citadel/modules/rssclient/rss_atom_parser.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 #include "event_client.h"
  59 #include "rss_atom_parser.h"
  60
  61 HashList *StartHandlers = NULL;
  62 HashList *EndHandlers = NULL;
  63 HashList *KnownNameSpaces = NULL;
  64 void AddRSSStartHandler(rss_handler_func Handler, int Flags, const char *key, long len)
  65 {
  66         rss_xml_handler *h;
  67         h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
  68         h->Flags = Flags;
  69         h->Handler = Handler;
  70         Put(StartHandlers, key, len, h, NULL);
  71 }
  72 void AddRSSEndHandler(rss_handler_func Handler, int Flags, const char *key, long len)
  73 {
  74         rss_xml_handler *h;
  75         h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
  76         h->Flags = Flags;
  77         h->Handler = Handler;
  78         Put(EndHandlers, key, len, h, NULL);
  79 }
  80
  81
  82
  83 /*
  84  * Convert an RDF/RSS datestamp into a time_t
  85  */
  86 time_t rdf_parsedate(const char *p)
  87 {
  88         struct tm tm;
  89         time_t t = 0;
  90
  91         if (!p) return 0L;
  92         if (strlen(p) < 10) return 0L;
  93
  94         memset(&tm, 0, sizeof tm);
  95
  96         /*
  97          * If the timestamp appears to be in W3C datetime format, try to
  98          * parse it.  See also: http://www.w3.org/TR/NOTE-datetime
  99          *
 100          * This code, along with parsedate.c, is a potential candidate for
 101          * moving into libcitadel.
 102          */
 103         if ( (p[4] == '-') && (p[7] == '-') ) {
 104                 tm.tm_year = atoi(&p[0]) - 1900;
 105                 tm.tm_mon = atoi(&p[5]) - 1;
 106                 tm.tm_mday = atoi(&p[8]);
 107                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 108                         tm.tm_hour = atoi(&p[11]);
 109                         tm.tm_min = atoi(&p[14]);
 110                 }
 111                 return mktime(&tm);
 112         }
 113
 114         /* hmm... try RFC822 date stamp format */
 115
 116         t = parsedate(p);
 117         if (t > 0) return(t);
 118
 119         /* yeesh.  ok, just return the current date and time. */
 120         return(time(NULL));
 121 }
 122
 123 void flush_rss_item(rss_item *ri)
 124 {
 125         /* Initialize the feed item data structure */
 126         FreeStrBuf(&ri->guid);
 127         FreeStrBuf(&ri->title);
 128         FreeStrBuf(&ri->link);
 129         FreeStrBuf(&ri->author_or_creator);
 130         FreeStrBuf(&ri->author_email);
 131         FreeStrBuf(&ri->author_url);
 132         FreeStrBuf(&ri->description);
 133 }
 134
 135 void rss_xml_start(void *data, const char *supplied_el, const char **attr)
 136 {
 137         rss_xml_handler *h;
 138         rsscollection   *rssc = (rsscollection*) data;
 139         rssnetcfg       *Cfg = rssc->Cfg;
 140         rss_item        *ri = rssc->Item;
 141         void            *pv;
 142         const char      *pel;
 143         char            *sep = NULL;
 144
 145         /* Axe the namespace, we don't care about it */
 146 ///     CtdlLogPrintf(0, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
 147         pel = supplied_el;
 148         while (sep = strchr(pel, ':'), sep) {
 149                 pel = sep + 1;
 150         }
 151
 152         if (pel != supplied_el)
 153         {
 154                 void *v;
 155
 156                 if (!GetHash(KnownNameSpaces,
 157                              supplied_el,
 158                              pel - supplied_el - 1,
 159                              &v))
 160                 {
 161 #ifdef DEBUG_RSS
 162                         CtdlLogPrintf(0, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
 163                                       supplied_el);
 164 #endif
 165                         return;
 166                 }
 167         }
 168
 169         StrBufPlain(rssc->Key, pel, -1);
 170         StrBufLowerCase(rssc->Key);
 171         if (GetHash(StartHandlers, SKEY(rssc->Key), &pv))
 172         {
 173                 rssc->Current = h = (rss_xml_handler*) pv;
 174
 175                 if (((h->Flags & RSS_UNSET) != 0) &&
 176                     (Cfg->ItemType == RSS_UNSET))
 177                 {
 178                         h->Handler(rssc->CData, ri, Cfg, attr);
 179                 }
 180                 else if (((h->Flags & RSS_RSS) != 0) &&
 181                     (Cfg->ItemType == RSS_RSS))
 182                 {
 183                         h->Handler(rssc->CData, ri, Cfg, attr);
 184                 }
