2 * Bring external RSS feeds into rooms.
4 * Copyright (c) 2007-2010 by the citadel.org team
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.
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.
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
25 #if TIME_WITH_SYS_TIME
26 # include <sys/time.h>
30 # include <sys/time.h>
39 #include <sys/types.h>
42 #include <curl/curl.h>
43 #include <libcitadel.h>
46 #include "citserver.h"
50 #include "ctdl_module.h"
51 #include "clientsocket.h"
53 #include "parsedate.h"
55 #include "citadel_dirs.h"
60 typedef struct rssnetcfg rssnetcfg;
65 time_t last_error_when;
70 #define RSS_UNSET (1<<0)
71 #define RSS_RSS (1<<1)
72 #define RSS_ATOM (1<<2)
73 #define RSS_REQUIRE_BUF (1<<3)
75 typedef struct _rss_item {
86 StrBuf *channel_title;
88 StrBuf *author_or_creator;
94 typedef void (*rss_handler_func)(StrBuf *CData,
99 typedef struct __rss_xml_handler {
101 rss_handler_func Handler;
105 typedef struct _rsscollection {
112 rss_xml_handler *Current;
115 struct rssnetcfg *rnclist = NULL;
116 HashList *StartHandlers = NULL;
117 HashList *EndHandlers = NULL;
118 HashList *KnownNameSpaces = NULL;
119 HashList *UrlShorteners = NULL;
120 void AddRSSStartHandler(rss_handler_func Handler, int Flags, const char *key, long len)
123 h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
125 h->Handler = Handler;
126 Put(StartHandlers, key, len, h, NULL);
128 void AddRSSEndHandler(rss_handler_func Handler, int Flags, const char *key, long len)
131 h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
133 h->Handler = Handler;
134 Put(EndHandlers, key, len, h, NULL);
143 * dug this out of the trashcan of the midgard project, lets see if it works for us.
144 * original code by Alexander Bokovoy <bokovoy@avilink.ne> distributed under GPL V2 or later
148 >= 0 - successfull, 0 means conversion doesn't use multibyte sequences
149 -1 - error during iconv_open call
150 -2 - error during iconv_close call
151 ----------------------------------
152 This function expects that multibyte encoding in 'charset' wouldn't have
153 characters with more than 3 bytes. It is not intended to convert UTF-8 because
154 we'll never receive UTF-8 in our handler (it is handled by Exat itself).
157 fill_encoding_info (const char *charset, XML_Encoding * info)
159 iconv_t cd = (iconv_t)(-1);
161 syslog(LOG_EMERG, "RSS: fill encoding info ...\n");
163 #if G_BYTE_ORDER == G_LITTLE_ENDIAN
164 cd = iconv_open ("UCS-2LE", charset);
166 cd = iconv_open ("UCS-2BE", charset);
169 if (cd == (iconv_t) (-1))
175 unsigned short out = 0;
176 unsigned char buf[4];
177 unsigned int i0, i1, i2;
180 for (i0 = 0; i0 < 0x100; i0++)
184 //result = try (cd, buf, 1, &out);
194 for (i1 = 0; i1 < 0x100; i1++)
197 ///result = try (cd, buf, 2, &out);
208 for (i2 = 0; i2 < 0x100; i2++)
211 ////result = try (cd, buf, 3, &out);
227 if (iconv_close (cd) < 0)
235 iconv_convertor (void *data, const char *s)
237 XML_Encoding *info = data;
239 syslog(LOG_EMERG, "RSS: Converting ...\n");
245 result = g_byte_array_new ();
246 if (process_block (info->data, (char *) s, strlen (s), result) == 0)
248 res = *(result->data);
249 g_byte_array_free (result, TRUE);
252 g_byte_array_free (result, TRUE);
258 my_release (void *data)
260 iconv_t cd = (iconv_t) data;
261 if (iconv_close (cd) != 0)
263 /// TODO: uh no. exit (1);
267 handle_unknown_xml_encoding (void *encodingHandleData,
268 const XML_Char * name,
272 syslog(LOG_EMERG, "RSS: unknown encoding ...\n");
273 result = fill_encoding_info (name, info);
277 Special case: client asked for reverse conversion, we'll provide him with
278 iconv descriptor which handles it. Client should release it by himself.
