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;
309 * Commit a fetched and parsed RSS item to disk
311 void rss_save_item(rss_item *ri)
314 struct MD5Context md5context;
315 u_char rawdigest[MD5_DIGEST_LEN];
318 struct cdbdata *cdbut;
320 struct CtdlMessage *msg;
321 struct recptypes *recp = NULL;
324 recp = (struct recptypes *) malloc(sizeof(struct recptypes));
325 if (recp == NULL) return;
326 memset(recp, 0, sizeof(struct recptypes));
327 memset(&ut, 0, sizeof(struct UseTable));
328 recp->recp_room = strdup(ri->roomlist);
329 recp->num_room = num_tokens(ri->roomlist, '|');
330 recp->recptypes_magic = RECPTYPES_MAGIC;
332 /* Construct a GUID to use in the S_USETABLE table.
333 * If one is not present in the item itself, make one up.
335 if (ri->guid != NULL) {
336 StrBufSpaceToBlank(ri->guid);
337 StrBufTrim(ri->guid);
338 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ChrPtr(ri->guid));
341 MD5Init(&md5context);
342 if (ri->title != NULL) {
343 MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->title), StrLength(ri->title));
345 if (ri->link != NULL) {
346 MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->link), StrLength(ri->link));
348 MD5Final(rawdigest, &md5context);
349 for (i=0; i<MD5_DIGEST_LEN; i++) {
350 sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
351 utmsgid[i*2] = tolower(utmsgid[i*2]);
352 utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
354 strcat(utmsgid, "_rss2ctdl");
357 /* Find out if we've already seen this item */
359 cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
360 #ifdef DEBUG_RSS/////TODO ifndef
362 /* Item has already been seen */
363 CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
366 /* rewrite the record anyway, to update the timestamp */
367 strcpy(ut.ut_msgid, utmsgid);
368 ut.ut_timestamp = time(NULL);
369 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
374 /* Item has not been seen, so save it. */
375 CtdlLogPrintf(CTDL_DEBUG, "RSS: saving item...\n");
376 if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY(""));
377 StrBufSpaceToBlank(ri->description);
378 msg = malloc(sizeof(struct CtdlMessage));
379 memset(msg, 0, sizeof(struct CtdlMessage));
380 msg->cm_magic = CTDLMESSAGE_MAGIC;
381 msg->cm_anon_type = MES_NORMAL;
382 msg->cm_format_type = FMT_RFC822;
384 if (ri->guid != NULL) {
385 msg->cm_fields['E'] = strdup(ChrPtr(ri->guid));
388 if (ri->author_or_creator != NULL) {
390 StrBuf *Encoded, *QPEncoded;
392 StrBuf *EmailAddress;
396 UserName = NewStrBuf();
397 EmailAddress = NewStrBuf();
398 EncBuf = NewStrBuf();
400 From = html_to_ascii(ChrPtr(ri->author_or_creator),
401 StrLength(ri->author_or_creator),
403 FromAt = strchr(From, '@') != NULL;
404 if (!FromAt && StrLength (ri->author_email) > 0)
406 Encoded = NewStrBuf();
407 if (!IsEmptyStr(From))
409 StrBufPrintf(Encoded,
412 ChrPtr(ri->author_email));
416 StrBufPrintf(Encoded,
418 ChrPtr(ri->author_email));
424 Encoded = NewStrBufPlain(From, -1);
427 Encoded = NewStrBuf();
428 StrBufPrintf(Encoded,
431 "rss@localhost"); /// TODO: get hostname?
