/*
* Bring external RSS feeds into rooms.
*
- * Copyright (c) 2007-2012 by the citadel.org team
+ * Copyright (c) 2007-2015 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 "event_client.h"
#include "rss_atom_parser.h"
-void rss_save_item(rss_item *ri, rss_aggregator *Cfg);
+void rss_remember_item(rss_item *ri, rss_aggregator *Cfg);
int RSSAtomParserDebugEnabled = 0;
-#define N ((rss_aggregator*)IO->Data)->QRnumber
-
-#define DBGLOG(LEVEL) if ((LEVEL != LOG_DEBUG) || (RSSAtomParserDebugEnabled != 0))
-
-#define EVRSSATOM_syslog(LEVEL, FORMAT, ...) \
- DBGLOG(LEVEL) syslog(LEVEL, \
- "IO[%ld]CC[%d][%ld]RSSP" FORMAT, \
- IO->ID, CCID, N, __VA_ARGS__)
-
-#define EVRSSATOMM_syslog(LEVEL, FORMAT) \
- DBGLOG(LEVEL) syslog(LEVEL, \
- "IO[%ld]CC[%d][%ld]RSSP" FORMAT, \
- IO->ID, CCID, N)
-
-#define EVRSSATOMCS_syslog(LEVEL, FORMAT, ...) \
- DBGLOG(LEVEL) syslog(LEVEL, "IO[%ld][%ld]RSSP" FORMAT, \
- IO->ID, N, __VA_ARGS__)
-
-#define EVRSSATOMSM_syslog(LEVEL, FORMAT) \
- DBGLOG(LEVEL) syslog(LEVEL, "IO[%ld][%ld]RSSP" FORMAT, \
- IO->ID, N)
+#define N ((rss_aggregator*)IO->Data)->Cfg.QRnumber
/*
* Convert an RDF/RSS datestamp into a time_t
rss_aggregator *RSSAggr,
const char** Attr)
{
- AsyncIO *IO = &RSSAggr->IO;
- EVRSSATOMM_syslog(LOG_DEBUG, "RSS: This is an RSS feed.\n");
+ syslog(LOG_DEBUG, "RSS: This is an RSS feed.");
RSSAggr->ItemType = RSS_RSS;
}
rss_aggregator *RSSAggr,
const char** Attr)
{
- AsyncIO *IO = &RSSAggr->IO;
- EVRSSATOMM_syslog(LOG_DEBUG, "RSS: This is an RDF feed.\n");
+ syslog(LOG_DEBUG, "RSS: This is an RDF feed.");
RSSAggr->ItemType = RSS_RSS;
}
rss_aggregator *RSSAggr,
const char** Attr)
{
- AsyncIO *IO = &RSSAggr->IO;
- EVRSSATOMM_syslog(LOG_DEBUG, "RSS: This is an ATOM feed.\n");
+ syslog(LOG_DEBUG, "RSS: This is an ATOM feed.");
RSSAggr->ItemType = RSS_ATOM;
}
const char** Attr)
{
--ri->item_tag_nesting;
- rss_save_item(ri, RSSAggr);
+ rss_remember_item(ri, RSSAggr);
}
const char** Attr)
{
--ri->item_tag_nesting;
- rss_save_item(ri, RSSAggr);
+ rss_remember_item(ri, RSSAggr);
}
void RSS_item_rss_end(StrBuf *CData,
rss_aggregator *RSSAggr,
const char** Attr)
{
- AsyncIO *IO = &RSSAggr->IO;
- EVRSSATOMM_syslog(LOG_DEBUG, "End of feed detected. Closing parser.\n");
+ syslog(LOG_DEBUG, "End of feed detected. Closing parser.");
ri->done_parsing = 1;
}
rss_aggregator *RSSAggr,
const char** Attr)
{
- AsyncIO *IO = &RSSAggr->IO;
- EVRSSATOMM_syslog(LOG_DEBUG, "End of feed detected. Closing parser.\n");
+ syslog(LOG_DEBUG, "End of feed detected. Closing parser.");
ri->done_parsing = 1;
}
{
networker_save_message *Msg = (networker_save_message *) vMsg;
- CtdlFreeMessageContents(&Msg->Msg);
+ CM_FreeContents(&Msg->Msg);
FreeStrBuf(&Msg->Message);
FreeStrBuf(&Msg->MsgGUID);
- free(Msg);
-}
+ FreeStrBuf(&Msg->author_email);
+ FreeStrBuf(&Msg->author_or_creator);
+ FreeStrBuf(&Msg->title);
+ FreeStrBuf(&Msg->description);
-void AppendLink(StrBuf *Message,
- StrBuf *link,
- StrBuf *LinkTitle,
- const char *Title)
-{
- if (StrLength(link) > 0)
- {
- StrBufAppendBufPlain(Message, HKEY("<a href=\""), 0);
- StrBufAppendBuf(Message, link, 0);
- StrBufAppendBufPlain(Message, HKEY("\">"), 0);
- if (StrLength(LinkTitle) > 0)
- StrBufAppendBuf(Message, LinkTitle, 0);