280 if(encodingHandleData != NULL)
281 *((iconv_t *)encodingHandleData) = iconv_open(name, "UTF-8");
283 Optimization: we do not need conversion function if encoding is one-to-one,
284 info->map table will be enough
289 info->convert = NULL;
290 info->release = NULL;
294 We do need conversion function because this encoding uses multibyte sequences
296 info->data = (void *) iconv_open ("UTF-8", name);
297 if ((int)info->data == -1)
299 info->convert = iconv_convertor;
300 info->release = my_release;
303 if(encodingHandleData != NULL)
304 *(iconv_t *)encodingHandleData = NULL;
310 size_t GetLocationString( void *ptr, size_t size, size_t nmemb, void *userdata)
312 #define LOCATION "location"
313 if (strncasecmp((char*)ptr, LOCATION, sizeof(LOCATION) - 1) == 0)
315 StrBuf *pURL = (StrBuf*) userdata;
316 char *pch = (char*) ptr;
319 pche = pch + (size * nmemb);
320 pch += sizeof(LOCATION);
322 while (isspace(*pch) || (*pch == ':'))
325 while (isspace(*pche) || (*pche == '\0'))
329 StrBufPlain(pURL, pch, pche - pch + 1);
334 int LookupUrl(StrBuf *ShorterUrlStr)
337 char errmsg[1024] = "";
341 curl = curl_easy_init();
343 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
346 Answer = NewStrBufPlain(NULL, SIZ);
348 curl_easy_setopt(curl, CURLOPT_URL, ChrPtr(ShorterUrlStr));
349 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
350 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
351 curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
352 // curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
353 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
354 curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
355 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
356 #ifdef CURLOPT_HTTP_CONTENT_DECODING
357 curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
358 curl_easy_setopt(curl, CURLOPT_ENCODING, "");
360 curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
361 curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180); /* die after 180 seconds */
362 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 0);
364 curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION , GetLocationString);
365 curl_easy_setopt(curl, CURLOPT_WRITEHEADER, ShorterUrlStr);
369 (!IsEmptyStr(config.c_ip_addr))
370 && (strcmp(config.c_ip_addr, "*"))
371 && (strcmp(config.c_ip_addr, "::"))
372 && (strcmp(config.c_ip_addr, "0.0.0.0"))
374 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
377 if (server_shutting_down)
380 rc = curl_easy_perform(curl);
382 syslog(LOG_ALERT, "libcurl error %d: %s\n", rc, errmsg);
389 curl_easy_cleanup(curl);
397 void CrawlMessageForShorterUrls(HashList *pUrls, StrBuf *Message)
406 while (GetHash(UrlShorteners, IKEY(nShorter), &pv))
409 pch = ChrPtr(Message);
410 pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
411 while ((pUrl != NULL) && (nHits < 99))
413 pCUrl = malloc(sizeof(ConstStr));
416 pch = pUrl + StrLength((StrBuf*)pv);
417 while (isalnum(*pch)||(*pch == '-')||(*pch == '/'))
419 pCUrl->len = pch - pCUrl->Key;
421 Put(pUrls, IKEY(nHits), pCUrl, NULL);
423 pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
428 int SortConstStrByPosition(const void *Item1, const void *Item2)
430 const ConstStr *p1, *p2;
431 p1 = (const ConstStr*) Item1;
432 p2 = (const ConstStr*) Item2;
433 if (p1->Key == p2->Key)
435 if (p1->Key > p2->Key)
440 void ExpandShortUrls(StrBuf *Message)
448 /* we just suspect URL shorteners to be inside of feeds from twitter
449 * or other short content messages, so don't crawl through real blogs.
451 if (StrLength(Message) > 500)
454 pUrls = NewHash(1, Flathash);
455 CrawlMessageForShorterUrls(pUrls, Message);
457 if (GetCount(pUrls) > 0)
459 StrBuf *ShorterUrlStr;
465 Shadow = NewStrBufPlain(NULL, StrLength(Message));
466 SortByPayload (pUrls, SortConstStrByPosition);
468 ShorterUrlStr = NewStrBufPlain(NULL, StrLength(Message));
470 pch = ChrPtr(Message);
471 pche = pch + StrLength(Message);
472 Pos = GetNewHashPos(pUrls, 1);
473 while (GetNextHashPos(pUrls, Pos, &len, &RetrKey, &pv))
475 pCUrl = (ConstStr*) pv;
477 if (pch != pCUrl->Key)
478 StrBufAppendBufPlain(Shadow, pch, pCUrl->Key - pch, 0);
480 StrBufPlain(ShorterUrlStr, CKEY(*pCUrl));
481 if (LookupUrl(ShorterUrlStr))
483 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
484 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
485 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
486 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