436 QPEncoded = StrBufSanitizeEmailRecipientVector(Encoded, UserName, EmailAddress, EncBuf);
437 msg->cm_fields['A'] = SmashStrBuf(&QPEncoded);
439 FreeStrBuf(&Encoded);
440 FreeStrBuf(&UserName);
441 FreeStrBuf(&EmailAddress);
446 msg->cm_fields['A'] = strdup("rss");
449 msg->cm_fields['N'] = strdup(NODENAME);
450 if (ri->title != NULL) {
453 StrBuf *Encoded, *QPEncoded;
456 StrBufSpaceToBlank(ri->title);
457 len = StrLength(ri->title);
458 Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
459 Encoded = NewStrBufPlain(Sbj, -1);
463 StrBufRFC2047encode(&QPEncoded, Encoded);
465 msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
466 FreeStrBuf(&Encoded);
468 msg->cm_fields['T'] = malloc(64);
469 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
470 if (ri->channel_title != NULL) {
471 if (StrLength(ri->channel_title) > 0) {
472 msg->cm_fields['O'] = strdup(ChrPtr(ri->channel_title));
475 if (ri->link == NULL)
476 ri->link = NewStrBufPlain(HKEY(""));
477 msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
478 msg->cm_fields['M'] = malloc(msglen);
479 snprintf(msg->cm_fields['M'], msglen,
480 "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
483 "<a href=\"%s\">%s</a><br>\n"
484 "<a href=\"%s\">%s</a>\n"
487 ChrPtr(ri->description),
489 (StrLength(ri->linkTitle)>0)?ChrPtr(ri->linkTitle):ChrPtr(ri->link),
491 (StrLength(ri->reLinkTitle)>0)?ChrPtr(ri->reLinkTitle):"Reply to this"
494 CtdlSubmitMsg(msg, recp, NULL, 0);
495 CtdlFreeMessage(msg);
497 /* write the uidl to the use table so we don't store this item again */
498 strcpy(ut.ut_msgid, utmsgid);
499 ut.ut_timestamp = time(NULL);
500 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
502 free_recipients(recp);
508 * Convert an RDF/RSS datestamp into a time_t
510 time_t rdf_parsedate(const char *p)
516 if (strlen(p) < 10) return 0L;
518 memset(&tm, 0, sizeof tm);
521 * If the timestamp appears to be in W3C datetime format, try to
522 * parse it. See also: http://www.w3.org/TR/NOTE-datetime
524 * This code, along with parsedate.c, is a potential candidate for
525 * moving into libcitadel.
527 if ( (p[4] == '-') && (p[7] == '-') ) {
528 tm.tm_year = atoi(&p[0]) - 1900;
529 tm.tm_mon = atoi(&p[5]) - 1;
530 tm.tm_mday = atoi(&p[8]);
531 if ( (p[10] == 'T') && (p[13] == ':') ) {
532 tm.tm_hour = atoi(&p[11]);
533 tm.tm_min = atoi(&p[14]);
538 /* hmm... try RFC822 date stamp format */
541 if (t > 0) return(t);
543 /* yeesh. ok, just return the current date and time. */
547 void flush_rss_item(rss_item *ri)
549 /* Initialize the feed item data structure */
550 FreeStrBuf(&ri->guid);
551 FreeStrBuf(&ri->title);
552 FreeStrBuf(&ri->link);
553 FreeStrBuf(&ri->author_or_creator);
554 FreeStrBuf(&ri->author_email);
555 FreeStrBuf(&ri->author_url);
556 FreeStrBuf(&ri->description);
559 void rss_xml_start(void *data, const char *supplied_el, const char **attr)
562 rsscollection *rssc = (rsscollection*) data;
563 rssnetcfg *Cfg = rssc->Cfg;
564 rss_item *ri = rssc->Item;
569 /* Axe the namespace, we don't care about it */
570 /// CtdlLogPrintf(0, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
572 while (sep = strchr(pel, ':'), sep) {
576 if (pel != supplied_el)
580 if (!GetHash(KnownNameSpaces,
582 pel - supplied_el - 1,
586 CtdlLogPrintf(0, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
593 StrBufPlain(rssc->Key, pel, -1);
594 StrBufLowerCase(rssc->Key);
595 if (GetHash(StartHandlers, SKEY(rssc->Key), &pv))
597 rssc->Current = h = (rss_xml_handler*) pv;
599 if (((h->Flags & RSS_UNSET) != 0) &&
600 (Cfg->ItemType == RSS_UNSET))
602 h->Handler(rssc->CData, ri, Cfg, attr);
604 else if (((h->Flags & RSS_RSS) != 0) &&
605 (Cfg->ItemType == RSS_RSS))
607 h->Handler(rssc->CData, ri, Cfg, attr);
609 else if (((h->Flags & RSS_ATOM) != 0) &&
610 (Cfg->ItemType == RSS_ATOM))
612 h->Handler(rssc->CData, ri, Cfg, attr);
