2 * $Id: serv_rssclient.c 5652 2007-10-29 20:14:48Z ajc $
4 * Bring external RSS feeds into rooms.
12 #if TIME_WITH_SYS_TIME
13 # include <sys/time.h>
17 # include <sys/time.h>
26 #include <sys/types.h>
29 #include <curl/curl.h>
30 #include <libcitadel.h>
33 #include "citserver.h"
38 #include "ctdl_module.h"
39 #include "clientsocket.h"
41 #include "parsedate.h"
43 #include "citadel_dirs.h"
48 struct rssnetcfg *next;
63 char channel_title[256];
67 struct rssnetcfg *rnclist = NULL;
71 * Commit a fetched and parsed RSS item to disk
73 void rss_save_item(struct rss_item *ri) {
75 struct MD5Context md5context;
76 u_char rawdigest[MD5_DIGEST_LEN];
79 struct cdbdata *cdbut;
81 struct CtdlMessage *msg;
82 struct recptypes *recp = NULL;
85 recp = (struct recptypes *) malloc(sizeof(struct recptypes));
86 if (recp == NULL) return;
87 memset(recp, 0, sizeof(struct recptypes));
88 recp->recp_room = strdup(ri->roomlist);
89 recp->num_room = num_tokens(ri->roomlist, '|');
90 recp->recptypes_magic = RECPTYPES_MAGIC;
92 /* Construct a GUID to use in the S_USETABLE table.
93 * If one is not present in the item itself, make one up.
95 if (ri->guid != NULL) {
96 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ri->guid);
100 if (ri->title != NULL) {
101 MD5Update(&md5context, (unsigned char*)ri->title, strlen(ri->title));
103 if (ri->link != NULL) {
104 MD5Update(&md5context, (unsigned char*)ri->link, strlen(ri->link));
106 MD5Final(rawdigest, &md5context);
107 for (i=0; i<MD5_DIGEST_LEN; i++) {
108 sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
109 utmsgid[i*2] = tolower(utmsgid[i*2]);
110 utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
112 strcat(utmsgid, "_rss2ctdl");
115 /* Find out if we've already seen this item */
116 cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
118 /* Item has already been seen */
119 CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
122 /* rewrite the record anyway, to update the timestamp */
123 strcpy(ut.ut_msgid, utmsgid);
124 ut.ut_timestamp = time(NULL);
125 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
128 /* Item has not been seen, so save it. */
130 if (ri->description == NULL) ri->description = strdup("");
131 for (i=strlen(ri->description); i>=0; --i) {
132 if (isspace(ri->description[i])) {
133 ri->description[i] = ' ';
137 msg = malloc(sizeof(struct CtdlMessage));
138 memset(msg, 0, sizeof(struct CtdlMessage));
139 msg->cm_magic = CTDLMESSAGE_MAGIC;
140 msg->cm_anon_type = MES_NORMAL;
141 msg->cm_format_type = FMT_RFC822;
142 msg->cm_fields['A'] = strdup("rss");
143 msg->cm_fields['N'] = strdup(NODENAME);
144 msg->cm_fields['U'] = strdup(ri->title);
145 msg->cm_fields['T'] = malloc(64);
146 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
147 if (!IsEmptyStr(ri->channel_title)) {
148 msg->cm_fields['O'] = strdup(ri->channel_title);
151 msglen = 1024 + strlen(ri->link) + strlen(ri->description) ;
152 msg->cm_fields['M'] = malloc(msglen);
153 snprintf(msg->cm_fields['M'], msglen,
154 "Content-type: text/html\r\n\r\n"
157 "<a href=\"%s\">%s</a>\n"
164 CtdlSubmitMsg(msg, recp, NULL);
165 CtdlFreeMessage(msg);
167 /* write the uidl to the use table so we don't store this item again */
168 strcpy(ut.ut_msgid, utmsgid);
169 ut.ut_timestamp = time(NULL);
170 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
172 free_recipients(recp);
178 * Convert an RDF/RSS datestamp into a time_t
180 time_t rdf_parsedate(char *p)
186 if (strlen(p) < 10) return 0L;
188 memset(&tm, 0, sizeof tm);
190 /* YYYY-MM-DDTHH:MM format...
