/*
- * $Id$
+ * Bring external RSS and/or Atom feeds into rooms. This module implements a
+ * very loose parser that scrapes both kinds of feeds and is not picky about
+ * the standards compliance of the source data.
*
- * Bring external RSS feeds into rooms.
+ * Copyright (c) 2007-2020 by the citadel.org team
*
+ * This program is open source software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 3.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
*/
#include <stdlib.h>
# include <time.h>
#else
# if HAVE_SYS_TIME_H
-# include <sys/time.h>
+#include <sys/time.h>
# else
-# include <time.h>
+#include <time.h>
# endif
#endif
#include "support.h"
#include "config.h"
#include "threads.h"
-#include "room_ops.h"
#include "ctdl_module.h"
-#include "clientsocket.h"
#include "msgbase.h"
#include "parsedate.h"
#include "database.h"
#include "citadel_dirs.h"
-#include "md5.h"
+#include "context.h"
+#include "internet_addressing.h"
+struct rssroom {
+ struct rssroom *next;
+ char *room;
+};
-struct rssnetcfg {
- struct rssnetcfg *next;
- char url[256];
- char *rooms;
+struct rssurl {
+ struct rssurl *next;
+ char *url;
+ struct rssroom *rooms;
};
-struct rss_item {
- char *chardata;
- int chardata_len;
- char *roomlist;
- int done_parsing;
- char *guid;
- char *title;
+struct rssparser {
+ StrBuf *CData;
+ struct CtdlMessage *msg;
char *link;
char *description;
- time_t pubdate;
- char channel_title[256];
- int item_tag_nesting;
- char *author_or_creator;
+ char *item_id;
+ struct rssroom *rooms;
};
-struct rssnetcfg *rnclist = NULL;
+time_t last_run = 0L;
+struct rssurl *rsstodo = NULL;
-/*
- * Commit a fetched and parsed RSS item to disk
- */
-void rss_save_item(struct rss_item *ri) {
-
- struct MD5Context md5context;
- u_char rawdigest[MD5_DIGEST_LEN];
+// This handler is called whenever an XML tag opens.
+//
+void rss_start_element(void *data, const char *el, const char **attribute)
+{
+ struct rssparser *r = (struct rssparser *)data;
int i;
- char utmsgid[SIZ];
- struct cdbdata *cdbut;
- struct UseTable ut;
- struct CtdlMessage *msg;
- struct recptypes *recp = NULL;
- int msglen = 0;
-
- recp = (struct recptypes *) malloc(sizeof(struct recptypes));
- if (recp == NULL) return;
- memset(recp, 0, sizeof(struct recptypes));
- recp->recp_room = strdup(ri->roomlist);
- recp->num_room = num_tokens(ri->roomlist, '|');
- recp->recptypes_magic = RECPTYPES_MAGIC;
-
- /* Construct a GUID to use in the S_USETABLE table.
- * If one is not present in the item itself, make one up.
- */
- if (ri->guid != NULL) {
- snprintf(utmsgid, sizeof utmsgid, "rss/%s", ri->guid);
- }
- else {
- MD5Init(&md5context);
- if (ri->title != NULL) {
- MD5Update(&md5context, (unsigned char*)ri->title, strlen(ri->title));
- }
- if (ri->link != NULL) {
- MD5Update(&md5context, (unsigned char*)ri->link, strlen(ri->link));
- }
- MD5Final(rawdigest, &md5context);
- for (i=0; i<MD5_DIGEST_LEN; i++) {
- sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
- utmsgid[i*2] = tolower(utmsgid[i*2]);
- utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
- }
- strcat(utmsgid, "_rss2ctdl");
- }