616 CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
621 CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
625 void rss_xml_end(void *data, const char *supplied_el)
628 rsscollection *rssc = (rsscollection*) data;
629 rssnetcfg *Cfg = rssc->Cfg;
630 rss_item *ri = rssc->Item;
635 /* Axe the namespace, we don't care about it */
637 while (sep = strchr(pel, ':'), sep) {
640 // CtdlLogPrintf(0, "RSS: END %s...\n", el);
641 if (pel != supplied_el)
645 if (!GetHash(KnownNameSpaces,
647 pel - supplied_el - 1,
651 CtdlLogPrintf(0, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
652 supplied_el, ChrPtr(rssc->CData));
654 FlushStrBuf(rssc->CData);
659 StrBufPlain(rssc->Key, pel, -1);
660 StrBufLowerCase(rssc->Key);
661 if (GetHash(EndHandlers, SKEY(rssc->Key), &pv))
663 h = (rss_xml_handler*) pv;
665 if (((h->Flags & RSS_UNSET) != 0) &&
666 (Cfg->ItemType == RSS_UNSET))
668 h->Handler(rssc->CData, ri, Cfg, NULL);
670 else if (((h->Flags & RSS_RSS) != 0) &&
671 (Cfg->ItemType == RSS_RSS))
673 h->Handler(rssc->CData, ri, Cfg, NULL);
675 else if (((h->Flags & RSS_ATOM) != 0) &&
676 (Cfg->ItemType == RSS_ATOM))
678 h->Handler(rssc->CData, ri, Cfg, NULL);
682 CtdlLogPrintf(0, "RSS: END unhandled: [%s] [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
687 CtdlLogPrintf(0, "RSS: END unhandled: [%s] [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
689 FlushStrBuf(rssc->CData);
690 rssc->Current = NULL;
697 void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
699 CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RSS feed.\n");
700 Cfg->ItemType = RSS_RSS;
703 void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
705 CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RDF feed.\n");
706 Cfg->ItemType = RSS_RSS;
709 void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
711 CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an ATOM feed.\n");
712 Cfg->ItemType = RSS_ATOM;
716 void RSS_item_item_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
718 ri->item_tag_nesting ++;
722 void ATOM_item_entry_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
725 ri->item_tag_nesting ++;
729 void ATOM_item_link_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
732 const char *pHref = NULL;
733 const char *pType = NULL;
734 const char *pRel = NULL;
735 const char *pTitle = NULL;
737 for (i = 0; Attr[i] != NULL; i+=2)
739 if (!strcmp(Attr[i], "href"))
743 else if (!strcmp(Attr[i], "rel"))
747 else if (!strcmp(Attr[i], "type"))
751 else if (!strcmp(Attr[i], "title"))
757 return; /* WHUT? Pointing... where? */
758 if ((pType != NULL) && !strcasecmp(pType, "application/atom+xml"))
759 return; /* these just point to other rss resources, we're not interested in them. */
762 if (!strcasecmp (pRel, "replies"))
764 NewStrBufDupAppendFlush(&ri->reLink, NULL, pHref, -1);
765 StrBufTrim(ri->link);
766 NewStrBufDupAppendFlush(&ri->reLinkTitle, NULL, pTitle, -1);
768 else if (!strcasecmp(pRel, "alternate")) /* Alternative representation of this Item... */
770 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
771 StrBufTrim(ri->link);
772 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
775 #if 0 /* these are also defined, but dunno what to do with them.. */
776 else if (!strcasecmp(pRel, "related"))
779 else if (!strcasecmp(pRel, "self"))
782 else if (!strcasecmp(pRel, "enclosure"))
783 {/* this reference can get big, and is probably the full article... */
785 else if (!strcasecmp(pRel, "via"))
786 {/* this article was provided via... */
790 else if (StrLength(ri->link) == 0)
792 NewStrBufDupAppendFlush(&ri->link, NULL, pHref, -1);
793 StrBufTrim(ri->link);
794 NewStrBufDupAppendFlush(&ri->linkTitle, NULL, pTitle, -1);
801 void ATOMRSS_item_title_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
803 if ((ri->item_tag_nesting == 0) && (StrLength(CData) > 0)) {
