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         curl_easy_cleanup(curl);
 390
 391         return rc;
 392
 393 }
 394
 395
 396
 397 void CrawlMessageForShorterUrls(HashList *pUrls, StrBuf *Message)
 398 {
 399         int nHits = 0;
 400         void *pv;
 401         int nShorter = 0;
 402         const char *pch;
 403         const char *pUrl;
 404         ConstStr *pCUrl;
 405
 406         while (GetHash(UrlShorteners, IKEY(nShorter), &pv))
 407         {
 408                 nShorter++;
 409                 pch = ChrPtr(Message);
 410                 pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 411                 while ((pUrl != NULL) && (nHits < 99))
 412                 {
 413                         pCUrl = malloc(sizeof(ConstStr));
 414
 415                         pCUrl->Key = pUrl;
 416                         pch = pUrl + StrLength((StrBuf*)pv);
 417                         while (isalnum(*pch)||(*pch == '-')||(*pch == '/'))
 418                                 pch++;
 419                         pCUrl->len = pch - pCUrl->Key;
 420
 421                         Put(pUrls, IKEY(nHits), pCUrl, NULL);
 422                         nHits ++;
 423                         pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 424                 }
 425         }
 426 }
 427
 428 int SortConstStrByPosition(const void *Item1, const void *Item2)
 429 {
 430         const ConstStr *p1, *p2;
 431         p1 = (const ConstStr*) Item1;
 432         p2 = (const ConstStr*) Item2;
 433         if (p1->Key == p2->Key)
 434                 return 0;
 435         if (p1->Key > p2->Key)
 436                 return 1;
 437         return -1;
 438 }
 439
 440 void ExpandShortUrls(StrBuf *Message)
 441 {
 442         StrBuf *Shadow;
 443         HashList *pUrls;
 444         ConstStr *pCUrl;
 445         const char *pch;
 446         const char *pche;
 447
 448         /* we just suspect URL shorteners to be inside of feeds from twitter
 449          * or other short content messages, so don't crawl through real blogs.
 450          */
 451         if (StrLength(Message) > 500)
 452                 return;
 453
 454         pUrls = NewHash(1, Flathash);
 455         CrawlMessageForShorterUrls(pUrls, Message);
 456
 457         if (GetCount(pUrls) > 0)
 458         {
 459                 StrBuf *ShorterUrlStr;
 460                 HashPos *Pos;
 461                 const char *RetrKey;
 462                 void *pv;
 463                 long len;
 464
 465                 Shadow = NewStrBufPlain(NULL, StrLength(Message));
 466                 SortByPayload (pUrls, SortConstStrByPosition);
 467
 468                 ShorterUrlStr = NewStrBufPlain(NULL, StrLength(Message));
 469
 470                 pch = ChrPtr(Message);
 471                 pche = pch + StrLength(Message);
 472                 Pos = GetNewHashPos(pUrls, 1);
 473                 while (GetNextHashPos(pUrls, Pos, &len, &RetrKey, &pv))
 474                 {
 475                         pCUrl = (ConstStr*) pv;
 476
 477                         if (pch != pCUrl->Key)
 478                                 StrBufAppendBufPlain(Shadow, pch, pCUrl->Key - pch, 0);
 479
 480                         StrBufPlain(ShorterUrlStr, CKEY(*pCUrl));
 481                         if (LookupUrl(ShorterUrlStr))
 482                         {
 483                                 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 484                                 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 485                                 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 486                                 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 487                                 StrBufAppendBufPlain(Shadow, HKEY("["), 0);
 488                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 489                                 StrBufAppendBufPlain(Shadow, HKEY("]</a>"), 0);
 490                         }
 491                         else
 492                         {
 493                                 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 494                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 495                                 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 496                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 497                                 StrBufAppendBufPlain(Shadow, HKEY("</a>"), 0);
 498                         }
 499                         pch = pCUrl->Key + pCUrl->len + 1;
 500
 501                 }
 502                 if (pch < pche)
 503                         StrBufAppendBufPlain(Shadow, pch, pche - pch, 0);
 504                 FlushStrBuf(Message);
 505                 StrBufAppendBuf(Message, Shadow, 0);
 506
 507                 FreeStrBuf(&ShorterUrlStr);
 508                 FreeStrBuf(&Shadow);
 509                 DeleteHashPos(&Pos);
 510         }
 511
 512         DeleteHash(&pUrls);
 513 }
 514
 515
 516 void AppendLink(StrBuf *Message, StrBuf *link, StrBuf *LinkTitle, const char *Title)
 517 {
 518         if (StrLength(link) > 0)
 519         {
 520                 StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
 521                 StrBufAppendBuf(Message, link, 0);
 522                 StrBufAppendBufPlain(Message, HKEY("\">"), 0);
 523                 if (StrLength(LinkTitle) > 0)
 524                         StrBufAppendBuf(Message, LinkTitle, 0);
 525                 else if ((Title != NULL) && !IsEmptyStr(Title))
 526                         StrBufAppendBufPlain(Message, Title, -1, 0);
 527                 else
 528                         StrBufAppendBuf(Message, link, 0);
 529                 StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
 530         }
 531 }
 532 /*
 533  * Commit a fetched and parsed RSS item to disk
 534  */
 535 void rss_save_item(rss_item *ri)
 536 {
 537
 538         struct MD5Context md5context;
 539         u_char rawdigest[MD5_DIGEST_LEN];
 540         int i;
 541         char utmsgid[SIZ];
 542         struct cdbdata *cdbut;
 543         struct UseTable ut;
 544         struct CtdlMessage *msg;
 545         struct recptypes *recp = NULL;
 546         int msglen = 0;
 547         StrBuf *Message;
 548
 549         recp = (struct recptypes *) malloc(sizeof(struct recptypes));
 550         if (recp == NULL) return;
 551         memset(recp, 0, sizeof(struct recptypes));
 552         memset(&ut, 0, sizeof(struct UseTable));
 553         recp->recp_room = strdup(ri->roomlist);
 554         recp->num_room = num_tokens(ri->roomlist, '|');
 555         recp->recptypes_magic = RECPTYPES_MAGIC;
 556
 557         /* Construct a GUID to use in the S_USETABLE table.
