[tracker-miners/wip/carlosg/extract-native-uris] tracker-extract: Avoid non-native URIs

[Carlos Garnacho <carlosg src gnome org>]

commit 319930706c9700fb9325d6ca1bd20f815bc0a1e7
Author: Carlos Garnacho <carlosg gnome org>
Date:   Fri Oct 30 13:25:38 2020 +0100

    tracker-extract: Avoid non-native URIs
    
    The tracker-extract-3 process is only meant to deal with native files.
    Make it avoid non-native and other extraneous URIs, so we don't end
    up breaking assumptions down the line.
    
    Related: https://gitlab.gnome.org/GNOME/tracker-miners/-/issues/150

 src/tracker-extract/tracker-extract-decorator.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)
---
diff --git a/src/tracker-extract/tracker-extract-decorator.c b/src/tracker-extract/tracker-extract-decorator.c
index 9e6c1816b..b338168a6 100644
--- a/src/tracker-extract/tracker-extract-decorator.c
+++ b/src/tracker-extract/tracker-extract-decorator.c
@@ -251,6 +251,7 @@ decorator_next_item_cb (TrackerDecorator *decorator,
        GError *error = NULL;
        ExtractData *data;
        GTask *task;
+       GFile *file;
 
        priv = tracker_extract_decorator_get_instance_private (TRACKER_EXTRACT_DECORATOR (decorator));
        info = tracker_decorator_next_finish (decorator, result, &error);
@@ -281,10 +282,21 @@ decorator_next_item_cb (TrackerDecorator *decorator,
                return;
        }
 
+       file = g_file_new_for_uri (tracker_decorator_info_get_url (info));
+
+       if (!g_file_is_native (file)) {
+               g_warning ("URI '%s' is not native",
+                          tracker_decorator_info_get_url (info));
+               priv->n_extracting_files--;
+               tracker_decorator_info_unref (info);
+               decorator_get_next_file (decorator);
+               return;
+       }
+
        data = g_new0 (ExtractData, 1);
        data->decorator = decorator;
        data->decorator_info = info;
-       data->file = g_file_new_for_uri (tracker_decorator_info_get_url (info));
+       data->file = file;
        task = tracker_decorator_info_get_task (info);
 
        g_debug ("Extracting metadata for '%s'", tracker_decorator_info_get_url (info));