- /* Find out if we've already seen this item */
- cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
- if (cdbut != NULL) {
- /* Item has already been seen */
- CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
- cdb_free(cdbut);
-
- /* rewrite the record anyway, to update the timestamp */
- strcpy(ut.ut_msgid, utmsgid);
- ut.ut_timestamp = time(NULL);
- cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
- }
- else {
- /* Item has not been seen, so save it. */
-
- if (ri->description == NULL) ri->description = strdup("");
- for (i=strlen(ri->description); i>=0; --i) {
- if (isspace(ri->description[i])) {
- ri->description[i] = ' ';
- }
- }
-
- msg = malloc(sizeof(struct CtdlMessage));
- memset(msg, 0, sizeof(struct CtdlMessage));
- msg->cm_magic = CTDLMESSAGE_MAGIC;
- msg->cm_anon_type = MES_NORMAL;
- msg->cm_format_type = FMT_RFC822;
-
- if (ri->author_or_creator != NULL) {
- msg->cm_fields['A'] = html_to_ascii(ri->author_or_creator,
- strlen(ri->author_or_creator), 512, 0);
- striplt(msg->cm_fields['A']);
- }
- else {
- msg->cm_fields['A'] = strdup("rss");
+ if (
+ (!strcasecmp(el, "entry"))
+ || (!strcasecmp(el, "item"))
+ ) {
+ // this is the start of a new item(rss) or entry(atom)
+ if (r->msg != NULL) {
+ CM_Free(r->msg);
+ r->msg = NULL;
}
+ r->msg = malloc(sizeof(struct CtdlMessage));
+ memset(r->msg, 0, sizeof(struct CtdlMessage));
+ r->msg->cm_magic = CTDLMESSAGE_MAGIC;
+ r->msg->cm_anon_type = MES_NORMAL;
+ r->msg->cm_format_type = FMT_RFC822;
+ }
- msg->cm_fields['N'] = strdup(NODENAME);
- if (ri->title != NULL) {
- msg->cm_fields['U'] = html_to_ascii(ri->title, strlen(ri->title), 512, 0);
- striplt(msg->cm_fields['U']);
- }
- msg->cm_fields['T'] = malloc(64);
- snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
- if (ri->channel_title != NULL) {
- if (!IsEmptyStr(ri->channel_title)) {
- msg->cm_fields['O'] = strdup(ri->channel_title);
+ else if (!strcasecmp(el, "link")) { // atom feeds have the link as an attribute
+ for(i = 0; attribute[i]; i += 2) {
+ if (!strcasecmp(attribute[i], "href")) {
+ if (r->link != NULL) {
+ free(r->link);
+ r->link = NULL;
+ }
+ r->link = strdup(attribute[i+1]);
+ striplt(r->link);
}
}
-
- msglen = 1024 + strlen(ri->link) + strlen(ri->description) ;
- msg->cm_fields['M'] = malloc(msglen);
- snprintf(msg->cm_fields['M'], msglen,
- "Content-type: text/html\r\n\r\n"
- "<html><body>\n"
- "%s<br><br>\n"
- "<a href=\"%s\">%s</a>\n"
- "</body></html>\n"
- ,
- ri->description,
- ri->link, ri->link
- );
-
- CtdlSubmitMsg(msg, recp, NULL, 0);
- CtdlFreeMessage(msg);
-
- /* write the uidl to the use table so we don't store this item again */
- strcpy(ut.ut_msgid, utmsgid);
- ut.ut_timestamp = time(NULL);
- cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
}
- free_recipients(recp);
}
-
-/*
- * Convert an RDF/RSS datestamp into a time_t
- */
-time_t rdf_parsedate(char *p)
+// This handler is called whenever an XML tag closes.
+//
+void rss_end_element(void *data, const char *el)
{
- struct tm tm;
- time_t t = 0;
-
- if (!p) return 0L;
- if (strlen(p) < 10) return 0L;
-
- memset(&tm, 0, sizeof tm);
-
- /* YYYY-MM-DDTHH:MM format...