 558          * If one is not present in the item itself, make one up.
 559          */
 560         if (ri->guid != NULL) {
 561                 StrBufSpaceToBlank(ri->guid);
 562                 StrBufTrim(ri->guid);
 563                 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ChrPtr(ri->guid));
 564         }
 565         else {
 566                 MD5Init(&md5context);
 567                 if (ri->title != NULL) {
 568                         MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->title), StrLength(ri->title));
 569                 }
 570                 if (ri->link != NULL) {
 571                         MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->link), StrLength(ri->link));
 572                 }
 573                 MD5Final(rawdigest, &md5context);
 574                 for (i=0; i<MD5_DIGEST_LEN; i++) {
 575                         sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
 576                         utmsgid[i*2] = tolower(utmsgid[i*2]);
 577                         utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
 578                 }
 579                 strcat(utmsgid, "_rss2ctdl");
 580         }
 581
 582         /* Find out if we've already seen this item */
 583
 584         cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
 585 #ifndef DEBUG_RSS
 586         if (cdbut != NULL) {
 587                 /* Item has already been seen */
 588                 syslog(LOG_DEBUG, "%s has already been seen\n", utmsgid);
 589                 cdb_free(cdbut);
 590
 591                 /* rewrite the record anyway, to update the timestamp */
 592                 strcpy(ut.ut_msgid, utmsgid);
 593                 ut.ut_timestamp = time(NULL);
 594                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 595         }
 596         else
 597 #endif
 598 {
 599                 /* Item has not been seen, so save it. */
 600                 syslog(LOG_DEBUG, "RSS: saving item...\n");
 601                 if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY(""));
 602                 StrBufSpaceToBlank(ri->description);
 603                 msg = malloc(sizeof(struct CtdlMessage));
 604                 memset(msg, 0, sizeof(struct CtdlMessage));
 605                 msg->cm_magic = CTDLMESSAGE_MAGIC;
 606                 msg->cm_anon_type = MES_NORMAL;
 607                 msg->cm_format_type = FMT_RFC822;
 608
 609                 if (ri->guid != NULL) {
 610                         msg->cm_fields['E'] = strdup(ChrPtr(ri->guid));
 611                 }
 612
 613                 if (ri->author_or_creator != NULL) {
 614                         char *From;
 615                         StrBuf *Encoded = NULL;
 616                         int FromAt;
 617
 618                         From = html_to_ascii(ChrPtr(ri->author_or_creator),
 619                                              StrLength(ri->author_or_creator),
 620                                              512, 0);
 621                         StrBufPlain(ri->author_or_creator, From, -1);
 622                         StrBufTrim(ri->author_or_creator);
 623                         free(From);
 624
 625                         FromAt = strchr(ChrPtr(ri->author_or_creator), '@') != NULL;
 626                         if (!FromAt && StrLength (ri->author_email) > 0)
 627                         {
 628                                 StrBufRFC2047encode(&Encoded, ri->author_or_creator);
 629                                 msg->cm_fields['A'] = SmashStrBuf(&Encoded);
 630                                 msg->cm_fields['P'] = SmashStrBuf(&ri->author_email);
 631                         }
 632                         else
 633                         {
 634                                 if (FromAt)
 635                                 {
 636                                         msg->cm_fields['A'] = SmashStrBuf(&ri->author_or_creator);
 637                                         msg->cm_fields['P'] = strdup(msg->cm_fields['A']);
 638                                 }
 639                                 else
 640                                 {
 641                                         StrBufRFC2047encode(&Encoded, ri->author_or_creator);
 642                                         msg->cm_fields['A'] = SmashStrBuf(&Encoded);
 643                                         msg->cm_fields['P'] = strdup("rss@localhost");
 644                                 }
 645                         }
 646                 }
 647                 else {
 648                         msg->cm_fields['A'] = strdup("rss");
 649                 }
 650
 651                 msg->cm_fields['N'] = strdup(NODENAME);
 652                 if (ri->title != NULL) {
 653                         long len;
 654                         char *Sbj;
 655                         StrBuf *Encoded, *QPEncoded;
 656
 657                         QPEncoded = NULL;
 658                         StrBufSpaceToBlank(ri->title);
 659                         len = StrLength(ri->title);
 660                         Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
 661                         len = strlen(Sbj);
 662                         if (Sbj[len - 1] == '\n')
 663                         {
 664                                 len --;
 665                                 Sbj[len] = '\0';
 666                         }
 667                         Encoded = NewStrBufPlain(Sbj, len);
 668                         free(Sbj);
 669
 670                         StrBufTrim(Encoded);
 671                         StrBufRFC2047encode(&QPEncoded, Encoded);
 672
 673                         msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
 674                         FreeStrBuf(&Encoded);
 675                 }
 676
 677                 if (ri->pubdate <= 0) {
 678                         ri->pubdate = time(NULL);
 679                 }
 680                 msg->cm_fields['T'] = malloc(64);
 681                 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
 682
 683                 if (ri->channel_title != NULL) {
 684                         if (StrLength(ri->channel_title) > 0) {
 685                                 msg->cm_fields['O'] = strdup(ChrPtr(ri->channel_title));
 686                         }
 687                 }
 688                 if (ri->link == NULL)
