 921
 922 void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 923 {
 924         syslog(LOG_DEBUG, "RSS: This is an RDF feed.\n");
 925         Cfg->ItemType = RSS_RSS;
 926 }
 927
 928 void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 929 {
 930         syslog(LOG_DEBUG, "RSS: This is an ATOM feed.\n");
 931         Cfg->ItemType = RSS_ATOM;
 932 }
 933
 934
 935 void RSS_item_item_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 936 {
 937         ri->item_tag_nesting ++;
 938         flush_rss_item(ri);
 939 }
 940
 941 void ATOM_item_entry_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 942 {
 943 /* Atom feed... */
 944         ri->item_tag_nesting ++;
 945         flush_rss_item(ri);
 946 }
 947
 948 void ATOM_item_link_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 949 {
 950         int i;
 951         const char *pHref = NULL;
 952         const char *pType = NULL;
 953         const char *pRel = NULL;
 954         const char *pTitle = NULL;
 955
 956         for (i = 0; Attr[i] != NULL; i+=2)
 957         {
 958                 if (!strcmp(Attr[i], "href"))
 959                 {
 960                         pHref = Attr[i+1];
 961                 }
 962                 else if (!strcmp(Attr[i], "rel"))
 963                 {
 964                         pRel = Attr[i+1];
 965                 }
 966                 else if (!strcmp(Attr[i], "type"))
 967                 {
 968                         pType = Attr[i+1];
 969                 }
 970                 else if (!strcmp(Attr[i], "title"))
 971                 {
 972                         pTitle = Attr[i+1];
 973                 }
 974         }
 975         if (pHref == NULL)
 976                 return; /* WHUT? Pointing... where? */
 977         if ((pType != NULL) && !strcasecmp(pType, "application/atom+xml"))
 978                 return; /* these just point to other rss resources, we're not interested in them. */
 979         if (pRel != NULL)
 980         {
 981                 if (!strcasecmp (pRel, "replies"))
 982                 {
 983                         NewStrBufDupAppendFlush(&ri->reLink, NULL, pHref, -1);
 984                         StrBufTrim(ri->link);
 985                         NewStrBufDupAppendFlush(&ri->reLinkTitle, NULL, pTitle, -1);
 986                 }
 987                 else if (!strcasecmp(pRel, "alternate")) /* Alternative representation of this Item... */
 988                 {
 989                         NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 990                         StrBufTrim(ri->link);
 991                         NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
 992
 993                 }
 994 #if 0 /* these are also defined, but dunno what to do with them.. */
 995                 else if (!strcasecmp(pRel, "related"))
 996                 {
 997                 }
 998                 else if (!strcasecmp(pRel, "self"))
 999                 {
1000                 }
1001                 else if (!strcasecmp(pRel, "enclosure"))
1002                 {/* this reference can get big, and is probably the full article... */
1003                 }
1004                 else if (!strcasecmp(pRel, "via"))
1005                 {/* this article was provided via... */
1006                 }
1007 #endif
1008         }
1009         else if (StrLength(ri->link) == 0)
1010         {
1011                 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
1012                 StrBufTrim(ri->link);
1013                 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
1014         }
1015 }
1016
1017
1018
1019
1020 void ATOMRSS_item_title_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1021 {
1022         if ((ri->item_tag_nesting == 0) && (StrLength(CData) > 0)) {
1023                 NewStrBufDupAppendFlush(&ri->channel_title, CData, NULL, 0);
1024                 StrBufTrim(ri->channel_title);
1025         }
1026 }
1027
1028 void RSS_item_guid_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1029 {
1030         if (StrLength(CData) > 0) {
1031                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
1032         }
1033 }
1034
1035 void ATOM_item_id_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1036 {
1037         if (StrLength(CData) > 0) {
1038                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
1039         }
1040 }
1041
1042
1043 void RSS_item_link_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1044 {
1045         if (StrLength(CData) > 0) {
1046                 NewStrBufDupAppendFlush(&ri->link, CData, NULL, 0);
1047                 StrBufTrim(ri->link);
1048         }
1049 }
1050 void RSS_item_relink_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1051 {
1052         if (StrLength(CData) > 0) {
1053                 NewStrBufDupAppendFlush(&ri->reLink, CData, NULL, 0);
1054                 StrBufTrim(ri->reLink);
1055         }
1056 }
1057
1058 void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1059 {
1060         if (StrLength(CData) > 0) {
1061                 NewStrBufDupAppendFlush(&ri->title, CData, NULL, 0);
1062                 StrBufTrim(ri->title);
1063         }
1064 }
1065
1066 void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1067 {
