tracker r2069 - in branches/indexer-split: . src/libtracker-common src/tracker-indexer

[carlosg svn gnome org]

Author: carlosg
Date: Wed Aug 13 14:26:35 2008
New Revision: 2069
URL: http://svn.gnome.org/viewvc/tracker?rev=2069&view=rev

Log:
2008-08-12  Carlos Garnacho  <carlos imendio com>

        * src/libtracker-common/tracker-parser.c
        (tracker_parser_text_to_string): Check we're dealing with sane UTF8.
        * src/tracker-indexer/tracker-indexer.c (index_metadata_item): Handle
        the case where tracker_parser_text_to_string() returns NULL.
        (delete_item) (create_update_item): Add debug messages that give a
        better clue about what is it doing at that moment.
        (process_file): Remove debug message from here.


Modified:
   branches/indexer-split/ChangeLog
   branches/indexer-split/src/libtracker-common/tracker-parser.c
   branches/indexer-split/src/tracker-indexer/tracker-indexer.c

Modified: branches/indexer-split/src/libtracker-common/tracker-parser.c
==============================================================================
--- branches/indexer-split/src/libtracker-common/tracker-parser.c	(original)
+++ branches/indexer-split/src/libtracker-common/tracker-parser.c	Wed Aug 13 14:26:35 2008
@@ -353,6 +353,7 @@
 	gchar       *parsed_text;
 	gchar       *word = NULL;
 	guint32      i = 0;
+        gint         len;
 
         g_return_val_if_fail (language != NULL, NULL);
 
@@ -360,6 +361,13 @@
                 return NULL;
         }
 
+        len = strlen (txt);
+        len = MIN (len, 500);
+
+        if (!g_utf8_validate (txt, len, NULL)) {
+                return NULL;
+        }
+
         if (text_needs_pango (txt)) {
                 /* CJK text does not need stemming or other
                  * treatment.

Modified: branches/indexer-split/src/tracker-indexer/tracker-indexer.c
==============================================================================
--- branches/indexer-split/src/tracker-indexer/tracker-indexer.c	(original)
+++ branches/indexer-split/src/tracker-indexer/tracker-indexer.c	Wed Aug 13 14:26:35 2008
@@ -790,6 +790,11 @@
 						      tracker_field_get_filtered (field),
 						      tracker_field_get_filtered (field),
 						      tracker_field_get_delimited (field));
+
+	if (!parsed_value) {
+		return;
+	}
+
 	arr = g_strsplit (parsed_value, " ", -1);
 
 	for (i = 0; arr[i]; i++) {
@@ -1019,6 +1024,8 @@
 	tracker_indexer_module_file_get_uri (info->module, info->file, &dirname, &basename);
 	id = tracker_db_check_service (service_def, dirname, basename);
 
+	g_debug ("Processing item:'%s/%s'", dirname, basename);
+
 	/* FIXME: should check mtime and reindex if necessary */
 
 	if (id == 0) {
@@ -1120,6 +1127,8 @@
 		return;
 	}
 
+	g_debug ("Deleting item:'%s/%s'", dirname, basename);
+
 	if (!service_type || !service_type[0]) {
 		gchar *name;
 
@@ -1145,7 +1154,7 @@
 	g_free (basename);
 
 	if (service_id < 1) {
-		g_message ("Can not delete file, it doesnt exist in DB");
+		g_debug ("Can not delete file, it doesnt exist in DB");
 		return;
 	}
 
@@ -1414,8 +1423,6 @@
 {
 	TrackerMetadata *metadata;
 
-	g_debug ("Processing file:'%s'", info->file->path);
-
 	/* Set the current module */
 	g_free (indexer->private->current_module_name);
 	indexer->private->current_module_name = g_strdup (info->file->module_name);