- */
- if ( (p[4] == '-') && (p[7] == '-') ) {
- tm.tm_year = atoi(&p[0]) - 1900;
- tm.tm_mon = atoi(&p[5]) - 1;
- tm.tm_mday = atoi(&p[8]);
- if ( (p[10] == 'T') && (p[13] == ':') ) {
- tm.tm_hour = atoi(&p[11]);
- tm.tm_min = atoi(&p[14]);
- }
- return mktime(&tm);
- }
-
- /* hmm... try RFC822 date stamp format */
+ struct rssparser *r = (struct rssparser *)data;
+ StrBuf *encoded_field;
- t = parsedate(p);
- if (t > 0) return(t);
-
- /* yeesh. ok, just return the current date and time. */
- return(time(NULL));
-}
+ if (StrLength(r->CData) > 0) { // strip leading/trailing whitespace from field
+ StrBufTrim(r->CData);
+ }
+ if ( // end of a new item(rss) or entry(atom)
+ (!strcasecmp(el, "entry"))
+ || (!strcasecmp(el, "item"))
+ ) {
+ if (r->msg != NULL) { // Save the message to the rooms
+ // use the link as an item id if nothing else is available
+ if ((r->item_id == NULL) && (r->link != NULL)) {
+ r->item_id = strdup(r->link);
+ }
-void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
- struct rss_item *ri = (struct rss_item *) data;
- char el[256];
- char *sep = NULL;
+ // check the use table
+ StrBuf *u = NewStrBuf();
+ StrBufAppendPrintf(u, "rss/%s", r->item_id);
+ int already_seen = CheckIfAlreadySeen(u);
+ FreeStrBuf(&u);
+
+ if (already_seen == 0) {
+
+ // Compose the message text
+ StrBuf *TheMessage = NewStrBuf();
+ StrBufAppendPrintf(TheMessage,
+ "Content-type: text/html\n\n"
+ "\n\n"
+ "<html><head></head><body>"
+ );
+
+ if (r->description != NULL) {
+ StrBufAppendPrintf(TheMessage, "%s<br><br>\r\n", r->description);
+ free(r->description);
+ r->description = NULL;
+ }
+
+ if (r->link != NULL) {
+ StrBufAppendPrintf(TheMessage, "<a href=\"%s\">%s</a>\r\n", r->link, r->link);
+ free(r->link);
+ r->link = NULL;
+ }
+
+ StrBufAppendPrintf(TheMessage, "</body></html>\r\n");
+ CM_SetField(r->msg, eMesageText, ChrPtr(TheMessage), StrLength(TheMessage));
+ FreeStrBuf(&TheMessage);
+
+ if (CM_IsEmpty(r->msg, eAuthor)) {
+ CM_SetField(r->msg, eAuthor, HKEY("rss"));
+ }
+
+ if (CM_IsEmpty(r->msg, eTimestamp)) {
+ CM_SetFieldLONG(r->msg, eTimestamp, time(NULL));
+ }
+
+ // Save it to the room(s)
+ struct rssroom *rr = NULL;
+ long msgnum = (-1);
+ for (rr=r->rooms; rr!=NULL; rr=rr->next) {
+ if (rr == r->rooms) {
+ msgnum = CtdlSubmitMsg(r->msg, NULL, rr->room); // in first room, save msg
+ }
+ else {
+ CtdlSaveMsgPointerInRoom(rr->room, msgnum, 0, NULL); // elsewhere, save a pointer
+ }
+ syslog(LOG_DEBUG, "rssclient: saved message %ld to %s", msgnum, rr->room);
+ }
+ }
+ else {
+ syslog(LOG_DEBUG, "rssclient: already seen %s", r->item_id);
+ }
+
+ CM_Free(r->msg);
+ r->msg = NULL;
+ }
- /* Axe the namespace, we don't care about it */
- safestrncpy(el, supplied_el, sizeof el);
- while (sep = strchr(el, ':'), sep) {
- strcpy(el, ++sep);
+ if (r->item_id != NULL) {
+ free(r->item_id);
+ r->item_id = NULL;
+ }
}
- if (!strcasecmp(el, "item")) {
- ++ri->item_tag_nesting;
-
- /* Initialize the feed item data structure */
- if (ri->guid != NULL) free(ri->guid);
- ri->guid = NULL;
- if (ri->title != NULL) free(ri->title);
- ri->title = NULL;
- if (ri->link != NULL) free(ri->link);
- ri->link = NULL;
- if (ri->author_or_creator != NULL) free(ri->author_or_creator);
- ri->author_or_creator = NULL;
- if (ri->description != NULL) free(ri->description);
- ri->description = NULL;
-
- /* Throw away any existing character data */
- if (ri->chardata_len > 0) {
- free(ri->chardata);
- ri->chardata = 0;
- ri->chardata_len = 0;
+ else if (!strcasecmp(el, "title")) { // item subject (rss and atom)
+ if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eMsgSubject))) {
+ encoded_field = NewStrBuf();
+ StrBufRFC2047encode(&encoded_field, r->CData);
+ CM_SetAsFieldSB(r->msg, eMsgSubject, &encoded_field);
}
}
-
-
-}
-
-void rss_xml_end(void *data, const char *supplied_el) {