1068         long olen = StrLength (ri->description);
1069         long clen = StrLength (CData);
1070         if (clen > 0)
1071         {
1072                 if (olen == 0) {
1073                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1074                         StrBufTrim(ri->description);
1075                 }
1076                 else if (olen < clen) {
1077                         FlushStrBuf(ri->description);
1078                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1079                         StrBufTrim(ri->description);
1080                 }
1081         }
1082 }
1083 void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1084 {
1085         /* this can contain an abstract of the article. but we don't want to verwrite a full document if we already have it. */
1086         if ((StrLength(CData) > 0) && (StrLength(ri->description) == 0))
1087         {
1088                 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1089                 StrBufTrim(ri->description);
1090         }
1091 }
1092
1093 void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1094 {
1095         long olen = StrLength (ri->description);
1096         long clen = StrLength (CData);
1097         if (clen > 0)
1098         {
1099                 if (olen == 0) {
1100                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1101                         StrBufTrim(ri->description);
1102                 }
1103                 else if (olen < clen) {
1104                         FlushStrBuf(ri->description);
1105                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1106                         StrBufTrim(ri->description);
1107                 }
1108         }
1109 }
1110
1111 void ATOM_item_published_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1112 {
1113         if (StrLength(CData) > 0) {
1114                 StrBufTrim(CData);
1115                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1116         }
1117 }
1118
1119 void ATOM_item_updated_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1120 {
1121         if (StrLength(CData) > 0) {
1122                 StrBufTrim(CData);
1123                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1124         }
1125 }
1126
1127 void RSS_item_pubdate_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1128 {
1129         if (StrLength(CData) > 0) {
1130                 StrBufTrim(CData);
1131                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1132         }
1133 }
1134
1135
1136 void RSS_item_date_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1137 {
1138         if (StrLength(CData) > 0) {
1139                 StrBufTrim(CData);
1140                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1141         }
1142 }
1143
1144
1145
1146 void RSS_item_author_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1147 {
1148         if (StrLength(CData) > 0) {
1149                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1150                 StrBufTrim(ri->author_or_creator);
1151         }
1152 }
1153
1154
1155 void ATOM_item_name_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1156 {
1157         if (StrLength(CData) > 0) {
1158                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1159                 StrBufTrim(ri->author_or_creator);
1160         }
1161 }
1162
1163 void ATOM_item_email_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1164 {
1165         if (StrLength(CData) > 0) {
1166                 NewStrBufDupAppendFlush(&ri->author_email, CData, NULL, 0);
1167                 StrBufTrim(ri->author_email);
1168         }
1169 }
1170
1171 void RSS_item_creator_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1172 {
1173         if ((StrLength(CData) > 0) &&
1174             (StrLength(ri->author_or_creator) == 0))
1175         {
1176                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1177                 StrBufTrim(ri->author_or_creator);
1178         }
1179 }
1180
1181
1182 void ATOM_item_uri_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1183 {
1184         if (StrLength(CData) > 0) {
1185                 NewStrBufDupAppendFlush(&ri->author_url, CData, NULL, 0);
1186                 StrBufTrim(ri->author_url);
1187         }
1188 }
1189
1190 void RSS_item_item_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1191 {
1192         --ri->item_tag_nesting;
1193         rss_save_item(ri);
1194 }
1195
1196
1197 void ATOM_item_entry_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1198 {
1199         --ri->item_tag_nesting;
1200         rss_save_item(ri);
1201 }
1202
1203 void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1204 {
1205 //              syslog(LOG_DEBUG, "End of feed detected.  Closing parser.\n");
1206         ri->done_parsing = 1;
1207
1208 }
1209 void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1210 {
1211 //              syslog(LOG_DEBUG, "End of feed detected.  Closing parser.\n");
1212         ri->done_parsing = 1;
1213 }
1214
1215
1216 void RSSATOM_item_ignore(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1217 {
1218 }
1219
1220
1221
1222 /*
1223  * This callback stores up the data which appears in between tags.