487 StrBufAppendBufPlain(Shadow, HKEY("["), 0);
488 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
489 StrBufAppendBufPlain(Shadow, HKEY("]</a>"), 0);
493 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
494 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
495 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
496 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
497 StrBufAppendBufPlain(Shadow, HKEY("</a>"), 0);
499 pch = pCUrl->Key + pCUrl->len + 1;
503 StrBufAppendBufPlain(Shadow, pch, pche - pch, 0);
504 FlushStrBuf(Message);
505 StrBufAppendBuf(Message, Shadow, 0);
507 FreeStrBuf(&ShorterUrlStr);
516 void AppendLink(StrBuf *Message, StrBuf *link, StrBuf *LinkTitle, const char *Title)
518 if (StrLength(link) > 0)
520 StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
521 StrBufAppendBuf(Message, link, 0);
522 StrBufAppendBufPlain(Message, HKEY("\">"), 0);
523 if (StrLength(LinkTitle) > 0)
524 StrBufAppendBuf(Message, LinkTitle, 0);
525 else if ((Title != NULL) && !IsEmptyStr(Title))
526 StrBufAppendBufPlain(Message, Title, -1, 0);
528 StrBufAppendBuf(Message, link, 0);
529 StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
533 * Commit a fetched and parsed RSS item to disk
535 void rss_save_item(rss_item *ri)
538 struct MD5Context md5context;
539 u_char rawdigest[MD5_DIGEST_LEN];
542 struct cdbdata *cdbut;
544 struct CtdlMessage *msg;
545 struct recptypes *recp = NULL;
549 recp = (struct recptypes *) malloc(sizeof(struct recptypes));
550 if (recp == NULL) return;
551 memset(recp, 0, sizeof(struct recptypes));
552 memset(&ut, 0, sizeof(struct UseTable));
553 recp->recp_room = strdup(ri->roomlist);
554 recp->num_room = num_tokens(ri->roomlist, '|');
555 recp->recptypes_magic = RECPTYPES_MAGIC;
557 /* Construct a GUID to use in the S_USETABLE table.
558 * If one is not present in the item itself, make one up.
560 if (ri->guid != NULL) {
561 StrBufSpaceToBlank(ri->guid);
562 StrBufTrim(ri->guid);
563 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ChrPtr(ri->guid));
566 MD5Init(&md5context);
567 if (ri->title != NULL) {
568 MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->title), StrLength(ri->title));
570 if (ri->link != NULL) {
571 MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->link), StrLength(ri->link));
573 MD5Final(rawdigest, &md5context);
574 for (i=0; i<MD5_DIGEST_LEN; i++) {
575 sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
576 utmsgid[i*2] = tolower(utmsgid[i*2]);
577 utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
579 strcat(utmsgid, "_rss2ctdl");
582 /* Find out if we've already seen this item */
584 cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
587 /* Item has already been seen */
588 syslog(LOG_DEBUG, "%s has already been seen\n", utmsgid);
591 /* rewrite the record anyway, to update the timestamp */
592 strcpy(ut.ut_msgid, utmsgid);
593 ut.ut_timestamp = time(NULL);
594 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
599 /* Item has not been seen, so save it. */
600 syslog(LOG_DEBUG, "RSS: saving item...\n");
601 if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY(""));
602 StrBufSpaceToBlank(ri->description);
603 msg = malloc(sizeof(struct CtdlMessage));
604 memset(msg, 0, sizeof(struct CtdlMessage));
605 msg->cm_magic = CTDLMESSAGE_MAGIC;
606 msg->cm_anon_type = MES_NORMAL;
607 msg->cm_format_type = FMT_RFC822;
609 if (ri->guid != NULL) {
610 msg->cm_fields['E'] = strdup(ChrPtr(ri->guid));
613 if (ri->author_or_creator != NULL) {
615 StrBuf *Encoded = NULL;
618 From = html_to_ascii(ChrPtr(ri->author_or_creator),
619 StrLength(ri->author_or_creator),
621 StrBufPlain(ri->author_or_creator, From, -1);
622 StrBufTrim(ri->author_or_creator);
625 FromAt = strchr(ChrPtr(ri->author_or_creator), '@') != NULL;
626 if (!FromAt && StrLength (ri->author_email) > 0)
628 StrBufRFC2047encode(&Encoded, ri->author_or_creator);
629 msg->cm_fields['A'] = SmashStrBuf(&Encoded);
630 msg->cm_fields['P'] = SmashStrBuf(&ri->author_email);
636 msg->cm_fields['A'] = SmashStrBuf(&ri->author_or_creator);
637 msg->cm_fields['P'] = strdup(msg->cm_fields['A']);
641 StrBufRFC2047encode(&Encoded, ri->author_or_creator);
642 msg->cm_fields['A'] = SmashStrBuf(&Encoded);
643 msg->cm_fields['P'] = strdup("rss@localhost");
648 msg->cm_fields['A'] = strdup("rss");
651 msg->cm_fields['N'] = strdup(NODENAME);
652 if (ri->title != NULL) {
655 StrBuf *Encoded, *QPEncoded;
658 StrBufSpaceToBlank(ri->title);
659 len = StrLength(ri->title);