 185                 else if (((h->Flags & RSS_ATOM) != 0) &&
 186                          (Cfg->ItemType == RSS_ATOM))
 187                 {
 188                         h->Handler(rssc->CData, ri, Cfg, attr);
 189                 }
 190 #ifdef DEBUG_RSS
 191                 else
 192                         CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
 193 #endif
 194         }
 195 #ifdef DEBUG_RSS
 196         else
 197                 CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel,  supplied_el);
 198 #endif
 199 }
 200
 201 void rss_xml_end(void *data, const char *supplied_el)
 202 {
 203         rss_xml_handler *h;
 204         rsscollection   *rssc = (rsscollection*) data;
 205         rssnetcfg       *Cfg = rssc->Cfg;
 206         rss_item        *ri = rssc->Item;
 207         const char      *pel;
 208         char            *sep = NULL;
 209         void            *pv;
 210
 211         /* Axe the namespace, we don't care about it */
 212         pel = supplied_el;
 213         while (sep = strchr(pel, ':'), sep) {
 214                 pel = sep + 1;
 215         }
 216 //      CtdlLogPrintf(0, "RSS: END %s...\n", el);
 217         if (pel != supplied_el)
 218         {
 219                 void *v;
 220
 221                 if (!GetHash(KnownNameSpaces,
 222                              supplied_el,
 223                              pel - supplied_el - 1,
 224                              &v))
 225                 {
 226 #ifdef DEBUG_RSS
 227                         CtdlLogPrintf(0, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
 228                                       supplied_el, ChrPtr(rssc->CData));
 229 #endif
 230                         FlushStrBuf(rssc->CData);
 231                         return;
 232                 }
 233         }
 234
 235         StrBufPlain(rssc->Key, pel, -1);
 236         StrBufLowerCase(rssc->Key);
 237         if (GetHash(EndHandlers, SKEY(rssc->Key), &pv))
 238         {
 239                 h = (rss_xml_handler*) pv;
 240
 241                 if (((h->Flags & RSS_UNSET) != 0) &&
 242                     (Cfg->ItemType == RSS_UNSET))
 243                 {
 244                         h->Handler(rssc->CData, ri, Cfg, NULL);
 245                 }
 246                 else if (((h->Flags & RSS_RSS) != 0) &&
 247                     (Cfg->ItemType == RSS_RSS))
 248                 {
 249                         h->Handler(rssc->CData, ri, Cfg, NULL);
 250                 }
 251                 else if (((h->Flags & RSS_ATOM) != 0) &&
 252                          (Cfg->ItemType == RSS_ATOM))
 253                 {
 254                         h->Handler(rssc->CData, ri, Cfg, NULL);
 255                 }
 256 #ifdef DEBUG_RSS
 257                 else
 258                         CtdlLogPrintf(0, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 259 #endif
 260         }
 261 #ifdef DEBUG_RSS
 262         else
 263                 CtdlLogPrintf(0, "RSS: END   unhandled: [%s]  [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
 264 #endif
 265         FlushStrBuf(rssc->CData);
 266         rssc->Current = NULL;
 267 }
 268
 269
 270
 271
 272
 273 void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 274 {
 275         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RSS feed.\n");
 276         Cfg->ItemType = RSS_RSS;
 277 }
 278
 279 void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 280 {
 281         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RDF feed.\n");
 282         Cfg->ItemType = RSS_RSS;
 283 }
 284
 285 void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 286 {
 287         CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an ATOM feed.\n");
 288         Cfg->ItemType = RSS_ATOM;
 289 }
 290
 291
 292 void RSS_item_item_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 293 {
 294         ri->item_tag_nesting ++;
 295         flush_rss_item(ri);
 296 }
 297
 298 void ATOM_item_entry_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 299 {
 300 /* Atom feed... */
 301         ri->item_tag_nesting ++;
 302         flush_rss_item(ri);
 303 }
 304
 305 void ATOM_item_link_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 306 {
 307         int i;
 308         const char *pHref = NULL;
 309         const char *pType = NULL;
 310         const char *pRel = NULL;
 311         const char *pTitle = NULL;
 312
 313         for (i = 0; Attr[i] != NULL; i+=2)
 314         {
 315                 if (!strcmp(Attr[i], "href"))
 316                 {
 317                         pHref = Attr[i+1];
 318                 }
 319                 else if (!strcmp(Attr[i], "rel"))