1224  */
1225 void rss_xml_cdata_start(void *data)
1226 {
1227         rsscollection *rssc = (rsscollection*) data;
1228
1229         FlushStrBuf(rssc->CData);
1230 }
1231
1232 void rss_xml_cdata_end(void *data)
1233 {
1234 }
1235 void rss_xml_chardata(void *data, const XML_Char *s, int len)
1236 {
1237         rsscollection *rssc = (rsscollection*) data;
1238
1239         StrBufAppendBufPlain (rssc->CData, s, len, 0);
1240 }
1241
1242 /*
1243  * Callback function for passing libcurl's output to expat for parsing
1244  */
1245 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
1246 {
1247         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
1248         return (size*nmemb);
1249 }
1250
1251
1252
1253 /*
1254  * Begin a feed parse
1255  */
1256 void rss_do_fetching(rssnetcfg *Cfg) {
1257         rsscollection rssc;
1258         rss_item ri;
1259         XML_Parser xp = NULL;
1260         StrBuf *Answer;
1261
1262         CURL *curl;
1263         CURLcode res;
1264         char errmsg[1024] = "";
1265         char *ptr;
1266         const char *at;
1267         long len;
1268
1269         time_t now;
1270
1271         now = time(NULL);
1272
1273         if ((Cfg->next_poll != 0) && (now < Cfg->next_poll))
1274                 return;
1275         memset(&ri, 0, sizeof(rss_item));
1276         rssc.Item = &ri;
1277         rssc.Cfg = Cfg;
1278
1279         syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
1280
1281         curl = curl_easy_init();
1282         if (!curl) {
1283                 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
1284                 return;
1285         }
1286         Answer = NewStrBufPlain(NULL, SIZ);
1287
1288         curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
1289         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
1290         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
1291         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
1292 //      curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
1293         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
1294         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
1295         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
1296 #ifdef CURLOPT_HTTP_CONTENT_DECODING
1297         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
1298         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
1299 #endif
1300         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
1301         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
1302         if (
1303                 (!IsEmptyStr(config.c_ip_addr))
1304                 && (strcmp(config.c_ip_addr, "*"))
1305                 && (strcmp(config.c_ip_addr, "::"))
1306                 && (strcmp(config.c_ip_addr, "0.0.0.0"))
1307         ) {
1308                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
1309         }
1310
1311         if (server_shutting_down)
1312         {
1313                 curl_easy_cleanup(curl);
1314                 return;
1315         }
1316
1317         if (server_shutting_down)
1318                 goto shutdown ;
1319
1320         res = curl_easy_perform(curl);
1321         if (res) {
1322                 syslog(LOG_ALERT, "libcurl error %d: %s\n", res, errmsg);
1323         }
1324
1325         if (server_shutting_down)
1326                 goto shutdown ;
1327
1328
1329
1330
1331         memset(&ri, 0, sizeof(rss_item));
1332         ri.roomlist = Cfg->rooms;
1333         rssc.CData = NewStrBufPlain(NULL, SIZ);