660 Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
662 if (Sbj[len - 1] == '\n')
667 Encoded = NewStrBufPlain(Sbj, len);
671 StrBufRFC2047encode(&QPEncoded, Encoded);
673 msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
674 FreeStrBuf(&Encoded);
676 msg->cm_fields['T'] = malloc(64);
677 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
678 if (ri->channel_title != NULL) {
679 if (StrLength(ri->channel_title) > 0) {
680 msg->cm_fields['O'] = strdup(ChrPtr(ri->channel_title));
683 if (ri->link == NULL)
684 ri->link = NewStrBufPlain(HKEY(""));
685 ExpandShortUrls(ri->description);
686 msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
688 Message = NewStrBufPlain(NULL, StrLength(ri->description));
690 StrBufPlain(Message, HKEY(
691 "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
694 StrBufAppendBuf(Message, ri->description, 0);
695 StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
697 AppendLink(Message, ri->link, ri->linkTitle, NULL);
698 AppendLink(Message, ri->reLink, ri->reLinkTitle, "Reply to this");
699 StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
701 msg->cm_fields['M'] = SmashStrBuf(&Message);
703 CtdlSubmitMsg(msg, recp, NULL, 0);
704 CtdlFreeMessage(msg);
706 /* write the uidl to the use table so we don't store this item again */
707 strcpy(ut.ut_msgid, utmsgid);
708 ut.ut_timestamp = time(NULL);
709 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
711 free_recipients(recp);
717 * Convert an RDF/RSS datestamp into a time_t
719 time_t rdf_parsedate(const char *p)
725 if (strlen(p) < 10) return 0L;
727 memset(&tm, 0, sizeof tm);
730 * If the timestamp appears to be in W3C datetime format, try to
731 * parse it. See also: http://www.w3.org/TR/NOTE-datetime
733 * This code, along with parsedate.c, is a potential candidate for
734 * moving into libcitadel.
736 if ( (p[4] == '-') && (p[7] == '-') ) {
737 tm.tm_year = atoi(&p[0]) - 1900;
738 tm.tm_mon = atoi(&p[5]) - 1;
739 tm.tm_mday = atoi(&p[8]);
740 if ( (p[10] == 'T') && (p[13] == ':') ) {
741 tm.tm_hour = atoi(&p[11]);
742 tm.tm_min = atoi(&p[14]);
747 /* hmm... try RFC822 date stamp format */
750 if (t > 0) return(t);
752 /* yeesh. ok, just return the current date and time. */
756 void flush_rss_item(rss_item *ri)
758 /* Initialize the feed item data structure */
759 FreeStrBuf(&ri->guid);
760 FreeStrBuf(&ri->title);
761 FreeStrBuf(&ri->link);
762 FreeStrBuf(&ri->linkTitle);
763 FreeStrBuf(&ri->reLink);
764 FreeStrBuf(&ri->reLinkTitle);
765 FreeStrBuf(&ri->description);
766 FreeStrBuf(&ri->channel_title);
767 FreeStrBuf(&ri->author_or_creator);
768 FreeStrBuf(&ri->author_url);
769 FreeStrBuf(&ri->author_email);
772 void rss_xml_start(void *data, const char *supplied_el, const char **attr)
775 rsscollection *rssc = (rsscollection*) data;
776 rssnetcfg *Cfg = rssc->Cfg;
777 rss_item *ri = rssc->Item;
782 /* Axe the namespace, we don't care about it */
783 /// syslog(LOG_EMERG, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
785 while (sep = strchr(pel, ':'), sep) {
789 if (pel != supplied_el)
793 if (!GetHash(KnownNameSpaces,
795 pel - supplied_el - 1,
799 syslog(LOG_EMERG, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
806 StrBufPlain(rssc->Key, pel, -1);
807 StrBufLowerCase(rssc->Key);
808 if (GetHash(StartHandlers, SKEY(rssc->Key), &pv))
810 rssc->Current = h = (rss_xml_handler*) pv;
812 if (((h->Flags & RSS_UNSET) != 0) &&
813 (Cfg->ItemType == RSS_UNSET))
815 h->Handler(rssc->CData, ri, Cfg, attr);
817 else if (((h->Flags & RSS_RSS) != 0) &&
818 (Cfg->ItemType == RSS_RSS))
820 h->Handler(rssc->CData, ri, Cfg, attr);
822 else if (((h->Flags & RSS_ATOM) != 0) &&
823 (Cfg->ItemType == RSS_ATOM))
825 h->Handler(rssc->CData, ri, Cfg, attr);
829 syslog(LOG_EMERG, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
834 syslog(LOG_EMERG, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
838 void rss_xml_end(void *data, const char *supplied_el)
841 rsscollection *rssc = (rsscollection*) data;
842 rssnetcfg *Cfg = rssc->Cfg;
843 rss_item *ri = rssc->Item;
848 /* Axe the namespace, we don't care about it */
850 while (sep = strchr(pel, ':'), sep) {
853 // syslog(LOG_EMERG, "RSS: END %s...\n", el);
854 if (pel != supplied_el)
858 if (!GetHash(KnownNameSpaces,
860 pel - supplied_el - 1,
864 syslog(LOG_EMERG, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
865 supplied_el, ChrPtr(rssc->CData));
867 FlushStrBuf(rssc->CData);
872 StrBufPlain(rssc->Key, pel, -1);