804 NewStrBufDupAppendFlush(&ri->channel_title, CData, NULL, 0);
805 StrBufTrim(ri->channel_title);
809 void RSS_item_guid_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
811 if (StrLength(CData) > 0) {
812 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
816 void ATOM_item_id_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
818 if (StrLength(CData) > 0) {
819 NewStrBufDupAppendFlush(&ri->guid, CData, NULL, 0);
824 void RSS_item_link_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
826 if (StrLength(CData) > 0) {
827 NewStrBufDupAppendFlush(&ri->link, CData, NULL, 0);
828 StrBufTrim(ri->link);
832 void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
834 if (StrLength(CData) > 0) {
835 NewStrBufDupAppendFlush(&ri->title, CData, NULL, 0);
836 StrBufTrim(ri->title);
840 void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
842 if (StrLength(CData) > 0) {
843 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
844 StrBufTrim(ri->description);
847 void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
849 /* this can contain an abstract of the article. but we don't want to verwrite a full document if we already have it. */
850 if ((StrLength(CData) > 0) && (StrLength(ri->description) == 0))
852 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
853 StrBufTrim(ri->description);
857 void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
859 if (StrLength(CData) > 0) {
860 NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
864 void ATOM_item_published_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
866 if (StrLength(CData) > 0) {
868 ri->pubdate = rdf_parsedate(ChrPtr(CData));
872 void ATOM_item_updated_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
874 if (StrLength(CData) > 0) {
876 ri->pubdate = rdf_parsedate(ChrPtr(CData));
880 void RSS_item_pubdate_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
882 if (StrLength(CData) > 0) {
884 ri->pubdate = rdf_parsedate(ChrPtr(CData));
889 void RSS_item_date_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
891 if (StrLength(CData) > 0) {
893 ri->pubdate = rdf_parsedate(ChrPtr(CData));
899 void RSS_item_author_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
901 if (StrLength(CData) > 0) {
902 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
903 StrBufTrim(ri->author_or_creator);
908 void ATOM_item_name_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
910 if (StrLength(CData) > 0) {
911 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
912 StrBufTrim(ri->author_or_creator);
916 void ATOM_item_email_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
918 if (StrLength(CData) > 0) {
919 NewStrBufDupAppendFlush(&ri->author_email, CData, NULL, 0);
920 StrBufTrim(ri->author_email);
924 void RSS_item_creator_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
926 if ((StrLength(CData) > 0) &&
927 (StrLength(ri->author_or_creator) == 0))
929 NewStrBufDupAppendFlush(&ri->author_or_creator, CData, NULL, 0);
930 StrBufTrim(ri->author_or_creator);
935 void ATOM_item_uri_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
937 if (StrLength(CData) > 0) {
938 NewStrBufDupAppendFlush(&ri->author_url, CData, NULL, 0);
939 StrBufTrim(ri->author_url);
943 void RSS_item_item_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
945 --ri->item_tag_nesting;
950 void ATOM_item_entry_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
952 --ri->item_tag_nesting;
956 void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
958 // CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
959 ri->done_parsing = 1;
962 void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
964 // CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
965 ri->done_parsing = 1;
969 void RSSATOM_item_ignore(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
976 * This callback stores up the data which appears in between tags.
