[tracker-miners/wip/carlosg/shuffle-libtracker-miner: 108/116] libtracker-miner: Ensure basename utf8-ness in filters

[Carlos Garnacho <carlosg src gnome org>]

commit 2b18687cdb9a9bf525b85eb2e30913dfab99bf46
Author: Carlos Garnacho <carlosg gnome org>
Date:   Wed Aug 7 14:39:08 2019 +0200

    libtracker-miner: Ensure basename utf8-ness in filters
    
    Filenames with invalid UTF-8 may cause invalid writes on
    g_pattern_match(), as patterns like '*a*a' will require the reverse of
    the string, and g_utf8_strreverse() might write past boundaries if
    the invalid UTF-8 happens near the end of the given string. See
    https://gitlab.gnome.org/GNOME/glib/issues/1863.
    
    Ensure the UTF-8-ness of the string by replacing those invalid characters,
    they presumably are irrelevant for matching purposes (as the patterns
    express valid UTF-8). Also reverse the string in place, so it's slightly
    faster at checking all the patterns.
    
    This seems a frequent crash report in Fedora FAF along time, eg:
    https://retrace.fedoraproject.org/faf/reports/2559134/
    https://retrace.fedoraproject.org/faf/reports/1598399/

 src/libtracker-miner/tracker-indexing-tree.c | 23 ++++++++++++++++-------
 1 file changed, 16 insertions(+), 7 deletions(-)
---
diff --git a/src/libtracker-miner/tracker-indexing-tree.c b/src/libtracker-miner/tracker-indexing-tree.c
index 564ba56d4..8d8a00290 100644
--- a/src/libtracker-miner/tracker-indexing-tree.c
+++ b/src/libtracker-miner/tracker-indexing-tree.c
@@ -748,7 +748,9 @@ tracker_indexing_tree_file_matches_filter (TrackerIndexingTree *tree,
 {
        TrackerIndexingTreePrivate *priv;
        GList *filters;
-       gchar *basename;
+       gchar *basename, *str, *reverse;
+       gboolean match = FALSE;
+       gint len;
 
        g_return_val_if_fail (TRACKER_IS_INDEXING_TREE (tree), FALSE);
        g_return_val_if_fail (G_IS_FILE (file), FALSE);
@@ -757,6 +759,10 @@ tracker_indexing_tree_file_matches_filter (TrackerIndexingTree *tree,
        filters = priv->filter_patterns;
        basename = g_file_get_basename (file);
 
+       str = g_utf8_make_valid (basename, -1);
+       len = strlen (str);
+       reverse = g_utf8_strreverse (str, len);
+
        while (filters) {
                PatternData *data = filters->data;
 
@@ -768,18 +774,21 @@ tracker_indexing_tree_file_matches_filter (TrackerIndexingTree *tree,
                if (data->file &&
                    (g_file_equal (file, data->file) ||
                     g_file_has_prefix (file, data->file))) {
-                       g_free (basename);
-                       return TRUE;
+                       match = TRUE;
+                       break;
                }
 
-               if (g_pattern_match_string (data->pattern, basename)) {
-                       g_free (basename);
-                       return TRUE;
+               if (g_pattern_match (data->pattern, len, str, reverse)) {
+                       match = TRUE;
+                       break;
                }
        }
 
        g_free (basename);
-       return FALSE;
+       g_free (str);
+       g_free (reverse);
+
+       return match;
 }
 
 static gboolean