1334         rssc.Key = NewStrBuf();
1335         at = NULL;
1336         StrBufSipLine(rssc.Key, Answer, &at);
1337         ptr = NULL;
1338
1339 #define encoding "encoding=\""
1340         ptr = strstr(ChrPtr(rssc.Key), encoding);
1341         if (ptr != NULL)
1342         {
1343                 char *pche;
1344
1345                 ptr += sizeof (encoding) - 1;
1346                 pche = strchr(ptr, '"');
1347                 if (pche != NULL)
1348                         StrBufCutAt(rssc.Key, -1, pche);
1349                 else
1350                         ptr = "UTF-8";
1351         }
1352         else
1353                 ptr = "UTF-8";
1354
1355
1356         xp = XML_ParserCreateNS(ptr, ':');
1357         if (!xp) {
1358                 syslog(LOG_ALERT, "Cannot create XML parser!\n");
1359                 goto shutdown;
1360         }
1361         FlushStrBuf(rssc.Key);
1362 //#ifdef HAVE_ICONV
1363 #if 0
1364         XML_SetUnknownEncodingHandler(xp,
1365                                       handle_unknown_xml_encoding,
1366                                       &rssc);
1367 #endif
1368 //#endif
1369         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
1370         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
1371         XML_SetUserData(xp, &rssc);
1372         XML_SetCdataSectionHandler(xp,
1373                                    rss_xml_cdata_start,
1374                                    rss_xml_cdata_end);
1375
1376
1377         len = StrLength(Answer);
1378         ptr = SmashStrBuf(&Answer);
1379         XML_Parse(xp, ptr, len, 0);
1380         free (ptr);
1381         if (ri.done_parsing == 0)
1382                 XML_Parse(xp, "", 0, 1);
1383
1384
1385         syslog(LOG_ALERT, "RSS: XML Status [%s] \n",
1386                       XML_ErrorString(
1387                               XML_GetErrorCode(xp)));
1388
1389 shutdown:
1390         FreeStrBuf(&Answer);
1391         curl_easy_cleanup(curl);
1392         XML_ParserFree(xp);
1393
1394         flush_rss_item(&ri);
1395         FreeStrBuf(&rssc.CData);
1396         FreeStrBuf(&rssc.Key);
1397
1398         Cfg->next_poll = time(NULL) + config.c_net_freq;
1399 }
1400
1401
1402 /*
1403  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
1404  */
1405 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
1406 {
1407         char filename[PATH_MAX];
1408         char buf[1024];
1409         char instr[32];
1410         FILE *fp;
1411         char feedurl[256];
1412         rssnetcfg *rncptr = NULL;
1413         rssnetcfg *use_this_rncptr = NULL;
1414         int len = 0;
1415         char *ptr = NULL;
1416
1417         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
1418
1419         if (server_shutting_down)
1420                 return;
1421
1422         /* Only do net processing for rooms that have netconfigs */
1423         fp = fopen(filename, "r");
1424         if (fp == NULL) {
1425                 return;
1426         }
1427
1428         while (fgets(buf, sizeof buf, fp) != NULL && !server_shutting_down) {
1429                 buf[strlen(buf)-1] = 0;
1430
1431                 extract_token(instr, buf, 0, '|', sizeof instr);
1432                 if (!strcasecmp(instr, "rssclient")) {
1433
1434                         use_this_rncptr = NULL;
1435
1436                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
1437
1438                         /* If any other rooms have requested the same feed, then we will just add this
1439                          * room to the target list for that client request.
