X-Git-Url: https://code.citadel.org/?a=blobdiff_plain;f=citadel%2Fmodules%2Frssclient%2Fserv_rssclient.c;h=4466d7f621431e869eb11261fd91cb6a7fb0e36b;hb=f1ee61891901850ebbdee1e9440b363dc6df540a;hp=50ab007f25d1f449f54103245d6b233993a0b394;hpb=44c5b5cb18a4c425fd5aad3484d4ac9ee8a3cf43;p=citadel.git diff --git a/citadel/modules/rssclient/serv_rssclient.c b/citadel/modules/rssclient/serv_rssclient.c index 50ab007f2..4466d7f62 100644 --- a/citadel/modules/rssclient/serv_rssclient.c +++ b/citadel/modules/rssclient/serv_rssclient.c @@ -48,811 +48,499 @@ #include "config.h" #include "threads.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 "event_client.h" +#include "rss_atom_parser.h" -typedef struct rssnetcfg rssnetcfg; - -struct rssnetcfg { - rssnetcfg *next; - char url[256]; - char *rooms; - time_t last_error_when; - int ItemType; -}; - -#define RSS_UNSET 0 -#define RSS_RSS 1 -#define RSS_ATOM 2 - -typedef struct _rss_item { - char *chardata; - int chardata_len; - char *roomlist; - int done_parsing; - char *guid; - char *title; - char *link; - char *description; - time_t pubdate; - char channel_title[256]; - int item_tag_nesting; - char *author_or_creator; - char *author_url; - StrBuf *CData; -}rss_item; - - -typedef struct _rsscollection { - rss_item *Item; - rssnetcfg *Cfg; - - -} rsscollection; - -struct rssnetcfg *rnclist = NULL; +#define TMP_MSGDATA 0xFF +#define TMP_SHORTER_URL_OFFSET 0xFE +#define TMP_SHORTER_URLS 0xFD -#if 0 -#ifdef HAVE_ICONV -#include - - -/* - * dug this out of the trashcan of the midgard project, lets see if it works for us. - * original code by Alexander Bokovoy distributed under GPL V2 or later - */ - -/* Returns: - >= 0 - successfull, 0 means conversion doesn't use multibyte sequences - -1 - error during iconv_open call - -2 - error during iconv_close call - ---------------------------------- - This function expects that multibyte encoding in 'charset' wouldn't have - characters with more than 3 bytes. It is not intended to convert UTF-8 because - we'll never receive UTF-8 in our handler (it is handled by Exat itself). -*/ -static int -fill_encoding_info (const char *charset, XML_Encoding * info) -{ - iconv_t cd = (iconv_t)(-1); - int flag; - CtdlLogPrintf(0, "RSS: fill encoding info ...\n"); - -#if G_BYTE_ORDER == G_LITTLE_ENDIAN - cd = iconv_open ("UCS-2LE", charset); -#else - cd = iconv_open ("UCS-2BE", charset); -#endif - - if (cd == (iconv_t) (-1)) - { - return -1; - } - - { - unsigned short out; - unsigned char buf[4]; - unsigned int i0, i1, i2; - int result; - flag = 0; - for (i0 = 0; i0 < 0x100; i0++) - { - buf[0] = i0; - info->map[i0] = 0; - //result = try (cd, buf, 1, &out); - if (result < 0) - { - } - else if (result > 0) - { - info->map[i0] = out; - } - else - { - for (i1 = 0; i1 < 0x100; i1++) - { - buf[1] = i1; - ///result = try (cd, buf, 2, &out); - if (result < 0) - { - } - else if (result > 0) - { - flag++; - info->map[i0] = -2; - } - else - { - for (i2 = 0; i2 < 0x100; i2++) - { - buf[2] = i2; - ////result = try (cd, buf, 3, &out); - if (result < 0) - { - } - else if (result > 0) - { - flag++; - info->map[i0] = -3; - } - } - } - } - } - } - } - - if (iconv_close (cd) < 0) - { - return -2; - } - return flag; -} - -static int -iconv_convertor (void *data, const char *s) -{ - XML_Encoding *info = data; - int res; - CtdlLogPrintf(0, "RSS: Converting ...\n"); - - if (s == NULL) - return -1; -/* - GByteArray *result; - result = g_byte_array_new (); - if (process_block (info->data, (char *) s, strlen (s), result) == 0) - { - res = *(result->data); - g_byte_array_free (result, TRUE); - return res; - } - g_byte_array_free (result, TRUE); -*/ - return -1; -} - -static void -my_release (void *data) -{ - iconv_t cd = (iconv_t) data; - if (iconv_close (cd) != 0) - { -/// TODO: uh no. exit (1); - } -} -int -handle_unknown_xml_encoding (void *encodingHandleData, - const XML_Char * name, - XML_Encoding * info) -{ - int result; - CtdlLogPrintf(0, "RSS: unknown encoding ...