 320                 {
 321                         pRel = Attr[i+1];
 322                 }
 323                 else if (!strcmp(Attr[i], "type"))
 324                 {
 325                         pType = Attr[i+1];
 326                 }
 327                 else if (!strcmp(Attr[i], "title"))
 328                 {
 329                         pTitle = Attr[i+1];
 330                 }
 331         }
 332         if (pHref == NULL)
 333                 return; /* WHUT? Pointing... where? */
 334         if ((pType != NULL) && !strcasecmp(pType, "application/atom+xml"))
 335                 return; /* these just point to other rss resources, we're not interested in them. */
 336         if (pRel != NULL)
 337         {
 338                 if (!strcasecmp (pRel, "replies"))
 339                 {
 340                         NewStrBufDupAppendFlush(&ri->reLink, NULL, pHref, -1);
 341                         StrBufTrim(ri->link);
 342                         NewStrBufDupAppendFlush(&ri->reLinkTitle, NULL, pTitle, -1);
 343                 }
 344                 else if (!strcasecmp(pRel, "alternate")) /* Alternative representation of this Item... */
 345                 {
 346                         NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 347                         StrBufTrim(ri->link);
 348                         NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
 349
 350                 }
 351 #if 0 /* these are also defined, but dunno what to do with them.. */
 352                 else if (!strcasecmp(pRel, "related"))
 353                 {
 354                 }
 355                 else if (!strcasecmp(pRel, "self"))
 356                 {
 357                 }
 358                 else if (!strcasecmp(pRel, "enclosure"))
 359                 {/* this reference can get big, and is probably the full article... */
 360                 }
 361                 else if (!strcasecmp(pRel, "via"))
 362                 {/* this article was provided via... */
 363                 }
 364 #endif
 365         }
 366         else if (StrLength(ri->link) == 0)
 367         {
 368                 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
 369                 StrBufTrim(ri->link);
 370                 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
 371         }
 372 }
 373
 374
 375
 376
 377 void ATOMRSS_item_title_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 378 {
 379         if ((ri->item_tag_nesting == 0) && (StrLength(CData) > 0)) {
 380                 NewStrBufDupAppendFlush(&ri->channel_title, CData, NULL, 0);
 381                 StrBufTrim(ri->channel_title);
 382         }
 383 }
 384
 385 void RSS_item_guid_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 386 {
 387         if (StrLength(CData) > 0) {
 388                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
 389         }
 390 }
 391
 392 void ATOM_item_id_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 393 {
 394         if (StrLength(CData) > 0) {
 395                 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
 396         }
 397 }
 398
 399
 400 void RSS_item_link_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 401 {
 402         if (StrLength(CData) > 0) {
 403                 NewStrBufDupAppendFlush(&ri->link, CData, NULL, 0);
 404                 StrBufTrim(ri->link);
 405         }
 406 }
 407 void RSS_item_relink_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 408 {
 409         if (StrLength(CData) > 0) {
 410                 NewStrBufDupAppendFlush(&ri->reLink, CData, NULL, 0);
 411                 StrBufTrim(ri->reLink);
 412         }
 413 }
 414
 415 void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 416 {
 417         if (StrLength(CData) > 0) {
 418                 NewStrBufDupAppendFlush(&ri->title, CData, NULL, 0);
 419                 StrBufTrim(ri->title);
 420         }
 421 }
 422
 423 void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 424 {
 425         long olen = StrLength (ri->description);
 426         long clen = StrLength (CData);
 427         if (clen > 0)
 428         {
 429                 if (olen == 0) {
 430                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 431                         StrBufTrim(ri->description);
 432                 }
 433                 else if (olen < clen) {
 434                         FlushStrBuf(ri->description);
 435                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 436                         StrBufTrim(ri->description);
 437                 }
 438         }
 439 }
