4 * Bring external RSS feeds into rooms.
6 * Copyright (c) 2007-2009 by the citadel.org team
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 3 of the License, or
11 * (at your option) any later version.
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
18 * You should have received a copy of the GNU General Public License
19 * along with this program; if not, write to the Free Software
20 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
27 #if TIME_WITH_SYS_TIME
28 # include <sys/time.h>
32 # include <sys/time.h>
41 #include <sys/types.h>
44 #include <curl/curl.h>
45 #include <libcitadel.h>
48 #include "citserver.h"
52 #include "ctdl_module.h"
53 #include "clientsocket.h"
55 #include "parsedate.h"
57 #include "citadel_dirs.h"
61 typedef struct rssnetcfg rssnetcfg;
67 time_t last_error_when;
75 typedef struct _rss_item {
85 char channel_title[256];
87 char *author_or_creator;
92 typedef struct _rsscollection {
99 struct rssnetcfg *rnclist = NULL;
108 * dug this out of the trashcan of the midgard project, lets see if it works for us.
109 * original code by Alexander Bokovoy <bokovoy@avilink.ne> distributed under GPL V2 or later
113 >= 0 - successfull, 0 means conversion doesn't use multibyte sequences
114 -1 - error during iconv_open call
115 -2 - error during iconv_close call
116 ----------------------------------
117 This function expects that multibyte encoding in 'charset' wouldn't have
118 characters with more than 3 bytes. It is not intended to convert UTF-8 because
119 we'll never receive UTF-8 in our handler (it is handled by Exat itself).
122 fill_encoding_info (const char *charset, XML_Encoding * info)
124 iconv_t cd = (iconv_t)(-1);
126 CtdlLogPrintf(0, "RSS: fill encoding info ...\n");
128 #if G_BYTE_ORDER == G_LITTLE_ENDIAN
129 cd = iconv_open ("UCS-2LE", charset);
131 cd = iconv_open ("UCS-2BE", charset);
134 if (cd == (iconv_t) (-1))
141 unsigned char buf[4];
142 unsigned int i0, i1, i2;
145 for (i0 = 0; i0 < 0x100; i0++)
149 //result = try (cd, buf, 1, &out);
159 for (i1 = 0; i1 < 0x100; i1++)
162 ///result = try (cd, buf, 2, &out);
173 for (i2 = 0; i2 < 0x100; i2++)
176 ////result = try (cd, buf, 3, &out);
192 if (iconv_close (cd) < 0)
200 iconv_convertor (void *data, const char *s)
202 XML_Encoding *info = data;
204 CtdlLogPrintf(0, "RSS: Converting ...\n");
210 result = g_byte_array_new ();
211 if (process_block (info->data, (char *) s, strlen (s), result) == 0)
213 res = *(result->data);
214 g_byte_array_free (result, TRUE);
217 g_byte_array_free (result, TRUE);
223 my_release (void *data)
225 iconv_t cd = (iconv_t) data;
226 if (iconv_close (cd) != 0)
228 /// TODO: uh no. exit (1);
232 handle_unknown_xml_encoding (void *encodingHandleData,
233 const XML_Char * name,
237 CtdlLogPrintf(0, "RSS: unknown encoding ...\n");
238 result = fill_encoding_info (name, info);
242 Special case: client asked for reverse conversion, we'll provide him with
243 iconv descriptor which handles it. Client should release it by himself.