\n"); - result = fill_encoding_info (name, info); - if (result >= 0) - { - /* - Special case: client asked for reverse conversion, we'll provide him with - iconv descriptor which handles it. Client should release it by himself. - */ - if(encodingHandleData != NULL) - *((iconv_t *)encodingHandleData) = iconv_open(name, "UTF-8"); - /* - Optimization: we do not need conversion function if encoding is one-to-one, - info->map table will be enough - */ - if (result == 0) - { - info->data = NULL; - info->convert = NULL; - info->release = NULL; - return 1; - } - /* - We do need conversion function because this encoding uses multibyte sequences - */ - info->data = (void *) iconv_open ("UTF-8", name); - if ((int)info->data == -1) - return -1; - info->convert = iconv_convertor; - info->release = my_release; - return 1; - } - if(encodingHandleData != NULL) - *(iconv_t *)encodingHandleData = NULL; - return 0; -} +citthread_mutex_t RSSQueueMutex; /* locks the access to the following vars: */ +HashList *RSSQueueRooms = NULL; /* rss_room_counter */ +HashList *RSSFetchUrls = NULL; /* -> rss_aggregator; ->RefCount access to be locked too. */ -#endif -#endif +eNextState RSSAggregatorTerminate(AsyncIO *IO); -/* - * Commit a fetched and parsed RSS item to disk - */ -void rss_save_item(rsscollection *rssc) { - struct MD5Context md5context; - u_char rawdigest[MD5_DIGEST_LEN]; - int i; - char utmsgid[SIZ]; - struct cdbdata *cdbut; +struct rssnetcfg *rnclist = NULL; +void AppendLink(StrBuf *Message, StrBuf *link, StrBuf *LinkTitle, const char *Title) +{ + if (StrLength(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("
\n"), 0); + } +} +typedef struct __networker_save_message { + AsyncIO IO; + struct CtdlMessage *Msg; + struct recptypes *recp; + rss_aggregator *Cfg; + StrBuf *MsgGUID; + StrBuf *Message; struct UseTable ut; - struct CtdlMessage *msg; - struct recptypes *recp = NULL; - int msglen = 0; - rss_item *ri = rssc->Item; +} networker_save_message; - recp = (struct recptypes *) malloc(sizeof(struct recptypes)); - if (recp == NULL) return; - memset(recp, 0, sizeof(struct recptypes)); - memset(&ut, 0, sizeof(struct UseTable)); - 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; icount --; + if (pRoomC->count == 0) + DeleteEntryFromHash(RSSQueueRooms, At); } - else { - /* Item has not been seen, so save it. */ - CtdlLogPrintf(CTDL_DEBUG, "RSS: saving item...\n"); - 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; + DeleteHashPos(&At); +} - if (ri->guid != NULL) { - msg->cm_fields['E'] = strdup(ri->guid); +void UnlinkRooms(rss_aggregator *Cfg) +{ + + DeleteRoomReference(Cfg->QRnumber); + if (Cfg->OtherQRnumbers != NULL) + { + long HKLen; + const char *HK; + HashPos *At; + void *vData; + + At = GetNewHashPos(Cfg->OtherQRnumbers, 0); + while (GetNextHashPos(Cfg->OtherQRnumbers, At, &HKLen, &HK, &vData) && + (vData != NULL)) + { + long *lData = (long*) vData; + DeleteRoomReference(*lData); } - if (ri->author_or_creator != NULL) { - char *From; - StrBuf *Encoded, *QPEncoded; - StrBuf *UserName; - StrBuf *EmailAddress; - StrBuf *EncBuf; - - UserName = NewStrBuf(); - EmailAddress = NewStrBuf(); - EncBuf = NewStrBuf(); + DeleteHashPos(&At); + } - From = html_to_ascii(ri->author_or_creator, - strlen(ri->author_or_creator), - 512, 0); +} - Encoded = NewStrBufPlain(From, -1); - free(From); - StrBufTrim(Encoded); - QPEncoded = StrBufSanitizeEmailRecipientVector(Encoded, UserName, EmailAddress, EncBuf); - msg->cm_fields['A'] = SmashStrBuf(&QPEncoded); +void UnlinkRSSAggregator(rss_aggregator *Cfg) +{ + HashPos *At; - FreeStrBuf(&Encoded); - FreeStrBuf(&UserName); - FreeStrBuf(&EmailAddress); - FreeStrBuf(&EncBuf); + UnlinkRooms(Cfg); - } - else { - msg->cm_fields['A'] = strdup("rss"); - } + At = GetNewHashPos(RSSFetchUrls, 0); + if (GetHashPosFromKey(RSSFetchUrls, SKEY(Cfg->Url), At) == 0) + { + DeleteEntryFromHash(RSSFetchUrls, At); + } + DeleteHashPos(&At); +} - msg->cm_fields['N'] = strdup(NODENAME); - if (ri->title != NULL) { - long len; - char *Sbj; - StrBuf *Encoded, *QPEncoded; +eNextState FreeNetworkSaveMessage (AsyncIO *IO) +{ + networker_save_message *Ctx = (networker_save_message *) IO->Data; - QPEncoded = NULL; - len = strlen(ri->title); - Sbj = html_to_ascii(ri->title, len, 512, 0); - Encoded = NewStrBufPlain(Sbj, -1); - free(Sbj); + citthread_mutex_lock(&RSSQueueMutex); + Ctx->Cfg->RefCount --; - StrBufTrim(Encoded); - StrBufRFC2047encode(&QPEncoded, Encoded); + if (Ctx->Cfg->RefCount == 0) + { + UnlinkRSSAggregator(Ctx->Cfg); - msg->cm_fields['U'] = SmashStrBuf(&QPEncoded); - FreeStrBuf(&Encoded); - } - 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); - } - } - if (ri->link == NULL) - ri->link = strdup(""); - msglen += 1024 + strlen(ri->link) + strlen(ri->description) ; - msg->cm_fields['M'] = malloc(msglen); - snprintf(msg->cm_fields['M'], msglen, - "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n" - "\n" - "%s

\n" - "%s\n" - "\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); -} - + citthread_mutex_unlock(&RSSQueueMutex); + CtdlFreeMessage(Ctx->Msg); + free_recipients(Ctx->recp); + FreeStrBuf(&Ctx->Message); + FreeStrBuf(&Ctx->MsgGUID); + free(Ctx); + return eAbort; +} -/* - * Convert an RDF/RSS datestamp into a time_t - */ -time_t rdf_parsedate(char *p) +eNextState AbortNetworkSaveMessage (AsyncIO *IO) { - struct tm tm; - time_t t = 0; + return eAbort; ///TODO +} - if (!p) return 0L; - if (strlen(p) < 10) return 0L; +eNextState RSSSaveMessage(AsyncIO *IO) +{ + networker_save_message *Ctx = (networker_save_message *) IO->Data; - memset(&tm, 0, sizeof tm); + Ctx->Msg->cm_fields['M'] = SmashStrBuf(&Ctx->Message); - /* - * If the timestamp appears to be in W3C datetime format, try to - * parse it. See also: http://www.w3.org/TR/NOTE-datetime - * - * This code, along with parsedate.c, is a potential candidate for - * moving into libcitadel. - */ - 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 */ + CtdlSubmitMsg(Ctx->Msg, Ctx->recp, NULL, 0); - t = parsedate(p); - if (t > 0) return(t); + /* write the uidl to the use table so we don't store this item again */ + cdb_store(CDB_USETABLE, SKEY(Ctx->MsgGUID), &Ctx->ut, sizeof(struct UseTable) ); - /* yeesh. ok, just return the current date and time. */ - return(time(NULL)); + return eTerminateConnection; } -#define RSS_UNSET 0 -#define RSS_RSS 1 -#define RSS_ATOM 2 +// TODO: relink me: ExpandShortUrls(ri->description); -void flush_rss_ite(rss_item *ri) +eNextState RSS_FetchNetworkUsetableEntry(AsyncIO *IO) { - /* 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->author_url != NULL) free(ri->author_url); - ri->author_url = 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; - } - FreeStrBuf(&ri->CData); -} + struct cdbdata *cdbut; + networker_save_message *Ctx = (networker_save_message *) IO->Data; -void rss_xml_start(void *data, const char *supplied_el, const char **attr) { - rsscollection *rssc = (rsscollection*) data; - rss_item *ri = rssc->Item; - char el[256]; - 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); - safestrncpy(el, supplied_el, sizeof el); - while (sep = strchr(el, ':'), sep) { - strcpy(el, ++sep); - } + /* Find out if we've already seen this item */ + strcpy(Ctx->ut.ut_msgid, ChrPtr(Ctx->MsgGUID)); /// TODO + Ctx->ut.ut_timestamp = time(NULL); - if ((rssc->Cfg->ItemType == RSS_UNSET) && !strcasecmp(el, "rss")) - { - CtdlLogPrintf(9, "RSS: This is an RSS feed.\n"); - rssc->Cfg->ItemType = RSS_RSS; - } - if ((rssc->Cfg->ItemType == RSS_UNSET) && !strcasecmp(el, "rdf")) - { - CtdlLogPrintf(9, "RSS: This is an RDF feed.