192 if ( (p[4] == '-') && (p[7] == '-') ) {
193 tm.tm_year = atoi(&p[0]) - 1900;
194 tm.tm_mon = atoi(&p[5]) - 1;
195 tm.tm_mday = atoi(&p[8]);
196 if ( (p[10] == 'T') && (p[13] == ':') ) {
197 tm.tm_hour = atoi(&p[11]);
198 tm.tm_min = atoi(&p[14]);
203 /* hmm... try RFC822 date stamp format */
206 if (t > 0) return(t);
208 /* yeesh. ok, just return the current date and time. */
214 void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
215 struct rss_item *ri = (struct rss_item *) data;
219 /* Axe the namespace, we don't care about it */
220 safestrncpy(el, supplied_el, sizeof el);
221 while (sep = strchr(el, ':'), sep) {
225 if (!strcasecmp(el, "item")) {
226 ++ri->item_tag_nesting;
228 /* Initialize the feed item data structure */
229 if (ri->guid != NULL) free(ri->guid);
231 if (ri->title != NULL) free(ri->title);
233 if (ri->link != NULL) free(ri->link);
235 if (ri->description != NULL) free(ri->description);
236 ri->description = NULL;
238 /* Throw away any existing character data */
239 if (ri->chardata_len > 0) {
242 ri->chardata_len = 0;
250 void rss_xml_end(void *data, const char *supplied_el) {
251 struct rss_item *ri = (struct rss_item *) data;
255 /* Axe the namespace, we don't care about it */
256 safestrncpy(el, supplied_el, sizeof el);
257 while (sep = strchr(el, ':'), sep) {
261 if ( (!strcasecmp(el, "title")) && (ri->item_tag_nesting == 0) && (ri->chardata != NULL) ) {
262 safestrncpy(ri->channel_title, ri->chardata, sizeof ri->channel_title);
263 striplt(ri->channel_title);
266 if ( (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) {
267 if (ri->guid != NULL) free(ri->guid);
268 striplt(ri->chardata);
269 ri->guid = strdup(ri->chardata);
272 if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) {
273 if (ri->title != NULL) free(ri->title);
274 striplt(ri->chardata);
275 ri->title = strdup(ri->chardata);
278 if ( (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) {
279 if (ri->link != NULL) free(ri->link);
280 striplt(ri->chardata);
281 ri->link = strdup(ri->chardata);
284 if ( (!strcasecmp(el, "description")) && (ri->chardata != NULL) ) {
285 if (ri->description != NULL) free(ri->description);
286 ri->description = strdup(ri->chardata);
289 if ( ((!strcasecmp(el, "pubdate")) || (!strcasecmp(el, "date"))) && (ri->chardata != NULL) ) {
290 striplt(ri->chardata);
291 ri->pubdate = rdf_parsedate(ri->chardata);
294 if (!strcasecmp(el, "item")) {
295 --ri->item_tag_nesting;
299 if ( (!strcasecmp(el, "rss")) || (!strcasecmp(el, "rdf")) ) {
300 CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
301 ri->done_parsing = 1;
304 if (ri->chardata_len > 0) {
307 ri->chardata_len = 0;
314 * This callback stores up the data which appears in between tags.