- struct rss_item *ri = (struct rss_item *) data;
- char el[256];
- char *sep = NULL;
-
- /* Axe the namespace, we don't care about it */
- safestrncpy(el, supplied_el, sizeof el);
- while (sep = strchr(el, ':'), sep) {
- strcpy(el, ++sep);
+ else if (!strcasecmp(el, "creator")) { // <creator> can be used if <author> is not present
+ if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eAuthor))) {
+ encoded_field = NewStrBuf();
+ StrBufRFC2047encode(&encoded_field, r->CData);
+ CM_SetAsFieldSB(r->msg, eAuthor, &encoded_field);
+ }
}
- if ( (!strcasecmp(el, "title")) && (ri->item_tag_nesting == 0) && (ri->chardata != NULL) ) {
- safestrncpy(ri->channel_title, ri->chardata, sizeof ri->channel_title);
- striplt(ri->channel_title);
+ else if (!strcasecmp(el, "author")) { // <author> supercedes <creator> if both are present
+ if (r->msg != NULL) {
+ encoded_field = NewStrBuf();
+ StrBufRFC2047encode(&encoded_field, r->CData);
+ CM_SetAsFieldSB(r->msg, eAuthor, &encoded_field);
+ }
}
- if ( (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) {
- if (ri->guid != NULL) free(ri->guid);
- striplt(ri->chardata);
- ri->guid = strdup(ri->chardata);
+ else if (!strcasecmp(el, "pubdate")) { // date/time stamp (rss) Sat, 25 Feb 2017 14:28:01 EST
+ if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
+ CM_SetFieldLONG(r->msg, eTimestamp, parsedate(ChrPtr(r->CData)));
+ }
}
- if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) {
- if (ri->title != NULL) free(ri->title);
- striplt(ri->chardata);
- ri->title = strdup(ri->chardata);
+ else if (!strcasecmp(el, "updated")) { // date/time stamp (atom) 2003-12-13T18:30:02Z
+ if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
+ struct tm t;
+ char zulu;
+ memset(&t, 0, sizeof t);
+ sscanf(ChrPtr(r->CData), "%d-%d-%dT%d:%d:%d%c", &t.tm_year, &t.tm_mon, &t.tm_mday, &t.tm_hour, &t.tm_min, &t.tm_sec, &zulu);
+ t.tm_year -= 1900;
+ t.tm_mon -= 1;
+ CM_SetFieldLONG(r->msg, eTimestamp, mktime(&t));
+ }
}
- if ( (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) {
- if (ri->link != NULL) free(ri->link);
- striplt(ri->chardata);
- ri->link = strdup(ri->chardata);
+ else if (!strcasecmp(el, "link")) { // link to story (rss)
+ if (r->link != NULL) {
+ free(r->link);
+ r->link = NULL;
+ }
+ r->link = strdup(ChrPtr(r->CData));
}
- if ( (!strcasecmp(el, "description")) && (ri->chardata != NULL) ) {
- if (ri->description != NULL) free(ri->description);
- ri->description = strdup(ri->chardata);
+ else if (
+ (!strcasecmp(el, "guid")) // unique item id (rss)
+ || (!strcasecmp(el, "id")) // unique item id (atom)
+ ) {
+ if (r->item_id != NULL) {
+ free(r->item_id);
+ r->item_id = NULL;
+ }
+ r->item_id = strdup(ChrPtr(r->CData));
}
- if ( ((!strcasecmp(el, "pubdate")) || (!strcasecmp(el, "date"))) && (ri->chardata != NULL) ) {
- striplt(ri->chardata);
- ri->pubdate = rdf_parsedate(ri->chardata);
+ else if (
+ (!strcasecmp(el, "description")) // message text (rss)
+ || (!strcasecmp(el, "summary")) // message text (atom)
+ || (!strcasecmp(el, "content")) // message text (atom)
+ ) {
+ if (r->description != NULL) {
+ free(r->description);
+ r->description = NULL;
+ }
+ r->description = strdup(ChrPtr(r->CData));
}
- if ( ((!strcasecmp(el, "author")) || (!strcasecmp(el, "creator"))) && (ri->chardata != NULL) ) {
- if (ri->author_or_creator != NULL) free(ri->author_or_creator);
- striplt(ri->chardata);
- ri->author_or_creator = strdup(ri->chardata);
+ if (r->CData != NULL) {
+ FreeStrBuf(&r->CData);
+ r->CData = NULL;
}
+}
- if (!strcasecmp(el, "item")) {
- --ri->item_tag_nesting;
- rss_save_item(ri);
- }
- if ( (!strcasecmp(el, "rss")) || (!strcasecmp(el, "rdf")) ) {
- CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
- ri->done_parsing = 1;
- }
+// This handler is called whenever data appears between opening and closing tags.
