#include "md5.h"
#include "context.h"
+
typedef struct rssnetcfg rssnetcfg;
struct rssnetcfg {
rssnetcfg *next;
char *rooms;
time_t last_error_when;
int ItemType;
+ time_t next_poll;
};
#define RSS_UNSET (1<<0)
} rsscollection;
struct rssnetcfg *rnclist = NULL;
-HashList *StartHandlers;
-HashList *EndHandlers;
-HashList *KnownNameSpaces;
+HashList *StartHandlers = NULL;
+HashList *EndHandlers = NULL;
+HashList *KnownNameSpaces = NULL;
+HashList *UrlShorteners = NULL;
void AddRSSStartHandler(rss_handler_func Handler, int Flags, const char *key, long len)
{
rss_xml_handler *h;
Put(EndHandlers, key, len, h, NULL);
}
-///#if 0
+#if 0
//#ifdef HAVE_ICONV
#include <iconv.h>
{
iconv_t cd = (iconv_t)(-1);
int flag;
- CtdlLogPrintf(0, "RSS: fill encoding info ...\n");
+ syslog(LOG_EMERG, "RSS: fill encoding info ...\n");
#if G_BYTE_ORDER == G_LITTLE_ENDIAN
cd = iconv_open ("UCS-2LE", charset);
}
{
- unsigned short out;
+ unsigned short out = 0;
unsigned char buf[4];
unsigned int i0, i1, i2;
- int result;
+ int result = 0;
flag = 0;
for (i0 = 0; i0 < 0x100; i0++)
{
{
XML_Encoding *info = data;
int res;
- CtdlLogPrintf(0, "RSS: Converting ...\n");
+ syslog(LOG_EMERG, "RSS: Converting ...\n");
if (s == NULL)
return -1;
XML_Encoding * info)
{
int result;
- CtdlLogPrintf(0, "RSS: unknown encoding ...\n");
+ syslog(LOG_EMERG, "RSS: unknown encoding ...\n");
result = fill_encoding_info (name, info);
if (result >= 0)
{
}
///#endif
-//#endif
+#endif
+size_t GetLocationString( void *ptr, size_t size, size_t nmemb, void *userdata)
+{
+#define LOCATION "location"
+ if (strncasecmp((char*)ptr, LOCATION, sizeof(LOCATION) - 1) == 0)
+ {
+ StrBuf *pURL = (StrBuf*) userdata;
+ char *pch = (char*) ptr;
+ char *pche;
+
+ pche = pch + (size * nmemb);
+ pch += sizeof(LOCATION);
+
+ while (isspace(*pch) || (*pch == ':'))
+ pch ++;
+
+ while (isspace(*pche) || (*pche == '\0'))
+ pche--;
+
+ FlushStrBuf(pURL);
+ StrBufPlain(pURL, pch, pche - pch + 1);
+ }
+ return size * nmemb;
+}
+
+int LookupUrl(StrBuf *ShorterUrlStr)
+{
+ CURL *curl;
+ char errmsg[1024] = "";
+ StrBuf *Answer;
+ int rc = 0;
+
+ curl = curl_easy_init();
+ if (!curl) {
+ syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
+ return 0;
+ }
+ Answer = NewStrBufPlain(NULL, SIZ);
+
+ curl_easy_setopt(curl, CURLOPT_URL, ChrPtr(ShorterUrlStr));
+ curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
+ curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
+ curl_easy_setopt(curl, CURLOPT_WRITEDATA, Answer);
+// curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
+ curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback);
+ curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
+ curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
+#ifdef CURLOPT_HTTP_CONTENT_DECODING
+ curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
+ curl_easy_setopt(curl, CURLOPT_ENCODING, "");
+#endif
+ curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
+ curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180); /* die after 180 seconds */
+ curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 0);
+
+ curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION , GetLocationString);
+ curl_easy_setopt(curl, CURLOPT_WRITEHEADER, ShorterUrlStr);
+
+
+ if (
+ (!IsEmptyStr(config.c_ip_addr))
+ && (strcmp(config.c_ip_addr, "*"))
+ && (strcmp(config.c_ip_addr, "::"))
+ && (strcmp(config.c_ip_addr, "0.0.0.0"))
+ ) {
+ curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
+ }
+
+ if (server_shutting_down)
+ goto shutdown ;
+
+ rc = curl_easy_perform(curl);
+ if (rc) {
+ syslog(LOG_ALERT, "libcurl error %d: %s\n", rc, errmsg);
+ rc = 0;
+ }
+ else
+ rc = 1;
+
+shutdown:
+ FreeStrBuf(&Answer);
+ curl_easy_cleanup(curl);
+
+ return rc;
+
+}
+
+
+
+void CrawlMessageForShorterUrls(HashList *pUrls, StrBuf *Message)
+{
+ int nHits = 0;
+ void *pv;
+ int nShorter = 0;
+ const char *pch;
+ const char *pUrl;
+ ConstStr *pCUrl;
+
+ while (GetHash(UrlShorteners, IKEY(nShorter), &pv))
+ {
+ nShorter++;
+ pch = ChrPtr(Message);
+ pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
+ while ((pUrl != NULL) && (nHits < 99))
+ {
+ pCUrl = malloc(sizeof(ConstStr));
+
+ pCUrl->Key = pUrl;
+ pch = pUrl + StrLength((StrBuf*)pv);
+ while (isalnum(*pch)||(*pch == '-')||(*pch == '/'))
+ pch++;
+ pCUrl->len = pch - pCUrl->Key;
+
+ Put(pUrls, IKEY(nHits), pCUrl, NULL);
+ nHits ++;
+ pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
+ }
+ }
+}
+
+int SortConstStrByPosition(const void *Item1, const void *Item2)
+{
+ const ConstStr *p1, *p2;
+ p1 = (const ConstStr*) Item1;
+ p2 = (const ConstStr*) Item2;
+ if (p1->Key == p2->Key)
+ return 0;
+ if (p1->Key > p2->Key)
+ return 1;
+ return -1;
+}
+
+void ExpandShortUrls(StrBuf *Message)
+{
+ StrBuf *Shadow;
+ HashList *pUrls;
+ ConstStr *pCUrl;
+ const char *pch;
+ const char *pche;
+
+ /* we just suspect URL shorteners to be inside of feeds from twitter
+ * or other short content messages, so don't crawl through real blogs.
+ */
+ if (StrLength(Message) > 500)
+ return;
+
+ pUrls = NewHash(1, Flathash);
+ CrawlMessageForShorterUrls(pUrls, Message);
+
+ if (GetCount(pUrls) > 0)
+ {
+ StrBuf *ShorterUrlStr;
+ HashPos *Pos;
+ const char *RetrKey;
+ void *pv;
+ long len;
+
+ Shadow = NewStrBufPlain(NULL, StrLength(Message));
+ SortByPayload (pUrls, SortConstStrByPosition);
+
+ ShorterUrlStr = NewStrBufPlain(NULL, StrLength(Message));
+
+ pch = ChrPtr(Message);
+ pche = pch + StrLength(Message);
+ Pos = GetNewHashPos(pUrls, 1);
+ while (GetNextHashPos(pUrls, Pos, &len, &RetrKey, &pv))
+ {
+ pCUrl = (ConstStr*) pv;
+
+ if (pch != pCUrl->Key)
+ StrBufAppendBufPlain(Shadow, pch, pCUrl->Key - pch, 0);
+
+ StrBufPlain(ShorterUrlStr, CKEY(*pCUrl));
+ if (LookupUrl(ShorterUrlStr))
+ {
+ StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
+ StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
+ StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
+ StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
+ StrBufAppendBufPlain(Shadow, HKEY("["), 0);
+ StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
+ StrBufAppendBufPlain(Shadow, HKEY("]</a>"), 0);
+ }
+ else
+ {
+ StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
+ StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
+ StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
+ StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
+ StrBufAppendBufPlain(Shadow, HKEY("</a>"), 0);
+ }
+ pch = pCUrl->Key + pCUrl->len + 1;
+
+ }
+ if (pch < pche)
+ StrBufAppendBufPlain(Shadow, pch, pche - pch, 0);
+ FlushStrBuf(Message);
+ StrBufAppendBuf(Message, Shadow, 0);
+
+ FreeStrBuf(&ShorterUrlStr);
+ FreeStrBuf(&Shadow);
+ DeleteHashPos(&Pos);
+ }
+
+ DeleteHash(&pUrls);
+}
+
void AppendLink(StrBuf *Message, StrBuf *link, StrBuf *LinkTitle, const char *Title)
{
#ifndef DEBUG_RSS
if (cdbut != NULL) {
/* Item has already been seen */
- CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
+ syslog(LOG_DEBUG, "%s has already been seen\n", utmsgid);
cdb_free(cdbut);
/* rewrite the record anyway, to update the timestamp */
ut.ut_timestamp = time(NULL);
cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
}
- else
+ else
#endif
{
/* Item has not been seen, so save it. */
- CtdlLogPrintf(CTDL_DEBUG, "RSS: saving item...\n");
+ syslog(LOG_DEBUG, "RSS: saving item...\n");
if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY(""));
StrBufSpaceToBlank(ri->description);
msg = malloc(sizeof(struct CtdlMessage));
if (ri->author_or_creator != NULL) {
char *From;
- StrBuf *Encoded, *QPEncoded;
- StrBuf *UserName;
- StrBuf *EmailAddress;
- StrBuf *EncBuf;
+ StrBuf *Encoded = NULL;
int FromAt;
- UserName = NewStrBuf();
- EmailAddress = NewStrBuf();
- EncBuf = NewStrBuf();
-////TODO!
From = html_to_ascii(ChrPtr(ri->author_or_creator),
StrLength(ri->author_or_creator),
512, 0);
- FromAt = strchr(From, '@') != NULL;
+ StrBufPlain(ri->author_or_creator, From, -1);
+ StrBufTrim(ri->author_or_creator);
+ free(From);
+
+ FromAt = strchr(ChrPtr(ri->author_or_creator), '@') != NULL;
if (!FromAt && StrLength (ri->author_email) > 0)
{
- Encoded = NewStrBuf();
- if (!IsEmptyStr(From))
- {
- StrBufPrintf(Encoded,
- "\"%s\" <%s>",
- From,
- ChrPtr(ri->author_email));
- }
- else
- {
- StrBufPrintf(Encoded,
- "<%s>",
- ChrPtr(ri->author_email));
- }
+ StrBufRFC2047encode(&Encoded, ri->author_or_creator);
+ msg->cm_fields['A'] = SmashStrBuf(&Encoded);
+ msg->cm_fields['P'] = SmashStrBuf(&ri->author_email);
}
else
{
if (FromAt)
- Encoded = NewStrBufPlain(From, -1);
+ {
+ msg->cm_fields['A'] = SmashStrBuf(&ri->author_or_creator);
+ msg->cm_fields['P'] = strdup(msg->cm_fields['A']);
+ }
else
{
- Encoded = NewStrBuf();
- StrBufPrintf(Encoded,
- "\"%s\" <%s>",
- From,
- "rss@localhost"); /// TODO: get hostname?