873 StrBufLowerCase(rssc->Key);
874 if (GetHash(EndHandlers, SKEY(rssc->Key), &pv))
876 h = (rss_xml_handler*) pv;
878 if (((h->Flags & RSS_UNSET) != 0) &&
879 (Cfg->ItemType == RSS_UNSET))
881 h->Handler(rssc->CData, ri, Cfg, NULL);
883 else if (((h->Flags & RSS_RSS) != 0) &&
884 (Cfg->ItemType == RSS_RSS))
886 h->Handler(rssc->CData, ri, Cfg, NULL);
888 else if (((h->Flags & RSS_ATOM) != 0) &&
889 (Cfg->ItemType == RSS_ATOM))
891 h->Handler(rssc->CData, ri, Cfg, NULL);
895 syslog(LOG_EMERG, "RSS: END unhandled: [%s] [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
900 syslog(LOG_EMERG, "RSS: END unhandled: [%s] [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
902 FlushStrBuf(rssc->CData);
903 rssc->Current = NULL;
910 void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
912 syslog(LOG_DEBUG, "RSS: This is an RSS feed.\n");
913 Cfg->ItemType = RSS_RSS;
916 void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
918 syslog(LOG_DEBUG, "RSS: This is an RDF feed.\n");
919 Cfg->ItemType = RSS_RSS;
922 void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
924 syslog(LOG_DEBUG, "RSS: This is an ATOM feed.\n");
925 Cfg->ItemType = RSS_ATOM;
929 void RSS_item_item_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
931 ri->item_tag_nesting ++;
935 void ATOM_item_entry_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
938 ri->item_tag_nesting ++;
942 void ATOM_item_link_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
945 const char *pHref = NULL;
946 const char *pType = NULL;
947 const char *pRel = NULL;
948 const char *pTitle = NULL;
950 for (i = 0; Attr[i] != NULL; i+=2)
952 if (!strcmp(Attr[i], "href"))
956 else if (!strcmp(Attr[i], "rel"))
960 else if (!strcmp(Attr[i], "type"))
964 else if (!strcmp(Attr[i], "title"))
970 return; /* WHUT? Pointing... where? */
971 if ((pType != NULL) && !strcasecmp(pType, "application/atom+xml"))
972 return; /* these just point to other rss resources, we're not interested in them. */
975 if (!strcasecmp (pRel, "replies"))
977 NewStrBufDupAppendFlush(&ri->reLink, NULL, pHref, -1);
978 StrBufTrim(ri->link);
979 NewStrBufDupAppendFlush(&ri->reLinkTitle, NULL, pTitle, -1);
981 else if (!strcasecmp(pRel, "alternate")) /* Alternative representation of this Item... */
983 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
984 StrBufTrim(ri->link);
985 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
988 #if 0 /* these are also defined, but dunno what to do with them.. */
989 else if (!strcasecmp(pRel, "related"))
992 else if (!strcasecmp(pRel, "self"))
995 else if (!strcasecmp(pRel, "enclosure"))
996 {/* this reference can get big, and is probably the full article... */
998 else if (!strcasecmp(pRel, "via"))
999 {/* this article was provided via... */
1003 else if (StrLength(ri->link) == 0)
1005 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
1006 StrBufTrim(ri->link);
1007 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
1014 void ATOMRSS_item_title_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1016 if ((ri->item_tag_nesting == 0) && (StrLength(CData) > 0)) {
1017 NewStrBufDupAppendFlush(&ri->channel_title, CData, NULL, 0);
1018 StrBufTrim(ri->channel_title);
1022 void RSS_item_guid_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1024 if (StrLength(CData) > 0) {
1025 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
1029 void ATOM_item_id_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1031 if (StrLength(CData) > 0) {
1032 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
1037 void RSS_item_link_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1039 if (StrLength(CData) > 0) {
1040 NewStrBufDupAppendFlush(&ri->link, CData, NULL, 0);
1041 StrBufTrim(ri->link);
1044 void RSS_item_relink_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1046 if (StrLength(CData) > 0) {
1047 NewStrBufDupAppendFlush(&ri->reLink, CData, NULL, 0);
1048 StrBufTrim(ri->reLink);
1052 void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1054 if (StrLength(CData) > 0) {
1055 NewStrBufDupAppendFlush(&ri->title, CData, NULL, 0);
1056 StrBufTrim(ri->title);
1060 void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1062 long olen = StrLength (ri->description);
1063 long clen = StrLength (CData);
1067 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1068 StrBufTrim(ri->description);
1070 else if (olen < clen) {
1071 FlushStrBuf(ri->description);
