[tracker-miners/wip/carlosg/miner-fs-ancillary-files: 8/10] tracker-miner-fs: Repurpose crawling-interval setting

[Carlos Garnacho <carlosg src gnome org>]

commit f316415a026157103b72a4880a01014f51c9c067
Author: Carlos Garnacho <carlosg gnome org>
Date:   Sun Nov 8 20:37:58 2020 +0100

    tracker-miner-fs: Repurpose crawling-interval setting
    
    This setting (in days) was used to check whether tracker-miner-fs
    would crawl on startup, after checking a timestamp file. However, we
    are making crawling good and fast, and we don't like outdated databases.
    
    Make this setting set up a timeout for long-running sessions instead,
    and always perform crawling on startup.

 src/miners/fs/tracker-main.c        |  56 +---------------
 src/miners/fs/tracker-miner-files.c | 125 +++++++++++-------------------------
 src/miners/fs/tracker-miner-files.h |   4 --
 3 files changed, 37 insertions(+), 148 deletions(-)
---
diff --git a/src/miners/fs/tracker-main.c b/src/miners/fs/tracker-main.c
index c4574c2fd..03041685b 100644
--- a/src/miners/fs/tracker-main.c
+++ b/src/miners/fs/tracker-main.c
@@ -52,8 +52,6 @@
        "\n" \
        "  http://www.gnu.org/licenses/gpl.txt\n";
 
-#define SECONDS_PER_DAY 60 * 60 * 24
-
 #define DBUS_NAME_SUFFIX "Tracker3.Miner.Files"
 #define DBUS_PATH "/org/freedesktop/Tracker3/Miner/Files"
 #define LOCALE_FILENAME "locale-for-miner-apps.txt"
@@ -68,7 +66,6 @@ static gboolean no_daemon;
 static gchar *eligible;
 static gboolean version;
 static guint miners_timeout_id = 0;
-static gboolean do_crawling = FALSE;
 static gchar *domain_ontology_name = NULL;
 static gboolean dry_run = FALSE;
 
@@ -327,46 +324,6 @@ initialize_priority_and_scheduling (void)
        }
 }
 
-static gboolean
-should_crawl (TrackerMinerFiles *miner_files,
-              TrackerConfig     *config)
-{
-       gint crawling_interval;
-
-       crawling_interval = tracker_config_get_crawling_interval (config);
-
-       TRACKER_NOTE (CONFIG, g_message ("Checking whether to crawl file system based on configured crawling 
interval:"));
-
-       if (crawling_interval == -2) {
-               TRACKER_NOTE (CONFIG, g_message ("  Disabled"));
-               return FALSE;
-       } else if (crawling_interval == -1) {
-               TRACKER_NOTE (CONFIG, g_message ("  Maybe (depends on a clean last shutdown)"));
-               return TRUE;
-       } else if (crawling_interval == 0) {
-               TRACKER_NOTE (CONFIG, g_message ("  Forced"));
-               return TRUE;
-       } else {
-               guint64 then, now;
-
-               then = tracker_miner_files_get_last_crawl_done (miner_files);
-
-               if (then < 1) {
-                       return TRUE;
-               }
-
-               now = (guint64) time (NULL);
-
-               if (now < then + (crawling_interval * SECONDS_PER_DAY)) {
-                       TRACKER_NOTE (CONFIG, g_message ("  Postponed"));
-                       return FALSE;
-               } else {
-                       TRACKER_NOTE (CONFIG, g_message ("  (More than) %d days after last crawling, 
enabled", crawling_interval));
-                       return TRUE;
-               }
-       }
-}
-
 static void
 miner_do_start (TrackerMiner *miner)
 {
@@ -486,11 +443,6 @@ miner_finished_cb (TrackerMinerFS *fs,
                total_directories_found,
                total_files_found);
 
-       if (do_crawling && !dry_run) {
-               tracker_miner_files_set_last_crawl_done (TRACKER_MINER_FILES (fs),
-                                                        TRUE);
-       }
-
        cleanup_id = g_timeout_add_seconds (30, cleanup_cb, NULL);
 
        /* We're not sticking around for file updates, so stop
@@ -1101,11 +1053,6 @@ main (gint argc, gchar *argv[])
 
        g_free (dbus_name);
 
-       /* Check if we should crawl and if we should force mtime
-        * checking based on the config.
-        */
-       do_crawling = should_crawl (TRACKER_MINER_FILES (miner_files), config);
-
        g_signal_connect (miner_files, "started",
                          G_CALLBACK (miner_started_cb),
                          NULL);
@@ -1128,8 +1075,7 @@ main (gint argc, gchar *argv[])
                                                "CREATE SILENT GRAPH tracker:Video ",
                                                NULL, graphs_created_cb, miner_files);
 
-       if (do_crawling)
-               miner_start (miner_files, config);
+       miner_start (miner_files, config);
 
        initialize_signal_handler ();
 
diff --git a/src/miners/fs/tracker-miner-files.c b/src/miners/fs/tracker-miner-files.c
index 654759742..851d02ce5 100644
--- a/src/miners/fs/tracker-miner-files.c
+++ b/src/miners/fs/tracker-miner-files.c
@@ -50,7 +50,6 @@
 
 /* Stamp files to know crawling/indexing state */
 #define FIRST_INDEX_FILENAME          "first-index.txt"
-#define LAST_CRAWL_FILENAME           "last-crawl.txt"
 
 #define DEFAULT_GRAPH "tracker:FileSystem"
 
@@ -96,6 +95,7 @@ struct TrackerMinerFilesPrivate {
        GDBusConnection *connection;
 
        guint force_recheck_id;
+       guint crawl_interval_id;
 
        gboolean index_removable_devices;
        gboolean index_optical_discs;
@@ -765,6 +765,9 @@ miner_files_finalize (GObject *object)
                priv->force_recheck_id = 0;
        }
 