+ StrBufRFC2047encode(&Encoded, ri->author_or_creator);
+ msg->cm_fields['A'] = SmashStrBuf(&Encoded);
+ msg->cm_fields['P'] = strdup("rss@localhost");
}
}
- free(From);
- StrBufTrim(Encoded);
- QPEncoded = StrBufSanitizeEmailRecipientVector(Encoded, UserName, EmailAddress, EncBuf);
- msg->cm_fields['A'] = SmashStrBuf(&QPEncoded);
-
- FreeStrBuf(&Encoded);
- FreeStrBuf(&UserName);
- FreeStrBuf(&EmailAddress);
- FreeStrBuf(&EncBuf);
-
}
else {
msg->cm_fields['A'] = strdup("rss");
StrBufSpaceToBlank(ri->title);
len = StrLength(ri->title);
Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
- Encoded = NewStrBufPlain(Sbj, -1);
+ len = strlen(Sbj);
+ if (Sbj[len - 1] == '\n')
+ {
+ len --;
+ Sbj[len] = '\0';
+ }
+ Encoded = NewStrBufPlain(Sbj, len);
free(Sbj);
StrBufTrim(Encoded);
msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
FreeStrBuf(&Encoded);
}
+
+ if (ri->pubdate <= 0) {
+ ri->pubdate = time(NULL);
+ }
msg->cm_fields['T'] = malloc(64);
snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
+
if (ri->channel_title != NULL) {
if (StrLength(ri->channel_title) > 0) {
msg->cm_fields['O'] = strdup(ChrPtr(ri->channel_title));
}
if (ri->link == NULL)
ri->link = NewStrBufPlain(HKEY(""));
+#ifdef EXPERIMENTAL_SHORTER_URLS
+/* its rather hard to implement this libevent compatible, so we don't ship it. */
+ ExpandShortUrls(ri->description);
+#endif
msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
Message = NewStrBufPlain(NULL, StrLength(ri->description));
FreeStrBuf(&ri->guid);
FreeStrBuf(&ri->title);
FreeStrBuf(&ri->link);
+ FreeStrBuf(&ri->linkTitle);
+ FreeStrBuf(&ri->reLink);
+ FreeStrBuf(&ri->reLinkTitle);
+ FreeStrBuf(&ri->description);
+ FreeStrBuf(&ri->channel_title);
FreeStrBuf(&ri->author_or_creator);
- FreeStrBuf(&ri->author_email);
FreeStrBuf(&ri->author_url);
- FreeStrBuf(&ri->description);
+ FreeStrBuf(&ri->author_email);
}
void rss_xml_start(void *data, const char *supplied_el, const char **attr)
char *sep = NULL;
/* Axe the namespace, we don't care about it */
-/// CtdlLogPrintf(0, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
+/// syslog(LOG_EMERG, "RSS: supplied el %d: %s...\n", rssc->Cfg->ItemType, supplied_el);
pel = supplied_el;
while (sep = strchr(pel, ':'), sep) {
pel = sep + 1;
&v))
{
#ifdef DEBUG_RSS
- CtdlLogPrintf(0, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
+ syslog(LOG_EMERG, "RSS: START ignoring because of wrong namespace [%s] = [%s]\n",
supplied_el);
#endif
return;
}
#ifdef DEBUG_RSS
else
- CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
+ syslog(LOG_EMERG, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
#endif
}
#ifdef DEBUG_RSS
else
- CtdlLogPrintf(0, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
+ syslog(LOG_EMERG, "RSS: START unhandled: [%s] [%s]...\n", pel, supplied_el);
#endif
}
while (sep = strchr(pel, ':'), sep) {
pel = sep + 1;
}
-// CtdlLogPrintf(0, "RSS: END %s...\n", el);
+// syslog(LOG_EMERG, "RSS: END %s...\n", el);
if (pel != supplied_el)
{
void *v;
&v))
{
#ifdef DEBUG_RSS
- CtdlLogPrintf(0, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
+ syslog(LOG_EMERG, "RSS: END ignoring because of wrong namespace [%s] = [%s]\n",
supplied_el, ChrPtr(rssc->CData));
#endif
FlushStrBuf(rssc->CData);
}
#ifdef DEBUG_RSS
else
- CtdlLogPrintf(0, "RSS: END unhandled: [%s] [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
+ syslog(LOG_EMERG, "RSS: END unhandled: [%s] [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
#endif
}
#ifdef DEBUG_RSS
else