\n"); - rssc->Cfg->ItemType = RSS_RSS; - } - else if ((rssc->Cfg->ItemType == RSS_UNSET) && !strcasecmp(el, "feed")) - { - CtdlLogPrintf(9, "RSS: This is an ATOM feed.\n"); - rssc->Cfg->ItemType = RSS_ATOM; - } - else if ((rssc->Cfg->ItemType == RSS_RSS) && - !strcasecmp(el, "item")) - { - ri->item_tag_nesting ++ ; - flush_rss_ite(ri); - } - else if ( (rssc->Cfg->ItemType == RSS_ATOM) && - !strcasecmp(el, "entry")) - { /* Atom feed... */ - ++ri->item_tag_nesting; - flush_rss_ite(ri); + cdbut = cdb_fetch(CDB_USETABLE, SKEY(Ctx->MsgGUID)); +#ifndef DEBUG_RSS + if (cdbut != NULL) { + /* Item has already been seen */ + syslog(LOG_DEBUG, "%s has already been seen\n", ChrPtr(Ctx->MsgGUID)); + cdb_free(cdbut); + + /* rewrite the record anyway, to update the timestamp */ + cdb_store(CDB_USETABLE, + SKEY(Ctx->MsgGUID), + &Ctx->ut, sizeof(struct UseTable) ); + return eAbort; } - else if ((rssc->Cfg->ItemType == RSS_ATOM) && - !strcasecmp(el, "link")) + else +#endif { - int found ; - int i; - - for (found = 0, i = 0;!found && attr[i] != NULL; i+=2) - { - if (!strcmp(attr[i], "href")) - { - found = 1; - if (ri->link != NULL) - free(ri->link); - ri->link = strdup(attr[i+1]); - striplt(ri->link); - } - } - + NextDBOperation(IO, RSSSaveMessage); + return eSendMore; } +} +void RSSQueueSaveMessage(struct CtdlMessage *Msg, struct recptypes *recp, StrBuf *MsgGUID, StrBuf *MessageBody, rss_aggregator *Cfg) +{ + networker_save_message *Ctx; + Ctx = (networker_save_message *) malloc(sizeof(networker_save_message)); + memset(Ctx, 0, sizeof(networker_save_message)); + + Ctx->MsgGUID = MsgGUID; + Ctx->Message = MessageBody; + Ctx->Msg = Msg; + Ctx->Cfg = Cfg; + Ctx->recp = recp; + Ctx->IO.Data = Ctx; + Ctx->IO.CitContext = CloneContext(CC); + Ctx->IO.Terminate = FreeNetworkSaveMessage; + Ctx->IO.ShutdownAbort = AbortNetworkSaveMessage; + QueueDBOperation(&Ctx->IO, RSS_FetchNetworkUsetableEntry); } -void rss_xml_end(void *data, const char *supplied_el) { - rsscollection *rssc = (rsscollection*) data; - rss_item *ri = rssc->Item; - 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); - } -// CtdlLogPrintf(0, "RSS: END %s...\n", el); +/* + * Commit a fetched and parsed RSS item to disk + */ +void rss_save_item(rss_item *ri, rss_aggregator *Cfg) +{ - 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); - } + struct MD5Context md5context; + u_char rawdigest[MD5_DIGEST_LEN]; + struct CtdlMessage *msg; + struct recptypes *recp = NULL; + int msglen = 0; + StrBuf *Message; + StrBuf *guid; + StrBuf *Buf; - if ( (rssc->Cfg->ItemType == RSS_RSS) && - (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) { - if (ri->guid != NULL) free(ri->guid); - striplt(ri->chardata); - ri->guid = strdup(ri->chardata); + recp = (struct recptypes *) malloc(sizeof(struct recptypes)); + if (recp == NULL) return; + memset(recp, 0, sizeof(struct recptypes)); + Buf = NewStrBufDup(Cfg->rooms); + recp->recp_room = SmashStrBuf(&Buf); + recp->num_room = Cfg->roomlist_parts; + recp->recptypes_magic = RECPTYPES_MAGIC; + + Cfg->RefCount ++; + /* 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) { + StrBufSpaceToBlank(ri->guid); + StrBufTrim(ri->guid); + guid = NewStrBufPlain(HKEY("rss/")); + StrBufAppendBuf(guid, ri->guid, 0); } - else if ( (rssc->Cfg->ItemType == RSS_ATOM) && - (!strcasecmp(el, "id")) && (ri->chardata != NULL) ) { - if (ri->guid != NULL) free(ri->guid); - striplt(ri->chardata); - ri->guid = strdup(ri->chardata); + else { + MD5Init(&md5context); + if (ri->title != NULL) { + MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->title), StrLength(ri->title)); + } + if (ri->link != NULL) { + MD5Update(&md5context, (const unsigned char*)ChrPtr(ri->link), StrLength(ri->link)); + } + MD5Final(rawdigest, &md5context); + guid = NewStrBufPlain(NULL, MD5_DIGEST_LEN * 2 + 12 /* _rss2ctdl*/); + StrBufHexEscAppend(guid, NULL, rawdigest, MD5_DIGEST_LEN); + StrBufAppendBufPlain(guid, HKEY("_rss2ctdl"), 0); } - else if ( (rssc->Cfg->ItemType == RSS_RSS) && (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) { - if (ri->link != NULL) free(ri->link); - striplt(ri->chardata); - ri->link = strdup(ri->chardata); - } + /* translate Item into message. */ + syslog(LOG_DEBUG, "RSS: translating item...