#include <errno.h>
#include <sys/types.h>
#include <sys/stat.h>
+#include <expat.h>
+#include <curl/curl.h>
#include <libcitadel.h>
#include "citadel.h"
#include "server.h"
#include "citserver.h"
#include "support.h"
#include "config.h"
+#include "threads.h"
#include "room_ops.h"
#include "ctdl_module.h"
#include "clientsocket.h"
#include "citadel_dirs.h"
#include "md5.h"
-#ifdef HAVE_EXPAT
-#include <expat.h>
-
struct rssnetcfg {
struct rssnetcfg *next;
else {
MD5Init(&md5context);
if (ri->title != NULL) {
- MD5Update(&md5context, ri->title, strlen(ri->title));
+ MD5Update(&md5context, (unsigned char*)ri->title, strlen(ri->title));
}
if (ri->link != NULL) {
- MD5Update(&md5context, ri->link, strlen(ri->link));
+ MD5Update(&md5context, (unsigned char*)ri->link, strlen(ri->link));
}
MD5Final(rawdigest, &md5context);
for (i=0; i<MD5_DIGEST_LEN; i++) {
cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
if (cdbut != NULL) {
/* Item has already been seen */
- lprintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
+ CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
cdb_free(cdbut);
/* rewrite the record anyway, to update the timestamp */
ri->link, ri->link
);
- CtdlSubmitMsg(msg, recp, NULL);
+ CtdlSubmitMsg(msg, recp, NULL, 0);
CtdlFreeMessage(msg);
/* write the uidl to the use table so we don't store this item again */
}
if ( (!strcasecmp(el, "rss")) || (!strcasecmp(el, "rdf")) ) {
- lprintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
+ CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
ri->done_parsing = 1;
}
-/*
- * Parse a URL into host, port number, and resource identifier.
+/*
+ * Callback function for passing libcurl's output to expat for parsing
*/
-int parse_url(char *url, char *hostname, int *port, char *identifier)
+size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
{
- char protocol[1024];
- char scratch[1024];
- char *ptr = NULL;
- char *nptr = NULL;
-
- strcpy(scratch, url);
- ptr = (char *)strchr(scratch, ':');
- if (!ptr) {
- return(1); /* no protocol specified */
- }
-
- strcpy(ptr, "");
- strcpy(protocol, scratch);
- if (strcmp(protocol, "http")) {
- return(2); /* not HTTP */
- }
-
- strcpy(scratch, url);
- ptr = (char *) strstr(scratch, "//");
- if (!ptr) {
- return(3); /* no server specified */
- }
- ptr += 2;
-
- strcpy(hostname, ptr);
- nptr = (char *)strchr(ptr, ':');
- if (!nptr) {
- *port = 80; /* default */
- nptr = (char *)strchr(hostname, '/');
- }
- else {
- sscanf(nptr, ":%d", port);
- nptr = (char *)strchr(hostname, ':');
- }
-
- if (nptr) {
- *nptr = '\0';
- }
-
- nptr = (char *)strchr(ptr, '/');
-
- if (!nptr) {
- return(4); /* no url specified */
- }
-
- strcpy(identifier, nptr);
- return(0);
+ XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
+ return (size*nmemb);
}
+
/*
* Begin a feed parse
*/
void rss_do_fetching(char *url, char *rooms) {
- char buf[1024];
- char rsshost[1024];
- int rssport = 80;
- char rssurl[1024];
struct rss_item ri;
XML_Parser xp;
- int sock = (-1);
- int got_bytes = (-1);
- int redirect_count = 0;
- /* Parse the URL */
- if (parse_url(url, rsshost, &rssport, rssurl) != 0) {
- lprintf(CTDL_ALERT, "Invalid URL: %s\n", url);
+ CURL *curl;
+ CURLcode res;
+ char errmsg[1024] = "";
+
+ CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", url);
+
+ curl = curl_easy_init();
+ if (!curl) {
+ CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
+ return;
}
xp = XML_ParserCreateNS("UTF-8", ':');
if (!xp) {
- lprintf(CTDL_ALERT, "Cannot create XML parser!\n");
+ 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);
+ 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);
-retry: lprintf(CTDL_NOTICE, "Connecting to <%s>\n", rsshost);
- sprintf(buf, "%d", rssport);
- sock = sock_connect(rsshost, buf, "tcp");
- if (sock >= 0) {