 689                         ri->link = NewStrBufPlain(HKEY(""));
 690                 ExpandShortUrls(ri->description);
 691                 msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
 692
 693                 Message = NewStrBufPlain(NULL, StrLength(ri->description));
 694
 695                 StrBufPlain(Message, HKEY(
 696                          "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 697                          "<html><body>\n"));
 698
 699                 StrBufAppendBuf(Message, ri->description, 0);
 700                 StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
 701
 702                 AppendLink(Message, ri->link, ri->linkTitle, NULL);
 703                 AppendLink(Message, ri->reLink, ri->reLinkTitle, "Reply to this");
 704                 StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
 705
 706                 msg->cm_fields['M'] = SmashStrBuf(&Message);
 707
 708                 CtdlSubmitMsg(msg, recp, NULL, 0);
 709                 CtdlFreeMessage(msg);
 710
 711                 /* write the uidl to the use table so we don't store this item again */
 712                 strcpy(ut.ut_msgid, utmsgid);
 713                 ut.ut_timestamp = time(NULL);
 714                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 715         }
 716         free_recipients(recp);
 717 }
 718
 719
 720
 721 /*
 722  * Convert an RDF/RSS datestamp into a time_t
 723  */
 724 time_t rdf_parsedate(const char *p)
 725 {
 726         struct tm tm;
 727         time_t t = 0;
 728
 729         if (!p) return 0L;
 730         if (strlen(p) < 10) return 0L;
 731
 732         memset(&tm, 0, sizeof tm);
 733
 734         /*
 735          * If the timestamp appears to be in W3C datetime format, try to
 736          * parse it.  See also: http://www.w3.org/TR/NOTE-datetime
 737          *
 738          * This code, along with parsedate.c, is a potential candidate for
 739          * moving into libcitadel.
 740          */
 741         if ( (p[4] == '-') && (p[7] == '-') ) {
 742                 tm.tm_year = atoi(&p[0]) - 1900;
 743                 tm.tm_mon = atoi(&p[5]) - 1;
 744                 tm.tm_mday = atoi(&p[8]);
 745                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 746                         tm.tm_hour = atoi(&p[11]);
 747                         tm.tm_min = atoi(&p[14]);
 748                 }
 749                 return mktime(&tm);
 750         }
 751
 752         /* hmm... try RFC822 date stamp format */
 753
 754         t = parsedate(p);
 755         if (t > 0) return(t);
 756
 757         /* yeesh.  ok, just return the current date and time. */
 758         return(time(NULL));
 759 }
 760
 761 void flush_rss_item(rss_item *ri)
 762 {
 763         /* Initialize the feed item data structure */
 764         FreeStrBuf(&ri->guid);
 765         FreeStrBuf(&ri->title);
 766         FreeStrBuf(&ri->link);
 767         FreeStrBuf(&ri->linkTitle);
 768         FreeStrBuf(&ri->reLink);
 769         FreeStrBuf(&ri->reLinkTitle);
 770         FreeStrBuf(&ri->description);
 771         FreeStrBuf(&ri->channel_title);
 772         FreeStrBuf(&ri->author_or_creator);
 773         FreeStrBuf(&ri->author_url);
 774         FreeStrBuf(&ri->author_email);
 775 }
 776
 777 void rss_xml_start(void *data, const char *supplied_el, const char **attr)
 778 {
 779         rss_xml_handler *h;
 780         rsscollection   *rssc = (rsscollection*) data;
 781         rssnetcfg       *Cfg = rssc->Cfg;
 782         rss_item        *ri = rssc->Item;
 783         void            *pv;
 784         const char      *pel;
 785         char            *sep = NULL;
 786
 787         /* Axe the namespace, we don't care about it */
 788 ///     syslog(LOG_EMERG, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
 789         pel = supplied_el;
 790         while (sep = strchr(pel, ':'), sep) {
 791                 pel = sep + 1;
 792         }
 793
 794         if (pel != supplied_el)
 795         {
 796                 void *v;
 797
 798                 if (!GetHash(KnownNameSpaces,
 799                              supplied_el,
 800                              pel - supplied_el - 1,
 801                              &v))
 802                 {
 803 #ifdef DEBUG_RSS
 804                         syslog(LOG_EMERG, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
 805                                       supplied_el);
 806 #endif
 807                         return;
 808                 }
 809         }
 810
 811         StrBufPlain(rssc->Key, pel, -1);
 812         StrBufLowerCase(rssc->Key);
 813         if (GetHash(StartHandlers, SKEY(rssc->Key), &pv))
 814         {
 815                 rssc->Current = h = (rss_xml_handler*) pv;
 816
 817                 if (((h->Flags & RSS_UNSET) != 0) &&
 818                     (Cfg->ItemType == RSS_UNSET))
 819                 {
 820                         h->Handler(rssc->CData, ri, Cfg, attr);
 821                 }
 822                 else if (((h->Flags & RSS_RSS) != 0) &&
 823                     (Cfg->ItemType == RSS_RSS))
 824                 {
 825                         h->Handler(rssc->CData, ri, Cfg, attr);
 826                 }
 827                 else if (((h->Flags & RSS_ATOM) != 0) &&
 828                          (Cfg->ItemType == RSS_ATOM))
 829                 {
 830                         h->Handler(rssc->CData, ri, Cfg, attr);
 831                 }
 832 #ifdef DEBUG_RSS
 833                 else
 834                         syslog(LOG_EMERG, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
 835 #endif
 836         }
 837 #ifdef DEBUG_RSS
 838         else
 839                 syslog(LOG_EMERG, "RSS: START unhandled: [%s] [%s]...\n", pel,  supplied_el);
 840 #endif
 841 }
 842
 843 void rss_xml_end(void *data, const char *supplied_el)
 844 {
 845         rss_xml_handler *h;
 846         rsscollection   *rssc = (rsscollection*) data;
 847         rssnetcfg       *Cfg = rssc->Cfg;
 848         rss_item        *ri = rssc->Item;
 849         const char      *pel;
 850         char            *sep = NULL;
 851         void            *pv;
 852
 853         /* Axe the namespace, we don't care about it */
 854         pel = supplied_el;
 855         while (sep = strchr(pel, ':'), sep) {
 856                 pel = sep + 1;
 857         }
