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"
59 typedef struct rssnetcfg rssnetcfg;
64 time_t last_error_when;
68 #define RSS_UNSET (1<<0)
69 #define RSS_RSS (1<<1)
70 #define RSS_ATOM (1<<2)
71 #define RSS_REQUIRE_BUF (1<<3)
73 typedef struct _rss_item {
84 StrBuf *channel_title;
86 StrBuf *author_or_creator;
92 typedef void (*rss_handler_func)(StrBuf *CData,
97 typedef struct __rss_xml_handler {
99 rss_handler_func Handler;
103 typedef struct _rsscollection {
110 rss_xml_handler *Current;
113 struct rssnetcfg *rnclist = NULL;
114 HashList *StartHandlers;
115 HashList *EndHandlers;
116 HashList *KnownNameSpaces;
117 void AddRSSStartHandler(rss_handler_func Handler, int Flags, const char *key, long len)
120 h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
122 h->Handler = Handler;
123 Put(StartHandlers, key, len, h, NULL);
125 void AddRSSEndHandler(rss_handler_func Handler, int Flags, const char *key, long len)
128 h = (rss_xml_handler*) malloc(sizeof (rss_xml_handler));
130 h->Handler = Handler;
131 Put(EndHandlers, key, len, h, NULL);
140 * dug this out of the trashcan of the midgard project, lets see if it works for us.
141 * original code by Alexander Bokovoy <bokovoy@avilink.ne> distributed under GPL V2 or later
145 >= 0 - successfull, 0 means conversion doesn't use multibyte sequences
146 -1 - error during iconv_open call
147 -2 - error during iconv_close call
148 ----------------------------------
149 This function expects that multibyte encoding in 'charset' wouldn't have
150 characters with more than 3 bytes. It is not intended to convert UTF-8 because
151 we'll never receive UTF-8 in our handler (it is handled by Exat itself).
154 fill_encoding_info (const char *charset, XML_Encoding * info)
156 iconv_t cd = (iconv_t)(-1);
158 CtdlLogPrintf(0, "RSS: fill encoding info ...\n");
160 #if G_BYTE_ORDER == G_LITTLE_ENDIAN
161 cd = iconv_open ("UCS-2LE", charset);
163 cd = iconv_open ("UCS-2BE", charset);
166 if (cd == (iconv_t) (-1))
173 unsigned char buf[4];
174 unsigned int i0, i1, i2;
177 for (i0 = 0; i0 < 0x100; i0++)
181 //result = try (cd, buf, 1, &out);
191 for (i1 = 0; i1 < 0x100; i1++)
194 ///result = try (cd, buf, 2, &out);
205 for (i2 = 0; i2 < 0x100; i2++)
208 ////result = try (cd, buf, 3, &out);
224 if (iconv_close (cd) < 0)
232 iconv_convertor (void *data, const char *s)
234 XML_Encoding *info = data;
236 CtdlLogPrintf(0, "RSS: Converting ...\n");
242 result = g_byte_array_new ();
243 if (process_block (info->data, (char *) s, strlen (s), result) == 0)
245 res = *(result->data);
246 g_byte_array_free (result, TRUE);
249 g_byte_array_free (result, TRUE);
255 my_release (void *data)
257 iconv_t cd = (iconv_t) data;
258 if (iconv_close (cd) != 0)
260 /// TODO: uh no. exit (1);
264 handle_unknown_xml_encoding (void *encodingHandleData,
265 const XML_Char * name,
269 CtdlLogPrintf(0, "RSS: unknown encoding ...\n");
270 result = fill_encoding_info (name, info);
274 Special case: client asked for reverse conversion, we'll provide him with
275 iconv descriptor which handles it. Client should release it by himself.