- CtdlLogPrintf(0, "RSS: END unhandled: [%s] [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
+ syslog(LOG_EMERG, "RSS: END unhandled: [%s] [%s] = [%s]...\n", pel, supplied_el, ChrPtr(rssc->CData));
#endif
FlushStrBuf(rssc->CData);
rssc->Current = NULL;
void RSS_item_rss_start (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
{
- CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RSS feed.\n");
+ syslog(LOG_DEBUG, "RSS: This is an RSS feed.\n");
Cfg->ItemType = RSS_RSS;
}
void RSS_item_rdf_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
{
- CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an RDF feed.\n");
+ syslog(LOG_DEBUG, "RSS: This is an RDF feed.\n");
Cfg->ItemType = RSS_RSS;
}
void ATOM_item_feed_start(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
{
- CtdlLogPrintf(CTDL_DEBUG, "RSS: This is an ATOM feed.\n");
+ syslog(LOG_DEBUG, "RSS: This is an ATOM feed.\n");
Cfg->ItemType = RSS_ATOM;
}
StrBufTrim(ri->link);
}
}
+void RSS_item_relink_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
+{
+ if (StrLength(CData) > 0) {
+ NewStrBufDupAppendFlush(&ri->reLink, CData, NULL, 0);
+ StrBufTrim(ri->reLink);
+ }
+}
void RSSATOM_item_title_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
{
void ATOM_item_content_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
{
- if (StrLength(CData) > 0) {
- NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
- StrBufTrim(ri->description);
+ long olen = StrLength (ri->description);
+ long clen = StrLength (CData);
+ if (clen > 0)
+ {
+ if (olen == 0) {
+ NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
+ StrBufTrim(ri->description);
+ }
+ else if (olen < clen) {
+ FlushStrBuf(ri->description);
+ NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
+ StrBufTrim(ri->description);
+ }
}
}
void ATOM_item_summary_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
void RSS_item_description_end (StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
{
- if (StrLength(CData) > 0) {
- NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
+ long olen = StrLength (ri->description);
+ long clen = StrLength (CData);
+ if (clen > 0)
+ {
+ if (olen == 0) {
+ NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
+ StrBufTrim(ri->description);
+ }
+ else if (olen < clen) {
+ FlushStrBuf(ri->description);
+ NewStrBufDupAppendFlush(&ri->description, CData, NULL, 0);
+ StrBufTrim(ri->description);
+ }
}
}
void RSS_item_rss_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
{
-// CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
+// syslog(LOG_DEBUG, "End of feed detected. Closing parser.\n");
ri->done_parsing = 1;
}
void RSS_item_rdf_end(StrBuf *CData, rss_item *ri, rssnetcfg *Cfg, const char** Attr)
{
-// CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n");
+// syslog(LOG_DEBUG, "End of feed detected. Closing parser.\n");
ri->done_parsing = 1;
}
void rss_do_fetching(rssnetcfg *Cfg) {
rsscollection rssc;
rss_item ri;
- XML_Parser xp;
+ XML_Parser xp = NULL;
StrBuf *Answer;
CURL *curl;
const char *at;
long len;
+ time_t now;
+
+ now = time(NULL);
+
+ if ((Cfg->next_poll != 0) && (now < Cfg->next_poll))
+ return;
memset(&ri, 0, sizeof(rss_item));
rssc.Item = &ri;
rssc.Cfg = Cfg;
- CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
+ syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url);
curl = curl_easy_init();
if (!curl) {
- CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
+ syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
return;
}
Answer = NewStrBufPlain(NULL, SIZ);
curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
}
- if (CtdlThreadCheckStop())
+ if (server_shutting_down)
{
curl_easy_cleanup(curl);
return;
}
- if (CtdlThreadCheckStop())
+ if (server_shutting_down)
goto shutdown ;
res = curl_easy_perform(curl);
if (res) {
- CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
+ syslog(LOG_ALERT, "libcurl error %d: %s\n", res, errmsg);
}
- if (CtdlThreadCheckStop())
+ if (server_shutting_down)
goto shutdown ;
xp = XML_ParserCreateNS(ptr, ':');
if (!xp) {
- CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
+ syslog(LOG_ALERT, "Cannot create XML parser!\n");
goto shutdown;
}
FlushStrBuf(rssc.Key);
XML_Parse(xp, "", 0, 1);
- CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n",
+ syslog(LOG_ALERT, "RSS: XML Status [%s] \n",
XML_ErrorString(
XML_GetErrorCode(xp)));
shutdown:
+ FreeStrBuf(&Answer);
curl_easy_cleanup(curl);
XML_ParserFree(xp);
flush_rss_item(&ri);
FreeStrBuf(&rssc.CData);
FreeStrBuf(&rssc.Key);
+
+ Cfg->next_poll = time(NULL) + config.c_net_freq;
}
assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
- if (CtdlThreadCheckStop())
+ if (server_shutting_down)
return;
/* Only do net processing for rooms that have netconfigs */
return;
}
- while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
+ while (fgets(buf, sizeof buf, fp) != NULL && !server_shutting_down) {
buf[strlen(buf)-1] = 0;
extract_token(instr, buf, 0, '|', sizeof instr);
/* Otherwise create a new client request */
if (use_this_rncptr == NULL) {
rncptr = (rssnetcfg *) malloc(sizeof(rssnetcfg));
+ memset(rncptr, 0, sizeof(rssnetcfg));
rncptr->ItemType = RSS_UNSET;
if (rncptr != NULL) {
rncptr->next = rnclist;
/*
* Scan for rooms that have RSS client requests configured
*/
-void *rssclient_scan(void *args) {
+void rssclient_scan(void) {
static time_t last_run = 0L;
static int doing_rssclient = 0;
rssnetcfg *rptr = NULL;
- CitContext rssclientCC;
- /* Give this thread its own private CitContext */
- CtdlFillSystemContext(&rssclientCC, "rssclient");
- citthread_setspecific(MyConKey, (void *)&rssclientCC );
+ /* Run no more than once every 15 minutes. */
+ if ((time(NULL) - last_run) < 900) {
+ return;
+ }
/*
* This is a simple concurrency check to make sure only one rssclient run
* don't really require extremely fine granularity here, we'll do it
* with a static variable instead.
*/
- if (doing_rssclient) return NULL;
+ if (doing_rssclient) return;
doing_rssclient = 1;
- CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
+ syslog(LOG_DEBUG, "rssclient started\n");
CtdlForEachRoom(rssclient_scan_room, NULL);
- while (rnclist != NULL && !CtdlThreadCheckStop()) {
+ while (rnclist != NULL && !server_shutting_down) {
rss_do_fetching(rnclist);
rptr = rnclist;
rnclist = rnclist->next;
free(rptr);
}
- CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
+ syslog(LOG_DEBUG, "rssclient ended\n");
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");
- CtdlClearSystemContext();
- return NULL;
+ return;
}
+void LoadUrlShorteners(void)
+{
+ int i = 0;
+ int fd;
+ const char *POS = NULL;
+ const char *Err = NULL;
+ StrBuf *Content, *Line;
+
+
+ UrlShorteners = NewHash(0, Flathash);
+
+ fd = open(file_citadel_urlshorteners, 0);
+
+ if (fd != 0)
+ {
+ Content = NewStrBufPlain(NULL, SIZ);
+ Line = NewStrBuf();
+ while (POS != StrBufNOTNULL)
+ {
+ StrBufTCP_read_buffered_line_fast (Line, Content, &POS, &fd, 1, 1, &Err);
+ StrBufTrim(Line);
+ if ((*ChrPtr(Line) != '#') && (StrLength(Line) > 0))
+ {
+ Put(UrlShorteners, IKEY(i), Line, HFreeStrBuf);