978 void rss_xml_cdata_start(void *data)
980 rsscollection *rssc = (rsscollection*) data;
982 FlushStrBuf(rssc->CData);
985 void rss_xml_cdata_end(void *data)
988 void rss_xml_chardata(void *data, const XML_Char *s, int len)
990 rsscollection *rssc = (rsscollection*) data;
992 StrBufAppendBufPlain (rssc->CData, s, len, 0);
996 * Callback function for passing libcurl's output to expat for parsing
998 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
1000 XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
1001 return (size*nmemb);
1007 * Begin a feed parse
1009 void rss_do_fetching(rssnetcfg *Cfg) {
1017 char errmsg[1024] = "";
1022 memset(&ri, 0, sizeof(rss_item));
1026 CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
1028 curl = curl_easy_init();
1030 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
1033 Answer = NewStrBufPlain(NULL, SIZ);
1035 curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
1036 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
1037 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
1038 curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
1039 // curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
1040 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
1041 curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
1042 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
1043 #ifdef CURLOPT_HTTP_CONTENT_DECODING
1044 curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
1045 curl_easy_setopt(curl, CURLOPT_ENCODING, "");
1047 curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
1048 curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180); /* die after 180 seconds */
1050 (!IsEmptyStr(config.c_ip_addr))
1051 && (strcmp(config.c_ip_addr, "*"))
1052 && (strcmp(config.c_ip_addr, "::"))
1053 && (strcmp(config.c_ip_addr, "0.0.0.0"))
1055 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
1058 if (CtdlThreadCheckStop())
1060 curl_easy_cleanup(curl);
1064 if (CtdlThreadCheckStop())
1067 res = curl_easy_perform(curl);
1069 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
1072 if (CtdlThreadCheckStop())
1078 memset(&ri, 0, sizeof(rss_item));
1079 ri.roomlist = Cfg->rooms;
1080 rssc.CData = NewStrBufPlain(NULL, SIZ);
1081 rssc.Key = NewStrBuf();
1083 StrBufSipLine(rssc.Key, Answer, &at);
1086 #define encoding "encoding=\""
1087 ptr = strstr(ChrPtr(rssc.Key), encoding);
1092 ptr += sizeof (encoding) - 1;
1093 pche = strchr(ptr, '"');
1095 StrBufCutAt(rssc.Key, -1, pche);
1103 xp = XML_ParserCreateNS(ptr, ':');
1105 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
1108 FlushStrBuf(rssc.Key);
1111 XML_SetUnknownEncodingHandler(xp,
1112 handle_unknown_xml_encoding,
1116 XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
1117 XML_SetCharacterDataHandler(xp, rss_xml_chardata);
1118 XML_SetUserData(xp, &rssc);
1119 XML_SetCdataSectionHandler(xp,
1120 rss_xml_cdata_start,
1124 len = StrLength(Answer);
1125 ptr = SmashStrBuf(&Answer);
1126 XML_Parse(xp, ptr, len, 0);
1128 if (ri.done_parsing == 0)
1129 XML_Parse(xp, "", 0, 1);
1132 CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n",
1134 XML_GetErrorCode(xp)));
1137 curl_easy_cleanup(curl);
1140 flush_rss_item(&ri);
1141 FreeStrBuf(&rssc.CData);
1142 FreeStrBuf(&rssc.Key);
1147 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
1149 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
1151 char filename[PATH_MAX];
1156 rssnetcfg *rncptr = NULL;
1157 rssnetcfg *use_this_rncptr = NULL;
1161 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
1163 if (CtdlThreadCheckStop())
1166 /* Only do net processing for rooms that have netconfigs */
1167 fp = fopen(filename, "r");
1172 while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
1173 buf[strlen(buf)-1] = 0;
1175 extract_token(instr, buf, 0, '|', sizeof instr);
1176 if (!strcasecmp(instr, "rssclient")) {
1178 use_this_rncptr = NULL;
1180 extract_token(feedurl, buf, 1, '|', sizeof feedurl);
1182 /* If any other rooms have requested the same feed, then we will just add this
1183 * room to the target list for that client request.
1185 for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
1186 if (!strcmp(rncptr->url, feedurl)) {
1187 use_this_rncptr = rncptr;
1191 /* Otherwise create a new client request */
1192 if (use_this_rncptr == NULL) {
1193 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
1194 rncptr->ItemType = RSS_UNSET;
1195 if (rncptr != NULL) {
1196 rncptr->next = rnclist;
1197 safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
1198 rncptr->rooms = NULL;
1200 use_this_rncptr = rncptr;
1204 /* Add the room name to the request */
1205 if (use_this_rncptr != NULL) {
1206 if (use_this_rncptr->rooms == NULL) {
1207 rncptr->rooms = strdup(qrbuf->QRname);
1210 len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
1211 ptr = realloc(use_this_rncptr->rooms, len);
1214 strcat(ptr, qrbuf->QRname);
1215 use_this_rncptr->rooms = ptr;
1228 * Scan for rooms that have RSS client requests configured
1230 void *rssclient_scan(void *args) {
1231 static time_t last_run = 0L;
1232 static int doing_rssclient = 0;
1233 rssnetcfg *rptr = NULL;
1234 CitContext rssclientCC;
1236 /* Give this thread its own private CitContext */
1237 CtdlFillSystemContext(&rssclientCC, "rssclient");
1238 citthread_setspecific(MyConKey, (void *)&rssclientCC );
1241 * This is a simple concurrency check to make sure only one rssclient run
1242 * is done at a time. We could do this with a mutex, but since we
1243 * don't really require extremely fine granularity here, we'll do it
1244 * with a static variable instead.