\n"); + if (ri->description == NULL) ri->description = NewStrBufPlain(HKEY("")); + StrBufSpaceToBlank(ri->description); + 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; - else if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) { - if (ri->title != NULL) free(ri->title); - striplt(ri->chardata); - ri->title = strdup(ri->chardata); + if (ri->guid != NULL) { + msg->cm_fields['E'] = strdup(ChrPtr(ri->guid)); } - else if ((rssc->Cfg->ItemType == RSS_ATOM) && - (!strcasecmp(el, "content")) && - (ri->chardata != NULL) ) { - if (ri->description != NULL) free(ri->description); - ri->description = strdup(ri->chardata); - } - else if ( (rssc->Cfg->ItemType == RSS_RSS) && - (!strcasecmp(el, "description")) && - (ri->chardata != NULL) ) { - if (ri->description != NULL) free(ri->description); - ri->description = strdup(ri->chardata); + 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); + + FromAt = strchr(ChrPtr(ri->author_or_creator), '@') != NULL; + if (!FromAt && StrLength (ri->author_email) > 0) + { + StrBufRFC2047encode(&Encoded, ri->author_or_creator); + msg->cm_fields['A'] = SmashStrBuf(&Encoded); + msg->cm_fields['P'] = SmashStrBuf(&ri->author_email); + } + else + { + if (FromAt) + msg->cm_fields['P'] = SmashStrBuf(&ri->author_or_creator); + else + { + StrBufRFC2047encode(&Encoded, ri->author_or_creator); + msg->cm_fields['A'] = SmashStrBuf(&Encoded); + msg->cm_fields['P'] = strdup("rss@localhost"); + } + } } - - else if ((rssc->Cfg->ItemType == RSS_ATOM) && - ((!strcasecmp(el, "published")) || - (!strcasecmp(el, "updated"))) && - (ri->chardata != NULL) ) { - striplt(ri->chardata); - ri->pubdate = rdf_parsedate(ri->chardata); + else { + msg->cm_fields['A'] = strdup("rss"); } + msg->cm_fields['N'] = strdup(NODENAME); + if (ri->title != NULL) { + long len; + char *Sbj; + StrBuf *Encoded, *QPEncoded; - else if ((rssc->Cfg->ItemType == RSS_RSS) && - ((!strcasecmp(el, "pubdate")) || - (!strcasecmp(el, "date"))) && - (ri->chardata != NULL) ) { - striplt(ri->chardata); - ri->pubdate = rdf_parsedate(ri->chardata); - } - - else if ((rssc->Cfg->ItemType == RSS_RSS) && - ((!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); - } + QPEncoded = NULL; + StrBufSpaceToBlank(ri->title); + len = StrLength(ri->title); + Sbj = html_to_ascii(ChrPtr(ri->title), len, 512, 0); + len = strlen(Sbj); + if (Sbj[len - 1] == '\n') + { + len --; + Sbj[len] = '\0'; + } + Encoded = NewStrBufPlain(Sbj, len); + free(Sbj); - else if ((rssc->Cfg->ItemType == RSS_ATOM) && - (!strcasecmp(el, "name")) && - (ri->chardata != NULL) ) { - if (ri->author_or_creator != NULL) free(ri->author_or_creator); - striplt(ri->chardata); - ri->author_or_creator = strdup(ri->chardata); - } - else if ((rssc->Cfg->ItemType == RSS_ATOM) && - (!strcasecmp(el, "uri")) && - (ri->chardata != NULL) ) { - if (ri->author_url != NULL) free(ri->author_url); - striplt(ri->chardata); - ri->author_url = strdup(ri->chardata); - } + StrBufTrim(Encoded); + StrBufRFC2047encode(&QPEncoded, Encoded); - else if ((rssc->Cfg->ItemType == RSS_RSS) && - !strcasecmp(el, "item")) { - --ri->item_tag_nesting; - rss_save_item(rssc); + msg->cm_fields['U'] = SmashStrBuf(&QPEncoded); + FreeStrBuf(&Encoded); } - else if ((rssc->Cfg->ItemType == RSS_ATOM) && - !strcasecmp(el, "entry")) { - --ri->item_tag_nesting; - rss_save_item(rssc); + 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("")); - else if ( (!strcasecmp(el, "rss")) || - (!strcasecmp(el, "rdf")) ) { -// CtdlLogPrintf(CTDL_DEBUG, "End of feed detected. Closing parser.