+ i++;
+ Line = NewStrBuf();
+ }
+ else
+ FlushStrBuf(Line);
+ if (POS == NULL)
+ POS = StrBufNOTNULL;
+ }
+ FreeStrBuf(&Line);
+ FreeStrBuf(&Content);
+ }
+ close(fd);
+}
+
+void rss_cleanup(void)
+{
+ DeleteHash(&StartHandlers);
+ DeleteHash(&EndHandlers);
+ DeleteHash(&UrlShorteners);
+ DeleteHash(&KnownNameSpaces);
+}
CTDL_MODULE_INIT(rssclient)
{
if (threading)
{
- CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
- CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
+ syslog(LOG_INFO, "%s\n", curl_version());
+ CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER);
}
-
- StartHandlers = NewHash(1, NULL);
- EndHandlers = NewHash(1, NULL);
-
- AddRSSStartHandler(RSS_item_rss_start, RSS_UNSET, HKEY("rss"));
- AddRSSStartHandler(RSS_item_rdf_start, RSS_UNSET, HKEY("rdf"));
- AddRSSStartHandler(ATOM_item_feed_start, RSS_UNSET, HKEY("feed"));
- AddRSSStartHandler(RSS_item_item_start, RSS_RSS, HKEY("item"));
- AddRSSStartHandler(ATOM_item_entry_start, RSS_ATOM, HKEY("entry"));
- AddRSSStartHandler(ATOM_item_link_start, RSS_ATOM, HKEY("link"));
-
- AddRSSEndHandler(ATOMRSS_item_title_end, RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
- AddRSSEndHandler(RSS_item_guid_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
- AddRSSEndHandler(ATOM_item_id_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
- AddRSSEndHandler(RSS_item_link_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
- AddRSSEndHandler(RSSATOM_item_title_end, RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
- AddRSSEndHandler(ATOM_item_content_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
- AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("encoded"));
- AddRSSEndHandler(ATOM_item_summary_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
- AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
- AddRSSEndHandler(ATOM_item_published_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
- AddRSSEndHandler(ATOM_item_updated_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
- AddRSSEndHandler(RSS_item_pubdate_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
- AddRSSEndHandler(RSS_item_date_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
- AddRSSEndHandler(RSS_item_author_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
- AddRSSEndHandler(RSS_item_creator_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
+ else
+ {
+ LoadUrlShorteners ();
+
+ StartHandlers = NewHash(1, NULL);
+ EndHandlers = NewHash(1, NULL);
+
+ AddRSSStartHandler(RSS_item_rss_start, RSS_UNSET, HKEY("rss"));
+ AddRSSStartHandler(RSS_item_rdf_start, RSS_UNSET, HKEY("rdf"));
+ AddRSSStartHandler(ATOM_item_feed_start, RSS_UNSET, HKEY("feed"));
+ AddRSSStartHandler(RSS_item_item_start, RSS_RSS, HKEY("item"));
+ AddRSSStartHandler(ATOM_item_entry_start, RSS_ATOM, HKEY("entry"));
+ AddRSSStartHandler(ATOM_item_link_start, RSS_ATOM, HKEY("link"));
+
+ AddRSSEndHandler(ATOMRSS_item_title_end, RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
+ AddRSSEndHandler(RSS_item_guid_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("guid"));
+ AddRSSEndHandler(ATOM_item_id_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("id"));
+ AddRSSEndHandler(RSS_item_link_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("link"));
+#if 0
+// hm, rss to the comments of that blog, might be interesting in future, but...
+ AddRSSEndHandler(RSS_item_relink_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("commentrss"));
+// comment count...