+//
+void rss_handle_data(void *data, const char *content, int length)
+{
+ struct rssparser *r = (struct rssparser *)data;
- if (ri->chardata_len > 0) {
- free(ri->chardata);
- ri->chardata = 0;
- ri->chardata_len = 0;
+ if (r->CData == NULL) {
+ r->CData = NewStrBuf();
}
+ StrBufAppendBufPlain(r->CData, content, length, 0);
}
-/*
- * This callback stores up the data which appears in between tags.
- */
-void rss_xml_chardata(void *data, const XML_Char *s, int len) {
- struct rss_item *ri = (struct rss_item *) data;
- int old_len;
- int new_len;
- char *new_buffer;
-
- old_len = ri->chardata_len;
- new_len = old_len + len;
- new_buffer = realloc(ri->chardata, new_len + 1);
- if (new_buffer != NULL) {
- memcpy(&new_buffer[old_len], s, len);
- new_buffer[new_len] = 0;
- ri->chardata = new_buffer;
- ri->chardata_len = new_len;
- }
+// Feed has been downloaded, now parse it.
+//
+void rss_parse_feed(StrBuf *Feed, struct rssroom *rooms)
+{
+ struct rssparser r;
+
+ memset(&r, 0, sizeof r);
+ r.rooms = rooms;
+ XML_Parser p = XML_ParserCreate("UTF-8");
+ XML_SetElementHandler(p, rss_start_element, rss_end_element);
+ XML_SetCharacterDataHandler(p, rss_handle_data);
+ XML_SetUserData(p, (void *)&r);
+ XML_Parse(p, ChrPtr(Feed), StrLength(Feed), XML_TRUE);
+ XML_ParserFree(p);
}
-
-/*
- * Callback function for passing libcurl's output to expat for parsing
- */
-size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
+// Add a feed/room pair into the todo list
+//
+void rssclient_push_todo(char *rssurl, char *roomname)
{
- XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
- return (size*nmemb);
-}
+ struct rssurl *r = NULL;
+ struct rssurl *thisone = NULL;
+ struct rssroom *newroom = NULL;
+
+ syslog(LOG_DEBUG, "rssclient: will fetch %s to %s", rssurl, roomname);
+
+ for (r=rsstodo; r!=NULL; r=r->next) {
+ if (!strcasecmp(r->url, rssurl)) {
+ thisone = r;
+ }
+ }
+ if (thisone == NULL) {
+ thisone = malloc(sizeof(struct rssurl));
+ thisone->url = strdup(rssurl);
+ thisone->rooms = NULL;
+ thisone->next = rsstodo;
+ rsstodo = thisone;
+ }
+ newroom = malloc(sizeof(struct rssroom));
+ newroom->room = strdup(roomname);
+ newroom->next = thisone->rooms;
+ thisone->rooms = newroom;
+}
-/*
- * Begin a feed parse
- */
-void rss_do_fetching(char *url, char *rooms) {
- struct rss_item ri;
- XML_Parser xp;
+// pull one feed (possibly multiple rooms)
+//
+void rss_pull_one_feed(struct rssurl *url)
+{
CURL *curl;
CURLcode res;
- char errmsg[1024] = "";
+ StrBuf *Downloaded = NULL;
- CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", url);
+ syslog(LOG_DEBUG, "rssclient: fetching %s", url->url);
curl = curl_easy_init();
if (!curl) {
- CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
- return;
- }
-
- xp = XML_ParserCreateNS("UTF-8", ':');
- if (!xp) {
- CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
- curl_easy_cleanup(curl);
- return;
- }
-
- curl_easy_setopt(curl, CURLOPT_URL, url);
- curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
- curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
- curl_easy_setopt(curl, CURLOPT_WRITEDATA, xp);
- curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
- curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
- curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
- curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
- curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180); /* die after 180 seconds */
- if (!IsEmptyStr(config.c_ip_addr)) {
- curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
- }
-