1440                          */
1441                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
1442                                 if (!strcmp(rncptr->url, feedurl)) {
1443                                         use_this_rncptr = rncptr;
1444                                 }
1445                         }
1446
1447                         /* Otherwise create a new client request */
1448                         if (use_this_rncptr == NULL) {
1449                                 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
1450                                 memset(rncptr, 0, sizeof(rssnetcfg));
1451                                 rncptr->ItemType = RSS_UNSET;
1452                                 if (rncptr != NULL) {
1453                                         rncptr->next = rnclist;
1454                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
1455                                         rncptr->rooms = NULL;
1456                                         rnclist = rncptr;
1457                                         use_this_rncptr = rncptr;
1458                                 }
1459                         }
1460
1461                         /* Add the room name to the request */
1462                         if (use_this_rncptr != NULL) {
1463                                 if (use_this_rncptr->rooms == NULL) {
1464                                         rncptr->rooms = strdup(qrbuf->QRname);
1465                                 }
1466                                 else {
1467                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
1468                                         ptr = realloc(use_this_rncptr->rooms, len);
1469                                         if (ptr != NULL) {
1470                                                 strcat(ptr, "|");
1471                                                 strcat(ptr, qrbuf->QRname);
1472                                                 use_this_rncptr->rooms = ptr;
1473                                         }
1474                                 }
1475                         }
1476                 }
1477
1478         }
1479
1480         fclose(fp);
1481
1482 }
1483
1484 /*
1485  * Scan for rooms that have RSS client requests configured
1486  */
1487 void rssclient_scan(void) {
1488         static time_t last_run = 0L;
1489         static int doing_rssclient = 0;
1490         rssnetcfg *rptr = NULL;
1491
1492         /* Run no more than once every 15 minutes. */
1493         if ((time(NULL) - last_run) < 900) {
1494                 return;
1495         }
1496
1497         /*
1498          * This is a simple concurrency check to make sure only one rssclient run
1499          * is done at a time.  We could do this with a mutex, but since we
1500          * don't really require extremely fine granularity here, we'll do it
1501          * with a static variable instead.
1502          */
1503         if (doing_rssclient) return;
1504         doing_rssclient = 1;
1505
1506         syslog(LOG_DEBUG, "rssclient started\n");
1507         CtdlForEachRoom(rssclient_scan_room, NULL);
1508
1509         while (rnclist != NULL && !server_shutting_down) {
1510                 rss_do_fetching(rnclist);
1511                 rptr = rnclist;
1512                 rnclist = rnclist->next;
1513                 if (rptr->rooms != NULL) free(rptr->rooms);
1514                 free(rptr);
1515         }
1516
1517         syslog(LOG_DEBUG, "rssclient ended\n");
1518         last_run = time(NULL);
1519         doing_rssclient = 0;
1520         return;
1521 }
1522
1523 void LoadUrlShorteners(void)
1524 {
1525         int i = 0;
1526         int fd;
1527         const char *POS = NULL;
1528         const char *Err = NULL;
1529         StrBuf *Content, *Line;
1530
1531
1532         UrlShorteners = NewHash(0, Flathash);
1533
1534         fd = open(file_citadel_urlshorteners, 0);
1535
1536         if (fd != 0)
1537         {
1538                 Content = NewStrBufPlain(NULL, SIZ);
1539                 Line = NewStrBuf();
1540                 while (POS != StrBufNOTNULL)
1541                 {
1542                         StrBufTCP_read_buffered_line_fast (Line, Content, &POS, &fd, 1, 1, &Err);
1543                         StrBufTrim(Line);
1544                         if ((*ChrPtr(Line) != '#') && (StrLength(Line) > 0))
1545                         {
1546                                 Put(UrlShorteners, IKEY(i), Line, HFreeStrBuf);
1547                                 i++;
1548                                 Line = NewStrBuf();
1549                         }
1550                         else
1551                                 FlushStrBuf(Line);
1552                         if (POS == NULL)
1553                                 POS = StrBufNOTNULL;
1554                 }
1555                 FreeStrBuf(&Line);
1556                 FreeStrBuf(&Content);
1557         }
1558         close(fd);
1559 }
1560
1561 void rss_cleanup(void)
1562 {
1563         DeleteHash(&StartHandlers);
1564         DeleteHash(&EndHandlers);
1565         DeleteHash(&UrlShorteners);
1566         DeleteHash(&KnownNameSpaces);
1567 }
1568
1569 CTDL_MODULE_INIT(rssclient)
1570 {
1571         if (threading)
1572         {
1573                 syslog(LOG_INFO, "%s\n", curl_version());
1574                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
1575         }
1576         else
1577         {
1578                 LoadUrlShorteners ();
1579
1580                 StartHandlers = NewHash(1, NULL);
1581                 EndHandlers = NewHash(1, NULL);
1582
1583                 AddRSSStartHandler(RSS_item_rss_start,     RSS_UNSET, HKEY("rss"));
1584                 AddRSSStartHandler(RSS_item_rdf_start,     RSS_UNSET, HKEY("rdf"));
