Improve handling of feeds with missing data

Previously the description was used in the treeview if the title
was not present, however this is not possible in some cases, as
it may contain markup.
The the RSS code is changed so now if there is no title, the
description is used as the title, but with the html stripped from
it. We also have to consider that some description elements may
simply contain an html tag (and therefore no text). In that
case the URI is used as the title.

diff --git a/extensions/feed-panel/feed-parse.c b/extensions/feed-panel/feed-parse.c
index 4c09dc647eb5ab94e0b688bd56790122174faaa6..1c2cd3f32569943f711d7c62250fd795a990a6d5 100644 (file)
--- a/extensions/feed-panel/feed-parse.c
+++ b/extensions/feed-panel/feed-parse.c
@@ -33,35 +33,36 @@ feed_get_element_string (FeedParser* fparser)
     return (gchar*)xmlNodeListGetString (fparser->doc, node->children, 1);
 }
 
-void
+static void
 handle_markup_chars (void*          user_data,
                      const xmlChar* ch,
                      int            len)
 {
-    gchar** markup;
-    gchar* temp;
+    if (len > 0)
+    {
+        gchar** markup;
+        gchar* temp;
 
-    markup = (gchar**)user_data;
-    temp = g_strndup ((gchar*)ch, len);
-    *markup = (*markup) ? g_strconcat (*markup, temp, NULL) : g_strdup (temp);
-    g_free (temp);
+        markup = (gchar**)user_data;
+        temp = g_strndup ((gchar*)ch, len);
+        *markup = (*markup) ? g_strconcat (*markup, temp, NULL) : g_strdup (temp);
+        g_free (temp);
+    }
 }
 
 gchar*
-feed_get_element_markup (FeedParser* fparser)
+feed_remove_markup (gchar* markup)
 {
-    gchar* markup;
     const xmlChar* stag;
-
-    markup = feed_get_element_string (fparser);
-    if ((stag = xmlStrchr (BAD_CAST markup, '<')) && xmlStrchr (stag, '>'))
+    if (((stag = xmlStrchr (BAD_CAST markup, '<')) && xmlStrchr (stag, '>')) ||
+         xmlStrchr (BAD_CAST markup, '&'))
     {
         gchar* text = NULL;
         htmlSAXHandlerPtr psax;
 
         psax = g_new0 (htmlSAXHandler, 1);
         psax->characters = handle_markup_chars;
-        htmlSAXParseDoc (BAD_CAST markup, NULL, psax, &text);
+        htmlSAXParseDoc (BAD_CAST markup, "UTF-8", psax, &text);
         g_free (psax);
         g_free (markup);
         return text;
@@ -69,6 +70,15 @@ feed_get_element_markup (FeedParser* fparser)
     return markup;
 }
 
+gchar*
+feed_get_element_markup (FeedParser* fparser)
+{
+    gchar* markup;
+
+    markup = feed_get_element_string (fparser);
+    return feed_remove_markup (markup);
+}
+
 gint64
 feed_get_element_date (FeedParser* fparser)
 {

diff --git a/extensions/feed-panel/feed-parse.h b/extensions/feed-panel/feed-parse.h
index 84ae9f19d1d2a37b4b4f395c6b522977dc25ffca..31685feb3c8b96ec504ea603b131e62d78846bb3 100644 (file)
--- a/extensions/feed-panel/feed-parse.h
+++ b/extensions/feed-panel/feed-parse.h
@@ -57,6 +57,9 @@ typedef struct _FeedParser
 gchar*
 feed_get_element_string (FeedParser* fparser);
 
+gchar*
+feed_remove_markup (gchar* markup);
+
 gchar*
 feed_get_element_markup (FeedParser* fparser);
 

diff --git a/extensions/feed-panel/feed-rss.c b/extensions/feed-panel/feed-rss.c
index 03843df6c4f3f7f6a6dd65b3de01843975f67ea2..e80226342d5ea08eb4b1d490fe8b0088af684cb7 100644 (file)
--- a/extensions/feed-panel/feed-rss.c
+++ b/extensions/feed-panel/feed-rss.c
@@ -137,11 +137,30 @@ rss_postparse_item (FeedParser* fparser)
         * Verify that the required RSS elements are added
         * (as per the spec)
         */
-        if (!katze_item_get_name (fparser->item) &&
-            !katze_item_get_text (fparser->item))
+        if (!katze_item_get_name (fparser->item))
         {
-            feed_parser_set_error (fparser, FEED_PARSE_ERROR_MISSING_ELEMENT,
-                                   _("Failed to find required RSS \"item\" elements in XML data."));
+            gchar* desc;
+
+            desc = (gchar*)katze_item_get_text (fparser->item);
+            if (!desc)
+            {
+                feed_parser_set_error (fparser, FEED_PARSE_ERROR_MISSING_ELEMENT,
+                                       _("Failed to find required RSS \"item\" elements in XML data."));
+            }
+            else
+            {
+                desc = feed_remove_markup (g_strdup (desc));
+                if (desc)
+                {
+                    katze_item_set_name (fparser->item, desc);
+                    g_free (desc);
+                }
+                else
+                {
+                    if ((desc = (gchar*)katze_item_get_uri (fparser->item)))
+                        katze_item_set_name (fparser->item, desc);
+                }
+            }
         }
     }