277 if(encodingHandleData != NULL)
278 *((iconv_t *)encodingHandleData) = iconv_open(name, "UTF-8");
280 Optimization: we do not need conversion function if encoding is one-to-one,
281 info->map table will be enough
286 info->convert = NULL;
287 info->release = NULL;
291 We do need conversion function because this encoding uses multibyte sequences
293 info->data = (void *) iconv_open ("UTF-8", name);
294 if ((int)info->data == -1)
296 info->convert = iconv_convertor;
297 info->release = my_release;
300 if(encodingHandleData != NULL)
301 *(iconv_t *)encodingHandleData = NULL;
308 void AppendLink(StrBuf *Message, StrBuf *link, StrBuf *LinkTitle, const char *Title)
310 if (StrLength(link) > 0)
312 StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
313 StrBufAppendBuf(Message, link, 0);
314 StrBufAppendBufPlain(Message, HKEY("\">"), 0);
315 if (StrLength(LinkTitle) > 0)
316 StrBufAppendBuf(Message, LinkTitle, 0);
317 else if ((Title != NULL) && !IsEmptyStr(Title))
318 StrBufAppendBufPlain(Message, Title, -1, 0);
320 StrBufAppendBuf(Message, link, 0);
321 StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
325 * Commit a fetched and parsed RSS item to disk
327 void rss_save_item(rss_item *ri)
330 struct MD5Context md5context;
331 u_char rawdigest[MD5_DIGEST_LEN];
334 struct cdbdata *cdbut;
336 struct CtdlMessage *msg;
337 struct recptypes *recp = NULL;
341 recp = (struct recptypes *) malloc(sizeof(struct recptypes));
342 if (recp == NULL) return;
343 memset(recp, 0, sizeof(struct recptypes));
344 memset(&ut, 0, sizeof(struct UseTable));
345 recp->recp_room = strdup(ri->roomlist);
346 recp->num_room = num_tokens(ri->roomlist, '|');
347 recp->recptypes_magic = RECPTYPES_MAGIC;
349 /* Construct a GUID to use in the S_USETABLE table.
350 * If one is not present in the item itself, make one up.
352 if (ri->guid != NULL) {
353 StrBufSpaceToBlank(ri->guid);
354 StrBufTrim(ri->guid);
355 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ChrPtr(ri->guid));
358 MD5Init(&md5context);
359 if (ri->title != NULL) {
360 MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->title), StrLength(ri->title));
362 if (ri->link != NULL) {
363 MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->link), StrLength(ri->link));
365 MD5Final(rawdigest, &md5context);
366 for (i=0; i<MD5_DIGEST_LEN; i++) {
367 sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
368 utmsgid[i*2] = tolower(utmsgid[i*2]);
369 utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
371 strcat(utmsgid, "_rss2ctdl");
374 /* Find out if we've already seen this item */
376 cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
379 /* Item has already been seen */
380 CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
383 /* rewrite the record anyway, to update the timestamp */
384 strcpy(ut.ut_msgid, utmsgid);
385 ut.ut_timestamp = time(NULL);
386 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
391 /* Item has not been seen, so save it. */
392 CtdlLogPrintf(CTDL_DEBUG, "RSS: saving item...\n");
393 if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY(""));
394 StrBufSpaceToBlank(ri->description);
395 msg = malloc(sizeof(struct CtdlMessage));
396 memset(msg, 0, sizeof(struct CtdlMessage));
397 msg->cm_magic = CTDLMESSAGE_MAGIC;
398 msg->cm_anon_type = MES_NORMAL;
399 msg->cm_format_type = FMT_RFC822;
401 if (ri->guid != NULL) {
402 msg->cm_fields['E'] = strdup(ChrPtr(ri->guid));
405 if (ri->author_or_creator != NULL) {
407 StrBuf *Encoded, *QPEncoded;
409 StrBuf *EmailAddress;
413 UserName = NewStrBuf();
414 EmailAddress = NewStrBuf();
415 EncBuf = NewStrBuf();
417 From = html_to_ascii(ChrPtr(ri->author_or_creator),
418 StrLength(ri->author_or_creator),
420 FromAt = strchr(From, '@') != NULL;
421 if (!FromAt && StrLength (ri->author_email) > 0)
423 Encoded = NewStrBuf();
424 if (!IsEmptyStr(From))
426 StrBufPrintf(Encoded,
429 ChrPtr(ri->author_email));
433 StrBufPrintf(Encoded,
435 ChrPtr(ri->author_email));
441 Encoded = NewStrBufPlain(From, -1);
444 Encoded = NewStrBuf();
445 StrBufPrintf(Encoded,
448 "rss@localhost"); /// TODO: get hostname?