- else if ((Title != NULL) && !IsEmptyStr(Title))
- StrBufAppendBufPlain(Message, Title, -1, 0);
- else
- StrBufAppendBuf(Message, link, 0);
- StrBufAppendBufPlain(Message, HKEY("</a><br>\n"), 0);
- }
+ FreeStrBuf(&Msg->link);
+ FreeStrBuf(&Msg->linkTitle);
+
+ FreeStrBuf(&Msg->reLink);
+ FreeStrBuf(&Msg->reLinkTitle);
+
+ free(Msg);
}
+
/*
* Commit a fetched and parsed RSS item to disk
*/
-void rss_save_item(rss_item *ri, rss_aggregator *RSSAggr)
+void rss_remember_item(rss_item *ri, rss_aggregator *RSSAggr)
{
networker_save_message *SaveMsg;
struct MD5Context md5context;
u_char rawdigest[MD5_DIGEST_LEN];
- int msglen = 0;
- StrBuf *Message;
StrBuf *guid;
- AsyncIO *IO = &RSSAggr->IO;
int n;
-
- SaveMsg = (networker_save_message *) malloc(
- sizeof(networker_save_message));
+ SaveMsg = (networker_save_message *) malloc(sizeof(networker_save_message));
memset(SaveMsg, 0, sizeof(networker_save_message));
/* Construct a GUID to use in the S_USETABLE table.
else {
MD5Init(&md5context);
if (ri->title != NULL) {
- MD5Update(&md5context,
- (const unsigned char*)SKEY(ri->title));
+ MD5Update(&md5context, (const unsigned char*)SKEY(ri->title));
}
if (ri->link != NULL) {
- MD5Update(&md5context,
- (const unsigned char*)SKEY(ri->link));
+ MD5Update(&md5context, (const unsigned char*)SKEY(ri->link));
}
MD5Final(rawdigest, &md5context);
- guid = NewStrBufPlain(NULL,
- MD5_DIGEST_LEN * 2 + 12 /* _rss2ctdl*/);
+ guid = NewStrBufPlain(NULL, MD5_DIGEST_LEN * 2 + 12 /* _rss2ctdl*/);
StrBufHexEscAppend(guid, NULL, rawdigest, MD5_DIGEST_LEN);
StrBufAppendBufPlain(guid, HKEY("_rss2ctdl"), 0);
}
/* translate Item into message. */
- EVRSSATOMM_syslog(LOG_DEBUG, "RSS: translating item...\n");
+ syslog(LOG_DEBUG, "RSS: translating item...");
if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY(""));
StrBufSpaceToBlank(ri->description);
SaveMsg->Msg.cm_magic = CTDLMESSAGE_MAGIC;
SaveMsg->Msg.cm_anon_type = MES_NORMAL;
SaveMsg->Msg.cm_format_type = FMT_RFC822;
- if (ri->guid != NULL) {
- SaveMsg->Msg.cm_fields['E'] = strdup(ChrPtr(ri->guid));
- }
-
- if (ri->author_or_creator != NULL) {
- char *From;
- StrBuf *Encoded = NULL;
- int FromAt;
-
- From = html_to_ascii(ChrPtr(ri->author_or_creator),
- StrLength(ri->author_or_creator),
- 512, 0);
- StrBufPlain(ri->author_or_creator, From, -1);
- StrBufTrim(ri->author_or_creator);
- free(From);
+ /* gather the cheaply computed information now... */
- FromAt = strchr(ChrPtr(ri->author_or_creator), '@') != NULL;
- if (!FromAt && StrLength (ri->author_email) > 0)
- {
- StrBufRFC2047encode(&Encoded, ri->author_or_creator);
- SaveMsg->Msg.cm_fields['A'] = SmashStrBuf(&Encoded);
- SaveMsg->Msg.cm_fields['P'] =
- SmashStrBuf(&ri->author_email);
- }
- else
- {
- if (FromAt)
- {
- SaveMsg->Msg.cm_fields['A'] =
- SmashStrBuf(&ri->author_or_creator);
- SaveMsg->Msg.cm_fields['P'] =
- strdup(SaveMsg->Msg.cm_fields['A']);
- }
- else
- {
- StrBufRFC2047encode(&Encoded,
- ri->author_or_creator);
- SaveMsg->Msg.cm_fields['A'] =
- SmashStrBuf(&Encoded);
- SaveMsg->Msg.cm_fields['P'] =
- strdup("rss@localhost");
-
- }
- if (ri->pubdate <= 0) {
- ri->pubdate = time(NULL);
- }
- }
- }
- else {
- SaveMsg->Msg.cm_fields['A'] = strdup("rss");