316 void rss_xml_chardata(void *data, const XML_Char *s, int len) {
317 struct rss_item *ri = (struct rss_item *) data;
322 old_len = ri->chardata_len;
323 new_len = old_len + len;
324 new_buffer = realloc(ri->chardata, new_len + 1);
325 if (new_buffer != NULL) {
326 memcpy(&new_buffer[old_len], s, len);
327 new_buffer[new_len] = 0;
328 ri->chardata = new_buffer;
329 ri->chardata_len = new_len;
336 * Callback function for passing libcurl's output to expat for parsing
338 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
340 XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
349 void rss_do_fetching(char *url, char *rooms) {
355 char errmsg[1024] = "";
357 CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", url);
359 curl = curl_easy_init();
361 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
365 xp = XML_ParserCreateNS("UTF-8", ':');
367 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
368 curl_easy_cleanup(curl);
372 curl_easy_setopt(curl, CURLOPT_URL, url);
373 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
374 curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
375 curl_easy_setopt(curl, CURLOPT_WRITEDATA, xp);
376 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
377 curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
378 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
380 memset(&ri, 0, sizeof(struct rss_item));
382 XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
383 XML_SetCharacterDataHandler(xp, rss_xml_chardata);
384 XML_SetUserData(xp, &ri);
386 if (CtdlThreadCheckStop())
389 curl_easy_cleanup(curl);
393 if (CtdlThreadCheckStop())
396 res = curl_easy_perform(curl);
398 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
401 if (CtdlThreadCheckStop())
404 if (ri.done_parsing == 0) XML_Parse(xp, "", 0, 1);
408 curl_easy_cleanup(curl);
411 /* Free the feed item data structure */
412 if (ri.guid != NULL) free(ri.guid);
414 if (ri.title != NULL) free(ri.title);
416 if (ri.link != NULL) free(ri.link);
418 if (ri.description != NULL) free(ri.description);
419 ri.description = NULL;
420 if (ri.chardata_len > 0) {
429 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
431 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
433 char filename[PATH_MAX];
438 struct rssnetcfg *rncptr = NULL;
439 struct rssnetcfg *use_this_rncptr = NULL;
443 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
445 if (CtdlThreadCheckStop())
448 /* Only do net processing for rooms that have netconfigs */
449 fp = fopen(filename, "r");
454 while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
455 buf[strlen(buf)-1] = 0;
457 extract_token(instr, buf, 0, '|', sizeof instr);
458 if (!strcasecmp(instr, "rssclient")) {
460 use_this_rncptr = NULL;
462 extract_token(feedurl, buf, 1, '|', sizeof feedurl);
464 /* If any other rooms have requested the same feed, then we will just add this
465 * room to the target list for that client request.
467 for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
468 if (!strcmp(rncptr->url, feedurl)) {
469 use_this_rncptr = rncptr;
473 /* Otherwise create a new client request */
474 if (use_this_rncptr == NULL) {
475 rncptr = (struct rssnetcfg *) malloc(sizeof(struct rssnetcfg));
476 if (rncptr != NULL) {
477 rncptr->next = rnclist;
478 safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
479 rncptr->rooms = NULL;
481 use_this_rncptr = rncptr;
485 /* Add the room name to the request */
486 if (use_this_rncptr != NULL) {
487 if (use_this_rncptr->rooms == NULL) {
488 rncptr->rooms = strdup(qrbuf->QRname);
491 len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
492 ptr = realloc(use_this_rncptr->rooms, len);
495 strcat(ptr, qrbuf->QRname);
496 use_this_rncptr->rooms = ptr;
509 * Scan for rooms that have RSS client requests configured
511 void *rssclient_scan(void *args) {
512 static time_t last_run = 0L;
513 static int doing_rssclient = 0;
514 struct rssnetcfg *rptr = NULL;
515 struct CitContext rssclientCC;
517 /* Give this thread its own private CitContext */
518 memset(&rssclientCC, 0, sizeof(struct CitContext));
519 rssclientCC.internal_pgm = 1;
520 rssclientCC.cs_pid = 0;
521 pthread_setspecific(MyConKey, (void *)&rssclientCC );
523 CtdlThreadAllocTSD();
526 * This is a simple concurrency check to make sure only one rssclient run
527 * is done at a time. We could do this with a mutex, but since we
528 * don't really require extremely fine granularity here, we'll do it
529 * with a static variable instead.
531 if (doing_rssclient) return NULL;
534 CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
535 ForEachRoom(rssclient_scan_room, NULL);
537 while (rnclist != NULL && !CtdlThreadCheckStop()) {
538 rss_do_fetching(rnclist->url, rnclist->rooms);
540 rnclist = rnclist->next;
541 if (rptr->rooms != NULL) free(rptr->rooms);
545 CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
546 last_run = time(NULL);
548 if (!CtdlThreadCheckStop())
549 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
551 CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
556 CTDL_MODULE_INIT(rssclient)
560 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
562 /* return our Subversion id for the Log */
563 return "$Id: serv_rssclient.c 5652 2007-10-29 20:14:48Z ajc $";