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
  61 const char *LinkShortenerServices[] = {
  62 "http://bit.ly/",
  63 "http://krz.ch/",
  64 "http://flic.kr/",
  65 "http://sns.ly/",
  66 "http://wp.me/",
  67 "http://ow.ly/",
  68 "http://tinyurl.com/",
  69 NULL
  70 };
  71
  72 typedef struct rssnetcfg rssnetcfg;
  73 struct rssnetcfg {
  74         rssnetcfg *next;
  75         char url[256];
  76         char *rooms;
  77         time_t last_error_when;
  78         int ItemType;
  79 };
  80
  81 #define RSS_UNSET       (1<<0)
  82 #define RSS_RSS         (1<<1)
  83 #define RSS_ATOM        (1<<2)
  84 #define RSS_REQUIRE_BUF (1<<3)
  85
  86 typedef struct _rss_item {
  87         char *roomlist;
  88         int done_parsing;
  89         StrBuf *guid;
  90         StrBuf *title;
  91         StrBuf *link;
  92         StrBuf *linkTitle;
  93         StrBuf *reLink;
  94         StrBuf *reLinkTitle;
  95         StrBuf *description;
  96         time_t pubdate;
  97         StrBuf *channel_title;
  98         int item_tag_nesting;
  99         StrBuf *author_or_creator;
 100         StrBuf *author_url;
 101         StrBuf *author_email;
 102 }rss_item;
 103
 104
 105 typedef void (*rss_handler_func)(StrBuf *CData,
 106                                  rss_item *ri,
 107                                  rssnetcfg *Cfg,
 108                                  const char** Attr);
 109
 110 typedef struct __rss_xml_handler {
 111         int Flags;
 112         rss_handler_func Handler;
 113 }rss_xml_handler;
 114
 115
 116 typedef struct _rsscollection {
 117         StrBuf *CData;
 118         StrBuf *Key;
 119
 120         rss_item *Item;
 121         rssnetcfg *Cfg;
 122
 123         rss_xml_handler *Current;
 124 } rsscollection;
 125
 126 struct rssnetcfg *rnclist = NULL;
 127 HashList *StartHandlers;
 128 HashList *EndHandlers;
 129 HashList *KnownNameSpaces;
 130 void AddRSSStartHandler(rss_handler_func Handler, int Flags, const char *key, long len)
 131 {
 132         rss_xml_handler *h;
 133         h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
 134         h->Flags = Flags;
 135         h->Handler = Handler;
 136         Put(StartHandlers, key, len, h, NULL);
 137 }
 138 void AddRSSEndHandler(rss_handler_func Handler, int Flags, const char *key, long len)
 139 {
 140         rss_xml_handler *h;
 141         h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
 142         h->Flags = Flags;
 143         h->Handler = Handler;
 144         Put(EndHandlers, key, len, h, NULL);
 145 }
 146
 147 #if 0
 148 //#ifdef HAVE_ICONV
 149 #include <iconv.h>
 150
 151
 152 /*
 153  * dug this out of the trashcan of the midgard project, lets see if it works for us.
 154  * original code by Alexander Bokovoy <bokovoy@avilink.ne> distributed under GPL V2 or later
 155  */
 156
 157 /* Returns:
 158  >= 0 - successfull, 0 means conversion doesn't use multibyte sequences
 159    -1 - error during iconv_open call
 160    -2 - error during iconv_close call
 161    ----------------------------------
 162    This function expects that multibyte encoding in 'charset' wouldn't have
 163    characters with more than 3 bytes. It is not intended to convert UTF-8 because
 164    we'll never receive UTF-8 in our handler (it is handled by Exat itself).