 858 //      syslog(LOG_EMERG, "RSS: END %s...\n", el);
 859         if (pel != supplied_el)
 860         {
 861                 void *v;
 862
 863                 if (!GetHash(KnownNameSpaces,
 864                              supplied_el,
 865                              pel - supplied_el - 1,
 866                              &v))
 867                 {
 868 #ifdef DEBUG_RSS
 869                         syslog(LOG_EMERG, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
 870                                       supplied_el, ChrPtr(rssc->CData));
 871 #endif
 872                         FlushStrBuf(rssc->CData);
 873                         return;
 874                 }
 875         }
 876
 877         StrBufPlain(rssc->Key, pel, -1);
 878         StrBufLowerCase(rssc->Key);
 879         if (GetHash(EndHandlers, SKEY(rssc->Key), &pv))
 880         {
 881                 h = (rss_xml_handler*) pv;
 882
 883                 if (((h->Flags & RSS_UNSET) != 0) &&
 884                     (Cfg->ItemType == RSS_UNSET))
 885                 {
 886                         h->Handler(rssc->CData, ri, Cfg, NULL);
 887                 }
 888                 else if (((h->Flags & RSS_RSS) != 0) &&
 889                     (Cfg->ItemType == RSS_RSS))
 890                 {
 891                         h->Handler(rssc->CData, ri, Cfg, NULL);
 892                 }
 893                 else if (((h->Flags & RSS_ATOM) != 0) &&
 894                          (Cfg->ItemType == RSS_ATOM))
 895                 {
 896                         h->Handler(rssc->CData, ri, Cfg, NULL);
 897                 }
 898 #ifdef DEBUG_RSS
 899                 else
 900                         syslog(LOG_EMERG, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 901 #endif
 902         }
 903 #ifdef DEBUG_RSS
 904         else
 905                 syslog(LOG_EMERG, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 906 #endif
 907         FlushStrBuf(rssc->CData);
 908         rssc->Current = NULL;
 909 }
 910
 911
 912
 913
 914
 915 void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 916 {
 917         syslog(LOG_DEBUG, "RSS: This is an RSS feed.\n");
 918         Cfg->ItemType = RSS_RSS;
 919 }
 920
 921 void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 922 {
 923         syslog(LOG_DEBUG, "RSS: This is an RDF feed.\n");
 924         Cfg->ItemType = RSS_RSS;
 925 }
 926
 927 void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 928 {
 929         syslog(LOG_DEBUG, "RSS: This is an ATOM feed.\n");
 930         Cfg->ItemType = RSS_ATOM;
 931 }
 932
 933
 934 void RSS_item_item_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 935 {
 936         ri->item_tag_nesting ++;
 937         flush_rss_item(ri);
 938 }
 939
 940 void ATOM_item_entry_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 941 {
 942 /* Atom feed... */
 943         ri->item_tag_nesting ++;
 944         flush_rss_item(ri);
 945 }
 946
 947 void ATOM_item_link_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 948 {
 949         int i;
 950         const char *pHref = NULL;
 951         const char *pType = NULL;
 952         const char *pRel = NULL;
 953         const char *pTitle = NULL;
 954
 955         for (i = 0; Attr[i] != NULL; i+=2)
 956         {
 957                 if (!strcmp(Attr[i], "href"))
 958                 {
 959                         pHref = Attr[i+1];
 960                 }
 961                 else if (!strcmp(Attr[i], "rel"))
 962                 {
 963                         pRel = Attr[i+1];
 964                 }
 965                 else if (!strcmp(Attr[i], "type"))
 966                 {
 967                         pType = Attr[i+1];
 968                 }
 969                 else if (!strcmp(Attr[i], "title"))
 970                 {
 971                         pTitle = Attr[i+1];
 972                 }
 973         }
 974         if (pHref == NULL)
 975                 return; /* WHUT? Pointing... where? */
 976         if ((pType != NULL) && !strcasecmp(pType, "application/atom+xml"))
 977                 return; /* these just point to other rss resources, we're not interested in them. */
 978         if (pRel != NULL)
 979         {
 980                 if (!strcasecmp (pRel, "replies"))
 981                 {
 982                         NewStrBufDupAppendFlush(&ri->reLink, NULL, pHref, -1);
 983                         StrBufTrim(ri->link);
 984                         NewStrBufDupAppendFlush(&ri->reLinkTitle, NULL, pTitle, -1);
 985                 }
 986                 else if (!strcasecmp(pRel, "alternate")) /* Alternative representation of this Item... */
 987                 {
 988                         NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 989                         StrBufTrim(ri->link);
 990                         NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
 991
 992                 }
 993 #if 0 /* these are also defined, but dunno what to do with them.. */
 994                 else if (!strcasecmp(pRel, "related"))
 995                 {
 996                 }
 997                 else if (!strcasecmp(pRel, "self"))
 998                 {
 999                 }
1000                 else if (!strcasecmp(pRel, "enclosure"))
1001                 {/* this reference can get big, and is probably the full article... */
1002                 }
1003                 else if (!strcasecmp(pRel, "via"))
1004                 {/* this article was provided via... */
1005                 }
1006 #endif
1007         }
1008         else if (StrLength(ri->link) == 0)
1009         {
1010                 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
1011                 StrBufTrim(ri->link);
1012                 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
1013         }
1014 }
1015
1016
1017
1018
1019 void ATOMRSS_item_title_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1020 {
1021         if ((ri->item_tag_nesting == 0) && (StrLength(CData) > 0)) {
1022                 NewStrBufDupAppendFlush(&ri->channel_title, CData, NULL, 0);
1023                 StrBufTrim(ri->channel_title);
1024         }
1025 }
1026
1027 void RSS_item_guid_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1028 {
1029         if (StrLength(CData) > 0) {
1030                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
