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>
28 #include <libcitadel.h>
31 #include "citserver.h"
35 #include "ctdl_module.h"
36 #include "clientsocket.h"
39 #include "citadel_dirs.h"
47 struct rssnetcfg *next;
62 char channel_title[256];
66 struct rssnetcfg *rnclist = NULL;
70 * Commit a fetched and parsed RSS item to disk
72 void rss_save_item(struct rss_item *ri) {
74 struct MD5Context md5context;
75 u_char rawdigest[MD5_DIGEST_LEN];
78 struct cdbdata *cdbut;
80 struct CtdlMessage *msg;
81 struct recptypes *recp = NULL;
84 recp = (struct recptypes *) malloc(sizeof(struct recptypes));
85 if (recp == NULL) return;
86 memset(recp, 0, sizeof(struct recptypes));
87 recp->recp_room = strdup(ri->roomlist);
88 recp->num_room = num_tokens(ri->roomlist, '|');
89 recp->recptypes_magic = RECPTYPES_MAGIC;
91 /* Construct a GUID to use in the S_USETABLE table.
92 * If one is not present in the item itself, make one up.
94 if (ri->guid != NULL) {
95 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ri->guid);
99 if (ri->title != NULL) {
100 MD5Update(&md5context, ri->title, strlen(ri->title));
102 if (ri->link != NULL) {
103 MD5Update(&md5context, ri->link, strlen(ri->link));
105 MD5Final(rawdigest, &md5context);
106 for (i=0; i<MD5_DIGEST_LEN; i++) {
107 sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
108 utmsgid[i*2] = tolower(utmsgid[i*2]);
109 utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
111 strcat(utmsgid, "_rss2ctdl");
114 /* Find out if we've already seen this item */
115 cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
117 /* Item has already been seen */
118 lprintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
121 /* rewrite the record anyway, to update the timestamp */
122 strcpy(ut.ut_msgid, utmsgid);
123 ut.ut_timestamp = time(NULL);
124 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
127 /* Item has not been seen, so save it. */
129 if (ri->description == NULL) ri->description = strdup("");
130 for (i=strlen(ri->description); i>=0; --i) {
131 if (isspace(ri->description[i])) {
132 ri->description[i] = ' ';
136 msg = malloc(sizeof(struct CtdlMessage));
137 memset(msg, 0, sizeof(struct CtdlMessage));
138 msg->cm_magic = CTDLMESSAGE_MAGIC;
139 msg->cm_anon_type = MES_NORMAL;
140 msg->cm_format_type = FMT_RFC822;
141 msg->cm_fields['A'] = strdup("rss");
142 msg->cm_fields['N'] = strdup(NODENAME);
143 msg->cm_fields['U'] = strdup(ri->title);
144 msg->cm_fields['T'] = malloc(64);
145 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
146 if (!IsEmptyStr(ri->channel_title)) {
147 msg->cm_fields['O'] = strdup(ri->channel_title);
150 msglen = 1024 + strlen(ri->link) + strlen(ri->description) ;
151 msg->cm_fields['M'] = malloc(msglen);
152 snprintf(msg->cm_fields['M'], msglen,
153 "Content-type: text/html\r\n\r\n"
156 "<a href=\"%s\">%s</a>\n"
163 CtdlSubmitMsg(msg, recp, NULL);
164 CtdlFreeMessage(msg);
165 free_recipients(recp);
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) );
177 * Convert an RDF/RSS datestamp into a time_t
179 time_t rdf_parsedate(char *p)
184 if (strlen(p) < 10) return 0L;
186 memset(&tm, 0, sizeof tm);
188 /* YYYY-MM-DDTHH:MM format...
