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 CtdlLogPrintf(0, "RSS: saving item...\n");
293 recp = (struct recptypes *) malloc(sizeof(struct recptypes));
294 if (recp == NULL) return;
295 memset(recp, 0, sizeof(struct recptypes));
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. */
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->author_or_creator != NULL) {
352 msg->cm_fields['A'] = html_to_ascii(ri->author_or_creator,
353 strlen(ri->author_or_creator), 512, 0);
354 striplt(msg->cm_fields['A']);
357 msg->cm_fields['A'] = strdup("rss");
360 msg->cm_fields['N'] = strdup(NODENAME);
361 if (ri->title != NULL) {
362 msg->cm_fields['U'] = html_to_ascii(ri->title, strlen(ri->title), 512, 0);
363 striplt(msg->cm_fields['U']);
365 msg->cm_fields['T'] = malloc(64);
366 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
367 if (ri->channel_title != NULL) {
368 if (!IsEmptyStr(ri->channel_title)) {
369 msg->cm_fields['O'] = strdup(ri->channel_title);
372 if (ri->link == NULL)
373 ri->link = strdup("");
374 msglen += 1024 + strlen(ri->link) + strlen(ri->description) ;
375 msg->cm_fields['M'] = malloc(msglen);
376 snprintf(msg->cm_fields['M'], msglen,
377 "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
380 "<a href=\"%s\">%s</a>\n"
387 CtdlSubmitMsg(msg, recp, NULL, 0);
388 CtdlFreeMessage(msg);
390 /* write the uidl to the use table so we don't store this item again */
391 strcpy(ut.ut_msgid, utmsgid);
392 ut.ut_timestamp = time(NULL);
393 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
395 free_recipients(recp);
401 * Convert an RDF/RSS datestamp into a time_t
403 time_t rdf_parsedate(char *p)
409 if (strlen(p) < 10) return 0L;
411 memset(&tm, 0, sizeof tm);
414 * If the timestamp appears to be in W3C datetime format, try to
415 * parse it. See also: http://www.w3.org/TR/NOTE-datetime
417 * This code, along with parsedate.c, is a potential candidate for
418 * moving into libcitadel.
420 if ( (p[4] == '-') && (p[7] == '-') ) {
421 tm.tm_year = atoi(&p[0]) - 1900;
422 tm.tm_mon = atoi(&p[5]) - 1;
423 tm.tm_mday = atoi(&p[8]);
424 if ( (p[10] == 'T') && (p[13] == ':') ) {
425 tm.tm_hour = atoi(&p[11]);
426 tm.tm_min = atoi(&p[14]);
431 /* hmm... try RFC822 date stamp format */
434 if (t > 0) return(t);
436 /* yeesh. ok, just return the current date and time. */
444 void flush_rss_ite(rss_item *ri)
446 /* Initialize the feed item data structure */
447 if (ri->guid != NULL) free(ri->guid);
449 if (ri->title != NULL) free(ri->title);
451 if (ri->link != NULL) free(ri->link);
453 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
454 ri->author_or_creator = NULL;
455 if (ri->description != NULL) free(ri->description);
456 ri->description = NULL;
457 /* Throw away any existing character data */
458 if (ri->chardata_len > 0) {
461 ri->chardata_len = 0;
465 void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
466 rsscollection *rssc = (rsscollection*) data;
467 rss_item *ri = rssc->Item;
471 /* Axe the namespace, we don't care about it */
472 /// CtdlLogPrintf(0, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
473 safestrncpy(el, supplied_el, sizeof el);
474 while (sep = strchr(el, ':'), sep) {
478 if ((rssc->Cfg->ItemType == RSS_UNSET) &&
479 !strcasecmp(el, "rss"))
481 rssc->Cfg->ItemType = RSS_RSS;
483 else if ((rssc->Cfg->ItemType == RSS_UNSET) &&
484 !strcasecmp(el, "feed"))
486 rssc->Cfg->ItemType = RSS_ATOM;
488 else if ((rssc->Cfg->ItemType == RSS_RSS) &&
489 !strcasecmp(el, "item"))
491 ri->item_tag_nesting ++ ;
494 else if ( (rssc->Cfg->ItemType == RSS_ATOM) &&
495 !strcasecmp(el, "entry"))
497 CtdlLogPrintf(0, "RSS: found atom...\n");
498 ++ri->item_tag_nesting;
501 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