1072 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1073 StrBufTrim(ri->description);
1077 void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1079 /* this can contain an abstract of the article. but we don't want to verwrite a full document if we already have it. */
1080 if ((StrLength(CData) > 0) && (StrLength(ri->description) == 0))
1082 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1083 StrBufTrim(ri->description);
1087 void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1089 long olen = StrLength (ri->description);
1090 long clen = StrLength (CData);
1094 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1095 StrBufTrim(ri->description);
1097 else if (olen < clen) {
1098 FlushStrBuf(ri->description);
1099 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
1100 StrBufTrim(ri->description);
1105 void ATOM_item_published_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1107 if (StrLength(CData) > 0) {
1109 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1113 void ATOM_item_updated_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1115 if (StrLength(CData) > 0) {
1117 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1121 void RSS_item_pubdate_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1123 if (StrLength(CData) > 0) {
1125 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1130 void RSS_item_date_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1132 if (StrLength(CData) > 0) {
1134 ri->pubdate = rdf_parsedate(ChrPtr(CData));
1140 void RSS_item_author_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1142 if (StrLength(CData) > 0) {
1143 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1144 StrBufTrim(ri->author_or_creator);
1149 void ATOM_item_name_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1151 if (StrLength(CData) > 0) {
1152 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1153 StrBufTrim(ri->author_or_creator);
1157 void ATOM_item_email_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1159 if (StrLength(CData) > 0) {
1160 NewStrBufDupAppendFlush(&ri->author_email, CData, NULL, 0);
1161 StrBufTrim(ri->author_email);
1165 void RSS_item_creator_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1167 if ((StrLength(CData) > 0) &&
1168 (StrLength(ri->author_or_creator) == 0))
1170 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
1171 StrBufTrim(ri->author_or_creator);
1176 void ATOM_item_uri_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1178 if (StrLength(CData) > 0) {
1179 NewStrBufDupAppendFlush(&ri->author_url, CData, NULL, 0);
1180 StrBufTrim(ri->author_url);
1184 void RSS_item_item_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1186 --ri->item_tag_nesting;
1191 void ATOM_item_entry_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1193 --ri->item_tag_nesting;
1197 void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1199 // syslog(LOG_DEBUG, "End of feed detected. Closing parser.\n");
1200 ri->done_parsing = 1;
1203 void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1205 // syslog(LOG_DEBUG, "End of feed detected. Closing parser.\n");
1206 ri->done_parsing = 1;
1210 void RSSATOM_item_ignore(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1217 * This callback stores up the data which appears in between tags.
1219 void rss_xml_cdata_start(void *data)
1221 rsscollection *rssc = (rsscollection*) data;
1223 FlushStrBuf(rssc->CData);
1226 void rss_xml_cdata_end(void *data)
1229 void rss_xml_chardata(void *data, const XML_Char *s, int len)
1231 rsscollection *rssc = (rsscollection*) data;
1233 StrBufAppendBufPlain (rssc->CData, s, len, 0);
1237 * Callback function for passing libcurl's output to expat for parsing
1239 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
1241 XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
1242 return (size*nmemb);
1248 * Begin a feed parse
1250 void rss_do_fetching(rssnetcfg *Cfg) {
1253 XML_Parser xp = NULL;
1258 char errmsg[1024] = "";
1267 if ((Cfg->next_poll != 0) && (now < Cfg->next_poll))
1269 memset(&ri, 0, sizeof(rss_item));
1273 syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
1275 curl = curl_easy_init();
1277 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
1280 Answer = NewStrBufPlain(NULL, SIZ);
1282 curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
1283 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
1284 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
1285 curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