 165 */
 166 static int
 167 fill_encoding_info (const char *charset, XML_Encoding * info)
 168 {
 169   iconv_t cd = (iconv_t)(-1);
 170   int flag;
 171         CtdlLogPrintf(0, "RSS: fill encoding info ...\n");
 172
 173 #if G_BYTE_ORDER == G_LITTLE_ENDIAN
 174   cd = iconv_open ("UCS-2LE", charset);
 175 #else
 176   cd = iconv_open ("UCS-2BE", charset);
 177 #endif
 178
 179   if (cd == (iconv_t) (-1))
 180     {
 181       return -1;
 182     }
 183
 184   {
 185     unsigned short out = 0;
 186     unsigned char buf[4];
 187     unsigned int i0, i1, i2;
 188     int result = 0;
 189     flag = 0;
 190     for (i0 = 0; i0 < 0x100; i0++)
 191       {
 192         buf[0] = i0;
 193         info->map[i0] = 0;
 194         //result = try (cd, buf, 1, &out);
 195         if (result < 0)
 196           {
 197           }
 198         else if (result > 0)
 199           {
 200             info->map[i0] = out;
 201           }
 202         else
 203           {
 204             for (i1 = 0; i1 < 0x100; i1++)
 205               {
 206                 buf[1] = i1;
 207                 ///result = try (cd, buf, 2, &out);
 208                 if (result < 0)
 209                   {
 210                   }
 211                 else if (result > 0)
 212                   {
 213                     flag++;
 214                     info->map[i0] = -2;
 215                   }
 216                 else
 217                   {
 218                     for (i2 = 0; i2 < 0x100; i2++)
 219                       {
 220                         buf[2] = i2;
 221                         ////result = try (cd, buf, 3, &out);
 222                         if (result < 0)
 223                           {
 224                           }
 225                         else if (result > 0)
 226                           {
 227                             flag++;
 228                             info->map[i0] = -3;
 229                           }
 230                       }
 231                   }
 232               }
 233           }
 234       }
 235   }
 236
 237   if (iconv_close (cd) < 0)
 238     {
 239       return -2;
 240     }
 241   return flag;
 242 }
 243
 244 static int
 245 iconv_convertor (void *data, const char *s)
 246 {
 247   XML_Encoding *info = data;
 248   int res;
 249         CtdlLogPrintf(0, "RSS: Converting ...\n");
 250
 251   if (s == NULL)
 252     return -1;
 253 /*
 254   GByteArray *result;
 255   result = g_byte_array_new ();
 256   if (process_block (info->data, (char *) s, strlen (s), result) == 0)
 257     {
 258       res = *(result->data);
 259       g_byte_array_free (result, TRUE);
 260       return res;
 261     }
 262   g_byte_array_free (result, TRUE);
 263 */
 264   return -1;
 265 }
 266
 267 static void
 268 my_release (void *data)
 269 {
 270   iconv_t cd = (iconv_t) data;
 271   if (iconv_close (cd) != 0)
 272     {
 273 /// TODO: uh no.      exit (1);
 274     }
 275 }
 276 int
 277 handle_unknown_xml_encoding (void *encodingHandleData,
 278                              const XML_Char * name,
 279                              XML_Encoding * info)
 280 {
 281   int result;
 282   CtdlLogPrintf(0, "RSS: unknown encoding ...\n");
 283   result = fill_encoding_info (name, info);
 284   if (result >= 0)
 285     {
 286       /*
 287         Special case: client asked for reverse conversion, we'll provide him with
 288         iconv descriptor which handles it. Client should release it by himself.
 289       */
 290       if(encodingHandleData != NULL)
 291             *((iconv_t *)encodingHandleData) = iconv_open(name, "UTF-8");
 292       /*
 293          Optimization: we do not need conversion function if encoding is one-to-one,
 294          info->map table will be enough
 295        */
 296       if (result == 0)
 297         {
 298           info->data = NULL;
 299           info->convert = NULL;
 300           info->release = NULL;
 301           return 1;
 302         }
 303       /*
 304          We do need conversion function because this encoding uses multibyte sequences
 305        */
 306       info->data = (void *) iconv_open ("UTF-8", name);
 307       if ((int)info->data == -1)
 308         return -1;
 309       info->convert = iconv_convertor;
 310       info->release = my_release;
 311       return 1;
 312     }
 313   if(encodingHandleData != NULL)
 314     *(iconv_t *)encodingHandleData = NULL;
 315   return 0;
 316 }
 317
 318 ///#endif
 319 #endif
 320 size_t GetLocationString( void *ptr, size_t size, size_t nmemb, void *userdata)
 321 {
 322 #define LOCATION "location"
 323         if (strncasecmp((char*)ptr, LOCATION, sizeof(LOCATION) - 1) == 0)
 324         {
 325                 StrBuf *pURL = (StrBuf*) userdata;
 326                 char *pch = (char*) ptr;
 327                 char *pche;