245 if(encodingHandleData != NULL)
246 *((iconv_t *)encodingHandleData) = iconv_open(name, "UTF-8");
248 Optimization: we do not need conversion function if encoding is one-to-one,
249 info->map table will be enough
254 info->convert = NULL;
255 info->release = NULL;
259 We do need conversion function because this encoding uses multibyte sequences
261 info->data = (void *) iconv_open ("UTF-8", name);
262 if ((int)info->data == -1)
264 info->convert = iconv_convertor;
265 info->release = my_release;
268 if(encodingHandleData != NULL)
269 *(iconv_t *)encodingHandleData = NULL;
277 * Commit a fetched and parsed RSS item to disk
279 void rss_save_item(rsscollection *rssc) {
281 struct MD5Context md5context;
282 u_char rawdigest[MD5_DIGEST_LEN];
285 struct cdbdata *cdbut;
287 struct CtdlMessage *msg;
288 struct recptypes *recp = NULL;
290 rss_item *ri = rssc->Item;
292 recp = (struct recptypes *) malloc(sizeof(struct recptypes));
293 if (recp == NULL) return;
294 memset(recp, 0, sizeof(struct recptypes));
295 memset(&ut, 0, sizeof(struct UseTable));
296 recp->recp_room = strdup(ri->roomlist);
297 recp->num_room = num_tokens(ri->roomlist, '|');
298 recp->recptypes_magic = RECPTYPES_MAGIC;
300 /* Construct a GUID to use in the S_USETABLE table.
301 * If one is not present in the item itself, make one up.
303 if (ri->guid != NULL) {
304 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ri->guid);
307 MD5Init(&md5context);
308 if (ri->title != NULL) {
309 MD5Update(&md5context, (unsigned char*)ri->title, strlen(ri->title));
311 if (ri->link != NULL) {
312 MD5Update(&md5context, (unsigned char*)ri->link, strlen(ri->link));
314 MD5Final(rawdigest, &md5context);
315 for (i=0; i<MD5_DIGEST_LEN; i++) {
316 sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
317 utmsgid[i*2] = tolower(utmsgid[i*2]);
318 utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
320 strcat(utmsgid, "_rss2ctdl");
323 /* Find out if we've already seen this item */
324 cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
326 /* Item has already been seen */
327 CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
330 /* rewrite the record anyway, to update the timestamp */
331 strcpy(ut.ut_msgid, utmsgid);
332 ut.ut_timestamp = time(NULL);
333 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
336 /* Item has not been seen, so save it. */
337 CtdlLogPrintf(CTDL_DEBUG, "RSS: saving item...\n");
338 if (ri->description == NULL) ri->description = strdup("");
339 for (i=strlen(ri->description); i>=0; --i) {
340 if (isspace(ri->description[i])) {
341 ri->description[i] = ' ';
345 msg = malloc(sizeof(struct CtdlMessage));
346 memset(msg, 0, sizeof(struct CtdlMessage));
347 msg->cm_magic = CTDLMESSAGE_MAGIC;
348 msg->cm_anon_type = MES_NORMAL;
349 msg->cm_format_type = FMT_RFC822;
351 if (ri->guid != NULL) {
352 msg->cm_fields['E'] = strdup(ri->guid);
355 if (ri->author_or_creator != NULL) {
356 msg->cm_fields['A'] = html_to_ascii(ri->author_or_creator,
357 strlen(ri->author_or_creator), 512, 0);
358 striplt(msg->cm_fields['A']);
361 msg->cm_fields['A'] = strdup("rss");
364 msg->cm_fields['N'] = strdup(NODENAME);
365 if (ri->title != NULL) {
366 msg->cm_fields['U'] = html_to_ascii(ri->title, strlen(ri->title), 512, 0);
367 striplt(msg->cm_fields['U']);
369 msg->cm_fields['T'] = malloc(64);
370 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
371 if (ri->channel_title != NULL) {
372 if (!IsEmptyStr(ri->channel_title)) {
373 msg->cm_fields['O'] = strdup(ri->channel_title);
376 if (ri->link == NULL)
377 ri->link = strdup("");
378 msglen += 1024 + strlen(ri->link) + strlen(ri->description) ;
379 msg->cm_fields['M'] = malloc(msglen);
380 snprintf(msg->cm_fields['M'], msglen,
381 "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
384 "<a href=\"%s\">%s</a>\n"
391 CtdlSubmitMsg(msg, recp, NULL, 0);
392 CtdlFreeMessage(msg);
394 /* write the uidl to the use table so we don't store this item again */
395 strcpy(ut.ut_msgid, utmsgid);
396 ut.ut_timestamp = time(NULL);
397 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
399 free_recipients(recp);
405 * Convert an RDF/RSS datestamp into a time_t
407 time_t rdf_parsedate(char *p)
413 if (strlen(p) < 10) return 0L;
415 memset(&tm, 0, sizeof tm);
418 * If the timestamp appears to be in W3C datetime format, try to
419 * parse it. See also: http://www.w3.org/TR/NOTE-datetime
421 * This code, along with parsedate.c, is a potential candidate for
422 * moving into libcitadel.