- memset(&ri, 0, sizeof(struct rss_item));
- ri.roomlist = rooms;
- XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
- XML_SetCharacterDataHandler(xp, rss_xml_chardata);
- XML_SetUserData(xp, &ri);
-
- if (CtdlThreadCheckStop())
- {
- XML_ParserFree(xp);
- curl_easy_cleanup(curl);
return;
}
-
- if (CtdlThreadCheckStop())
- goto shutdown ;
- res = curl_easy_perform(curl);
- if (res) {
- CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
+ Downloaded = NewStrBuf();
+
+ curl_easy_setopt(curl, CURLOPT_URL, url->url);
+ curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0L);
+ curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0L);
+ curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1L); // Follow redirects
+ curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback); // What to do with downloaded data
+ curl_easy_setopt(curl, CURLOPT_WRITEDATA, Downloaded); // Give it our StrBuf to work with
+ curl_easy_setopt(curl, CURLOPT_TIMEOUT, 20L); // Time out after 20 seconds
+ res = curl_easy_perform(curl); // Perform the request
+ if (res != CURLE_OK) {
+ syslog(LOG_WARNING, "rssclient: failed to load feed: %s", curl_easy_strerror(res));
}
+ curl_easy_cleanup(curl);
- if (CtdlThreadCheckStop())
- goto shutdown ;
-
- if (ri.done_parsing == 0) XML_Parse(xp, "", 0, 1);
+ rss_parse_feed(Downloaded, url->rooms); // parse the feed
+ FreeStrBuf(&Downloaded); // free the downloaded feed data
+}
-shutdown:
- curl_easy_cleanup(curl);
- XML_ParserFree(xp);
-
- /* Free the feed item data structure */
- if (ri.guid != NULL) free(ri.guid);
- ri.guid = NULL;
- if (ri.title != NULL) free(ri.title);
- ri.title = NULL;
- if (ri.link != NULL) free(ri.link);
- ri.link = NULL;
- if (ri.author_or_creator != NULL) free(ri.author_or_creator);
- ri.author_or_creator = NULL;
- if (ri.description != NULL) free(ri.description);
- ri.description = NULL;
- if (ri.chardata_len > 0) {
- free(ri.chardata);
- ri.chardata = 0;
- ri.chardata_len = 0;
+// We have a list, now download the feeds
+//
+void rss_pull_feeds(void)
+{
+ struct rssurl *r;
+ struct rssroom *rr;
+
+ while (rsstodo != NULL) {
+ rss_pull_one_feed(rsstodo);
+ r = rsstodo;
+ rsstodo = rsstodo->next;
+ while (r->rooms != NULL) {
+ rr = r->rooms;
+ r->rooms = r->rooms->next;
+ free(rr->room);
+ free(rr);
+ }
+ free(r->url);
+ free(r);
}
}
-/*
- * Scan a room's netconfig to determine whether it is requesting any RSS feeds
- */
+// Scan a room's netconfig looking for RSS feed parsing requests
+//
void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
{
- char filename[PATH_MAX];
- char buf[1024];
- char instr[32];
- FILE *fp;
- char feedurl[256];
- struct rssnetcfg *rncptr = NULL;
- struct rssnetcfg *use_this_rncptr = NULL;
- int len = 0;
- char *ptr = NULL;
-
- assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
-
- if (CtdlThreadCheckStop())
- return;
-
- /* Only do net processing for rooms that have netconfigs */
- fp = fopen(filename, "r");
- if (fp == NULL) {
+ char *serialized_config = NULL;
+ int num_configs = 0;
+ char cfgline[SIZ];
+ int i = 0;
+
+ serialized_config = LoadRoomNetConfigFile(qrbuf->QRnumber);
+ if (!serialized_config) {
return;
}
- while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
- buf[strlen(buf)-1] = 0;
-
- extract_token(instr, buf, 0, '|', sizeof instr);
- if (!strcasecmp(instr, "rssclient")) {
-
- use_this_rncptr = NULL;
-
- extract_token(feedurl, buf, 1, '|', sizeof feedurl);
-
- /* If any other rooms have requested the same feed, then we will just add this
- * room to the target list for that client request.