 328
 329                 pche = pch + (size * nmemb);
 330                 pch += sizeof(LOCATION);
 331
 332                 while (isspace(*pch) || (*pch == ':'))
 333                         pch ++;
 334
 335                 while (isspace(*pche) || (*pche == '\0'))
 336                         pche--;
 337
 338                 FlushStrBuf(pURL);
 339                 StrBufPlain(pURL, pch, pche - pch + 1);
 340         }
 341         return size * nmemb;
 342 }
 343
 344 int LookupUrl(StrBuf *ShorterUrlStr)
 345 {
 346         CURL *curl;
 347         char errmsg[1024] = "";
 348         StrBuf *Answer;
 349         int rc = 0;
 350
 351         curl = curl_easy_init();
 352         if (!curl) {
 353                 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
 354                 return 0;
 355         }
 356         Answer = NewStrBufPlain(NULL, SIZ);
 357
 358         curl_easy_setopt(curl, CURLOPT_URL, ChrPtr(ShorterUrlStr));
 359         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
 360         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
 361         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
 362 //      curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
 363         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
 364         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
 365         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
 366 #ifdef CURLOPT_HTTP_CONTENT_DECODING
 367         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
 368         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
 369 #endif
 370         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
 371         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
 372         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 0);
 373
 374         curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION , GetLocationString);
 375         curl_easy_setopt(curl, CURLOPT_WRITEHEADER, ShorterUrlStr);
 376
 377
 378         if (
 379                 (!IsEmptyStr(config.c_ip_addr))
 380                 && (strcmp(config.c_ip_addr, "*"))
 381                 && (strcmp(config.c_ip_addr, "::"))
 382                 && (strcmp(config.c_ip_addr, "0.0.0.0"))
 383         ) {
 384                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
 385         }
 386
 387         if (CtdlThreadCheckStop())
 388                 goto shutdown ;
 389
 390         rc = curl_easy_perform(curl);
 391         if (rc) {
 392                 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", rc, errmsg);
 393                 rc = 0;
 394         }
 395         else
 396                 rc = 1;
 397
 398 shutdown:
 399         curl_easy_cleanup(curl);
 400
 401         return rc;
 402
 403 }
 404
 405
 406
 407 void CrawlMessageForShorterUrls(HashList *pUrls, StrBuf *Message)
 408 {
 409         int nHits = 0;
 410         const char *pShortenerService;
 411         int nShorter = 0;
 412         const char *pch;
 413         const char *pUrl;
 414         ConstStr *pCUrl;
 415
 416         pShortenerService = LinkShortenerServices[nShorter++];
 417         while (pShortenerService != NULL)
 418         {
 419                 pch = ChrPtr(Message);
 420                 pUrl = strstr(pch, pShortenerService);
 421                 while ((pUrl != NULL) && (nHits < 99))
 422                 {
 423                         pCUrl = malloc(sizeof(ConstStr));
 424
 425                         pCUrl->Key = pUrl;
 426                         pch = pUrl + strlen(pShortenerService);
 427                         while (isalnum(*pch)||(*pch == '-')||(*pch == '/'))
 428                                 pch++;
 429                         pCUrl->len = pch - pCUrl->Key;
 430
 431                         Put(pUrls, IKEY(nHits), pCUrl, NULL);
 432                         nHits ++;
 433                         pUrl = strstr(pch, pShortenerService);
 434                 }
 435                 pShortenerService = LinkShortenerServices[nShorter++];
 436         }
 437 }
 438
 439 int SortConstStrByPosition(const void *Item1, const void *Item2)
 440 {
 441         const ConstStr *p1, *p2;
 442         p1 = (const ConstStr*) Item1;
 443         p2 = (const ConstStr*) Item2;
 444         if (p1->Key == p2->Key)
 445                 return 0;
 446         if (p1->Key > p2->Key)
 447                 return 1;
 448         return -1;
 449 }
 450
 451 void ExpandShortUrls(StrBuf *Message)
 452 {
 453         StrBuf *Shadow;
 454         HashList *pUrls;
 455         ConstStr *pCUrl;
 456         const char *pch;
 457         const char *pche;
 458
 459         /* we just suspect URL shorteners to be inside of feeds from twitter
 460          * or other short content messages, so don't crawl through real blogs.