+ if (ri->guid != NULL) {
+ CM_SetField(&SaveMsg->Msg, eExclusiveID, SKEY(ri->guid));
}
- SaveMsg->Msg.cm_fields['N'] = strdup(NODENAME);
- if (ri->title != NULL) {
- long len;
- char *Sbj;
- StrBuf *Encoded, *QPEncoded;
-
- QPEncoded = NULL;
- StrBufSpaceToBlank(ri->title);
- len = StrLength(ri->title);
- Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0);
- len = strlen(Sbj);
- if ((len > 0) && (Sbj[len - 1] == '\n'))
- {
- len --;
- Sbj[len] = '\0';
- }
- Encoded = NewStrBufPlain(Sbj, len);
- free(Sbj);
-
- StrBufTrim(Encoded);
- StrBufRFC2047encode(&QPEncoded, Encoded);
+ SaveMsg->MsgGUID = guid;
- SaveMsg->Msg.cm_fields['U'] = SmashStrBuf(&QPEncoded);
- FreeStrBuf(&Encoded);
+ if (ri->pubdate <= 0) {
+ ri->pubdate = time(NULL);
}
- SaveMsg->Msg.cm_fields['T'] = malloc(64);
- snprintf(SaveMsg->Msg.cm_fields['T'], 64, "%ld", ri->pubdate);
+ CM_SetFieldLONG(&SaveMsg->Msg, eTimestamp, ri->pubdate);
if (ri->channel_title != NULL) {
if (StrLength(ri->channel_title) > 0) {
- SaveMsg->Msg.cm_fields['O'] =
- strdup(ChrPtr(ri->channel_title));
+ CM_SetField(&SaveMsg->Msg, eOriginalRoom, SKEY(ri->channel_title));
}
}
- if (ri->link == NULL)
- ri->link = NewStrBufPlain(HKEY(""));
-#if 0 /* temporarily disable shorter urls. */
- SaveMsg->Msg.cm_fields[TMP_SHORTER_URLS] =
- GetShorterUrls(ri->description);
-#endif
+ /* remember the ones for defferred processing to save computing power after we know if we realy need it. */
- msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ;
+ SaveMsg->author_or_creator = ri->author_or_creator;
+ ri->author_or_creator = NULL;
- Message = NewStrBufPlain(NULL, msglen);
+ SaveMsg->author_email = ri->author_email;
+ ri->author_email = NULL;
- StrBufPlain(Message, HKEY(
- "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
- "<html><body>\n"));
-#if 0 /* disable shorter url for now. */
- SaveMsg->Msg.cm_fields[TMP_SHORTER_URL_OFFSET] = StrLength(Message);
-#endif
- StrBufAppendBuf(Message, ri->description, 0);
- StrBufAppendBufPlain(Message, HKEY("<br><br>\n"), 0);
+ SaveMsg->title = ri->title;
+ ri->title = NULL;
- AppendLink(Message, ri->link, ri->linkTitle, NULL);
- AppendLink(Message, ri->reLink, ri->reLinkTitle, "Reply to this");
- StrBufAppendBufPlain(Message, HKEY("</body></html>\n"), 0);
+ SaveMsg->link = ri->link;
+ ri->link = NULL;
- SaveMsg->MsgGUID = guid;
- SaveMsg->Message = Message;
+ SaveMsg->description = ri->description;
+ ri->description = NULL;
+
+ SaveMsg->linkTitle = ri->linkTitle;
+ ri->linkTitle = NULL;
+
+ SaveMsg->reLink = ri->reLink;
+ ri->reLink = NULL;
+
+ SaveMsg->reLinkTitle = ri->reLinkTitle;
+ ri->reLinkTitle = NULL;
n = GetCount(RSSAggr->Messages) + 1;
Put(RSSAggr->Messages, IKEY(n), SaveMsg, FreeNetworkSaveMessage);
}
+
void rss_xml_start(void *data, const char *supplied_el, const char **attr)
{
rss_xml_handler *h;
rss_aggregator *RSSAggr = (rss_aggregator*) data;
- AsyncIO *IO = &RSSAggr->IO;
rss_item *ri = RSSAggr->Item;
void *pv;
const char *pel;
pel - supplied_el - 1,
&v))
{
- EVRSSATOM_syslog(LOG_DEBUG,
+ syslog(LOG_DEBUG,
"RSS: START ignoring "
- "because of wrong namespace [%s]\n",
- supplied_el);
+ "because of wrong namespace [%s]",
+ supplied_el
+ );
return;
}
}
attr);
}
else