1031         }
1032 }
1033
1034 void ATOM_item_id_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1035 {
1036         if (StrLength(CData) > 0) {
1037                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
1038         }
1039 }
1040
1041
1042 void RSS_item_link_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1043 {
1044         if (StrLength(CData) > 0) {
1045                 NewStrBufDupAppendFlush(&ri->link, CData, NULL, 0);
1046                 StrBufTrim(ri->link);
1047         }
1048 }
1049 void RSS_item_relink_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1050 {
1051         if (StrLength(CData) > 0) {
1052                 NewStrBufDupAppendFlush(&ri->reLink, CData, NULL, 0);
1053                 StrBufTrim(ri->reLink);
1054         }
1055 }
1056
1057 void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1058 {
1059         if (StrLength(CData) > 0) {
1060                 NewStrBufDupAppendFlush(&ri->title, CData, NULL, 0);
1061                 StrBufTrim(ri->title);
1062         }
1063 }
1064
1065 void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1066 {
1067         long olen = StrLength (ri->description);
1068         long clen = StrLength (CData);
1069         if (clen > 0)
1070         {
1071                 if (olen == 0) {
1072                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1073                         StrBufTrim(ri->description);
1074                 }
1075                 else if (olen < clen) {
1076                         FlushStrBuf(ri->description);
1077                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1078                         StrBufTrim(ri->description);
1079                 }
1080         }
1081 }
1082 void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1083 {
1084         /* this can contain an abstract of the article. but we don't want to verwrite a full document if we already have it. */
1085         if ((StrLength(CData) > 0) && (StrLength(ri->description) == 0))
1086         {
1087                 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1088                 StrBufTrim(ri->description);
1089         }
1090 }
1091
1092 void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1093 {
1094         long olen = StrLength (ri->description);
1095         long clen = StrLength (CData);
1096         if (clen > 0)
1097         {
1098                 if (olen == 0) {
1099                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1100                         StrBufTrim(ri->description);
1101                 }
1102                 else if (olen < clen) {
1103                         FlushStrBuf(ri->description);
1104                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1105                         StrBufTrim(ri->description);
1106                 }
1107         }
1108 }
1109
1110 void ATOM_item_published_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1111 {
1112         if (StrLength(CData) > 0) {
1113                 StrBufTrim(CData);
1114                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1115         }
1116 }
1117
1118 void ATOM_item_updated_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1119 {
1120         if (StrLength(CData) > 0) {
1121                 StrBufTrim(CData);
1122                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1123         }
1124 }
1125
1126 void RSS_item_pubdate_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1127 {
1128         if (StrLength(CData) > 0) {
1129                 StrBufTrim(CData);
1130                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1131         }
1132 }
1133
1134
1135 void RSS_item_date_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1136 {
1137         if (StrLength(CData) > 0) {
1138                 StrBufTrim(CData);
1139                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1140         }
1141 }
1142
1143
1144
1145 void RSS_item_author_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1146 {
1147         if (StrLength(CData) > 0) {
1148                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1149                 StrBufTrim(ri->author_or_creator);
1150         }
1151 }
1152
1153
1154 void ATOM_item_name_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1155 {
1156         if (StrLength(CData) > 0) {
1157                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1158                 StrBufTrim(ri->author_or_creator);
1159         }
1160 }
1161
1162 void ATOM_item_email_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1163 {
1164         if (StrLength(CData) > 0) {
1165                 NewStrBufDupAppendFlush(&ri->author_email, CData, NULL, 0);
1166                 StrBufTrim(ri->author_email);
1167         }
1168 }
1169
1170 void RSS_item_creator_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1171 {
1172         if ((StrLength(CData) > 0) &&
1173             (StrLength(ri->author_or_creator) == 0))
1174         {
1175                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1176                 StrBufTrim(ri->author_or_creator);
1177         }
1178 }
1179
1180
1181 void ATOM_item_uri_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1182 {
1183         if (StrLength(CData) > 0) {
1184                 NewStrBufDupAppendFlush(&ri->author_url, CData, NULL, 0);
1185                 StrBufTrim(ri->author_url);
1186         }
1187 }
1188
1189 void RSS_item_item_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1190 {
1191         --ri->item_tag_nesting;
1192         rss_save_item(ri);
1193 }
1194
1195
1196 void ATOM_item_entry_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1197 {
1198         --ri->item_tag_nesting;
1199         rss_save_item(ri);
1200 }
1201
1202 void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1203 {
1204 //              syslog(LOG_DEBUG, "End of feed detected.  Closing parser.\n");
1205         ri->done_parsing = 1;
1206
1207 }
1208 void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1209 {
1210 //              syslog(LOG_DEBUG, "End of feed detected.  Closing parser.\n");
1211         ri->done_parsing = 1;
1212 }
1213
1214
1215 void RSSATOM_item_ignore(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1216 {
1217 }
1218
1219
1220
1221 /*
1222  * This callback stores up the data which appears in between tags.