 440 void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 441 {
 442         /* this can contain an abstract of the article. but we don't want to verwrite a full document if we already have it. */
 443         if ((StrLength(CData) > 0) && (StrLength(ri->description) == 0))
 444         {
 445                 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 446                 StrBufTrim(ri->description);
 447         }
 448 }
 449
 450 void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 451 {
 452         long olen = StrLength (ri->description);
 453         long clen = StrLength (CData);
 454         if (clen > 0)
 455         {
 456                 if (olen == 0) {
 457                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 458                         StrBufTrim(ri->description);
 459                 }
 460                 else if (olen < clen) {
 461                         FlushStrBuf(ri->description);
 462                         NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
 463                         StrBufTrim(ri->description);
 464                 }
 465         }
 466 }
 467
 468 void ATOM_item_published_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 469 {
 470         if (StrLength(CData) > 0) {
 471                 StrBufTrim(CData);
 472                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 473         }
 474 }
 475
 476 void ATOM_item_updated_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 477 {
 478         if (StrLength(CData) > 0) {
 479                 StrBufTrim(CData);
 480                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 481         }
 482 }
 483
 484 void RSS_item_pubdate_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 485 {
 486         if (StrLength(CData) > 0) {
 487                 StrBufTrim(CData);
 488                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 489         }
 490 }
 491
 492
 493 void RSS_item_date_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 494 {
 495         if (StrLength(CData) > 0) {
 496                 StrBufTrim(CData);
 497                 ri->pubdate = rdf_parsedate(ChrPtr(CData));
 498         }
 499 }
 500
 501
 502
 503 void RSS_item_author_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 504 {
 505         if (StrLength(CData) > 0) {
 506                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
 507                 StrBufTrim(ri->author_or_creator);
 508         }
 509 }
 510
 511
 512 void ATOM_item_name_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 513 {
 514         if (StrLength(CData) > 0) {
 515                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
 516                 StrBufTrim(ri->author_or_creator);
 517         }
 518 }
 519
 520 void ATOM_item_email_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 521 {
 522         if (StrLength(CData) > 0) {
 523                 NewStrBufDupAppendFlush(&ri->author_email, CData, NULL, 0);
 524                 StrBufTrim(ri->author_email);
 525         }
 526 }
 527
 528 void RSS_item_creator_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 529 {
 530         if ((StrLength(CData) > 0) &&
 531             (StrLength(ri->author_or_creator) == 0))
 532         {
 533                 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
 534                 StrBufTrim(ri->author_or_creator);
 535         }
 536 }
 537
 538
 539 void ATOM_item_uri_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 540 {
 541         if (StrLength(CData) > 0) {
 542                 NewStrBufDupAppendFlush(&ri->author_url, CData, NULL, 0);
 543                 StrBufTrim(ri->author_url);
 544         }
 545 }
 546
 547 void RSS_item_item_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 548 {
 549         --ri->item_tag_nesting;
 550         rss_save_item(ri);
 551 }
 552
 553
 554 void ATOM_item_entry_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 555 {
 556         --ri->item_tag_nesting;
 557         rss_save_item(ri);
 558 }
 559
 560 void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 561 {
 562 //              CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
 563         ri->done_parsing = 1;
 564
 565 }
 566 void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 567 {
 568 //              CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
 569         ri->done_parsing = 1;
 570 }
 571
 572
 573 void RSSATOM_item_ignore(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
 574 {
 575 }
 576
 577
 578
 579 /*
 580  * This callback stores up the data which appears in between tags.