453 QPEncoded = StrBufSanitizeEmailRecipientVector(Encoded, UserName, EmailAddress, EncBuf);
454 msg->cm_fields['A'] = SmashStrBuf(&QPEncoded);
456 FreeStrBuf(&Encoded);
457 FreeStrBuf(&UserName);
458 FreeStrBuf(&EmailAddress);
463 msg->cm_fields['A'] = strdup("rss");
466 msg->cm_fields['N'] = strdup(NODENAME);
467 if (ri->title != NULL) {
470 StrBuf *Encoded, *QPEncoded;
473 StrBufSpaceToBlank(ri->title);
474 len = StrLength(ri->title);
475 Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
476 Encoded = NewStrBufPlain(Sbj, -1);
480 StrBufRFC2047encode(&QPEncoded, Encoded);
482 msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
483 FreeStrBuf(&Encoded);
485 msg->cm_fields['T'] = malloc(64);
486 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
487 if (ri->channel_title != NULL) {
488 if (StrLength(ri->channel_title) > 0) {
489 msg->cm_fields['O'] = strdup(ChrPtr(ri->channel_title));
492 if (ri->link == NULL)
493 ri->link = NewStrBufPlain(HKEY(""));
494 msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
496 Message = NewStrBufPlain(NULL, StrLength(ri->description));
498 StrBufPlain(Message, HKEY(
499 "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
502 StrBufAppendBuf(Message, ri->description, 0);
503 StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
505 AppendLink(Message, ri->link, ri->linkTitle, NULL);
506 AppendLink(Message, ri->reLink, ri->reLinkTitle, "Reply to this");
507 StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
509 msg->cm_fields['M'] = SmashStrBuf(&Message);
511 CtdlSubmitMsg(msg, recp, NULL, 0);
512 CtdlFreeMessage(msg);
514 /* write the uidl to the use table so we don't store this item again */
515 strcpy(ut.ut_msgid, utmsgid);
516 ut.ut_timestamp = time(NULL);
517 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
519 free_recipients(recp);
525 * Convert an RDF/RSS datestamp into a time_t
527 time_t rdf_parsedate(const char *p)
533 if (strlen(p) < 10) return 0L;
535 memset(&tm, 0, sizeof tm);
538 * If the timestamp appears to be in W3C datetime format, try to
539 * parse it. See also: http://www.w3.org/TR/NOTE-datetime
541 * This code, along with parsedate.c, is a potential candidate for
542 * moving into libcitadel.
544 if ( (p[4] == '-') && (p[7] == '-') ) {
545 tm.tm_year = atoi(&p[0]) - 1900;
546 tm.tm_mon = atoi(&p[5]) - 1;
547 tm.tm_mday = atoi(&p[8]);
548 if ( (p[10] == 'T') && (p[13] == ':') ) {
549 tm.tm_hour = atoi(&p[11]);
550 tm.tm_min = atoi(&p[14]);
555 /* hmm... try RFC822 date stamp format */
558 if (t > 0) return(t);
560 /* yeesh. ok, just return the current date and time. */
564 void flush_rss_item(rss_item *ri)
566 /* Initialize the feed item data structure */
567 FreeStrBuf(&ri->guid);
568 FreeStrBuf(&ri->title);
569 FreeStrBuf(&ri->link);
570 FreeStrBuf(&ri->author_or_creator);
571 FreeStrBuf(&ri->author_email);
572 FreeStrBuf(&ri->author_url);
573 FreeStrBuf(&ri->description);
576 void rss_xml_start(void *data, const char *supplied_el, const char **attr)
579 rsscollection *rssc = (rsscollection*) data;
580 rssnetcfg *Cfg = rssc->Cfg;
581 rss_item *ri = rssc->Item;
586 /* Axe the namespace, we don't care about it */
587 /// CtdlLogPrintf(0, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
589 while (sep = strchr(pel, ':'), sep) {
593 if (pel != supplied_el)
597 if (!GetHash(KnownNameSpaces,
599 pel - supplied_el - 1,
603 CtdlLogPrintf(0, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
610 StrBufPlain(rssc->Key, pel, -1);
611 StrBufLowerCase(rssc->Key);
612 if (GetHash(StartHandlers, SKEY(rssc->Key), &pv))
614 rssc->Current = h = (rss_xml_handler*) pv;
616 if (((h->Flags & RSS_UNSET) != 0) &&
617 (Cfg->ItemType == RSS_UNSET))
619 h->Handler(rssc->CData, ri, Cfg, attr);
621 else if (((h->Flags & RSS_RSS) != 0) &&
622 (Cfg->ItemType == RSS_RSS))
624 h->Handler(rssc->CData, ri, Cfg, attr);
626 else if (((h->Flags & RSS_ATOM) != 0) &&
627 (Cfg->ItemType == RSS_ATOM))
629 h->Handler(rssc->CData, ri, Cfg, attr);
633 CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