- */
- for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
- if (!strcmp(rncptr->url, feedurl)) {
- use_this_rncptr = rncptr;
- }
- }
-
- /* Otherwise create a new client request */
- if (use_this_rncptr == NULL) {
- rncptr = (struct rssnetcfg *) malloc(sizeof(struct rssnetcfg));
- if (rncptr != NULL) {
- rncptr->next = rnclist;
- safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
- rncptr->rooms = NULL;
- rnclist = rncptr;
- use_this_rncptr = rncptr;
- }
- }
-
- /* Add the room name to the request */
- if (use_this_rncptr != NULL) {
- if (use_this_rncptr->rooms == NULL) {
- rncptr->rooms = strdup(qrbuf->QRname);
- }
- else {
- len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
- ptr = realloc(use_this_rncptr->rooms, len);
- if (ptr != NULL) {
- strcat(ptr, "|");
- strcat(ptr, qrbuf->QRname);
- use_this_rncptr->rooms = ptr;
- }
- }
+ num_configs = num_tokens(serialized_config, '\n');
+ for (i=0; i<num_configs; ++i) {
+ extract_token(cfgline, serialized_config, i, '\n', sizeof cfgline);
+ if (!strncasecmp(cfgline, HKEY("rssclient|"))) {
+ strcpy(cfgline, &cfgline[10]);
+ char *vbar = strchr(cfgline, '|');
+ if (vbar != NULL) {
+ *vbar = 0;
}
+ rssclient_push_todo(cfgline, qrbuf->QRname);
}
-
}
- fclose(fp);
-
+ free(serialized_config);
}
+
/*
* Scan for rooms that have RSS client requests configured
*/
-void *rssclient_scan(void *args) {
- static time_t last_run = 0L;
- static int doing_rssclient = 0;
- struct rssnetcfg *rptr = NULL;
- struct CitContext rssclientCC;
-
- /* Give this thread its own private CitContext */
- CtdlFillSystemContext(&rssclientCC, "rssclient");
- citthread_setspecific(MyConKey, (void *)&rssclientCC );
-
- CtdlThreadAllocTSD();
-
- /*
- * This is a simple concurrency check to make sure only one rssclient run
- * is done at a time. We could do this with a mutex, but since we
- * don't really require extremely fine granularity here, we'll do it
- * with a static variable instead.
- */
- if (doing_rssclient) return NULL;
- doing_rssclient = 1;
-
- CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
- ForEachRoom(rssclient_scan_room, NULL);
-
- while (rnclist != NULL && !CtdlThreadCheckStop()) {
- rss_do_fetching(rnclist->url, rnclist->rooms);
- rptr = rnclist;
- rnclist = rnclist->next;
- if (rptr->rooms != NULL) free(rptr->rooms);
- free(rptr);
+void rssclient_scan(void) {
+ time_t now = time(NULL);
+
+ /* Run no more than once every 15 minutes. */
+ if ((now - last_run) < 900) {
+ syslog(LOG_DEBUG,
+ "rssclient: polling interval not yet reached; last run was %ldm%lds ago",
+ ((now - last_run) / 60),
+ ((now - last_run) % 60)
+ );
+ return;
}
- CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
+ syslog(LOG_DEBUG, "rssclient: started");
+ CtdlForEachRoom(rssclient_scan_room, NULL);
+ rss_pull_feeds();
+ syslog(LOG_DEBUG, "rssclient: ended");
last_run = time(NULL);
- doing_rssclient = 0;
- if (!CtdlThreadCheckStop())
- CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
- else
- CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
- return NULL;
+ return;
}
CTDL_MODULE_INIT(rssclient)
{
- if (threading)
+ if (!threading)
{
- CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
- CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
+ syslog(LOG_INFO, "rssclient: using %s", curl_version());
+ CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);
}
- /* return our Subversion id for the Log */
- return "$Id$";
+ return "rssclient";
}
+
projects / citadel.git / blobdiff