 581  */
 582 void rss_xml_cdata_start(void *data)
 583 {
 584         rsscollection *rssc = (rsscollection*) data;
 585
 586         FlushStrBuf(rssc->CData);
 587 }
 588
 589 void rss_xml_cdata_end(void *data)
 590 {
 591 }
 592 void rss_xml_chardata(void *data, const XML_Char *s, int len)
 593 {
 594         rsscollection *rssc = (rsscollection*) data;
 595
 596         StrBufAppendBufPlain (rssc->CData, s, len, 0);
 597 }
 598
 599 /*
 600  * Callback function for passing libcurl's output to expat for parsing
 601  */
 602 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
 603 {
 604         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
 605         return (size*nmemb);
 606 }
 607
 608
 609
 610 eNextState ParseRSSReply(AsyncIO *IO)
 611 {
 612         rsscollection *rssc;
 613         rss_item *ri;
 614         const char *at;
 615         char *ptr;
 616         long len;
 617
 618         rssc = IO->Data;
 619         ri = rssc->Item;
 620         rssc->CData = NewStrBufPlain(NULL, SIZ);
 621         rssc->Key = NewStrBuf();
 622         at = NULL;
 623         StrBufSipLine(rssc->Key, IO->HttpReq.ReplyData, &at);
 624         ptr = NULL;
 625
 626 #define encoding "encoding=\""
 627         ptr = strstr(ChrPtr(rssc->Key), encoding);
 628         if (ptr != NULL)
 629         {
 630                 char *pche;
 631
 632                 ptr += sizeof (encoding) - 1;
 633                 pche = strchr(ptr, '"');
 634                 if (pche != NULL)
 635                         StrBufCutAt(rssc->Key, -1, pche);
 636                 else
 637                         ptr = "UTF-8";
 638         }
 639         else
 640                 ptr = "UTF-8";
 641
 642         CtdlLogPrintf(CTDL_ALERT, "RSS: Now parsing [%s] \n", ChrPtr(rssc->Cfg->Url));
 643
 644         rssc->xp = XML_ParserCreateNS(ptr, ':');
 645         if (!rssc->xp) {
 646                 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
 647                 goto shutdown;
 648         }
 649         FlushStrBuf(rssc->Key);
 650
 651         XML_SetElementHandler(rssc->xp, rss_xml_start, rss_xml_end);
 652         XML_SetCharacterDataHandler(rssc->xp, rss_xml_chardata);
 653         XML_SetUserData(rssc->xp, rssc);
 654         XML_SetCdataSectionHandler(rssc->xp,
 655                                    rss_xml_cdata_start,
 656                                    rss_xml_cdata_end);
 657
 658
 659         len = StrLength(IO->HttpReq.ReplyData);
 660         ptr = SmashStrBuf(&IO->HttpReq.ReplyData);
 661         XML_Parse(rssc->xp, ptr, len, 0);
 662         free (ptr);
 663         if (ri->done_parsing == 0)
 664                 XML_Parse(rssc->xp, "", 0, 1);
 665
 666
 667         CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n",
 668                       XML_ErrorString(
 669                               XML_GetErrorCode(rssc->xp)));
 670
 671 shutdown:
 672         XML_ParserFree(rssc->xp);
 673
 674         flush_rss_item(ri);
 675         FreeStrBuf(&rssc->CData);
 676         FreeStrBuf(&rssc->Key);
 677
 678         ///Cfg->next_poll = time(NULL) + config.c_net_freq;
 679
 680         return eTerminateConnection;
 681 }
 682
 683
 684 void rss_cleanup(void)
 685 {
 686         DeleteHash(&StartHandlers);
 687         DeleteHash(&EndHandlers);
 688         DeleteHash(&KnownNameSpaces);
 689 }
 690
 691
 692 CTDL_MODULE_INIT(rssparser)
 693 {
 694         if (!threading)
 695         {
 696                 StartHandlers = NewHash(1, NULL);
 697                 EndHandlers = NewHash(1, NULL);
 698
 699                 AddRSSStartHandler(RSS_item_rss_start,     RSS_UNSET, HKEY("rss"));
 700                 AddRSSStartHandler(RSS_item_rdf_start,     RSS_UNSET, HKEY("rdf"));
 701                 AddRSSStartHandler(ATOM_item_feed_start,    RSS_UNSET, HKEY("feed"));
 702                 AddRSSStartHandler(RSS_item_item_start,    RSS_RSS, HKEY("item"));
 703                 AddRSSStartHandler(ATOM_item_entry_start,  RSS_ATOM, HKEY("entry"));
 704                 AddRSSStartHandler(ATOM_item_link_start,   RSS_ATOM, HKEY("link"));
 705
 706                 AddRSSEndHandler(ATOMRSS_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
 707                 AddRSSEndHandler(RSS_item_guid_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
 708                 AddRSSEndHandler(ATOM_item_id_end,         RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
 709                 AddRSSEndHandler(RSS_item_link_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
 710 #if 0
 711 // hm, rss to the comments of that blog, might be interesting in future, but...
 712                 AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("commentrss"));
 713 // comment count...
 714                 AddRSSEndHandler(RSS_item_relink_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("comments"));
 715 #endif