 461          */
 462         if (StrLength(Message) > 500)
 463                 return;
 464
 465         pUrls = NewHash(1, Flathash);
 466         CrawlMessageForShorterUrls(pUrls, Message);
 467
 468         if (GetCount(pUrls) > 0)
 469         {
 470                 StrBuf *ShorterUrlStr;
 471                 HashPos *Pos;
 472                 const char *Key;
 473                 void *pv;
 474                 long len;
 475
 476                 Shadow = NewStrBufPlain(NULL, StrLength(Message));
 477                 SortByPayload (pUrls, SortConstStrByPosition);
 478
 479                 ShorterUrlStr = NewStrBufPlain(NULL, StrLength(Message));
 480
 481                 pch = ChrPtr(Message);
 482                 pche = pch + StrLength(Message);
 483                 Pos = GetNewHashPos(pUrls, 1);
 484                 while (GetNextHashPos(pUrls, Pos, &len, &Key, &pv))
 485                 {
 486                         pCUrl = (ConstStr*) pv;
 487
 488                         if (pch != pCUrl->Key)
 489                                 StrBufAppendBufPlain(Shadow, pch, pCUrl->Key - pch, 0);
 490
 491                         StrBufPlain(ShorterUrlStr, CKEY(*pCUrl));
 492                         if (LookupUrl(ShorterUrlStr))
 493                         {
 494                                 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 495                                 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 496                                 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 497                                 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 498                                 StrBufAppendBufPlain(Shadow, HKEY("["), 0);
 499                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 500                                 StrBufAppendBufPlain(Shadow, HKEY("]</a>"), 0);
 501                         }
 502                         else
 503                         {
 504                                 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 505                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 506                                 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 507                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 508                                 StrBufAppendBufPlain(Shadow, HKEY("</a>"), 0);
 509                         }
 510                         pch = pCUrl->Key + pCUrl->len + 1;
 511
 512                 }
 513                 if (pch < pche)
 514                         StrBufAppendBufPlain(Shadow, pch, pche - pch, 0);
 515                 FlushStrBuf(Message);
 516                 StrBufAppendBuf(Message, Shadow, 0);
 517
 518                 FreeStrBuf(&ShorterUrlStr);
 519                 FreeStrBuf(&Shadow);
 520                 DeleteHashPos(&Pos);
 521         }
 522
 523         DeleteHash(&pUrls);
 524 }
 525
 526
 527 void AppendLink(StrBuf *Message, StrBuf *link, StrBuf *LinkTitle, const char *Title)
 528 {
 529         if (StrLength(link) > 0)
 530         {
 531                 StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
 532                 StrBufAppendBuf(Message, link, 0);
 533                 StrBufAppendBufPlain(Message, HKEY("\">"), 0);
 534                 if (StrLength(LinkTitle) > 0)
 535                         StrBufAppendBuf(Message, LinkTitle, 0);
 536                 else if ((Title != NULL) && !IsEmptyStr(Title))
 537                         StrBufAppendBufPlain(Message, Title, -1, 0);
 538                 else
 539                         StrBufAppendBuf(Message, link, 0);
 540                 StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
 541         }
 542 }
 543 /*
 544  * Commit a fetched and parsed RSS item to disk
 545  */
 546 void rss_save_item(rss_item *ri)
 547 {
 548
 549         struct MD5Context md5context;
 550         u_char rawdigest[MD5_DIGEST_LEN];
 551         int i;
 552         char utmsgid[SIZ];
 553         struct cdbdata *cdbut;
 554         struct UseTable ut;
 555         struct CtdlMessage *msg;
 556         struct recptypes *recp = NULL;
 557         int msglen = 0;
 558         StrBuf *Message;
 559
 560         recp = (struct recptypes *) malloc(sizeof(struct recptypes));
 561         if (recp == NULL) return;
 562         memset(recp, 0, sizeof(struct recptypes));
 563         memset(&ut, 0, sizeof(struct UseTable));
 564         recp->recp_room = strdup(ri->roomlist);
 565         recp->num_room = num_tokens(ri->roomlist, '|');
 566         recp->recptypes_magic = RECPTYPES_MAGIC;
 567
 568         /* Construct a GUID to use in the S_USETABLE table.
 569          * If one is not present in the item itself, make one up.