1286 // curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
1287 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
1288 curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
1289 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
1290 #ifdef CURLOPT_HTTP_CONTENT_DECODING
1291 curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
1292 curl_easy_setopt(curl, CURLOPT_ENCODING, "");
1294 curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
1295 curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180); /* die after 180 seconds */
1297 (!IsEmptyStr(config.c_ip_addr))
1298 && (strcmp(config.c_ip_addr, "*"))
1299 && (strcmp(config.c_ip_addr, "::"))
1300 && (strcmp(config.c_ip_addr, "0.0.0.0"))
1302 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
1305 if (server_shutting_down)
1307 curl_easy_cleanup(curl);
1311 if (server_shutting_down)
1314 res = curl_easy_perform(curl);
1316 syslog(LOG_ALERT, "libcurl error %d: %s\n", res, errmsg);
1319 if (server_shutting_down)
1325 memset(&ri, 0, sizeof(rss_item));
1326 ri.roomlist = Cfg->rooms;
1327 rssc.CData = NewStrBufPlain(NULL, SIZ);
1328 rssc.Key = NewStrBuf();
1330 StrBufSipLine(rssc.Key, Answer, &at);
1333 #define encoding "encoding=\""
1334 ptr = strstr(ChrPtr(rssc.Key), encoding);
1339 ptr += sizeof (encoding) - 1;
1340 pche = strchr(ptr, '"');
1342 StrBufCutAt(rssc.Key, -1, pche);
1350 xp = XML_ParserCreateNS(ptr, ':');
1352 syslog(LOG_ALERT, "Cannot create XML parser!\n");
1355 FlushStrBuf(rssc.Key);
1358 XML_SetUnknownEncodingHandler(xp,
1359 handle_unknown_xml_encoding,
1363 XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
1364 XML_SetCharacterDataHandler(xp, rss_xml_chardata);
1365 XML_SetUserData(xp, &rssc);
1366 XML_SetCdataSectionHandler(xp,
1367 rss_xml_cdata_start,
1371 len = StrLength(Answer);
1372 ptr = SmashStrBuf(&Answer);
1373 XML_Parse(xp, ptr, len, 0);
1375 if (ri.done_parsing == 0)
1376 XML_Parse(xp, "", 0, 1);
1379 syslog(LOG_ALERT, "RSS: XML Status [%s] \n",
1381 XML_GetErrorCode(xp)));
1384 FreeStrBuf(&Answer);
1385 curl_easy_cleanup(curl);
1388 flush_rss_item(&ri);
1389 FreeStrBuf(&rssc.CData);
1390 FreeStrBuf(&rssc.Key);
1392 Cfg->next_poll = time(NULL) + config.c_net_freq;
1397 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
1399 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
1401 char filename[PATH_MAX];
1406 rssnetcfg *rncptr = NULL;
1407 rssnetcfg *use_this_rncptr = NULL;
1411 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
1413 if (server_shutting_down)
1416 /* Only do net processing for rooms that have netconfigs */
1417 fp = fopen(filename, "r");
1422 while (fgets(buf, sizeof buf, fp) != NULL && !server_shutting_down) {
1423 buf[strlen(buf)-1] = 0;
1425 extract_token(instr, buf, 0, '|', sizeof instr);
1426 if (!strcasecmp(instr, "rssclient")) {
1428 use_this_rncptr = NULL;
1430 extract_token(feedurl, buf, 1, '|', sizeof feedurl);
1432 /* If any other rooms have requested the same feed, then we will just add this
1433 * room to the target list for that client request.
1435 for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
1436 if (!strcmp(rncptr->url, feedurl)) {
1437 use_this_rncptr = rncptr;
1441 /* Otherwise create a new client request */
1442 if (use_this_rncptr == NULL) {
1443 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
1444 memset(rncptr, 0, sizeof(rssnetcfg));
1445 rncptr->ItemType = RSS_UNSET;
1446 if (rncptr != NULL) {
1447 rncptr->next = rnclist;
1448 safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
1449 rncptr->rooms = NULL;
1451 use_this_rncptr = rncptr;
1455 /* Add the room name to the request */
1456 if (use_this_rncptr != NULL) {
1457 if (use_this_rncptr->rooms == NULL) {
1458 rncptr->rooms = strdup(qrbuf->QRname);
1461 len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
1462 ptr = realloc(use_this_rncptr->rooms, len);
1465 strcat(ptr, qrbuf->QRname);
1466 use_this_rncptr->rooms = ptr;
1479 * Scan for rooms that have RSS client requests configured
1481 void rssclient_scan(void) {
1482 static time_t last_run = 0L;
1483 static int doing_rssclient = 0;
1484 rssnetcfg *rptr = NULL;
1486 /* Run no more than once every 15 minutes. */
1487 if ((time(NULL) - last_run) < 900) {
1492 * This is a simple concurrency check to make sure only one rssclient run
1493 * is done at a time. We could do this with a mutex, but since we
1494 * don't really require extremely fine granularity here, we'll do it
1495 * with a static variable instead.