\n"); - ri->done_parsing = 1; - } +#if 0 /* temporarily disable shorter urls. */ + msg->cm_fields[TMP_SHORTER_URLS] = GetShorterUrls(ri->description); +#endif - if (ri->chardata_len > 0) { - free(ri->chardata); - ri->chardata = 0; - ri->chardata_len = 0; - } + msglen += 1024 + StrLength(ri->link) + StrLength(ri->description) ; -} + Message = NewStrBufPlain(NULL, StrLength(ri->description)); + StrBufPlain(Message, HKEY( + "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n" + "\n")); +#if 0 /* disable shorter url for now. */ + msg->cm_fields[TMP_SHORTER_URL_OFFSET] = StrLength(Message); +#endif + StrBufAppendBuf(Message, ri->description, 0); + StrBufAppendBufPlain(Message, HKEY("

\n"), 0); -/* - * This callback stores up the data which appears in between tags. - */ -void rss_xml_cdata_start(void *data) { - rsscollection *rssc = (rsscollection*) data; - rss_item *ri = rssc->Item; + AppendLink(Message, ri->link, ri->linkTitle, NULL); + AppendLink(Message, ri->reLink, ri->reLinkTitle, "Reply to this"); + StrBufAppendBufPlain(Message, HKEY("\n"), 0); - ri->CData = NewStrBuf(); + RSSQueueSaveMessage(msg, recp, guid, Message, Cfg); } -void rss_xml_cdata_end(void *data) { - rsscollection *rssc = (rsscollection*) data; - rss_item *ri = rssc->Item; - ri->description = SmashStrBuf(&ri->CData); -} -void rss_xml_chardata(void *data, const XML_Char *s, int len) { - rsscollection *rssc = (rsscollection*) data; - rss_item *ri = rssc->Item; - int old_len; - int new_len; - char *new_buffer; - - if (ri->CData != NULL) { - StrBufAppendBufPlain (ri->CData, s, len, 0); - } else { - 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; - } - } -} /* - * Callback function for passing libcurl's output to expat for parsing + * Begin a feed parse */ -size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream) +int rss_do_fetching(rss_aggregator *Cfg) { - XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0); - return (size*nmemb); -} - + rss_item *ri; + + time_t now; + AsyncIO *IO; + now = time(NULL); -/* - * Begin a feed parse - */ -void rss_do_fetching(rssnetcfg *Cfg) { - rsscollection rssc; - rss_item ri; - XML_Parser xp; + if ((Cfg->next_poll != 0) && (now < Cfg->next_poll)) + return 0; + Cfg->RefCount = 1; - CURL *curl; - CURLcode res; - char errmsg[1024] = ""; + ri = (rss_item*) malloc(sizeof(rss_item)); + memset(ri, 0, sizeof(rss_item)); + Cfg->Item = ri; + IO = &Cfg->IO; + IO->CitContext = CloneContext(CC); + IO->Data = Cfg; - rssc.Item = &ri; - rssc.Cfg = Cfg; - CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", Cfg->url); + syslog(LOG_DEBUG, "Fetching RSS feed <%s>\n", ChrPtr(Cfg->Url)); + ParseURL(&IO->ConnectMe, Cfg->Url, 80); + CurlPrepareURL(IO->ConnectMe); - curl = curl_easy_init(); - if (!curl) { - CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n"); - return; + if (! evcurl_init(IO, +// Ctx, + NULL, + "Citadel RSS Client", + ParseRSSReply, + RSSAggregatorTerminate)) + { + syslog(LOG_DEBUG, "Unable to initialize libcurl.\n"); + return 0; } - xp = XML_ParserCreateNS("UTF-8", ':'); - if (!xp) { - CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n"); - curl_easy_cleanup(curl); - return; - } + evcurl_handle_start(IO); + return 1; +} - curl_easy_setopt(curl, CURLOPT_URL, Cfg->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); -#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 */ - 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); - } - memset(&ri, 0, sizeof(rss_item)); - ri.roomlist = Cfg->rooms; -#ifdef HAVE_ICONV -#if 0 - XML_SetUnknownEncodingHandler(xp, - handle_unknown_xml_encoding, - NULL); -#endif -#endif - XML_SetElementHandler(xp, rss_xml_start, rss_xml_end); - XML_SetCharacterDataHandler(xp, rss_xml_chardata); - XML_SetUserData(xp, &rssc); - XML_SetCdataSectionHandler(xp, - rss_xml_cdata_start, - rss_xml_cdata_end); - 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); - } +void