 570          */
 571         if (ri->guid != NULL) {
 572                 StrBufSpaceToBlank(ri->guid);
 573                 StrBufTrim(ri->guid);
 574                 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ChrPtr(ri->guid));
 575         }
 576         else {
 577                 MD5Init(&md5context);
 578                 if (ri->title != NULL) {
 579                         MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->title), StrLength(ri->title));
 580                 }
 581                 if (ri->link != NULL) {
 582                         MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->link), StrLength(ri->link));
 583                 }
 584                 MD5Final(rawdigest, &md5context);
 585                 for (i=0; i<MD5_DIGEST_LEN; i++) {
 586                         sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
 587                         utmsgid[i*2] = tolower(utmsgid[i*2]);
 588                         utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
 589                 }
 590                 strcat(utmsgid, "_rss2ctdl");
 591         }
 592
 593         /* Find out if we've already seen this item */
 594
 595         cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
 596 #ifndef DEBUG_RSS
 597         if (cdbut != NULL) {
 598                 /* Item has already been seen */
 599                 CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
 600                 cdb_free(cdbut);
 601
 602                 /* rewrite the record anyway, to update the timestamp */
 603                 strcpy(ut.ut_msgid, utmsgid);
 604                 ut.ut_timestamp = time(NULL);
 605                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 606         }
 607         else
 608 #endif
 609 {
 610                 /* Item has not been seen, so save it. */
 611                 CtdlLogPrintf(CTDL_DEBUG, "RSS: saving item...\n");
 612                 if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY(""));
 613                 StrBufSpaceToBlank(ri->description);
 614                 msg = malloc(sizeof(struct CtdlMessage));
 615                 memset(msg, 0, sizeof(struct CtdlMessage));
 616                 msg->cm_magic = CTDLMESSAGE_MAGIC;
 617                 msg->cm_anon_type = MES_NORMAL;
 618                 msg->cm_format_type = FMT_RFC822;
 619
 620                 if (ri->guid != NULL) {
 621                         msg->cm_fields['E'] = strdup(ChrPtr(ri->guid));
 622                 }
 623
 624                 if (ri->author_or_creator != NULL) {
 625                         char *From;
 626                         StrBuf *Encoded = NULL;
 627                         int FromAt;
 628                         int FromLen;
 629
 630                         From = html_to_ascii(ChrPtr(ri->author_or_creator),
 631                                              StrLength(ri->author_or_creator),
 632                                              512, 0);
 633                         StrBufPlain(ri->author_or_creator, From, -1);
 634                         StrBufTrim(ri->author_or_creator);
 635                         free(From);
 636
 637                         FromAt = strchr(ChrPtr(ri->author_or_creator), '@') != NULL;
 638                         if (!FromAt && StrLength (ri->author_email) > 0)
 639                         {
 640                                 StrBufRFC2047encode(&Encoded, ri->author_or_creator);
 641                                 msg->cm_fields['A'] = SmashStrBuf(&Encoded);
 642                                 msg->cm_fields['P'] = SmashStrBuf(&ri->author_email);
 643                         }
 644                         else
 645                         {
 646                                 if (FromAt)
 647                                         msg->cm_fields['P'] = SmashStrBuf(&ri->author_or_creator);
 648                                 else
 649                                 {
 650                                         StrBufRFC2047encode(&Encoded, ri->author_or_creator);
 651                                         msg->cm_fields['A'] = SmashStrBuf(&Encoded);
 652                                         msg->cm_fields['P'] = strdup("rss@localhost");
 653                                 }
 654                         }
 655                 }
 656                 else {
 657                         msg->cm_fields['A'] = strdup("rss");
 658                 }
 659
 660                 msg->cm_fields['N'] = strdup(NODENAME);
 661                 if (ri->title != NULL) {
 662                         long len;
 663                         char *Sbj;
 664                         StrBuf *Encoded, *QPEncoded;
 665
 666                         QPEncoded = NULL;
 667                         StrBufSpaceToBlank(ri->title);
 668                         len = StrLength(ri->title);
 669                         Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
 670                         len = strlen(Sbj);
 671                         if (Sbj[len - 1] == '\n')
 672                         {
 673                                 len --;
 674                                 Sbj[len] = '\0';
 675                         }
 676                         Encoded = NewStrBufPlain(Sbj, len);
 677                         free(Sbj);
 678
 679                         StrBufTrim(Encoded);
 680                         StrBufRFC2047encode(&QPEncoded, Encoded);
 681
 682                         msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
 683                         FreeStrBuf(&Encoded);
 684                 }
 685                 msg->cm_fields['T'] = malloc(64);
 686                 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
 687                 if (ri->channel_title != NULL) {
 688                         if (StrLength(ri->channel_title) > 0) {
 689                                 msg->cm_fields['O'] = strdup(ChrPtr(ri->channel_title));
 690                         }
 691                 }
 692                 if (ri->link == NULL)
 693                         ri->link = NewStrBufPlain(HKEY(""));
 694                 ExpandShortUrls(ri->description);
 695                 msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
 696
 697                 Message = NewStrBufPlain(NULL, StrLength(ri->description));
 698
 699                 StrBufPlain(Message, HKEY(
 700                          "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 701                          "<html><body>\n"));
 702
 703                 StrBufAppendBuf(Message, ri->description, 0);
 704                 StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
 705
 706                 AppendLink(Message, ri->link, ri->linkTitle, NULL);
 707                 AppendLink(Message, ri->reLink, ri->reLinkTitle, "Reply to this");
 708                 StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
 709
 710                 msg->cm_fields['M'] = SmashStrBuf(&Message);
 711
 712                 CtdlSubmitMsg(msg, recp, NULL, 0);
 713                 CtdlFreeMessage(msg);
 714
 715                 /* write the uidl to the use table so we don't store this item again */
 716                 strcpy(ut.ut_msgid, utmsgid);
 717                 ut.ut_timestamp = time(NULL);
 718                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 719         }
 720         free_recipients(recp);
 721 }
 722
 723
 724
 725 /*
 726  * Convert an RDF/RSS datestamp into a time_t
 727  */
 728 time_t rdf_parsedate(const char *p)
 729 {
 730         struct tm tm;
 731         time_t t = 0;
 732
 733         if (!p) return 0L;
 734         if (strlen(p) < 10) return 0L;
 735
 736         memset(&tm, 0, sizeof tm);
 737
 738         /*
 739          * If the timestamp appears to be in W3C datetime format, try to
 740          * parse it.  See also: http://www.w3.org/TR/NOTE-datetime
 741          *
 742          * This code, along with parsedate.c, is a potential candidate for
 743          * moving into libcitadel.