638 CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
642 void rss_xml_end(void *data, const char *supplied_el)
645 rsscollection *rssc = (rsscollection*) data;
646 rssnetcfg *Cfg = rssc->Cfg;
647 rss_item *ri = rssc->Item;
652 /* Axe the namespace, we don't care about it */
654 while (sep = strchr(pel, ':'), sep) {
657 // CtdlLogPrintf(0, "RSS: END %s...\n", el);
658 if (pel != supplied_el)
662 if (!GetHash(KnownNameSpaces,
664 pel - supplied_el - 1,
668 CtdlLogPrintf(0, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
669 supplied_el, ChrPtr(rssc->CData));
671 FlushStrBuf(rssc->CData);
676 StrBufPlain(rssc->Key, pel, -1);
677 StrBufLowerCase(rssc->Key);
678 if (GetHash(EndHandlers, SKEY(rssc->Key), &pv))
680 h = (rss_xml_handler*) pv;
682 if (((h->Flags & RSS_UNSET) != 0) &&
683 (Cfg->ItemType == RSS_UNSET))
685 h->Handler(rssc->CData, ri, Cfg, NULL);
687 else if (((h->Flags & RSS_RSS) != 0) &&
688 (Cfg->ItemType == RSS_RSS))
690 h->Handler(rssc->CData, ri, Cfg, NULL);
692 else if (((h->Flags & RSS_ATOM) != 0) &&
693 (Cfg->ItemType == RSS_ATOM))
695 h->Handler(rssc->CData, ri, Cfg, NULL);
699 CtdlLogPrintf(0, "RSS: END unhandled: [%s] [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
704 CtdlLogPrintf(0, "RSS: END unhandled: [%s] [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
706 FlushStrBuf(rssc->CData);
707 rssc->Current = NULL;
714 void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
716 CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RSS feed.\n");
717 Cfg->ItemType = RSS_RSS;
720 void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
722 CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RDF feed.\n");
723 Cfg->ItemType = RSS_RSS;
726 void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
728 CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an ATOM feed.\n");
729 Cfg->ItemType = RSS_ATOM;
733 void RSS_item_item_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
735 ri->item_tag_nesting ++;
739 void ATOM_item_entry_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
742 ri->item_tag_nesting ++;
746 void ATOM_item_link_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
749 const char *pHref = NULL;
750 const char *pType = NULL;
751 const char *pRel = NULL;
752 const char *pTitle = NULL;
754 for (i = 0; Attr[i] != NULL; i+=2)
756 if (!strcmp(Attr[i], "href"))
760 else if (!strcmp(Attr[i], "rel"))
764 else if (!strcmp(Attr[i], "type"))
768 else if (!strcmp(Attr[i], "title"))
774 return; /* WHUT? Pointing... where? */
775 if ((pType != NULL) && !strcasecmp(pType, "application/atom+xml"))
776 return; /* these just point to other rss resources, we're not interested in them. */
779 if (!strcasecmp (pRel, "replies"))
781 NewStrBufDupAppendFlush(&ri->reLink, NULL, pHref, -1);
782 StrBufTrim(ri->link);
783 NewStrBufDupAppendFlush(&ri->reLinkTitle, NULL, pTitle, -1);
785 else if (!strcasecmp(pRel, "alternate")) /* Alternative representation of this Item... */
787 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
788 StrBufTrim(ri->link);
789 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
792 #if 0 /* these are also defined, but dunno what to do with them.. */
793 else if (!strcasecmp(pRel, "related"))
796 else if (!strcasecmp(pRel, "self"))
799 else if (!strcasecmp(pRel, "enclosure"))
800 {/* this reference can get big, and is probably the full article... */
802 else if (!strcasecmp(pRel, "via"))
803 {/* this article was provided via... */
807 else if (StrLength(ri->link) == 0)
809 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
810 StrBufTrim(ri->link);
811 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
818 void ATOMRSS_item_title_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
820 if ((ri->item_tag_nesting == 0) && (StrLength(CData) > 0)) {
821 NewStrBufDupAppendFlush(&ri->channel_title, CData, NULL, 0);
822 StrBufTrim(ri->channel_title);
826 void RSS_item_guid_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
828 if (StrLength(CData) > 0) {
829 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