1223  */
1224 void rss_xml_cdata_start(void *data)
1225 {
1226         rsscollection *rssc = (rsscollection*) data;
1227
1228         FlushStrBuf(rssc->CData);
1229 }
1230
1231 void rss_xml_cdata_end(void *data)
1232 {
1233 }
1234 void rss_xml_chardata(void *data, const XML_Char *s, int len)
1235 {
1236         rsscollection *rssc = (rsscollection*) data;
1237
1238         StrBufAppendBufPlain (rssc->CData, s, len, 0);
1239 }
1240
1241 /*
1242  * Callback function for passing libcurl's output to expat for parsing
1243  */
1244 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
1245 {
1246         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
1247         return (size*nmemb);
1248 }
1249
1250
1251
1252 /*
1253  * Begin a feed parse
1254  */
1255 void rss_do_fetching(rssnetcfg *Cfg) {
1256         rsscollection rssc;
1257         rss_item ri;
1258         XML_Parser xp = NULL;
1259         StrBuf *Answer;
1260
1261         CURL *curl;
1262         CURLcode res;
1263         char errmsg[1024] = "";
1264         char *ptr;
1265         const char *at;
1266         long len;
1267
1268         time_t now;
1269
1270         now = time(NULL);
1271
1272         if ((Cfg->next_poll != 0) && (now < Cfg->next_poll))
1273                 return;
1274         memset(&ri, 0, sizeof(rss_item));
1275         rssc.Item = &ri;
1276         rssc.Cfg = Cfg;
1277
1278         syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
1279
1280         curl = curl_easy_init();
1281         if (!curl) {
1282                 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
1283                 return;
1284         }
1285         Answer = NewStrBufPlain(NULL, SIZ);
1286
1287         curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
1288         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
1289         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
1290         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
1291 //      curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
1292         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
1293         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
1294         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
1295 #ifdef CURLOPT_HTTP_CONTENT_DECODING
1296         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
1297         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
1298 #endif
1299         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
1300         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
1301         if (
1302                 (!IsEmptyStr(config.c_ip_addr))
1303                 && (strcmp(config.c_ip_addr, "*"))
1304                 && (strcmp(config.c_ip_addr, "::"))
1305                 && (strcmp(config.c_ip_addr, "0.0.0.0"))
1306         ) {
1307                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
1308         }
1309
1310         if (server_shutting_down)
1311         {
1312                 curl_easy_cleanup(curl);
1313                 return;
1314         }
1315
1316         if (server_shutting_down)
1317                 goto shutdown ;
1318
1319         res = curl_easy_perform(curl);
1320         if (res) {
1321                 syslog(LOG_ALERT, "libcurl error %d: %s\n", res, errmsg);
1322         }
1323
1324         if (server_shutting_down)
1325                 goto shutdown ;
1326
1327
1328
1329
1330         memset(&ri, 0, sizeof(rss_item));
1331         ri.roomlist = Cfg->rooms;
1332         rssc.CData = NewStrBufPlain(NULL, SIZ);
1333         rssc.Key = NewStrBuf();
1334         at = NULL;
1335         StrBufSipLine(rssc.Key, Answer, &at);
1336         ptr = NULL;
1337
1338 #define encoding "encoding=\""
1339         ptr = strstr(ChrPtr(rssc.Key), encoding);
1340         if (ptr != NULL)
1341         {
1342                 char *pche;
1343
1344                 ptr += sizeof (encoding) - 1;
1345                 pche = strchr(ptr, '"');
1346                 if (pche != NULL)
1347                         StrBufCutAt(rssc.Key, -1, pche);
1348                 else
1349                         ptr = "UTF-8";
1350         }
1351         else
1352                 ptr = "UTF-8";
1353
1354
1355         xp = XML_ParserCreateNS(ptr, ':');
1356         if (!xp) {
1357                 syslog(LOG_ALERT, "Cannot create XML parser!\n");
1358                 goto shutdown;
1359         }
1360         FlushStrBuf(rssc.Key);
1361 //#ifdef HAVE_ICONV
1362 #if 0
1363         XML_SetUnknownEncodingHandler(xp,
1364                                       handle_unknown_xml_encoding,
1365                                       &rssc);
1366 #endif
1367 //#endif
1368         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
1369         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
1370         XML_SetUserData(xp, &rssc);
1371         XML_SetCdataSectionHandler(xp,
1372                                    rss_xml_cdata_start,
1373                                    rss_xml_cdata_end);
1374
1375