 744          */
 745         if ( (p[4] == '-') && (p[7] == '-') ) {
 746                 tm.tm_year = atoi(&p[0]) - 1900;
 747                 tm.tm_mon = atoi(&p[5]) - 1;
 748                 tm.tm_mday = atoi(&p[8]);
 749                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 750                         tm.tm_hour = atoi(&p[11]);
 751                         tm.tm_min = atoi(&p[14]);
 752                 }
 753                 return mktime(&tm);
 754         }
 755
 756         /* hmm... try RFC822 date stamp format */
 757
 758         t = parsedate(p);
 759         if (t > 0) return(t);
 760
 761         /* yeesh.  ok, just return the current date and time. */
 762         return(time(NULL));
 763 }
 764
 765 void flush_rss_item(rss_item *ri)
 766 {
 767         /* Initialize the feed item data structure */
 768         FreeStrBuf(&ri->guid);
 769         FreeStrBuf(&ri->title);
 770         FreeStrBuf(&ri->link);
 771         FreeStrBuf(&ri->author_or_creator);
 772         FreeStrBuf(&ri->author_email);
 773         FreeStrBuf(&ri->author_url);
 774         FreeStrBuf(&ri->description);
 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 ///     CtdlLogPrintf(0, "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                         CtdlLogPrintf(0, "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                         CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
 835 #endif
 836         }
 837 #ifdef DEBUG_RSS
 838         else
 839                 CtdlLogPrintf(0, "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 //      CtdlLogPrintf(0, "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                         CtdlLogPrintf(0, "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                         CtdlLogPrintf(0, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 901 #endif
 902         }
 903 #ifdef DEBUG_RSS
 904         else
 905                 CtdlLogPrintf(0, "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         CtdlLogPrintf(CTDL_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         CtdlLogPrintf(CTDL_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         CtdlLogPrintf(CTDL_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 //              CtdlLogPrintf(CTDL_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 //              CtdlLogPrintf(CTDL_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         memset(&ri, 0, sizeof(rss_item));
1269         rssc.Item = &ri;
1270         rssc.Cfg = Cfg;
1271
1272         CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
1273
1274         curl = curl_easy_init();
1275         if (!curl) {
1276                 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
1277                 return;
1278         }
1279         Answer = NewStrBufPlain(NULL, SIZ);
1280
1281         curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
1282         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
1283         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
1284         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
1285 //      curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
1286         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
1287         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
1288         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
1289 #ifdef CURLOPT_HTTP_CONTENT_DECODING
1290         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
1291         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
1292 #endif
1293         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
1294         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
1295         if (
1296                 (!IsEmptyStr(config.c_ip_addr))
1297                 && (strcmp(config.c_ip_addr, "*"))
1298                 && (strcmp(config.c_ip_addr, "::"))
1299                 && (strcmp(config.c_ip_addr, "0.0.0.0"))
1300         ) {
1301                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
1302         }
1303
1304         if (CtdlThreadCheckStop())
1305         {
1306                 curl_easy_cleanup(curl);
1307                 return;
1308         }
1309
1310         if (CtdlThreadCheckStop())
1311                 goto shutdown ;
1312
1313         res = curl_easy_perform(curl);
1314         if (res) {
1315                 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
1316         }
1317
1318         if (CtdlThreadCheckStop())
1319                 goto shutdown ;
1320
1321
1322
1323
1324         memset(&ri, 0, sizeof(rss_item));
1325         ri.roomlist = Cfg->rooms;
1326         rssc.CData = NewStrBufPlain(NULL, SIZ);
1327         rssc.Key = NewStrBuf();
1328         at = NULL;
1329         StrBufSipLine(rssc.Key, Answer, &at);
1330         ptr = NULL;
1331
1332 #define encoding "encoding=\""
1333         ptr = strstr(ChrPtr(rssc.Key), encoding);
1334         if (ptr != NULL)
1335         {
1336                 char *pche;
1337
1338                 ptr += sizeof (encoding) - 1;
1339                 pche = strchr(ptr, '"');
1340                 if (pche != NULL)
1341                         StrBufCutAt(rssc.Key, -1, pche);
1342                 else
1343                         ptr = "UTF-8";
1344         }
1345         else
1346                 ptr = "UTF-8";
1347
1348
1349         xp = XML_ParserCreateNS(ptr, ':');
1350         if (!xp) {
