* rss_save_item(): encode sender & subject, so webcit will correctly display umlauts...

[citadel.git] / citadel / modules / rssclient / serv_rssclient.c

diff --git a/citadel/modules/rssclient/serv_rssclient.c b/citadel/modules/rssclient/serv_rssclient.c
index 9293f637ab87aa604aaa2e4c012180fc44dc99ec..3661f52e4bf057589c1b4843b169630883520ace 100644 (file)
--- a/citadel/modules/rssclient/serv_rssclient.c
+++ b/citadel/modules/rssclient/serv_rssclient.c
@@ -289,10 +289,10 @@ void rss_save_item(rsscollection *rssc) {
        int msglen = 0;
        rss_item *ri = rssc->Item;
 
-       CtdlLogPrintf(0, "RSS: saving item...\n");
        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;
@@ -334,7 +334,7 @@ void rss_save_item(rsscollection *rssc) {
        }
        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])) {
@@ -348,10 +348,35 @@ void rss_save_item(rsscollection *rssc) {
                msg->cm_anon_type = MES_NORMAL;
                msg->cm_format_type = FMT_RFC822;
 
+               if (ri->guid != NULL) {
+                       msg->cm_fields['E'] = strdup(ri->guid);
+               }
+
                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']);
+                       char *From;
+                       StrBuf *Encoded, *QPEncoded;
+                       StrBuf *UserName;
+                       StrBuf *EmailAddress;
+                       StrBuf *EncBuf;
+                       
+                       UserName = NewStrBuf();
+                       EmailAddress = NewStrBuf();
+                       EncBuf = NewStrBuf();
+
+                       From = html_to_ascii(ri->author_or_creator,
+                                            strlen(ri->author_or_creator), 
+                                            512, 0);
+
+                       Encoded = NewStrBufPlain(From, -1);
+                       free(From);
+                       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");
@@ -359,8 +384,21 @@ void rss_save_item(rsscollection *rssc) {
 
                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']);
+                       long len;
+                       char *Sbj;
+                       StrBuf *Encoded, *QPEncoded;
+
+                       QPEncoded = NULL;
+                       len = strlen(ri->title);
+                       Sbj = html_to_ascii(ri->title, len, 512, 0);
+                       Encoded = NewStrBufPlain(Sbj, -1);
+                       free(Sbj);
+
+                       StrBufTrim(Encoded);
+                       StrBufRFC2047encode(&QPEncoded, Encoded);
+
+                       msg->cm_fields['U'] = SmashStrBuf(&QPEncoded);
+                       FreeStrBuf(&Encoded);
                }
                msg->cm_fields['T'] = malloc(64);
                snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
@@ -452,6 +490,8 @@ void flush_rss_ite(rss_item *ri)
        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 */
@@ -475,14 +515,19 @@ void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
                strcpy(el, ++sep);
        }
 
-       if ((rssc->Cfg->ItemType == RSS_UNSET) &&
-           !strcasecmp(el, "rss")) 
+       if ((rssc->Cfg->ItemType == RSS_UNSET) && !strcasecmp(el, "rss")) 
        {
+               CtdlLogPrintf(9, "RSS: This is an RSS feed.\n");
                rssc->Cfg->ItemType = RSS_RSS;
        }
-       else if ((rssc->Cfg->ItemType == RSS_UNSET) &&
-           !strcasecmp(el, "feed")) 
+       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) &&
@@ -494,7 +539,6 @@ void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
        else if ( (rssc->Cfg->ItemType == RSS_ATOM) &&
                 !strcasecmp(el, "entry")) 
        { /* Atom feed... */
-               CtdlLogPrintf(0, "RSS: found atom...\n");
                ++ri->item_tag_nesting;
                flush_rss_ite(ri);
        }
@@ -864,8 +908,6 @@ void *rssclient_scan(void *args) {
        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
@@ -893,6 +935,7 @@ void *rssclient_scan(void *args) {
                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;
 }