502 !strcasecmp(el, "link"))
507 for (found = 0, i = 0;!found && attr[i] != NULL; i+=2)
509 if (!strcmp(attr[i], "href"))
512 if (ri->link != NULL)
514 ri->link = strdup(attr[i+1]);
523 void rss_xml_end(void *data, const char *supplied_el) {
524 rsscollection *rssc = (rsscollection*) data;
525 rss_item *ri = rssc->Item;
529 /* Axe the namespace, we don't care about it */
530 safestrncpy(el, supplied_el, sizeof el);
531 while (sep = strchr(el, ':'), sep) {
534 // CtdlLogPrintf(0, "RSS: END %s...\n", el);
536 if ( (!strcasecmp(el, "title")) && (ri->item_tag_nesting == 0) && (ri->chardata != NULL) ) {
537 safestrncpy(ri->channel_title, ri->chardata, sizeof ri->channel_title);
538 striplt(ri->channel_title);
541 if ( (rssc->Cfg->ItemType == RSS_RSS) &&
542 (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) {
543 if (ri->guid != NULL) free(ri->guid);
544 striplt(ri->chardata);
545 ri->guid = strdup(ri->chardata);
547 else if ( (rssc->Cfg->ItemType == RSS_ATOM) &&
548 (!strcasecmp(el, "id")) && (ri->chardata != NULL) ) {
549 if (ri->guid != NULL) free(ri->guid);
550 striplt(ri->chardata);
551 ri->guid = strdup(ri->chardata);
554 else if ( (rssc->Cfg->ItemType == RSS_RSS) && (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) {
555 if (ri->link != NULL) free(ri->link);
556 striplt(ri->chardata);
557 ri->link = strdup(ri->chardata);
560 else if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) {
561 if (ri->title != NULL) free(ri->title);
562 striplt(ri->chardata);
563 ri->title = strdup(ri->chardata);
566 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
567 (!strcasecmp(el, "content")) &&
568 (ri->chardata != NULL) ) {
569 if (ri->description != NULL) free(ri->description);
570 ri->description = strdup(ri->chardata);
572 else if ( (rssc->Cfg->ItemType == RSS_RSS) &&
573 (!strcasecmp(el, "description")) &&
574 (ri->chardata != NULL) ) {
575 if (ri->description != NULL) free(ri->description);
576 ri->description = strdup(ri->chardata);
579 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
580 ((!strcasecmp(el, "published")) ||
581 (!strcasecmp(el, "updated"))) &&
582 (ri->chardata != NULL) ) {
583 striplt(ri->chardata);
584 ri->pubdate = rdf_parsedate(ri->chardata);
588 else if ((rssc->Cfg->ItemType == RSS_RSS) &&
589 ((!strcasecmp(el, "pubdate")) ||
590 (!strcasecmp(el, "date"))) &&
591 (ri->chardata != NULL) ) {
592 striplt(ri->chardata);
593 ri->pubdate = rdf_parsedate(ri->chardata);
596 else if ((rssc->Cfg->ItemType == RSS_RSS) &&
597 ((!strcasecmp(el, "author")) ||
598 (!strcasecmp(el, "creator"))) &&
599 (ri->chardata != NULL) ) {
600 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
601 striplt(ri->chardata);
602 ri->author_or_creator = strdup(ri->chardata);
605 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
606 (!strcasecmp(el, "name")) &&
607 (ri->chardata != NULL) ) {
608 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
609 striplt(ri->chardata);
610 ri->author_or_creator = strdup(ri->chardata);
612 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
613 (!strcasecmp(el, "uri")) &&
614 (ri->chardata != NULL) ) {
615 if (ri->author_url != NULL) free(ri->author_url);
616 striplt(ri->chardata);
617 ri->author_url = strdup(ri->chardata);
620 else if ((rssc->Cfg->ItemType == RSS_RSS) &&
621 !strcasecmp(el, "item")) {
622 --ri->item_tag_nesting;
625 else if ((rssc->Cfg->ItemType == RSS_ATOM) &&
626 !strcasecmp(el, "entry")) {
627 --ri->item_tag_nesting;
631 else if ( (!strcasecmp(el, "rss")) ||
632 (!strcasecmp(el, "rdf")) ) {
633 // CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
634 ri->done_parsing = 1;
637 if (ri->chardata_len > 0) {
640 ri->chardata_len = 0;
647 * This callback stores up the data which appears in between tags.