833 void ATOM_item_id_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
835 if (StrLength(CData) > 0) {
836 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
841 void RSS_item_link_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
843 if (StrLength(CData) > 0) {
844 NewStrBufDupAppendFlush(&ri->link, CData, NULL, 0);
845 StrBufTrim(ri->link);
849 void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
851 if (StrLength(CData) > 0) {
852 NewStrBufDupAppendFlush(&ri->title, CData, NULL, 0);
853 StrBufTrim(ri->title);
857 void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
859 long olen = StrLength (ri->description);
860 long clen = StrLength (CData);
864 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
865 StrBufTrim(ri->description);
867 else if (olen < clen) {
868 FlushStrBuf(ri->description);
869 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
870 StrBufTrim(ri->description);
874 void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
876 /* this can contain an abstract of the article. but we don't want to verwrite a full document if we already have it. */
877 if ((StrLength(CData) > 0) && (StrLength(ri->description) == 0))
879 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
880 StrBufTrim(ri->description);
884 void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
886 long olen = StrLength (ri->description);
887 long clen = StrLength (CData);
891 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
892 StrBufTrim(ri->description);
894 else if (olen < clen) {
895 FlushStrBuf(ri->description);
896 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
897 StrBufTrim(ri->description);
902 void ATOM_item_published_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
904 if (StrLength(CData) > 0) {
906 ri->pubdate = rdf_parsedate(ChrPtr(CData));
910 void ATOM_item_updated_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
912 if (StrLength(CData) > 0) {
914 ri->pubdate = rdf_parsedate(ChrPtr(CData));
918 void RSS_item_pubdate_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
920 if (StrLength(CData) > 0) {
922 ri->pubdate = rdf_parsedate(ChrPtr(CData));
927 void RSS_item_date_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
929 if (StrLength(CData) > 0) {
931 ri->pubdate = rdf_parsedate(ChrPtr(CData));
937 void RSS_item_author_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
939 if (StrLength(CData) > 0) {
940 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
941 StrBufTrim(ri->author_or_creator);
946 void ATOM_item_name_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
948 if (StrLength(CData) > 0) {
949 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
950 StrBufTrim(ri->author_or_creator);
954 void ATOM_item_email_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
956 if (StrLength(CData) > 0) {
957 NewStrBufDupAppendFlush(&ri->author_email, CData, NULL, 0);
958 StrBufTrim(ri->author_email);
962 void RSS_item_creator_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
964 if ((StrLength(CData) > 0) &&
965 (StrLength(ri->author_or_creator) == 0))
967 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
968 StrBufTrim(ri->author_or_creator);
973 void ATOM_item_uri_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
975 if (StrLength(CData) > 0) {
976 NewStrBufDupAppendFlush(&ri->author_url, CData, NULL, 0);
977 StrBufTrim(ri->author_url);
981 void RSS_item_item_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
983 --ri->item_tag_nesting;
988 void ATOM_item_entry_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
990 --ri->item_tag_nesting;
994 void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
996 // CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
997 ri->done_parsing = 1;
1000 void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1002 // CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
1003 ri->done_parsing = 1;
1007 void RSSATOM_item_ignore(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
1014 * This callback stores up the data which appears in between tags.
