[tracker/subtree-crawling: 1/4] libtracker-miner: Make crawler take a maximum recursion depth

[Carlos Garnacho <carlosg src gnome org>]

commit cc56e0c5b116f55456f87b4d8dac08205d22918c
Author: Carlos Garnacho <carlosg gnome org>
Date:   Sat Mar 22 23:09:11 2014 +0100

    libtracker-miner: Make crawler take a maximum recursion depth
    
    TrackerCrawler gets now the maximum depth to recurse into as a parameter
    to tracker_crawler_start(), this can be used to divide crawling into
    units bigger than individual folders. A value of -1 can be passed
    to have it recurse with no restrictions as with the previous recurse=TRUE
    argument.

 src/libtracker-miner/tracker-crawler.c        |   35 +++++++++---------------
 src/libtracker-miner/tracker-crawler.h        |    2 +-
 src/libtracker-miner/tracker-file-notifier.c  |    2 +-
 tests/libtracker-miner/tracker-crawler-test.c |   14 +++++-----
 4 files changed, 22 insertions(+), 31 deletions(-)
---
diff --git a/src/libtracker-miner/tracker-crawler.c b/src/libtracker-miner/tracker-crawler.c
index 0978996..d4a1dba 100644
--- a/src/libtracker-miner/tracker-crawler.c
+++ b/src/libtracker-miner/tracker-crawler.c
@@ -55,7 +55,7 @@ struct DirectoryProcessingData {
 struct DirectoryRootInfo {
        GFile *directory;
        GNode *tree;
-       guint recurse : 1;
+       gint max_depth;
 
        GQueue *directory_processing_queue;
 
@@ -79,8 +79,6 @@ struct TrackerCrawlerPrivate {
 
        gchar          *file_attributes;
 
-       gboolean        recurse;
-
        /* Statistics */
        GTimer         *timer;
 
@@ -370,9 +368,9 @@ directory_processing_data_add_child (DirectoryProcessingData *data,
 }
 
 static DirectoryRootInfo *
-directory_root_info_new (GFile    *file,
-                         gboolean  recurse,
-                         gchar    *file_attributes)
+directory_root_info_new (GFile *file,
+                         gint   max_depth,
+                         gchar *file_attributes)
 {
        DirectoryRootInfo *info;
        DirectoryProcessingData *dir_info;
@@ -380,7 +378,7 @@ directory_root_info_new (GFile    *file,
        info = g_slice_new0 (DirectoryRootInfo);
 
        info->directory = g_object_ref (file);
-       info->recurse = recurse;
+       info->max_depth = max_depth;
        info->directory_processing_queue = g_queue_new ();
 
        info->tree = g_node_new (g_object_ref (file));
@@ -461,19 +459,13 @@ process_func (gpointer data)
        }
 
        if (dir_data) {
-               /* One directory inside the tree hierarchy is being inspected */
-               if (!dir_data->was_inspected) {
-                       gboolean iterate;
+               gint depth = g_node_depth (dir_data->node) - 1;
+               gboolean iterate;
 
-                       if (G_NODE_IS_ROOT (dir_data->node)) {
-                               iterate = check_directory (crawler, info, dir_data->node->data);
-                       } else {
-                               /* Directory has been already checked in the block below, so
-                                * so obey the settings for the current directory root.
-                                */
-                               iterate = info->recurse;
-                       }
+               iterate = (info->max_depth >= 0) ? depth < info->max_depth : TRUE;
 
+               /* One directory inside the tree hierarchy is being inspected */
+               if (!dir_data->was_inspected) {
                        dir_data->was_inspected = TRUE;
 
                        /* Crawler may have been already stopped while we were waiting for the
@@ -511,7 +503,7 @@ process_func (gpointer data)
                                                                  g_object_ref (child_data->child));
                        }
 
-                       if (info->recurse && priv->is_running &&
+                       if (iterate && priv->is_running &&
                            child_node && child_data->is_dir) {
                                DirectoryProcessingData *child_dir_data;
 
@@ -839,7 +831,7 @@ file_enumerate_children (TrackerCrawler          *crawler,
 gboolean
 tracker_crawler_start (TrackerCrawler *crawler,
                        GFile          *file,
-                       gboolean        recurse)
+                       gint            max_depth)
 {
        TrackerCrawlerPrivate *priv;
        DirectoryRootInfo *info;
@@ -857,7 +849,6 @@ tracker_crawler_start (TrackerCrawler *crawler,
        }
 
        priv->was_started = TRUE;
-       priv->recurse = recurse;
 
        /* Time the event */
        if (priv->timer) {
@@ -874,7 +865,7 @@ tracker_crawler_start (TrackerCrawler *crawler,
        priv->is_running = TRUE;
        priv->is_finished = FALSE;
 