1376         len = StrLength(Answer);
1377         ptr = SmashStrBuf(&Answer);
1378         XML_Parse(xp, ptr, len, 0);
1379         free (ptr);
1380         if (ri.done_parsing == 0)
1381                 XML_Parse(xp, "", 0, 1);
1382
1383
1384         syslog(LOG_ALERT, "RSS: XML Status [%s] \n",
1385                       XML_ErrorString(
1386                               XML_GetErrorCode(xp)));
1387
1388 shutdown:
1389         FreeStrBuf(&Answer);
1390         curl_easy_cleanup(curl);
1391         XML_ParserFree(xp);
1392
1393         flush_rss_item(&ri);
1394         FreeStrBuf(&rssc.CData);
1395         FreeStrBuf(&rssc.Key);
1396
1397         Cfg->next_poll = time(NULL) + config.c_net_freq;
1398 }
1399
1400
1401 /*
1402  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
1403  */
1404 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
1405 {
1406         char filename[PATH_MAX];
1407         char buf[1024];
1408         char instr[32];
1409         FILE *fp;
1410         char feedurl[256];
1411         rssnetcfg *rncptr = NULL;
1412         rssnetcfg *use_this_rncptr = NULL;
1413         int len = 0;
1414         char *ptr = NULL;
1415
1416         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
1417
1418         if (server_shutting_down)
1419                 return;
1420
1421         /* Only do net processing for rooms that have netconfigs */
1422         fp = fopen(filename, "r");
1423         if (fp == NULL) {
1424                 return;
1425         }
1426
1427         while (fgets(buf, sizeof buf, fp) != NULL && !server_shutting_down) {
1428                 buf[strlen(buf)-1] = 0;
1429
1430                 extract_token(instr, buf, 0, '|', sizeof instr);
1431                 if (!strcasecmp(instr, "rssclient")) {
1432
1433                         use_this_rncptr = NULL;
1434
1435                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
1436
1437                         /* If any other rooms have requested the same feed, then we will just add this
1438                          * room to the target list for that client request.
1439                          */
1440                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
1441                                 if (!strcmp(rncptr->url, feedurl)) {
1442                                         use_this_rncptr = rncptr;
1443                                 }
1444                         }
1445
1446                         /* Otherwise create a new client request */
1447                         if (use_this_rncptr == NULL) {
1448                                 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
1449                                 memset(rncptr, 0, sizeof(rssnetcfg));
1450                                 rncptr->ItemType = RSS_UNSET;
1451                                 if (rncptr != NULL) {
1452                                         rncptr->next = rnclist;
1453                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
1454                                         rncptr->rooms = NULL;
1455                                         rnclist = rncptr;
1456                                         use_this_rncptr = rncptr;
1457                                 }
1458                         }
1459
1460                         /* Add the room name to the request */
1461                         if (use_this_rncptr != NULL) {
1462                                 if (use_this_rncptr->rooms == NULL) {
1463                                         rncptr->rooms = strdup(qrbuf->QRname);
1464                                 }
1465                                 else {
1466                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
1467                                         ptr = realloc(use_this_rncptr->rooms, len);
1468                                         if (ptr != NULL) {
1469                                                 strcat(ptr, "|");
1470                                                 strcat(ptr, qrbuf->QRname);
1471                                                 use_this_rncptr->rooms = ptr;
1472                                         }
1473                                 }
1474                         }
1475                 }
1476
1477         }
1478
1479         fclose(fp);
1480
1481 }
1482
1483 /*
1484  * Scan for rooms that have RSS client requests configured
1485  */
1486 void rssclient_scan(void) {
1487         static time_t last_run = 0L;
1488         static int doing_rssclient = 0;
1489         rssnetcfg *rptr = NULL;
1490
1491         /* Run no more than once every 15 minutes. */
1492         if ((time(NULL) - last_run) < 900) {
1493                 return;
1494         }
1495
1496         /*
1497          * This is a simple concurrency check to make sure only one rssclient run
1498          * is done at a time.  We could do this with a mutex, but since we
1499          * don't really require extremely fine granularity here, we'll do it
1500          * with a static variable instead.