424 if ( (p[4] == '-') && (p[7] == '-') ) {
425 tm.tm_year = atoi(&p[0]) - 1900;
426 tm.tm_mon = atoi(&p[5]) - 1;
427 tm.tm_mday = atoi(&p[8]);
428 if ( (p[10] == 'T') && (p[13] == ':') ) {
429 tm.tm_hour = atoi(&p[11]);
430 tm.tm_min = atoi(&p[14]);
435 /* hmm... try RFC822 date stamp format */
438 if (t > 0) return(t);
440 /* yeesh. ok, just return the current date and time. */
448 void flush_rss_ite(rss_item *ri)
450 /* Initialize the feed item data structure */
451 if (ri->guid != NULL) free(ri->guid);
453 if (ri->title != NULL) free(ri->title);
455 if (ri->link != NULL) free(ri->link);
457 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
458 ri->author_or_creator = NULL;
459 if (ri->author_url != NULL) free(ri->author_url);
460 ri->author_url = NULL;
461 if (ri->description != NULL) free(ri->description);
462 ri->description = NULL;
463 /* Throw away any existing character data */
464 if (ri->chardata_len > 0) {
467 ri->chardata_len = 0;
471 void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
472 rsscollection *rssc = (rsscollection*) data;
473 rss_item *ri = rssc->Item;
477 /* Axe the namespace, we don't care about it */
478 /// CtdlLogPrintf(0, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
479 safestrncpy(el, supplied_el, sizeof el);
480 while (sep = strchr(el, ':'), sep) {
484 if ((rssc->Cfg->ItemType == RSS_UNSET) && !strcasecmp(el, "rss"))
486 CtdlLogPrintf(9, "RSS: This is an RSS feed.\n");
487 rssc->Cfg->ItemType = RSS_RSS;
489 if ((rssc->Cfg->ItemType == RSS_UNSET) && !strcasecmp(el, "rdf"))
491 CtdlLogPrintf(9, "RSS: This is an RDF feed.\n");
492 rssc->Cfg->ItemType = RSS_RSS;
494 else if ((rssc->Cfg->ItemType == RSS_UNSET) && !strcasecmp(el, "feed"))
496 CtdlLogPrintf(9, "RSS: This is an ATOM feed.\n");
497 rssc->Cfg->ItemType = RSS_ATOM;
499 else if ((rssc->Cfg->ItemType == RSS_RSS) &&
500 !strcasecmp(el, "item"))
502 ri->item_tag_nesting ++ ;
505 else if ( (rssc->Cfg->ItemType == RSS_ATOM) &&
506 !strcasecmp(el, "entry"))
508 ++ri->item_tag_nesting;
511 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
512 !strcasecmp(el, "link"))
517 for (found = 0, i = 0;!found && attr[i] != NULL; i+=2)
519 if (!strcmp(attr[i], "href"))
522 if (ri->link != NULL)
524 ri->link = strdup(attr[i+1]);
533 void rss_xml_end(void *data, const char *supplied_el) {
534 rsscollection *rssc = (rsscollection*) data;
535 rss_item *ri = rssc->Item;
539 /* Axe the namespace, we don't care about it */
540 safestrncpy(el, supplied_el, sizeof el);
541 while (sep = strchr(el, ':'), sep) {
544 // CtdlLogPrintf(0, "RSS: END %s...\n", el);
546 if ( (!strcasecmp(el, "title")) && (ri->item_tag_nesting == 0) && (ri->chardata != NULL) ) {
547 safestrncpy(ri->channel_title, ri->chardata, sizeof ri->channel_title);
548 striplt(ri->channel_title);
551 if ( (rssc->Cfg->ItemType == RSS_RSS) &&
552 (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) {