- EVRSSATOM_syslog(LOG_DEBUG,
- "RSS: START unhandled: [%s] [%s]...\n",
+ syslog(LOG_DEBUG,
+ "RSS: START unhandled: [%s] [%s]...",
pel,
- supplied_el);
+ supplied_el
+ );
}
else
- EVRSSATOM_syslog(LOG_DEBUG,
- "RSS: START unhandled: [%s] [%s]...\n",
+ syslog(LOG_DEBUG,
+ "RSS: START unhandled: [%s] [%s]...",
pel,
- supplied_el);
+ supplied_el
+ );
}
void rss_xml_end(void *data, const char *supplied_el)
{
rss_xml_handler *h;
rss_aggregator *RSSAggr = (rss_aggregator*) data;
- AsyncIO *IO = &RSSAggr->IO;
rss_item *ri = RSSAggr->Item;
const char *pel;
char *sep = NULL;
while (sep = strchr(pel, ':'), sep) {
pel = sep + 1;
}
- EVRSSATOM_syslog(LOG_DEBUG, "RSS: END %s...\n", supplied_el);
+ syslog(LOG_DEBUG, "RSS: END %s...", supplied_el);
if (pel != supplied_el)
{
void *v;
pel - supplied_el - 1,
&v))
{
- EVRSSATOM_syslog(LOG_DEBUG,
+ syslog(LOG_DEBUG,
"RSS: END ignoring because of wrong namespace"
- "[%s] = [%s]\n",
+ "[%s] = [%s]",
supplied_el,
ChrPtr(RSSAggr->CData));
FlushStrBuf(RSSAggr->CData);
h->Handler(RSSAggr->CData, ri, RSSAggr, NULL);
}
else
- EVRSSATOM_syslog(LOG_DEBUG,
- "RSS: END unhandled: [%s] [%s] = [%s]...\n",
+ syslog(LOG_DEBUG,
+ "RSS: END unhandled: [%s] [%s] = [%s]...",
pel,
supplied_el,
ChrPtr(RSSAggr->CData));
}
else
- EVRSSATOM_syslog(LOG_DEBUG,
- "RSS: END unhandled: [%s] [%s] = [%s]...\n",
+ syslog(LOG_DEBUG,
+ "RSS: END unhandled: [%s] [%s] = [%s]...",
pel,
supplied_el,
ChrPtr(RSSAggr->CData));
-/*
- * Callback function for passing libcurl's output to expat for parsing
- * we don't do streamed parsing so expat can handle non-utf8 documents
-size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
-{
- XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
- return (size*nmemb);
-}
- */
-
-
-
eNextState RSSAggregator_ParseReply(AsyncIO *IO)
{
StrBuf *Buf;
else
ptr = "UTF-8";
- EVRSSATOM_syslog(LOG_DEBUG, "RSS: Now parsing [%s] \n", ChrPtr(RSSAggr->Url));
+ syslog(LOG_DEBUG, "RSS: Now parsing [%s]", ChrPtr(RSSAggr->Url));
RSSAggr->xp = XML_ParserCreateNS(ptr, ':');
if (!RSSAggr->xp) {
- EVRSSATOMM_syslog(LOG_ALERT, "Cannot create XML parser!\n");
+ syslog(LOG_ALERT, "Cannot create XML parser!");
return eAbort;
}
FlushStrBuf(RSSAggr->Key);
XML_SetUserData(RSSAggr->xp, RSSAggr);
XML_SetCdataSectionHandler(RSSAggr->xp,
rss_xml_cdata_start,
- rss_xml_cdata_end);
-
+ rss_xml_cdata_end
+ );
len = StrLength(IO->HttpReq.ReplyData);
ptr = SmashStrBuf(&IO->HttpReq.ReplyData);
XML_Parse(RSSAggr->xp, ptr, len, 0);
free (ptr);
- if (ri->done_parsing == 0)
+ if (ri->done_parsing == 0) {
XML_Parse(RSSAggr->xp, "", 0, 1);
+ }
-
- EVRSSATOM_syslog(LOG_DEBUG, "RSS: XML Status [%s] \n",
- XML_ErrorString(XML_GetErrorCode(RSSAggr->xp)));
+ syslog(LOG_DEBUG, "RSS: XML Status [%s]", XML_ErrorString(XML_GetErrorCode(RSSAggr->xp)));
XML_ParserFree(RSSAggr->xp);
flush_rss_item(ri);
RSSAggr->Pos = GetNewHashPos(RSSAggr->Messages, 1);
-//RSSAggr->next_poll = time(NULL) + config.c_net_freq;
+#if 0
+// FIXME ajc
if (GetNextHashPos(RSSAggr->Messages,
RSSAggr->Pos,
&len,
&Key,
- (void**) &RSSAggr->ThisMsg))
+ (void**) &RSSAggr->ThisMsg)) {
return NextDBOperation(IO, RSS_FetchNetworkUsetableEntry);
- else
+ }
+ else {
+#endif
return eAbort;
+#if 0
+ }
+#endif
}