1585                 AddRSSStartHandler(ATOM_item_feed_start,    RSS_UNSET, HKEY("feed"));
1586                 AddRSSStartHandler(RSS_item_item_start,    RSS_RSS, HKEY("item"));
1587                 AddRSSStartHandler(ATOM_item_entry_start,  RSS_ATOM, HKEY("entry"));
1588                 AddRSSStartHandler(ATOM_item_link_start,   RSS_ATOM, HKEY("link"));
1589
1590                 AddRSSEndHandler(ATOMRSS_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1591                 AddRSSEndHandler(RSS_item_guid_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
1592                 AddRSSEndHandler(ATOM_item_id_end,         RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
1593                 AddRSSEndHandler(RSS_item_link_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
1594 #if 0
1595 // hm, rss to the comments of that blog, might be interesting in future, but...
1596                 AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("commentrss"));
1597 // comment count...
1598                 AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("comments"));
1599 #endif
1600                 AddRSSEndHandler(RSSATOM_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1601                 AddRSSEndHandler(ATOM_item_content_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
1602                 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_ATOM|RSS_REQUIRE_BUF, HKEY("encoded"));
1603                 AddRSSEndHandler(ATOM_item_summary_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
1604                 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
1605                 AddRSSEndHandler(ATOM_item_published_end,  RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
1606                 AddRSSEndHandler(ATOM_item_updated_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
1607                 AddRSSEndHandler(RSS_item_pubdate_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
1608                 AddRSSEndHandler(RSS_item_date_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
1609                 AddRSSEndHandler(RSS_item_author_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
1610                 AddRSSEndHandler(RSS_item_creator_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
1611 /* <author> */
1612                 AddRSSEndHandler(ATOM_item_email_end,      RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
1613                 AddRSSEndHandler(ATOM_item_name_end,       RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
1614                 AddRSSEndHandler(ATOM_item_uri_end,        RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
1615 /* </author> */
1616                 AddRSSEndHandler(RSS_item_item_end,        RSS_RSS, HKEY("item"));
1617                 AddRSSEndHandler(RSS_item_rss_end,         RSS_RSS, HKEY("rss"));
1618                 AddRSSEndHandler(RSS_item_rdf_end,         RSS_RSS, HKEY("rdf"));
1619                 AddRSSEndHandler(ATOM_item_entry_end,      RSS_ATOM, HKEY("entry"));
1620
1621
1622 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
1623                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1624                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1625                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1626                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1627
1628 /* links to other feed generators... */
1629                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1630                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1631                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1632                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1633
1634                 KnownNameSpaces = NewHash(1, NULL);
1635                 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
1636                 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
1637                 Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
1638                 Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
1639                 Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
1640                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1641                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
1642                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
1643                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
1644                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1645                 Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
1646                 Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
1647                 Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
1648                 Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
1649                 Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
1650                 Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
1651                 Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
1652                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
1653                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1654                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1655                 Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
1656                 Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
1657                 Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
1658 #if 0
1659                 /* we don't like these namespaces because of they shadow our usefull parameters. */
1660                 Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
1661 #endif
1662                 CtdlRegisterCleanupHook(rss_cleanup);
1663         }
1664         return "rssclient";
1665 }