1351                 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
1352                 goto shutdown;
1353         }
1354         FlushStrBuf(rssc.Key);
1355 //#ifdef HAVE_ICONV
1356 #if 0
1357         XML_SetUnknownEncodingHandler(xp,
1358                                       handle_unknown_xml_encoding,
1359                                       &rssc);
1360 #endif
1361 //#endif
1362         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
1363         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
1364         XML_SetUserData(xp, &rssc);
1365         XML_SetCdataSectionHandler(xp,
1366                                    rss_xml_cdata_start,
1367                                    rss_xml_cdata_end);
1368
1369
1370         len = StrLength(Answer);
1371         ptr = SmashStrBuf(&Answer);
1372         XML_Parse(xp, ptr, len, 0);
1373         free (ptr);
1374         if (ri.done_parsing == 0)
1375                 XML_Parse(xp, "", 0, 1);
1376
1377
1378         CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n",
1379                       XML_ErrorString(
1380                               XML_GetErrorCode(xp)));
1381
1382 shutdown:
1383         curl_easy_cleanup(curl);
1384         XML_ParserFree(xp);
1385
1386         flush_rss_item(&ri);
1387         FreeStrBuf(&rssc.CData);
1388         FreeStrBuf(&rssc.Key);
1389 }
1390
1391
1392 /*
1393  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
1394  */
1395 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
1396 {
1397         char filename[PATH_MAX];
1398         char buf[1024];
1399         char instr[32];
1400         FILE *fp;
1401         char feedurl[256];
1402         rssnetcfg *rncptr = NULL;
1403         rssnetcfg *use_this_rncptr = NULL;
1404         int len = 0;
1405         char *ptr = NULL;
1406
1407         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
1408
1409         if (CtdlThreadCheckStop())
1410                 return;
1411
1412         /* Only do net processing for rooms that have netconfigs */
1413         fp = fopen(filename, "r");
1414         if (fp == NULL) {
1415                 return;
1416         }
1417
1418         while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
1419                 buf[strlen(buf)-1] = 0;
1420
1421                 extract_token(instr, buf, 0, '|', sizeof instr);
1422                 if (!strcasecmp(instr, "rssclient")) {
1423
1424                         use_this_rncptr = NULL;
1425
1426                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
1427
1428                         /* If any other rooms have requested the same feed, then we will just add this
1429                          * room to the target list for that client request.
1430                          */
1431                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
1432                                 if (!strcmp(rncptr->url, feedurl)) {
1433                                         use_this_rncptr = rncptr;
1434                                 }
1435                         }
1436
1437                         /* Otherwise create a new client request */
1438                         if (use_this_rncptr == NULL) {
1439                                 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
1440                                 rncptr->ItemType = RSS_UNSET;
1441                                 if (rncptr != NULL) {
1442                                         rncptr->next = rnclist;
1443                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
1444                                         rncptr->rooms = NULL;
1445                                         rnclist = rncptr;
1446                                         use_this_rncptr = rncptr;
1447                                 }
1448                         }
1449
1450                         /* Add the room name to the request */
1451                         if (use_this_rncptr != NULL) {
1452                                 if (use_this_rncptr->rooms == NULL) {
1453                                         rncptr->rooms = strdup(qrbuf->QRname);
1454                                 }
1455                                 else {
1456                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
1457                                         ptr = realloc(use_this_rncptr->rooms, len);
1458                                         if (ptr != NULL) {
1459                                                 strcat(ptr, "|");
1460                                                 strcat(ptr, qrbuf->QRname);
1461                                                 use_this_rncptr->rooms = ptr;
1462                                         }
1463                                 }
1464                         }
1465                 }
1466
1467         }
1468
1469         fclose(fp);
1470
1471 }
1472
1473 /*
1474  * Scan for rooms that have RSS client requests configured
1475  */
1476 void rssclient_scan(void) {
1477         static time_t last_run = 0L;
1478         static int doing_rssclient = 0;
1479         rssnetcfg *rptr = NULL;
1480
1481         /*
1482          * This is a simple concurrency check to make sure only one rssclient run
1483          * is done at a time.  We could do this with a mutex, but since we
1484          * don't really require extremely fine granularity here, we'll do it
1485          * with a static variable instead.