649 void rss_xml_chardata(void *data, const XML_Char *s, int len) {
650 rsscollection *rssc = (rsscollection*) data;
651 rss_item *ri = rssc->Item;
656 old_len = ri->chardata_len;
657 new_len = old_len + len;
658 new_buffer = realloc(ri->chardata, new_len + 1);
659 if (new_buffer != NULL) {
660 memcpy(&new_buffer[old_len], s, len);
661 new_buffer[new_len] = 0;
662 ri->chardata = new_buffer;
663 ri->chardata_len = new_len;
670 * Callback function for passing libcurl's output to expat for parsing
672 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
674 XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
683 void rss_do_fetching(rssnetcfg *Cfg) {
690 char errmsg[1024] = "";
695 CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
697 curl = curl_easy_init();
699 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
703 xp = XML_ParserCreateNS("UTF-8", ':');
705 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
706 curl_easy_cleanup(curl);
710 curl_easy_setopt(curl, CURLOPT_URL, Cfg->url);
711 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
712 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
713 curl_easy_setopt(curl, CURLOPT_WRITEDATA, xp);
714 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
715 curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
716 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
717 #ifdef CURLOPT_HTTP_CONTENT_DECODING
718 curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
719 curl_easy_setopt(curl, CURLOPT_ENCODING, "");
721 curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
722 curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180); /* die after 180 seconds */
723 if (!IsEmptyStr(config.c_ip_addr)) {
724 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
727 memset(&ri, 0, sizeof(rss_item));
728 ri.roomlist = Cfg->rooms;
731 XML_SetUnknownEncodingHandler(xp,
732 handle_unknown_xml_encoding,
736 XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
737 XML_SetCharacterDataHandler(xp, rss_xml_chardata);
738 XML_SetUserData(xp, &rssc);
739 if (CtdlThreadCheckStop())
742 curl_easy_cleanup(curl);
746 if (CtdlThreadCheckStop())
749 res = curl_easy_perform(curl);
751 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
754 if (CtdlThreadCheckStop())
757 if (ri.done_parsing == 0)
758 XML_Parse(xp, "", 0, 1);
761 CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n",
763 XML_GetErrorCode(xp)));
766 curl_easy_cleanup(curl);
774 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
776 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
778 char filename[PATH_MAX];
783 rssnetcfg *rncptr = NULL;
784 rssnetcfg *use_this_rncptr = NULL;
788 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
790 if (CtdlThreadCheckStop())
793 /* Only do net processing for rooms that have netconfigs */
794 fp = fopen(filename, "r");
799 while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
800 buf[strlen(buf)-1] = 0;
802 extract_token(instr, buf, 0, '|', sizeof instr);
803 if (!strcasecmp(instr, "rssclient")) {
805 use_this_rncptr = NULL;
807 extract_token(feedurl, buf, 1, '|', sizeof feedurl);
809 /* If any other rooms have requested the same feed, then we will just add this
810 * room to the target list for that client request.
812 for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
813 if (!strcmp(rncptr->url, feedurl)) {
814 use_this_rncptr = rncptr;
818 /* Otherwise create a new client request */
819 if (use_this_rncptr == NULL) {
820 rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
821 rncptr->ItemType = RSS_UNSET;
822 if (rncptr != NULL) {
823 rncptr->next = rnclist;
824 safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
825 rncptr->rooms = NULL;
827 use_this_rncptr = rncptr;
831 /* Add the room name to the request */
832 if (use_this_rncptr != NULL) {
833 if (use_this_rncptr->rooms == NULL) {
834 rncptr->rooms = strdup(qrbuf->QRname);
837 len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
838 ptr = realloc(use_this_rncptr->rooms, len);
841 strcat(ptr, qrbuf->QRname);
842 use_this_rncptr->rooms = ptr;
855 * Scan for rooms that have RSS client requests configured
857 void *rssclient_scan(void *args) {
858 static time_t last_run = 0L;
859 static int doing_rssclient = 0;
860 rssnetcfg *rptr = NULL;
861 CitContext rssclientCC;
863 /* Give this thread its own private CitContext */
864 CtdlFillSystemContext(&rssclientCC, "rssclient");
865 citthread_setspecific(MyConKey, (void *)&rssclientCC );
867 CtdlThreadAllocTSD();
870 * This is a simple concurrency check to make sure only one rssclient run
871 * is done at a time. We could do this with a mutex, but since we
872 * don't really require extremely fine granularity here, we'll do it
873 * with a static variable instead.
875 if (doing_rssclient) return NULL;
878 CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
879 CtdlForEachRoom(rssclient_scan_room, NULL);
881 while (rnclist != NULL && !CtdlThreadCheckStop()) {
882 rss_do_fetching(rnclist);
884 rnclist = rnclist->next;
885 if (rptr->rooms != NULL) free(rptr->rooms);
889 CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
890 last_run = time(NULL);
892 if (!CtdlThreadCheckStop())
893 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
895 CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
900 CTDL_MODULE_INIT(rssclient)
904 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
905 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
907 /* return our Subversion id for the Log */