553 if (ri->guid != NULL) free(ri->guid);
554 striplt(ri->chardata);
555 ri->guid = strdup(ri->chardata);
557 else if ( (rssc->Cfg->ItemType == RSS_ATOM) &&
558 (!strcasecmp(el, "id")) && (ri->chardata != NULL) ) {
559 if (ri->guid != NULL) free(ri->guid);
560 striplt(ri->chardata);
561 ri->guid = strdup(ri->chardata);
564 else if ( (rssc->Cfg->ItemType == RSS_RSS) && (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) {
565 if (ri->link != NULL) free(ri->link);
566 striplt(ri->chardata);
567 ri->link = strdup(ri->chardata);
570 else if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) {
571 if (ri->title != NULL) free(ri->title);
572 striplt(ri->chardata);
573 ri->title = strdup(ri->chardata);
576 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
577 (!strcasecmp(el, "content")) &&
578 (ri->chardata != NULL) ) {
579 if (ri->description != NULL) free(ri->description);
580 ri->description = strdup(ri->chardata);
582 else if ( (rssc->Cfg->ItemType == RSS_RSS) &&
583 (!strcasecmp(el, "description")) &&
584 (ri->chardata != NULL) ) {
585 if (ri->description != NULL) free(ri->description);
586 ri->description = strdup(ri->chardata);
589 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
590 ((!strcasecmp(el, "published")) ||
591 (!strcasecmp(el, "updated"))) &&
592 (ri->chardata != NULL) ) {
593 striplt(ri->chardata);
594 ri->pubdate = rdf_parsedate(ri->chardata);
598 else if ((rssc->Cfg->ItemType == RSS_RSS) &&
599 ((!strcasecmp(el, "pubdate")) ||
600 (!strcasecmp(el, "date"))) &&
601 (ri->chardata != NULL) ) {
602 striplt(ri->chardata);
603 ri->pubdate = rdf_parsedate(ri->chardata);
606 else if ((rssc->Cfg->ItemType == RSS_RSS) &&
607 ((!strcasecmp(el, "author")) ||
608 (!strcasecmp(el, "creator"))) &&
609 (ri->chardata != NULL) ) {
610 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
611 striplt(ri->chardata);
612 ri->author_or_creator = strdup(ri->chardata);
615 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
616 (!strcasecmp(el, "name")) &&
617 (ri->chardata != NULL) ) {
618 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
619 striplt(ri->chardata);
620 ri->author_or_creator = strdup(ri->chardata);
622 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
623 (!strcasecmp(el, "uri")) &&
624 (ri->chardata != NULL) ) {
625 if (ri->author_url != NULL) free(ri->author_url);
626 striplt(ri->chardata);
627 ri->author_url = strdup(ri->chardata);
630 else if ((rssc->Cfg->ItemType == RSS_RSS) &&
631 !strcasecmp(el, "item")) {
632 --ri->item_tag_nesting;
635 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
636 !strcasecmp(el, "entry")) {
637 --ri->item_tag_nesting;
641 else if ( (!strcasecmp(el, "rss")) ||
642 (!strcasecmp(el, "rdf")) ) {
643 // CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
644 ri->done_parsing = 1;
647 if (ri->chardata_len > 0) {
650 ri->chardata_len = 0;
657 * This callback stores up the data which appears in between tags.