+ AddRSSEndHandler(RSS_item_relink_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("comments"));
+#endif
+ AddRSSEndHandler(RSSATOM_item_title_end, RSS_ATOM|RSS_RSS|RSS_REQUIRE_BUF, HKEY("title"));
+ AddRSSEndHandler(ATOM_item_content_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("content"));
+ AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_ATOM|RSS_REQUIRE_BUF, HKEY("encoded"));
+ AddRSSEndHandler(ATOM_item_summary_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("summary"));
+ AddRSSEndHandler(RSS_item_description_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("description"));
+ AddRSSEndHandler(ATOM_item_published_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("published"));
+ AddRSSEndHandler(ATOM_item_updated_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("updated"));
+ AddRSSEndHandler(RSS_item_pubdate_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("pubdate"));
+ AddRSSEndHandler(RSS_item_date_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("date"));
+ AddRSSEndHandler(RSS_item_author_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("author"));
+ AddRSSEndHandler(RSS_item_creator_end, RSS_RSS|RSS_REQUIRE_BUF, HKEY("creator"));
/* <author> */
- AddRSSEndHandler(ATOM_item_email_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
- AddRSSEndHandler(ATOM_item_name_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
- AddRSSEndHandler(ATOM_item_uri_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
+ AddRSSEndHandler(ATOM_item_email_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("email"));
+ AddRSSEndHandler(ATOM_item_name_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("name"));
+ AddRSSEndHandler(ATOM_item_uri_end, RSS_ATOM|RSS_REQUIRE_BUF, HKEY("uri"));
/* </author> */
- AddRSSEndHandler(RSS_item_item_end, RSS_RSS, HKEY("item"));
- AddRSSEndHandler(RSS_item_rss_end, RSS_RSS, HKEY("rss"));
- AddRSSEndHandler(RSS_item_rdf_end, RSS_RSS, HKEY("rdf"));
- AddRSSEndHandler(ATOM_item_entry_end, RSS_ATOM, HKEY("entry"));
+ AddRSSEndHandler(RSS_item_item_end, RSS_RSS, HKEY("item"));
+ AddRSSEndHandler(RSS_item_rss_end, RSS_RSS, HKEY("rss"));
+ AddRSSEndHandler(RSS_item_rdf_end, RSS_RSS, HKEY("rdf"));
+ AddRSSEndHandler(ATOM_item_entry_end, RSS_ATOM, HKEY("entry"));
/* at the start of atoms: <seq> <li>link to resource</li></seq> ignore them. */
- AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("seq"));
- AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("seq"));
- AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("li"));
- AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("li"));
+ AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("seq"));
+ AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("seq"));
+ AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("li"));
+ AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("li"));
/* links to other feed generators... */
- AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("feedflare"));
- AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("feedflare"));
- AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
- AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
-
- KnownNameSpaces = NewHash(1, NULL);
- Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
- Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
+ AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("feedflare"));
+ AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("feedflare"));
+ AddRSSStartHandler(RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
+ AddRSSEndHandler (RSSATOM_item_ignore, RSS_RSS|RSS_ATOM, HKEY("browserfriendly"));
+
+ KnownNameSpaces = NewHash(1, NULL);
+ Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearch/1.1/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://a9.com/-/spec/opensearchrss/1.0/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://backend.userland.com/creativeCommonsRssModule"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://purl.org/atom/ns#"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://purl.org/dc/elements/1.1/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/content/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/slash/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/modules/syndication/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://purl.org/rss/1.0/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://purl.org/syndication/thread/1.0"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://rssnamespace.org/feedburner/ext/1.0"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://schemas.google.com/g/2005"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://webns.net/mvcb/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://web.resource.org/cc/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://wellformedweb.org/CommentAPI/"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://www.georss.org/georss"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/xhtml"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://www.w3.org/1999/02/22-rdf-syntax-ns#"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://www.w3.org/2003/01/geo/wgs84_pos#"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("http://www.w3.org/2005/Atom"), NULL, reference_free_handler);
+ Put(KnownNameSpaces, HKEY("urn:flickr:"), NULL, reference_free_handler);
#if 0
- /* we don't like these namespaces because of they shadow our usefull parameters. */
- Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
+ /* we don't like these namespaces because of they shadow our usefull parameters. */
+ Put(KnownNameSpaces, HKEY("http://search.yahoo.com/mrss/"), NULL, reference_free_handler);
#endif
+ CtdlRegisterCleanupHook(rss_cleanup);
+ }
return "rssclient";
}
projects / citadel.git / blobdiff