-       info = directory_root_info_new (file, recurse, priv->file_attributes);
+       info = directory_root_info_new (file, max_depth, priv->file_attributes);
        g_queue_push_tail (priv->directories, info);
 
        process_func_start (crawler);
diff --git a/src/libtracker-miner/tracker-crawler.h b/src/libtracker-miner/tracker-crawler.h
index 1bae0f6..669343f 100644
--- a/src/libtracker-miner/tracker-crawler.h
+++ b/src/libtracker-miner/tracker-crawler.h
@@ -73,7 +73,7 @@ GType           tracker_crawler_get_type     (void);
 TrackerCrawler *tracker_crawler_new          (void);
 gboolean        tracker_crawler_start        (TrackerCrawler *crawler,
                                               GFile          *file,
-                                              gboolean        recurse);
+                                             gint            max_depth);
 void            tracker_crawler_stop         (TrackerCrawler *crawler);
 void            tracker_crawler_pause        (TrackerCrawler *crawler);
 void            tracker_crawler_resume       (TrackerCrawler *crawler);
diff --git a/src/libtracker-miner/tracker-file-notifier.c b/src/libtracker-miner/tracker-file-notifier.c
index 96b8c1a..fdfda92 100644
--- a/src/libtracker-miner/tracker-file-notifier.c
+++ b/src/libtracker-miner/tracker-file-notifier.c
@@ -596,7 +596,7 @@ crawl_directories_start (TrackerFileNotifier *notifier)
                if ((flags & TRACKER_DIRECTORY_FLAG_IGNORE) == 0 &&
                    tracker_crawler_start (priv->crawler,
                                           directory,
-                                          (flags & TRACKER_DIRECTORY_FLAG_RECURSE) != 0)) {
+                                          (flags & TRACKER_DIRECTORY_FLAG_RECURSE) ? -1 : 1)) {
                        gchar *uri;
 
                        sparql_file_query_start (notifier, directory,
diff --git a/tests/libtracker-miner/tracker-crawler-test.c b/tests/libtracker-miner/tracker-crawler-test.c
index 69a9630..4f44ede 100644
--- a/tests/libtracker-miner/tracker-crawler-test.c
+++ b/tests/libtracker-miner/tracker-crawler-test.c
@@ -126,7 +126,7 @@ test_crawler_crawl (void)
 
        file = g_file_new_for_path (TEST_DATA_DIR);
 
-       started = tracker_crawler_start (crawler, file, TRUE);
+       started = tracker_crawler_start (crawler, file, -1);
 
        g_assert_cmpint (started, ==, 1);
 
@@ -153,7 +153,7 @@ test_crawler_crawl_interrupted (void)
 
        file = g_file_new_for_path (TEST_DATA_DIR);
 
-       started = tracker_crawler_start (crawler, file, TRUE);
+       started = tracker_crawler_start (crawler, file, -1);
 
        g_assert_cmpint (started, ==, 1);
 
@@ -175,7 +175,7 @@ test_crawler_crawl_nonexisting (void)
        crawler = tracker_crawler_new ();
        file = g_file_new_for_path (TEST_DATA_DIR "-idontexist");
 
-       started = tracker_crawler_start (crawler, file, TRUE);
+       started = tracker_crawler_start (crawler, file, -1);
 
        g_assert_cmpint (started, ==, 0);
 
@@ -200,7 +200,7 @@ test_crawler_crawl_recursive (void)
 
        file = g_file_new_for_path (TEST_DATA_DIR);
 
-       tracker_crawler_start (crawler, file, TRUE);
+       tracker_crawler_start (crawler, file, -1);
 
        g_main_loop_run (test.main_loop);
 
@@ -232,7 +232,7 @@ test_crawler_crawl_non_recursive (void)
 
        file = g_file_new_for_path (TEST_DATA_DIR);
 
-       tracker_crawler_start (crawler, file, FALSE);
+       tracker_crawler_start (crawler, file, 1);
 
        g_main_loop_run (test.main_loop);
 
@@ -270,7 +270,7 @@ test_crawler_crawl_n_signals (void)
 
        file = g_file_new_for_path (TEST_DATA_DIR);
 
-       tracker_crawler_start (crawler, file, TRUE);
+       tracker_crawler_start (crawler, file, -1);
 
        g_main_loop_run (test.main_loop);
 
@@ -308,7 +308,7 @@ test_crawler_crawl_n_signals_non_recursive (void)
 
        file = g_file_new_for_path (TEST_DATA_DIR);
 
-       tracker_crawler_start (crawler, file, FALSE);
+       tracker_crawler_start (crawler, file, 1);
 
        g_main_loop_run (test.main_loop);