1246 if (doing_rssclient) return NULL;
1247 doing_rssclient = 1;
1249 CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
1250 CtdlForEachRoom(rssclient_scan_room, NULL);
1252 while (rnclist != NULL && !CtdlThreadCheckStop()) {
1253 rss_do_fetching(rnclist);
1255 rnclist = rnclist->next;
1256 if (rptr->rooms != NULL) free(rptr->rooms);
1260 CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
1261 last_run = time(NULL);
1262 doing_rssclient = 0;
1263 if (!CtdlThreadCheckStop())
1264 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
1266 CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
1267 CtdlClearSystemContext();
1272 CTDL_MODULE_INIT(rssclient)
1276 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
1277 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
1280 StartHandlers = NewHash(1, NULL);
1281 EndHandlers = NewHash(1, NULL);
1283 AddRSSStartHandler(RSS_item_rss_start, RSS_UNSET, HKEY("rss"));
1284 AddRSSStartHandler(RSS_item_rdf_start, RSS_UNSET, HKEY("rdf"));
1285 AddRSSStartHandler(ATOM_item_feed_start, RSS_UNSET, HKEY("feed"));
1286 AddRSSStartHandler(RSS_item_item_start, RSS_RSS, HKEY("item"));
1287 AddRSSStartHandler(ATOM_item_entry_start, RSS_ATOM, HKEY("entry"));
1288 AddRSSStartHandler(ATOM_item_link_start, RSS_ATOM, HKEY("link"));
1290 AddRSSEndHandler(ATOMRSS_item_title_end, RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1291 AddRSSEndHandler(RSS_item_guid_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
1292 AddRSSEndHandler(ATOM_item_id_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
1293 AddRSSEndHandler(RSS_item_link_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
1294 AddRSSEndHandler(RSSATOM_item_title_end, RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
1295 AddRSSEndHandler(ATOM_item_content_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
1296 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("encoded"));
1297 AddRSSEndHandler(ATOM_item_summary_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
1298 AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
1299 AddRSSEndHandler(ATOM_item_published_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
1300 AddRSSEndHandler(ATOM_item_updated_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
1301 AddRSSEndHandler(RSS_item_pubdate_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
1302 AddRSSEndHandler(RSS_item_date_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
1303 AddRSSEndHandler(RSS_item_author_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
1304 AddRSSEndHandler(RSS_item_creator_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
1306 AddRSSEndHandler(ATOM_item_email_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
1307 AddRSSEndHandler(ATOM_item_name_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
1308 AddRSSEndHandler(ATOM_item_uri_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
1310 AddRSSEndHandler(RSS_item_item_end, RSS_RSS, HKEY("item"));
1311 AddRSSEndHandler(RSS_item_rss_end, RSS_RSS, HKEY("rss"));
1312 AddRSSEndHandler(RSS_item_rdf_end, RSS_RSS, HKEY("rdf"));
1313 AddRSSEndHandler(ATOM_item_entry_end, RSS_ATOM, HKEY("entry"));
1316 /* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
1317 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("seq"));
1318 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("seq"));
1319 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("li"));
1320 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("li"));
1322 /* links to other feed generators... */
1323 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1324 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("feedflare"));
1325 AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1326 AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
1328 KnownNameSpaces = NewHash(1, NULL);
1329 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
1330 Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
1331 Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
1332 Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
1333 Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
1334 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1335 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
1336 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
1337 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
1338 Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
1339 Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
1340 Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
1341 Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
1342 Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
1343 Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
1344 Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
1345 Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
1346 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
1347 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1348 Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
1349 Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
1350 Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
1351 Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
1353 /* we don't like these namespaces because of they shadow our usefull parameters. */
1354 Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);