1486          */
1487         if (doing_rssclient) return;
1488         doing_rssclient = 1;
1489
1490         CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
1491         CtdlForEachRoom(rssclient_scan_room, NULL);
1492
1493         while (rnclist != NULL && !CtdlThreadCheckStop()) {
1494                 rss_do_fetching(rnclist);
1495                 rptr = rnclist;
1496                 rnclist = rnclist->next;
1497                 if (rptr->rooms != NULL) free(rptr->rooms);
1498                 free(rptr);
1499         }
1500
1501         CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
1502         last_run = time(NULL);
1503         doing_rssclient = 0;
1504         return;
1505 }
1506
1507
1508 CTDL_MODULE_INIT(rssclient)
1509 {
1510         if (threading)
1511         {
1512                 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
1513                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
1514         }
1515
1516         StartHandlers = NewHash(1, NULL);
1517         EndHandlers = NewHash(1, NULL);
1518
1519         AddRSSStartHandler(RSS_item_rss_start,     RSS_UNSET, HKEY("rss"));
1520         AddRSSStartHandler(RSS_item_rdf_start,     RSS_UNSET, HKEY("rdf"));
1521         AddRSSStartHandler(ATOM_item_feed_start,    RSS_UNSET, HKEY("feed"));
1522         AddRSSStartHandler(RSS_item_item_start,    RSS_RSS, HKEY("item"));
1523         AddRSSStartHandler(ATOM_item_entry_start,  RSS_ATOM, HKEY("entry"));
1524         AddRSSStartHandler(ATOM_item_link_start,   RSS_ATOM, HKEY("link"));
1525
1526         AddRSSEndHandler(ATOMRSS_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1527         AddRSSEndHandler(RSS_item_guid_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
1528         AddRSSEndHandler(ATOM_item_id_end,         RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
1529         AddRSSEndHandler(RSS_item_link_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
1530 #if 0
1531 // hm, rss to the comments of that blog, might be interesting in future, but...
1532         AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("commentrss"));
1533 // comment count...
1534         AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("comments"));
1535 #endif
1536         AddRSSEndHandler(RSSATOM_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1537         AddRSSEndHandler(ATOM_item_content_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
1538         AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_ATOM|RSS_REQUIRE_BUF, HKEY("encoded"));
1539         AddRSSEndHandler(ATOM_item_summary_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
1540         AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
1541         AddRSSEndHandler(ATOM_item_published_end,  RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
1542         AddRSSEndHandler(ATOM_item_updated_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
1543         AddRSSEndHandler(RSS_item_pubdate_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
1544         AddRSSEndHandler(RSS_item_date_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
1545         AddRSSEndHandler(RSS_item_author_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
1546         AddRSSEndHandler(RSS_item_creator_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
1547 /* <author> */
1548         AddRSSEndHandler(ATOM_item_email_end,      RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
1549         AddRSSEndHandler(ATOM_item_name_end,       RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
1550         AddRSSEndHandler(ATOM_item_uri_end,        RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
1551 /* </author> */
1552         AddRSSEndHandler(RSS_item_item_end,        RSS_RSS, HKEY("item"));
1553         AddRSSEndHandler(RSS_item_rss_end,         RSS_RSS, HKEY("rss"));
1554         AddRSSEndHandler(RSS_item_rdf_end,         RSS_RSS, HKEY("rdf"));
1555         AddRSSEndHandler(ATOM_item_entry_end,      RSS_ATOM, HKEY("entry"));
1556
1557
1558 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
1559         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1560         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
1561         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1562         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
1563
1564 /* links to other feed generators... */
1565         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1566         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1567         AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1568         AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1569
1570         KnownNameSpaces = NewHash(1, NULL);
1571         Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
1572         Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
1573         Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
1574         Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
1575         Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
1576         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1577         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
1578         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
1579         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
1580         Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1581         Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
1582         Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
1583         Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
1584         Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
1585         Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
1586         Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
1587         Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
1588         Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
1589         Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1590         Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1591         Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
1592         Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
1593         Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
1594 #if 0
1595         /* we don't like these namespaces because of they shadow our usefull parameters. */
1596         Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
1597 #endif
1598         return "rssclient";
1599 }