190 if ( (p[4] == '-') && (p[7] == '-') ) {
191 tm.tm_year = atoi(&p[0]) - 1900;
192 tm.tm_mon = atoi(&p[5]) - 1;
193 tm.tm_mday = atoi(&p[8]);
194 if ( (p[10] == 'T') && (p[13] == ':') ) {
195 tm.tm_hour = atoi(&p[11]);
196 tm.tm_min = atoi(&p[14]);
201 /* FIXME try an imap timestamp conversion */
209 void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
210 struct rss_item *ri = (struct rss_item *) data;
214 /* Axe the namespace, we don't care about it */
215 safestrncpy(el, supplied_el, sizeof el);
216 while (sep = strchr(el, ':'), sep) {
220 if (!strcasecmp(el, "item")) {
221 ++ri->item_tag_nesting;
223 /* Initialize the feed item data structure */
224 if (ri->guid != NULL) free(ri->guid);
226 if (ri->title != NULL) free(ri->title);
228 if (ri->link != NULL) free(ri->link);
230 if (ri->description != NULL) free(ri->description);
231 ri->description = NULL;
233 /* Throw away any existing character data */
234 if (ri->chardata_len > 0) {
237 ri->chardata_len = 0;
245 void rss_xml_end(void *data, const char *supplied_el) {
246 struct rss_item *ri = (struct rss_item *) data;
250 /* Axe the namespace, we don't care about it */
251 safestrncpy(el, supplied_el, sizeof el);
252 while (sep = strchr(el, ':'), sep) {
256 if ( (!strcasecmp(el, "title")) && (ri->item_tag_nesting == 0) && (ri->chardata != NULL) ) {
257 safestrncpy(ri->channel_title, ri->chardata, sizeof ri->channel_title);
258 striplt(ri->channel_title);
261 if ( (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) {
262 if (ri->guid != NULL) free(ri->guid);
263 striplt(ri->chardata);
264 ri->guid = strdup(ri->chardata);
267 if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) {
268 if (ri->title != NULL) free(ri->title);
269 striplt(ri->chardata);
270 ri->title = strdup(ri->chardata);
273 if ( (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) {
274 if (ri->link != NULL) free(ri->link);
275 striplt(ri->chardata);
276 ri->link = strdup(ri->chardata);
279 if ( (!strcasecmp(el, "description")) && (ri->chardata != NULL) ) {
280 if (ri->description != NULL) free(ri->description);
281 ri->description = strdup(ri->chardata);
284 if ( ((!strcasecmp(el, "pubdate")) || (!strcasecmp(el, "date"))) && (ri->chardata != NULL) ) {
285 striplt(ri->chardata);
286 ri->pubdate = rdf_parsedate(ri->chardata);
289 if (!strcasecmp(el, "item")) {
290 --ri->item_tag_nesting;
294 if ( (!strcasecmp(el, "rss")) || (!strcasecmp(el, "rdf")) ) {
295 lprintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
296 ri->done_parsing = 1;
299 if (ri->chardata_len > 0) {
302 ri->chardata_len = 0;
309 * This callback stores up the data which appears in between tags.
311 void rss_xml_chardata(void *data, const XML_Char *s, int len) {
312 struct rss_item *ri = (struct rss_item *) data;
317 old_len = ri->chardata_len;
318 new_len = old_len + len;
319 new_buffer = realloc(ri->chardata, new_len + 1);
320 if (new_buffer != NULL) {
321 memcpy(&new_buffer[old_len], s, len);
322 new_buffer[new_len] = 0;
323 ri->chardata = new_buffer;
324 ri->chardata_len = new_len;
331 * Parse a URL into host, port number, and resource identifier.
333 int parse_url(char *url, char *hostname, int *port, char *identifier)
340 strcpy(scratch, url);
341 ptr = (char *)strchr(scratch, ':');
343 return(1); /* no protocol specified */
347 strcpy(protocol, scratch);
348 if (strcmp(protocol, "http")) {
349 return(2); /* not HTTP */
352 strcpy(scratch, url);
353 ptr = (char *) strstr(scratch, "//");
355 return(3); /* no server specified */
359 strcpy(hostname, ptr);
360 nptr = (char *)strchr(ptr, ':');
362 *port = 80; /* default */
363 nptr = (char *)strchr(hostname, '/');
366 sscanf(nptr, ":%d", port);
367 nptr = (char *)strchr(hostname, ':');
374 nptr = (char *)strchr(ptr, '/');
377 return(4); /* no url specified */
380 strcpy(identifier, nptr);
388 void rss_do_fetching(char *url, char *rooms) {
396 int got_bytes = (-1);
397 int redirect_count = 0;
400 if (parse_url(url, rsshost, &rssport, rssurl) != 0) {
401 lprintf(CTDL_ALERT, "Invalid URL: %s\n", url);
404 xp = XML_ParserCreateNS("UTF-8", ':');
406 lprintf(CTDL_ALERT, "Cannot create XML parser!\n");
410 memset(&ri, 0, sizeof(struct rss_item));
412 XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
413 XML_SetCharacterDataHandler(xp, rss_xml_chardata);
414 XML_SetUserData(xp, &ri);
416 retry: lprintf(CTDL_NOTICE, "Connecting to <%s>\n", rsshost);
417 sprintf(buf, "%d", rssport);
418 sock = sock_connect(rsshost, buf, "tcp");
420 lprintf(CTDL_DEBUG, "Connected!\n");
422 snprintf(buf, sizeof buf, "GET %s HTTP/1.0", rssurl);
423 lprintf(CTDL_DEBUG, "<%s\n", buf);