+       if (priv->crawl_interval_id != 0)
+               g_source_remove (priv->crawl_interval_id);
+
        if (priv->stale_volumes_check_id) {
                g_source_remove (priv->stale_volumes_check_id);
                priv->stale_volumes_check_id = 0;
@@ -1688,6 +1691,33 @@ miner_files_force_recheck_idle (gpointer user_data)
        return G_SOURCE_REMOVE;
 }
 
+static gboolean
+crawl_interval_idle (gpointer user_data)
+{
+       TrackerMinerFiles *miner_files = user_data;
+
+       tracker_miner_files_trigger_check (miner_files);
+
+       return G_SOURCE_CONTINUE;
+}
+
+static void
+tracker_miner_files_update_crawl_interval (TrackerMinerFiles *mf)
+{
+       guint interval;
+
+       if (mf->private->crawl_interval_id != 0)
+               g_source_remove (mf->private->crawl_interval_id);
+
+       interval = tracker_config_get_crawling_interval (mf->private->config);
+
+       if (interval > 0) {
+               mf->private->crawl_interval_id =
+                       g_timeout_add_seconds (interval * SECONDS_PER_DAY,
+                                              crawl_interval_idle, mf);
+       }
+}
+
 static void
 trigger_recheck_cb (GObject    *gobject,
                     GParamSpec *arg1,
@@ -2238,9 +2268,12 @@ miner_files_finished (TrackerMinerFS *fs,
                       gint            files_found,
                       gint            files_ignored)
 {
-       tracker_miner_files_set_last_crawl_done (TRACKER_MINER_FILES (fs), TRUE);
+       TrackerMinerFiles *mf = TRACKER_MINER_FILES (fs);
 
-       tracker_miner_files_check_unextracted (TRACKER_MINER_FILES (fs));
+       tracker_miner_files_check_unextracted (mf);
+
+       /* Schedule a long-term periodic check */
+       tracker_miner_files_update_crawl_interval (mf);
 }
 
 static gchar *
@@ -2714,89 +2747,3 @@ tracker_miner_files_set_first_index_done (TrackerMinerFiles *mf,
 
        g_free (filename);
 }
-
-static inline gchar *
-get_last_crawl_filename (TrackerMinerFiles *mf)
-{
-       GFile *file;
-       gchar *prefix, *path;
-
-       file = get_cache_dir (mf);
-       prefix = g_file_get_path (file);
-
-       path = g_build_filename (prefix,
-                                LAST_CRAWL_FILENAME,
-                                NULL);
-       g_free (prefix);
-       g_object_unref (file);
-
-       return path;
-}
-
-/**
- * tracker_miner_files_get_last_crawl_done:
- *
- * Check when last crawl was performed.
- *
- * Returns: time_t() value when last crawl occurred, otherwise 0.
- **/
-guint64
-tracker_miner_files_get_last_crawl_done (TrackerMinerFiles *mf)
-{
-       gchar *filename;
-       gchar *content;
-       guint64 then;
-
-       filename = get_last_crawl_filename (mf);
-
-       if (!g_file_get_contents (filename, &content, NULL, NULL)) {
-               g_info ("  No previous timestamp, crawling forced");
-               return 0;
-       }
-
-       then = g_ascii_strtoull (content, NULL, 10);
-       g_free (content);
-
-       return then;
-}
-
-/**
- * tracker_miner_files_set_last_crawl_done:
- *
- * Set the time stamp of the last full index of files.
- **/
-void
-tracker_miner_files_set_last_crawl_done (TrackerMinerFiles *mf,
-                                        gboolean           done)
-{
-       gboolean already_exists;
-       gchar *filename;
-
-       filename = get_last_crawl_filename (mf);
-       already_exists = g_file_test (filename, G_FILE_TEST_EXISTS);
-
-       if (done) {
-               GError *error = NULL;
-               gchar *content;
-               content = g_strdup_printf ("%" G_GUINT64_FORMAT, (guint64) time (NULL));
-               if (already_exists) {
-                       g_info ("  Overwriting last crawl file:'%s'", filename);
-               } else {
-                       g_info ("  Creating last crawl file:'%s'", filename);
-               }
-               /* Create/update time stamp file */
-               if (!g_file_set_contents (filename, content, -1, &error)) {
-                       g_warning ("  Could not create/overwrite file:'%s' failed, %s",
-                                  filename,
-                                  error->message);
-                       g_error_free (error);
-               } else {
-                       g_info ("  Last crawl file:'%s' updated", filename);
-               }
-
-               g_free (content);
-       } else {
-               g_info ("  Crawl not done yet, doesn't update last crawl file.");
-       }
-       g_free (filename);
-}
diff --git a/src/miners/fs/tracker-miner-files.h b/src/miners/fs/tracker-miner-files.h
index 6053c9b10..16e8deb1e 100644
--- a/src/miners/fs/tracker-miner-files.h
+++ b/src/miners/fs/tracker-miner-files.h
@@ -58,10 +58,6 @@ gboolean tracker_miner_files_get_first_index_done (TrackerMinerFiles *mf);
 void     tracker_miner_files_set_first_index_done (TrackerMinerFiles *mf,
                                                    gboolean           done);
 
-guint64  tracker_miner_files_get_last_crawl_done  (TrackerMinerFiles *mf);
-void     tracker_miner_files_set_last_crawl_done  (TrackerMinerFiles *mf,
-                                                   gboolean           done);
-
 G_END_DECLS
 
 #endif /* __TRACKER_MINER_FS_FILES_H__ */