1497 if (doing_rssclient) return;
1498 doing_rssclient = 1;
1500 syslog(LOG_DEBUG, "rssclient started\n");
1501 CtdlForEachRoom(rssclient_scan_room, NULL);
1503 while (rnclist != NULL && !server_shutting_down) {
1504 rss_do_fetching(rnclist);
1506 rnclist = rnclist->next;
1507 if (rptr->rooms != NULL) free(rptr->rooms);
1511 syslog(LOG_DEBUG, "rssclient ended\n");
1512 last_run = time(NULL);
1513 doing_rssclient = 0;
1517 void LoadUrlShorteners(void)
1521 const char *POS = NULL;
1522 const char *Err = NULL;
1523 StrBuf *Content, *Line;
1526 UrlShorteners = NewHash(0, Flathash);
1528 fd = open(file_citadel_urlshorteners, 0);
1532 Content = NewStrBufPlain(NULL, SIZ);
1534 while (POS != StrBufNOTNULL)
1536 StrBufTCP_read_buffered_line_fast (Line, Content, &POS, &fd, 1, 1, &Err);
1538 if ((*ChrPtr(Line) != '#') && (StrLength(Line) > 0))
1540 Put(UrlShorteners, IKEY(i), Line, HFreeStrBuf);
1547 POS = StrBufNOTNULL;
1550 FreeStrBuf(&Content);
1555 void rss_cleanup(void)
1557 DeleteHash(&StartHandlers);
1558 DeleteHash(&EndHandlers);
1559 DeleteHash(&UrlShorteners);
1560 DeleteHash(&KnownNameSpaces);
1563 CTDL_MODULE_INIT(rssclient)
1567 syslog(LOG_INFO, "%s\n", curl_version());
1568 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
1572 LoadUrlShorteners ();
1574 StartHandlers = NewHash(1, NULL);
1575 EndHandlers = NewHash(1, NULL);
1577 AddRSSStartHandler(RSS_item_rss_start, RSS_UNSET, HKEY("rss"));
1578 AddRSSStartHandler(RSS_item_rdf_start, RSS_UNSET, HKEY("rdf"));
1579 AddRSSStartHandler(ATOM_item_feed_start, RSS_UNSET, HKEY("feed"));
1580 AddRSSStartHandler(RSS_item_item_start, RSS_RSS, HKEY("item"));
1581 AddRSSStartHandler(ATOM_item_entry_start, RSS_ATOM, HKEY("entry"));
1582 AddRSSStartHandler(ATOM_item_link_start, RSS_ATOM, HKEY("link"));
1584 AddRSSEndHandler(ATOMRSS_item_title_end, RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1585 AddRSSEndHandler(RSS_item_guid_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
1586 AddRSSEndHandler(ATOM_item_id_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
1587 AddRSSEndHandler(RSS_item_link_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
1589 // hm, rss to the comments of that blog, might be interesting in future, but...
1590 AddRSSEndHandler(RSS_item_relink_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("commentrss"));
1592 AddRSSEndHandler(RSS_item_relink_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("comments"));
1594 AddRSSEndHandler(RSSATOM_item_title_end, RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1595 AddRSSEndHandler(ATOM_item_content_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
1596 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_ATOM|RSS_REQUIRE_BUF, HKEY("encoded"));
1597 AddRSSEndHandler(ATOM_item_summary_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
1598 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
1599 AddRSSEndHandler(ATOM_item_published_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
1600 AddRSSEndHandler(ATOM_item_updated_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
1601 AddRSSEndHandler(RSS_item_pubdate_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
1602 AddRSSEndHandler(RSS_item_date_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
1603 AddRSSEndHandler(RSS_item_author_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
1604 AddRSSEndHandler(RSS_item_creator_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
1606 AddRSSEndHandler(ATOM_item_email_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
1607 AddRSSEndHandler(ATOM_item_name_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
1608 AddRSSEndHandler(ATOM_item_uri_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
1610 AddRSSEndHandler(RSS_item_item_end, RSS_RSS, HKEY("item"));
1611 AddRSSEndHandler(RSS_item_rss_end, RSS_RSS, HKEY("rss"));
1612 AddRSSEndHandler(RSS_item_rdf_end, RSS_RSS, HKEY("rdf"));
1613 AddRSSEndHandler(ATOM_item_entry_end, RSS_ATOM, HKEY("entry"));
1616 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
1617 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("seq"));
1618 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("seq"));
1619 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("li"));
1620 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("li"));
1622 /* links to other feed generators... */
1623 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1624 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1625 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1626 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1628 KnownNameSpaces = NewHash(1, NULL);
1629 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
1630 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
1631 Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
1632 Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
1633 Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
1634 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1635 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
1636 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
1637 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
1638 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1639 Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
1640 Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
1641 Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
1642 Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
1643 Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
1644 Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
1645 Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
1646 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
1647 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1648 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1649 Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
1650 Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
1651 Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
1653 /* we don't like these namespaces because of they shadow our usefull parameters. */
1654 Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
1656 CtdlRegisterCleanupHook(rss_cleanup);