424 sock_puts(sock, buf);
426 snprintf(buf, sizeof buf, "Host: %s", rsshost);
427 lprintf(CTDL_DEBUG, "<%s\n", buf);
428 sock_puts(sock, buf);
430 snprintf(buf, sizeof buf, "User-Agent: %s", CITADEL);
431 lprintf(CTDL_DEBUG, "<%s\n", buf);
432 sock_puts(sock, buf);
434 snprintf(buf, sizeof buf, "Accept: */*");
435 lprintf(CTDL_DEBUG, "<%s\n", buf);
436 sock_puts(sock, buf);
440 if (sock_getln(sock, buf, sizeof buf) >= 0) {
441 lprintf(CTDL_DEBUG, ">%s\n", buf);
442 remove_token(buf, 0, ' ');
447 while (got_bytes = sock_getln(sock, buf, sizeof buf),
448 (got_bytes >= 0 && (strcmp(buf, "")) && (strcmp(buf, "\r"))) ) {
449 /* discard headers */
452 while (got_bytes = sock_read(sock, buf, sizeof buf, 0),
453 ((got_bytes>=0) && (ri.done_parsing == 0)) ) {
454 XML_Parse(xp, buf, got_bytes, 0);
456 if (ri.done_parsing == 0) XML_Parse(xp, "", 0, 1);
460 else if ( (!strncmp(buf, "30", 2)) && (redirect_count < 16) ) {
461 while (got_bytes = sock_getln(sock, buf, sizeof buf),
462 (got_bytes >= 0 && (strcmp(buf, "")) && (strcmp(buf, "\r"))) ) {
463 if (!strncasecmp(buf, "Location:", 9)) {
465 strcpy(buf, &buf[9]);
467 if (parse_url(buf, rsshost, &rssport, rssurl) == 0) {
471 lprintf(CTDL_ALERT, "Invalid URL: %s\n", buf);
481 lprintf(CTDL_ERR, "Could not connect: %s\n", strerror(errno));
486 /* Free the feed item data structure */
487 if (ri.guid != NULL) free(ri.guid);
489 if (ri.title != NULL) free(ri.title);
491 if (ri.link != NULL) free(ri.link);
493 if (ri.description != NULL) free(ri.description);
494 ri.description = NULL;
495 if (ri.chardata_len > 0) {
504 * Scan a room's netconfig to determine whether it is requesting any RSS feeds
506 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
508 char filename[PATH_MAX];
513 struct rssnetcfg *rncptr = NULL;
514 struct rssnetcfg *use_this_rncptr = NULL;
518 assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
520 /* Only do net processing for rooms that have netconfigs */
521 fp = fopen(filename, "r");
526 while (fgets(buf, sizeof buf, fp) != NULL) {
527 buf[strlen(buf)-1] = 0;
529 extract_token(instr, buf, 0, '|', sizeof instr);
530 if (!strcasecmp(instr, "rssclient")) {
532 use_this_rncptr = NULL;
534 extract_token(feedurl, buf, 1, '|', sizeof feedurl);
536 /* If any other rooms have requested the same feed, then we will just add this
537 * room to the target list for that client request.
539 for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
540 if (!strcmp(rncptr->url, feedurl)) {
541 use_this_rncptr = rncptr;
545 /* Otherwise create a new client request */
546 if (use_this_rncptr == NULL) {
547 rncptr = (struct rssnetcfg *) malloc(sizeof(struct rssnetcfg));
548 if (rncptr != NULL) {
549 rncptr->next = rnclist;
550 safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
551 rncptr->rooms = NULL;
553 use_this_rncptr = rncptr;
557 /* Add the room name to the request */
558 if (use_this_rncptr != NULL) {
559 if (use_this_rncptr->rooms == NULL) {
560 rncptr->rooms = strdup(qrbuf->QRname);
563 len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
564 ptr = realloc(use_this_rncptr->rooms, len);
567 strcat(ptr, qrbuf->QRname);
568 use_this_rncptr->rooms = ptr;
581 * Scan for rooms that have RSS client requests configured
583 void *rssclient_scan(void *args) {
584 static time_t last_run = 0L;
585 static int doing_rssclient = 0;
586 struct rssnetcfg *rptr = NULL;
588 CtdlThreadAllocTSD();
590 * Run RSS client no more frequently than once every n seconds
592 // if ( (time(NULL) - last_run) < config.c_net_freq ) {
597 * This is a simple concurrency check to make sure only one rssclient run
598 * is done at a time. We could do this with a mutex, but since we
599 * don't really require extremely fine granularity here, we'll do it
600 * with a static variable instead.
602 if (doing_rssclient) return NULL;
605 lprintf(CTDL_DEBUG, "rssclient started\n");
606 ForEachRoom(rssclient_scan_room, NULL);
608 while (rnclist != NULL) {
609 rss_do_fetching(rnclist->url, rnclist->rooms);
611 rnclist = rnclist->next;
612 if (rptr->rooms != NULL) free(rptr->rooms);
616 lprintf(CTDL_DEBUG, "rssclient ended\n");
617 last_run = time(NULL);
619 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
624 #endif /* HAVE_EXPAT */
626 CTDL_MODULE_INIT(rssclient)
631 // CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
632 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
634 lprintf(CTDL_INFO, "This server is missing the Expat XML parser. RSS client will be disabled.\n");
638 /* return our Subversion id for the Log */
639 return "$Id: serv_rssclient.c 5652 2007-10-29 20:14:48Z ajc $";