DeleteRssCfg(void *vptr) +{ + rss_aggregator *rncptr = (rss_aggregator *)vptr; - if (CtdlThreadCheckStop()) - goto shutdown ; + FreeStrBuf(&rncptr->Url); + FreeStrBuf(&rncptr->rooms); + FreeStrBuf(&rncptr->CData); + FreeStrBuf(&rncptr->Key); - if (ri.done_parsing == 0) - XML_Parse(xp, "", 0, 1); + DeleteHash(&rncptr->OtherQRnumbers); + if (rncptr->Item != NULL) + { + FreeStrBuf(&rncptr->Item->guid); + FreeStrBuf(&rncptr->Item->title); + FreeStrBuf(&rncptr->Item->link); + FreeStrBuf(&rncptr->Item->linkTitle); + FreeStrBuf(&rncptr->Item->reLink); + FreeStrBuf(&rncptr->Item->reLinkTitle); + FreeStrBuf(&rncptr->Item->description); + FreeStrBuf(&rncptr->Item->channel_title); + FreeStrBuf(&rncptr->Item->author_or_creator); + FreeStrBuf(&rncptr->Item->author_url); + FreeStrBuf(&rncptr->Item->author_email); + + free(rncptr->Item); + } + free(rncptr); +} + +eNextState RSSAggregatorTerminate(AsyncIO *IO) +{ + rss_aggregator *rncptr = (rss_aggregator *)IO->Data; + HashPos *At; + long HKLen; + const char *HK; + void *vData; + + citthread_mutex_lock(&RSSQueueMutex); + rncptr->RefCount --; + if (rncptr->RefCount == 0) + { + UnlinkRSSAggregator(rncptr); - CtdlLogPrintf(CTDL_ALERT, "RSS: XML Status [%s] \n", - XML_ErrorString( - XML_GetErrorCode(xp))); + } + citthread_mutex_unlock(&RSSQueueMutex); +/* + At = GetNewHashPos(RSSFetchUrls, 0); -shutdown: - curl_easy_cleanup(curl); - XML_ParserFree(xp); + citthread_mutex_lock(&RSSQueueMutex); + GetHashPosFromKey(RSSFetchUrls, SKEY(rncptr->Url), At); + GetHashPos(RSSFetchUrls, At, &HKLen, &HK, &vData); + DeleteEntryFromHash(RSSFetchUrls, At); + citthread_mutex_unlock(&RSSQueueMutex); - flush_rss_ite(&ri); + DeleteHashPos(&At); +*/ + return eAbort; } - /* * Scan a room's netconfig to determine whether it is requesting any RSS feeds */ void rssclient_scan_room(struct ctdlroom *qrbuf, void *data) { + StrBuf *CfgData; + StrBuf *CfgType; + StrBuf *Line; + rss_room_counter *Count = NULL; + struct stat statbuf; char filename[PATH_MAX]; - char buf[1024]; - char instr[32]; - FILE *fp; - char feedurl[256]; - rssnetcfg *rncptr = NULL; - rssnetcfg *use_this_rncptr = NULL; - int len = 0; - char *ptr = NULL; + int fd; + int Done; + rss_aggregator *rncptr = NULL; + rss_aggregator *use_this_rncptr = NULL; + void *vptr; + const char *CfgPtr, *lPtr; + const char *Err; + + citthread_mutex_lock(&RSSQueueMutex); + if (GetHash(RSSQueueRooms, LKEY(qrbuf->QRnumber), &vptr)) + { + syslog(LOG_DEBUG, + "rssclient: [%ld] %s already in progress.\n", + qrbuf->QRnumber, + qrbuf->QRname); + citthread_mutex_unlock(&RSSQueueMutex); + return; + } + citthread_mutex_unlock(&RSSQueueMutex); assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir); @@ -860,78 +548,122 @@ void rssclient_scan_room(struct ctdlroom *qrbuf, void *data) return; /* Only do net processing for rooms that have netconfigs */ - fp = fopen(filename, "r"); - if (fp == NULL) { + fd = open(filename, 0); + if (fd <= 0) { + //syslog(LOG_DEBUG, "rssclient: %s no config.\n", qrbuf->QRname); 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 = (rssnetcfg *) malloc(sizeof(rssnetcfg)); - rncptr->ItemType = RSS_UNSET; - 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; - } - } - } + if (CtdlThreadCheckStop()) + return; + if (fstat(fd, &statbuf) == -1) { + syslog(LOG_DEBUG, "ERROR: could not stat configfile '%s' - %s\n", + filename, strerror(errno)); + return; + } + if (CtdlThreadCheckStop()) + return; + CfgData = NewStrBufPlain(NULL, statbuf.st_size + 1); + if (StrBufReadBLOB(CfgData, &fd, 1, statbuf.st_size, &Err) < 0) { + close(fd); + FreeStrBuf(&CfgData); + syslog(LOG_DEBUG, "ERROR: reading config '%s' - %s