- lprintf(CTDL_DEBUG, "Connected!\n");
-
- snprintf(buf, sizeof buf, "GET %s HTTP/1.0", rssurl);
- lprintf(CTDL_DEBUG, "<%s\n", buf);
- sock_puts(sock, buf);
-
- snprintf(buf, sizeof buf, "Host: %s", rsshost);
- lprintf(CTDL_DEBUG, "<%s\n", buf);
- sock_puts(sock, buf);
-
- snprintf(buf, sizeof buf, "User-Agent: %s", CITADEL);
- lprintf(CTDL_DEBUG, "<%s\n", buf);
- sock_puts(sock, buf);
-
- snprintf(buf, sizeof buf, "Accept: */*");
- lprintf(CTDL_DEBUG, "<%s\n", buf);
- sock_puts(sock, buf);
-
- sock_puts(sock, "");
+ if (CtdlThreadCheckStop())
+ {
+ XML_ParserFree(xp);
+ curl_easy_cleanup(curl);
+ return;
+ }
+
+ if (CtdlThreadCheckStop())
+ goto shutdown ;
- if (sock_getln(sock, buf, sizeof buf) >= 0) {
- lprintf(CTDL_DEBUG, ">%s\n", buf);
- remove_token(buf, 0, ' ');
+ res = curl_easy_perform(curl);
+ if (res) {
+ CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
+ }
- /* 200 OK */
- if (buf[0] == '2') {
+ if (CtdlThreadCheckStop())
+ goto shutdown ;
- while (got_bytes = sock_getln(sock, buf, sizeof buf),
- (got_bytes >= 0 && (strcmp(buf, "")) && (strcmp(buf, "\r"))) ) {
- /* discard headers */
- }
+ if (ri.done_parsing == 0) XML_Parse(xp, "", 0, 1);
- while (got_bytes = sock_read(sock, buf, sizeof buf, 0),
- ((got_bytes>=0) && (ri.done_parsing == 0)) ) {
- XML_Parse(xp, buf, got_bytes, 0);
- }
- if (ri.done_parsing == 0) XML_Parse(xp, "", 0, 1);
- }
-
- /* 30X redirect */
- else if ( (!strncmp(buf, "30", 2)) && (redirect_count < 16) ) {
- while (got_bytes = sock_getln(sock, buf, sizeof buf),
- (got_bytes >= 0 && (strcmp(buf, "")) && (strcmp(buf, "\r"))) ) {
- if (!strncasecmp(buf, "Location:", 9)) {
- ++redirect_count;
- strcpy(buf, &buf[9]);
- striplt(buf);
- if (parse_url(buf, rsshost, &rssport, rssurl) == 0) {
- goto retry;
- }
- else {
- lprintf(CTDL_ALERT, "Invalid URL: %s\n", buf);
- }
- }
- }
- }
-
- }
- sock_close(sock);
- }
- else {
- lprintf(CTDL_ERR, "Could not connect: %s\n", strerror(errno));
- }
+shutdown:
+ curl_easy_cleanup(curl);
XML_ParserFree(xp);
/* Free the feed item data structure */
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) {
return;
}
- while (fgets(buf, sizeof buf, fp) != NULL) {
+ while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
buf[strlen(buf)-1] = 0;
extract_token(instr, buf, 0, '|', sizeof instr);
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 */
+ CtdlFillPrivateContext(&rssclientCC, "rssclient");
+ citthread_setspecific(MyConKey, (void *)&rssclientCC );
CtdlThreadAllocTSD();
- /*
- * Run RSS client no more frequently than once every n seconds
- */
-// if ( (time(NULL) - last_run) < config.c_net_freq ) {
-// return;
-// }
/*
* This is a simple concurrency check to make sure only one rssclient run
if (doing_rssclient) return NULL;
doing_rssclient = 1;
- lprintf(CTDL_DEBUG, "rssclient started\n");
+ CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
ForEachRoom(rssclient_scan_room, NULL);
- while (rnclist != NULL) {
+ while (rnclist != NULL && !CtdlThreadCheckStop()) {
rss_do_fetching(rnclist->url, rnclist->rooms);
rptr = rnclist;
rnclist = rnclist->next;
free(rptr);
}
- lprintf(CTDL_DEBUG, "rssclient ended\n");
+ CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
last_run = time(NULL);
doing_rssclient = 0;
- CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
+ 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;
}
-#endif /* HAVE_EXPAT */
-
CTDL_MODULE_INIT(rssclient)
{
if (threading)
{
-#ifdef HAVE_EXPAT
-// CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
+ CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
-#else
- lprintf(CTDL_INFO, "This server is missing the Expat XML parser. RSS client will be disabled.\n");
-#endif
}
-
/* return our Subversion id for the Log */
return "$Id: serv_rssclient.c 5652 2007-10-29 20:14:48Z ajc $";
}