 716                 AddRSSEndHandler(RSSATOM_item_title_end,   RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
 717                 AddRSSEndHandler(ATOM_item_content_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
 718                 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_ATOM|RSS_REQUIRE_BUF, HKEY("encoded"));
 719                 AddRSSEndHandler(ATOM_item_summary_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
 720                 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
 721                 AddRSSEndHandler(ATOM_item_published_end,  RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
 722                 AddRSSEndHandler(ATOM_item_updated_end,    RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
 723                 AddRSSEndHandler(RSS_item_pubdate_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
 724                 AddRSSEndHandler(RSS_item_date_end,        RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
 725                 AddRSSEndHandler(RSS_item_author_end,      RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
 726                 AddRSSEndHandler(RSS_item_creator_end,     RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
 727 /* <author> */
 728                 AddRSSEndHandler(ATOM_item_email_end,      RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
 729                 AddRSSEndHandler(ATOM_item_name_end,       RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
 730                 AddRSSEndHandler(ATOM_item_uri_end,        RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
 731 /* </author> */
 732                 AddRSSEndHandler(RSS_item_item_end,        RSS_RSS, HKEY("item"));
 733                 AddRSSEndHandler(RSS_item_rss_end,         RSS_RSS, HKEY("rss"));
 734                 AddRSSEndHandler(RSS_item_rdf_end,         RSS_RSS, HKEY("rdf"));
 735                 AddRSSEndHandler(ATOM_item_entry_end,      RSS_ATOM, HKEY("entry"));
 736
 737
 738 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
 739                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
 740                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("seq"));
 741                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
 742                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("li"));
 743
 744 /* links to other feed generators... */
 745                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
 746                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("feedflare"));
 747                 AddRSSStartHandler(RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
 748                 AddRSSEndHandler  (RSSATOM_item_ignore,      RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
 749
 750                 KnownNameSpaces = NewHash(1, NULL);
 751                 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
 752                 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
 753                 Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
 754                 Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
 755                 Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
 756                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
 757                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
 758                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
 759                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
 760                 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
 761                 Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
 762                 Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
 763                 Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
 764                 Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
 765                 Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
 766                 Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
 767                 Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
 768                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
 769                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
 770                 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
 771                 Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
 772                 Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
 773                 Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
 774 #if 0
 775                 /* we don't like these namespaces because of they shadow our usefull parameters. */
 776                 Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
 777 #endif
 778                 CtdlRegisterCleanupHook(rss_cleanup);
 779         }
 780         return "rssparser";
 781 }