1016 void rss_xml_cdata_start(void *data)
1018 rsscollection *rssc = (rsscollection*) data;
1020 FlushStrBuf(rssc->CData);
1023 void rss_xml_cdata_end(void *data)
1026 void rss_xml_chardata(void *data, const XML_Char *s, int len)
1028 rsscollection *rssc = (rsscollection*) data;
1030 StrBufAppendBufPlain (rssc->CData, s, len, 0);
1034 * Callback function for passing libcurl's output to expat for parsing
1036 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
1038 XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
1039 return (size*nmemb);
1045 * Begin a feed parse
1047 void rss_do_fetching(rssnetcfg *Cfg) {
1055 char errmsg[1024] = "";
1060 memset(&ri, 0, sizeof(rss_item));
1064 CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
1066 curl = curl_easy_init();
1068 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
1071 Answer = NewStrBufPlain(NULL, SIZ);
1073 curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
1074 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
1075 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
1076 curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
1077 // curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
1078 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
1079 curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
1080 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
1081 #ifdef CURLOPT_HTTP_CONTENT_DECODING
1082 curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
1083 curl_easy_setopt(curl, CURLOPT_ENCODING, "");
1085 curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
1086 curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180); /* die after 180 seconds */
1088 (!IsEmptyStr(config.c_ip_addr))
1089 && (strcmp(config.c_ip_addr, "*"))
1090 && (strcmp(config.c_ip_addr, "::"))
1091 && (strcmp(config.c_ip_addr, "0.0.0.0"))
1093 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
1096 if (CtdlThreadCheckStop())
1098 curl_easy_cleanup(curl);
1102 if (CtdlThreadCheckStop())
1105 res = curl_easy_perform(curl);
1107 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
1110 if (CtdlThreadCheckStop())
1116 memset(&ri, 0, sizeof(rss_item));
1117 ri.roomlist = Cfg->rooms;
1118 rssc.CData = NewStrBufPlain(NULL, SIZ);
1119 rssc.Key = NewStrBuf();
1121 StrBufSipLine(rssc.Key, Answer, &at);
1124 #define encoding "encoding=\""
1125 ptr = strstr(ChrPtr(rssc.Key), encoding);
1130 ptr += sizeof (encoding) - 1;
1131 pche = strchr(ptr, '"');
1133 StrBufCutAt(rssc.Key, -1, pche);
1141 xp = XML_ParserCreateNS(ptr, ':');
1143 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
1146 FlushStrBuf(rssc.Key);
1149 XML_SetUnknownEncodingHandler(xp,
1150 handle_unknown_xml_encoding,
1154 XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
1155 XML_SetCharacterDataHandler(xp, rss_xml_chardata);
1156 XML_SetUserData(xp, &rssc);
1157 XML_SetCdataSectionHandler(xp,
1158 rss_xml_cdata_start,
1162 len = StrLength(Answer);
1163 ptr = SmashStrBuf(&Answer);
1164 XML_Parse(xp, ptr, len, 0);
1166 if (ri.done_parsing == 0)
1167 XML_Parse(xp, "", 0, 1);
1170 CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n",
1172 XML_GetErrorCode(xp)));
1175 curl_easy_cleanup(curl);
1178 flush_rss_item(&ri);
1179 FreeStrBuf(&rssc.CData);
1180 FreeStrBuf(&rssc.Key);
1185 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
1187 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
1189 char filename[PATH_MAX];
1194 rssnetcfg *rncptr = NULL;
1195 rssnetcfg *use_this_rncptr = NULL;
1199 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
1201 if (CtdlThreadCheckStop())
1204 /* Only do net processing for rooms that have netconfigs */
1205 fp = fopen(filename, "r");
1210 while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
1211 buf[strlen(buf)-1] = 0;
1213 extract_token(instr, buf, 0, '|', sizeof instr);
1214 if (!strcasecmp(instr, "rssclient")) {
1216 use_this_rncptr = NULL;
1218 extract_token(feedurl, buf, 1, '|', sizeof feedurl);
1220 /* If any other rooms have requested the same feed, then we will just add this
1221 * room to the target list for that client request.
1223 for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
1224 if (!strcmp(rncptr->url, feedurl)) {
1225 use_this_rncptr = rncptr;
1229 /* Otherwise create a new client request */
1230 if (use_this_rncptr == NULL) {
1231 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
1232 rncptr->ItemType = RSS_UNSET;
1233 if (rncptr != NULL) {
1234 rncptr->next = rnclist;
1235 safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
1236 rncptr->rooms = NULL;
1238 use_this_rncptr = rncptr;
1242 /* Add the room name to the request */
1243 if (use_this_rncptr != NULL) {
1244 if (use_this_rncptr->rooms == NULL) {
1245 rncptr->rooms = strdup(qrbuf->QRname);
1248 len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
1249 ptr = realloc(use_this_rncptr->rooms, len);
1252 strcat(ptr, qrbuf->QRname);
1253 use_this_rncptr->rooms = ptr;
1266 * Scan for rooms that have RSS client requests configured
1268 void rssclient_scan(void *args) {
1269 static time_t last_run = 0L;
1270 static int doing_rssclient = 0;
1271 rssnetcfg *rptr = NULL;
1272 CitContext rssclientCC;
1274 /* Give this thread its own private CitContext */
1275 CtdlFillSystemContext(&rssclientCC, "rssclient");
1276 citthread_setspecific(MyConKey, (void *)&rssclientCC );
1279 * This is a simple concurrency check to make sure only one rssclient run
1280 * is done at a time. We could do this with a mutex, but since we
1281 * don't really require extremely fine granularity here, we'll do it
1282 * with a static variable instead.
