Merge lp:~rainct/zeitgeist/ignore-ubuntuone-actor into lp:~zeitgeist/zeitgeist/bluebird

[Siegfried Gevatter]

<mhr3_> ok, so it seems we decided to skip the u1 events when indexing
<mhr3_> (they can be uniquely identified by actor)

[Michal Hruby (mhr3)]

Can we add an envvar which would control the ignoring (ignore by default)? Plus can we change the primary template when doing reindex, so we don't grab the U1 events in the first place?

review: Needs Fixing

[Mikkel Kamstrup Erlandsen (kamstrup)]

8 + // Blacklist Ubuntu One events...
9 + const gchar *actor;
10 + actor = zeitgeist_event_get_actor (event);
11 + if (strcmp(actor, "dbus://com.ubuntuone.SyncDaemon.service") == 0)
12 + return;
13 + if (strcmp(actor, "dbus://org.desktopcouch.CouchDB.service") == 0)
14 + return;

Can we move this out into a separate function, to make this hack less impacting on the main codepath?

Ala if (CheckEventBlacklisted(event)) return;

[Mikkel Kamstrup Erlandsen (kamstrup)]

Another benefit of a separate function is also that we can unit test the function directly as well.

[lp:~rainct/zeitgeist/ignore-ubuntuone-actor]

458. By Siegfried Gevatter on 2012-04-03

FTS++: Ubuntu One events blacklist:
 - Move blacklist check to separate function.
 - Exclude blacklisted events from the Zeitgeist query when reindexing.
 - Add environment variable to disable blacklisting.

459. By Siegfried Gevatter on 2012-04-03

Bump FTS++ index version.

[Michal Hruby (mhr3)]

Great job as always, thanks!

review: Approve

=== modified file 'extensions/fts++/controller.cpp'
--- extensions/fts++/controller.cpp	2012-03-26 15:09:54 +0000
+++ extensions/fts++/controller.cpp	2012-04-03 19:08:21 +0000
@@ -39,9 +39,23 @@
   GError *error = NULL;
   guint32 *event_ids;
   gint event_ids_size;
-  GPtrArray *templates = g_ptr_array_new ();
+  GPtrArray *templates = g_ptr_array_new_with_free_func (g_object_unref);
+  ZeitgeistEvent *event;
   ZeitgeistTimeRange *time_range = zeitgeist_time_range_new_anytime ();
 
+  if (g_getenv ("ZEITGEIST_FTS_DISABLE_EVENT_BLACKLIST") == NULL)
+  {
+    // Blacklist Ubuntu One events...
+
+    event = zeitgeist_event_new ();
+    zeitgeist_event_set_actor (event, "!dbus://com.ubuntuone.SyncDaemon.service");
+    g_ptr_array_add (templates, event);
+
+    event = zeitgeist_event_new ();
+    zeitgeist_event_set_actor (event, "!dbus://org.desktopcouch.CouchDB.service");
+    g_ptr_array_add (templates, event);
+  }
+
   g_debug ("asking reader for all events");
   event_ids = zeitgeist_db_reader_find_event_ids (zg_reader,
                                                   time_range,

=== modified file 'extensions/fts++/indexer.cpp'
--- extensions/fts++/indexer.cpp	2012-03-26 16:43:07 +0000
+++ extensions/fts++/indexer.cpp	2012-04-03 19:08:21 +0000
@@ -1301,8 +1301,25 @@
   g_assert (digest_size == NULL || *digest_size == HASH_LENGTH);
 }
 
+static bool
+CheckEventBlacklisted (ZeitgeistEvent *event)
+{
+  // Blacklist Ubuntu One events...
+  const gchar *actor;
+  actor = zeitgeist_event_get_actor (event);
+  if (g_strcmp0(actor, "dbus://com.ubuntuone.SyncDaemon.service") == 0)
+    return true;
+  if (g_strcmp0(actor, "dbus://org.desktopcouch.CouchDB.service") == 0)
+    return true;
+
+  return false;
+}
+
 void Indexer::IndexEvent (ZeitgeistEvent *event)
 {
+  if (blacklisting_enabled and CheckEventBlacklisted (event))
+    return;
+
   try
   {
     const gchar *val;