1501          */
1502         if (doing_rssclient) return;
1503         doing_rssclient = 1;
1504
1505         syslog(LOG_DEBUG, "rssclient started\n");
1506         CtdlForEachRoom(rssclient_scan_room, NULL);
1507
1508         while (rnclist != NULL && !server_shutting_down) {
1509                 rss_do_fetching(rnclist);
1510                 rptr = rnclist;
1511                 rnclist = rnclist->next;
1512                 if (rptr->rooms != NULL) free(rptr->rooms);
1513                 free(rptr);
1514         }
1515
1516         syslog(LOG_DEBUG, "rssclient ended\n");
1517         last_run = time(NULL);
1518         doing_rssclient = 0;
1519         return;
1520 }
1521
1522 void LoadUrlShorteners(void)
1523 {
1524         int i = 0;
1525         int fd;
1526         const char *POS = NULL;
1527         const char *Err = NULL;
1528         StrBuf *Content, *Line;
1529
1530
1531         UrlShorteners = NewHash(0, Flathash);
1532
1533         fd = open(file_citadel_urlshorteners, 0);
1534
1535         if (fd != 0)
1536         {
1537                 Content = NewStrBufPlain(NULL, SIZ);
1538                 Line = NewStrBuf();
1539                 while (POS != StrBufNOTNULL)
1540                 {
1541                         StrBufTCP_read_buffered_line_fast (Line, Content, &POS, &fd, 1, 1, &Err);
1542                         StrBufTrim(Line);
1543                         if ((*ChrPtr(Line) != '#') && (StrLength(Line) > 0))
1544                         {
1545                                 Put(UrlShorteners, IKEY(i), Line, HFreeStrBuf);
1546                                 i++;
1547                                 Line = NewStrBuf();
1548                         }
1549                         else
1550                                 FlushStrBuf(Line);
1551                         if (POS == NULL)
1552                                 POS = StrBufNOTNULL;
1553                 }
1554                 FreeStrBuf(&Line);
1555                 FreeStrBuf(&Content);
1556         }
1557         close(fd);
1558 }
1559
1560 void rss_cleanup(void)
1561 {
1562         DeleteHash(&StartHandlers);
1563         DeleteHash(&EndHandlers);
1564         DeleteHash(&UrlShorteners);
1565         DeleteHash(&KnownNameSpaces);
1566 }
1567
1568 CTDL_MODULE_INIT(rssclient)
1569 {
1570         if (threading)
1571         {
1572                 syslog(LOG_INFO, "%s\n", curl_version());
1573                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
1574         }
1575         else
1576         {
1577                 LoadUrlShorteners ();
1578
1579                 StartHandlers = NewHash(1, NULL);
1580                 EndHandlers = NewHash(1, NULL);
1581
1582                 AddRSSStartHandler(RSS_item_rss_start,     RSS_UNSET, HKEY("rss"));
1583                 AddRSSStartHandler(RSS_item_rdf_start,     RSS_UNSET, HKEY("rdf"));
1584                 AddRSSStartHandler(ATOM_item_feed_start,    RSS_UNSET, HKEY("feed"));
1585                 AddRSSStartHandler(RSS_item_item_start,    RSS_RSS, HKEY("item"));
1586                 AddRSSStartHandler(ATOM_item_entry_start,  RSS_ATOM, HKEY("entry"));
1587                 AddRSSStartHandler(ATOM_item_link_start,   RSS_ATOM, HKEY("link"));
1588
1589                 AddRSSEndHandler(ATOMRSS_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1590                 AddRSSEndHandler(RSS_item_guid_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
1591                 AddRSSEndHandler(ATOM_item_id_end,         RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
1592                 AddRSSEndHandler(RSS_item_link_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
1593 #if 0
1594 // hm, rss to the comments of that blog, might be interesting in future, but...
1595                 AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("commentrss"));
1596 // comment count...
1597                 AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("comments"));
1598 #endif
1599                 AddRSSEndHandler(RSSATOM_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1600                 AddRSSEndHandler(ATOM_item_content_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
1601                 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_ATOM|RSS_REQUIRE_BUF, HKEY("encoded"));
1602                 AddRSSEndHandler(ATOM_item_summary_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
1603                 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
1604                 AddRSSEndHandler(ATOM_item_published_end,  RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
1605                 AddRSSEndHandler(ATOM_item_updated_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
1606                 AddRSSEndHandler(RSS_item_pubdate_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
1607                 AddRSSEndHandler(RSS_item_date_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
1608                 AddRSSEndHandler(RSS_item_author_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
1609                 AddRSSEndHandler(RSS_item_creator_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
1610 /* <author> */
1611                 AddRSSEndHandler(ATOM_item_email_end,      RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
1612                 AddRSSEndHandler(ATOM_item_name_end,       RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
1613                 AddRSSEndHandler(ATOM_item_uri_end,        RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
1614 /* </author> */
1615                 AddRSSEndHandler(RSS_item_item_end,        RSS_RSS, HKEY("item"));
1616                 AddRSSEndHandler(RSS_item_rss_end,         RSS_RSS, HKEY("rss"));
1617                 AddRSSEndHandler(RSS_item_rdf_end,         RSS_RSS, HKEY("rdf"));
1618                 AddRSSEndHandler(ATOM_item_entry_end,      RSS_ATOM, HKEY("entry"));
1619
1620
1621 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
1622                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1623                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1624                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1625                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1626
1627 /* links to other feed generators... */
1628                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1629                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1630                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1631                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1632
1633                 KnownNameSpaces = NewHash(1, NULL);
1634                 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
1635                 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
1636                 Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
1637                 Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
1638                 Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
1639                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1640                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
1641                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
1642                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
1643                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1644                 Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
1645                 Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
1646                 Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
1647                 Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
1648                 Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
1649                 Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
1650                 Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
1651                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
1652                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), 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/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
1655                 Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
1656                 Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
1657 #if 0
1658                 /* we don't like these namespaces because of they shadow our usefull parameters. */
1659                 Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
1660 #endif
1661                 CtdlRegisterCleanupHook(rss_cleanup);
1662         }
1663         return "rssclient";
1664 }