1284 if (doing_rssclient) return NULL;
1285 doing_rssclient = 1;
1287 CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
1288 CtdlForEachRoom(rssclient_scan_room, NULL);
1290 while (rnclist != NULL && !CtdlThreadCheckStop()) {
1291 rss_do_fetching(rnclist);
1293 rnclist = rnclist->next;
1294 if (rptr->rooms != NULL) free(rptr->rooms);
1298 CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
1299 last_run = time(NULL);
1300 doing_rssclient = 0;
1301 CtdlClearSystemContext();
1306 CTDL_MODULE_INIT(rssclient)
1310 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
1311 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
1314 StartHandlers = NewHash(1, NULL);
1315 EndHandlers = NewHash(1, NULL);
1317 AddRSSStartHandler(RSS_item_rss_start, RSS_UNSET, HKEY("rss"));
1318 AddRSSStartHandler(RSS_item_rdf_start, RSS_UNSET, HKEY("rdf"));
1319 AddRSSStartHandler(ATOM_item_feed_start, RSS_UNSET, HKEY("feed"));
1320 AddRSSStartHandler(RSS_item_item_start, RSS_RSS, HKEY("item"));
1321 AddRSSStartHandler(ATOM_item_entry_start, RSS_ATOM, HKEY("entry"));
1322 AddRSSStartHandler(ATOM_item_link_start, RSS_ATOM, HKEY("link"));
1324 AddRSSEndHandler(ATOMRSS_item_title_end, RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1325 AddRSSEndHandler(RSS_item_guid_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
1326 AddRSSEndHandler(ATOM_item_id_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
1327 AddRSSEndHandler(RSS_item_link_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
1328 AddRSSEndHandler(RSSATOM_item_title_end, RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1329 AddRSSEndHandler(ATOM_item_content_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
1330 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_ATOM|RSS_REQUIRE_BUF, HKEY("encoded"));
1331 AddRSSEndHandler(ATOM_item_summary_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
1332 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
1333 AddRSSEndHandler(ATOM_item_published_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
1334 AddRSSEndHandler(ATOM_item_updated_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
1335 AddRSSEndHandler(RSS_item_pubdate_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
1336 AddRSSEndHandler(RSS_item_date_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
1337 AddRSSEndHandler(RSS_item_author_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
1338 AddRSSEndHandler(RSS_item_creator_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
1340 AddRSSEndHandler(ATOM_item_email_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
1341 AddRSSEndHandler(ATOM_item_name_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
1342 AddRSSEndHandler(ATOM_item_uri_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
1344 AddRSSEndHandler(RSS_item_item_end, RSS_RSS, HKEY("item"));
1345 AddRSSEndHandler(RSS_item_rss_end, RSS_RSS, HKEY("rss"));
1346 AddRSSEndHandler(RSS_item_rdf_end, RSS_RSS, HKEY("rdf"));
1347 AddRSSEndHandler(ATOM_item_entry_end, RSS_ATOM, HKEY("entry"));
1350 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
1351 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("seq"));
1352 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("seq"));
1353 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("li"));
1354 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("li"));
1356 /* links to other feed generators... */
1357 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1358 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1359 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1360 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1362 KnownNameSpaces = NewHash(1, NULL);
1363 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
1364 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
1365 Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
1366 Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
1367 Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
1368 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1369 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
1370 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
1371 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
1372 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1373 Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
1374 Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
1375 Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
1376 Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
1377 Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
1378 Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
1379 Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
1380 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
1381 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1382 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1383 Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
1384 Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
1385 Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
1387 /* we don't like these namespaces because of they shadow our usefull parameters. */
1388 Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);