\n", + filename, strerror(errno)); + return; + } + close(fd); + if (CtdlThreadCheckStop()) + return; + + CfgPtr = NULL; + CfgType = NewStrBuf(); + Line = NewStrBufPlain(NULL, StrLength(CfgData)); + Done = 0; + while (!Done) + { + Done = StrBufSipLine(Line, CfgData, &CfgPtr) == 0; + if (StrLength(Line) > 0) + { + lPtr = NULL; + StrBufExtract_NextToken(CfgType, Line, &lPtr, '|'); + if (!strcasecmp("rssclient", ChrPtr(CfgType))) + { + if (Count == NULL) + { + Count = malloc(sizeof(rss_room_counter)); + Count->count = 0; + } + Count->count ++; + rncptr = (rss_aggregator *) malloc(sizeof(rss_aggregator)); + memset (rncptr, 0, sizeof(rss_aggregator)); + rncptr->roomlist_parts = 1; + rncptr->Url = NewStrBuf(); + StrBufExtract_NextToken(rncptr->Url, Line, &lPtr, '|'); + + citthread_mutex_lock(&RSSQueueMutex); + GetHash(RSSFetchUrls, SKEY(rncptr->Url), &vptr); + use_this_rncptr = (rss_aggregator *)vptr; + if (use_this_rncptr != NULL) + { + /* mustn't attach to an active session */ + if (use_this_rncptr->RefCount > 0) + { + DeleteRssCfg(rncptr); + Count->count--; + } + else + { + long *QRnumber; + StrBufAppendBufPlain(use_this_rncptr->rooms, + qrbuf->QRname, + -1, 0); + if (use_this_rncptr->roomlist_parts == 1) + { + use_this_rncptr->OtherQRnumbers = NewHash(1, lFlathash); + } + QRnumber = (long*)malloc(sizeof(long)); + *QRnumber = qrbuf->QRnumber; + Put(use_this_rncptr->OtherQRnumbers, LKEY(qrbuf->QRnumber), QRnumber, NULL); + use_this_rncptr->roomlist_parts++; + } + citthread_mutex_unlock(&RSSQueueMutex); + continue; + } + citthread_mutex_unlock(&RSSQueueMutex); + + rncptr->ItemType = RSS_UNSET; + + rncptr->rooms = NewStrBufPlain(qrbuf->QRname, -1); + + citthread_mutex_lock(&RSSQueueMutex); + Put(RSSFetchUrls, SKEY(rncptr->Url), rncptr, DeleteRssCfg); + citthread_mutex_unlock(&RSSQueueMutex); } - + } } - - fclose(fp); - + if (Count != NULL) + { + Count->QRnumber = qrbuf->QRnumber; + citthread_mutex_lock(&RSSQueueMutex); + syslog(LOG_DEBUG, "rssclient: [%ld] %s now starting.\n", + qrbuf->QRnumber, qrbuf->QRname); + Put(RSSQueueRooms, LKEY(qrbuf->QRnumber), Count, NULL); + citthread_mutex_unlock(&RSSQueueMutex); + } + FreeStrBuf(&CfgData); + FreeStrBuf(&CfgType); + FreeStrBuf(&Line); } /* * Scan for rooms that have RSS client requests configured */ -void *rssclient_scan(void *args) { - static time_t last_run = 0L; +void rssclient_scan(void) { 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 ); + rss_aggregator *rptr = NULL; + void *vrptr = NULL; + HashPos *it; + long len; + const char *Key; /* * This is a simple concurrency check to make sure only one rssclient run @@ -939,29 +671,35 @@ void *rssclient_scan(void *args) { * 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()) { - rss_do_fetching(rnclist); - rptr = rnclist; - rnclist = rnclist->next; - if (rptr->rooms != NULL) free(rptr->rooms); - free(rptr); + citthread_mutex_lock(&RSSQueueMutex); + + it = GetNewHashPos(RSSFetchUrls, 0); + while (GetNextHashPos(RSSFetchUrls, it, &len, &Key, &vrptr) && + (vrptr != NULL)) { + rptr = (rss_aggregator *)vrptr; + if (rptr->RefCount == 0) + if (!rss_do_fetching(rptr)) + UnlinkRSSAggregator(rptr); } + DeleteHashPos(&it); + citthread_mutex_unlock(&RSSQueueMutex); - CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n"); - last_run = time(NULL); + syslog(LOG_DEBUG, "rssclient ended\n"); 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 rss_cleanup(void) +{ + citthread_mutex_destroy(&RSSQueueMutex); + DeleteHash(&RSSFetchUrls); + DeleteHash(&RSSQueueRooms); } @@ -969,9 +707,12 @@ CTDL_MODULE_INIT(rssclient) { if (threading) { - CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version()); - CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0); - } - /* return our Subversion id for the Log */ - return "rssclient"; + citthread_mutex_init(&RSSQueueMutex, NULL); + RSSQueueRooms = NewHash(1, lFlathash); + RSSFetchUrls = NewHash(1, NULL); + syslog(LOG_INFO, "%s\n", curl_version()); + CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER); + CtdlRegisterCleanupHook(rss_cleanup); + } + return "rssclient"; }