=== modified file 'extensions/fts++/indexer.h'
--- extensions/fts++/indexer.h	2012-03-19 21:42:52 +0000
+++ extensions/fts++/indexer.h	2012-04-03 19:08:21 +0000
@@ -29,7 +29,7 @@
 
 namespace ZeitgeistFTS {
 
-const std::string INDEX_VERSION = "3";
+const std::string INDEX_VERSION = "4";
 
 class Indexer
 {
@@ -48,6 +48,7 @@
   {
     const gchar *home_dir = g_get_home_dir ();
     home_dir_path = home_dir != NULL ? home_dir : "/home";
+    blacklisting_enabled = g_getenv ("ZEITGEIST_FTS_DISABLE_EVENT_BLACKLIST") == NULL;
   }
 
   ~Indexer ()
@@ -129,6 +130,7 @@
 
   guint                     clear_failed_id;
   std::string               home_dir_path;
+  bool                      blacklisting_enabled;
 };
 
 }

=== modified file 'extensions/fts++/test/test-indexer.cpp'
--- extensions/fts++/test/test-indexer.cpp	2012-03-20 12:25:59 +0000
+++ extensions/fts++/test/test-indexer.cpp	2012-04-03 19:08:21 +0000
@@ -270,6 +270,15 @@
   return event;
 }
 
+static void
+process_pending (Fixture *fix)
+{
+  while (zeitgeist_indexer_has_pending_tasks (fix->indexer))
+  {
+    zeitgeist_indexer_process_task (fix->indexer);
+  }
+}
+
 // Steals the event, ref it if you want to keep it
 static guint
 index_event (Fixture *fix, ZeitgeistEvent *event)
@@ -295,10 +304,7 @@
   zeitgeist_indexer_index_events (fix->indexer, events);
   g_ptr_array_unref (events);
 
-  while (zeitgeist_indexer_has_pending_tasks (fix->indexer))
-  {
-    zeitgeist_indexer_process_task (fix->indexer);
-  }
+  process_pending (fix);
 
   // sleep for 1 msec to make sure the next event will have a
   // different timestamp
@@ -1074,6 +1080,42 @@
   assert_nth_result_has_id (results, 2, event_id6);
 }
 
+static void
+test_index_ignore_ubuntu_one (Fixture *fix, gconstpointer data)
+{
+  guint matches;
+  ZeitgeistEvent *event;
+  GPtrArray *results;
+
+  // add test events to DBs
+  index_event (fix, create_test_event_simple ("ubuntuone:uuid", "failme"));
+  event = create_test_event_simple ("file:///nice%20uri", "failme");
+  zeitgeist_event_set_actor (event, "dbus://com.ubuntuone.SyncDaemon.service");
+  index_event (fix, event);
+
+  results = search_simple (fix, "failme", NULL,
+          ZEITGEIST_RESULT_TYPE_MOST_RECENT_EVENTS, &matches);
+
+  g_assert_cmpuint (results->len, ==, 0);
+
+  // disabling blacklisting
+  g_setenv ("ZEITGEIST_FTS_DISABLE_EVENT_BLACKLIST", "1", true);
+
+  // create a new FTS instance
+  zeitgeist_indexer_free (fix->indexer);
+  GError *error = NULL;
+  fix->indexer = zeitgeist_indexer_new (fix->db, &error);
+  g_assert (error == NULL);
+
+  // wait for it to rebuild the index
+  process_pending (fix);
+
+  results = search_simple (fix, "failme", NULL,
+          ZEITGEIST_RESULT_TYPE_MOST_RECENT_EVENTS, &matches);
+
+  g_assert_cmpuint (results->len, ==, 1); // we still don't want ubuntuone:uuid
+}
+
 G_BEGIN_DECLS
 
 static void discard_message (const gchar *domain,
@@ -1138,6 +1180,8 @@
   g_test_add ("/Zeitgeist/FTS/Indexer/Query/MostPopularSubjects", Fixture, 0,
               setup, test_query_most_popular_subjects, teardown);
   */
+  g_test_add ("/Zeitgeist/FTS/Indexer/Index/IgnoreUbuntuOne", Fixture, 0,
+              setup, test_index_ignore_ubuntu_one, teardown);
 
   // get rid of the "rebuilding index..." messages
   g_log_set_handler (NULL, G_LOG_LEVEL_MESSAGE, discard_message, NULL);