659 void rss_xml_chardata(void *data, const XML_Char *s, int len) {
660 rsscollection *rssc = (rsscollection*) data;
661 rss_item *ri = rssc->Item;
666 old_len = ri->chardata_len;
667 new_len = old_len + len;
668 new_buffer = realloc(ri->chardata, new_len + 1);
669 if (new_buffer != NULL) {
670 memcpy(&new_buffer[old_len], s, len);
671 new_buffer[new_len] = 0;
672 ri->chardata = new_buffer;
673 ri->chardata_len = new_len;
680 * Callback function for passing libcurl's output to expat for parsing
682 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
684 XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
693 void rss_do_fetching(rssnetcfg *Cfg) {
700 char errmsg[1024] = "";
705 CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
707 curl = curl_easy_init();
709 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
713 xp = XML_ParserCreateNS("UTF-8", ':');
715 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
716 curl_easy_cleanup(curl);
720 curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
721 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
722 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
723 curl_easy_setopt(curl, CURLOPT_WRITEDATA, xp);
724 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
725 curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
726 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
727 #ifdef CURLOPT_HTTP_CONTENT_DECODING
728 curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
729 curl_easy_setopt(curl, CURLOPT_ENCODING, "");
731 curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
732 curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180); /* die after 180 seconds */
733 if (!IsEmptyStr(config.c_ip_addr)) {
734 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
737 memset(&ri, 0, sizeof(rss_item));
738 ri.roomlist = Cfg->rooms;
741 XML_SetUnknownEncodingHandler(xp,
742 handle_unknown_xml_encoding,
746 XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
747 XML_SetCharacterDataHandler(xp, rss_xml_chardata);
748 XML_SetUserData(xp, &rssc);
749 if (CtdlThreadCheckStop())
752 curl_easy_cleanup(curl);
756 if (CtdlThreadCheckStop())
759 res = curl_easy_perform(curl);
761 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
764 if (CtdlThreadCheckStop())
767 if (ri.done_parsing == 0)
768 XML_Parse(xp, "", 0, 1);
771 CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n",
773 XML_GetErrorCode(xp)));
776 curl_easy_cleanup(curl);
784 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
786 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
788 char filename[PATH_MAX];
793 rssnetcfg *rncptr = NULL;
794 rssnetcfg *use_this_rncptr = NULL;
798 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
800 if (CtdlThreadCheckStop())
803 /* Only do net processing for rooms that have netconfigs */
804 fp = fopen(filename, "r");
809 while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
810 buf[strlen(buf)-1] = 0;
812 extract_token(instr, buf, 0, '|', sizeof instr);
813 if (!strcasecmp(instr, "rssclient")) {
815 use_this_rncptr = NULL;
817 extract_token(feedurl, buf, 1, '|', sizeof feedurl);
819 /* If any other rooms have requested the same feed, then we will just add this
820 * room to the target list for that client request.
822 for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
823 if (!strcmp(rncptr->url, feedurl)) {
824 use_this_rncptr = rncptr;
828 /* Otherwise create a new client request */
829 if (use_this_rncptr == NULL) {
830 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
831 rncptr->ItemType = RSS_UNSET;
832 if (rncptr != NULL) {
833 rncptr->next = rnclist;
834 safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
835 rncptr->rooms = NULL;
837 use_this_rncptr = rncptr;
841 /* Add the room name to the request */
842 if (use_this_rncptr != NULL) {
843 if (use_this_rncptr->rooms == NULL) {
844 rncptr->rooms = strdup(qrbuf->QRname);
847 len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
848 ptr = realloc(use_this_rncptr->rooms, len);
851 strcat(ptr, qrbuf->QRname);
852 use_this_rncptr->rooms = ptr;
865 * Scan for rooms that have RSS client requests configured
867 void *rssclient_scan(void *args) {
868 static time_t last_run = 0L;
869 static int doing_rssclient = 0;
870 rssnetcfg *rptr = NULL;
871 CitContext rssclientCC;
873 /* Give this thread its own private CitContext */
874 CtdlFillSystemContext(&rssclientCC, "rssclient");
875 citthread_setspecific(MyConKey, (void *)&rssclientCC );
878 * This is a simple concurrency check to make sure only one rssclient run
879 * is done at a time. We could do this with a mutex, but since we
880 * don't really require extremely fine granularity here, we'll do it
881 * with a static variable instead.
883 if (doing_rssclient) return NULL;
886 CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
887 CtdlForEachRoom(rssclient_scan_room, NULL);
889 while (rnclist != NULL && !CtdlThreadCheckStop()) {
890 rss_do_fetching(rnclist);
892 rnclist = rnclist->next;
893 if (rptr->rooms != NULL) free(rptr->rooms);
897 CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
898 last_run = time(NULL);
900 if (!CtdlThreadCheckStop())
901 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
903 CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
904 CtdlClearSystemContext();
909 CTDL_MODULE_INIT(rssclient)
913 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
914 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
916 /* return our Subversion id for the Log */