matroskademux: store cluster positions provided by SeekHead

... and use those, if available, to locate a cluster rather than scanning.

2 files changed, 57 insertions, 0 deletions

diff --git a/gst/matroska/matroska-demux.c b/gst/matroska/matroska-demux.c
index f1b2b783f..76431193c 100644
--- a/gst/matroska/matroska-demux.c
+++ b/gst/matroska/matroska-demux.c
@@ -396,12 +396,17 @@ gst_matroska_demux_reset (GstElement * element)
   /* reset indexes */
   if (demux->index) {
     g_array_free (demux->index, TRUE);
     demux->index = NULL;
   }
 
+  if (demux->clusters) {
+    g_array_free (demux->clusters, TRUE);
+    demux->clusters = NULL;
+  }
+
   /* reset timers */
   demux->clock = NULL;
   demux->time_scale = 1000000;
   demux->created = G_MININT64;
 
   demux->index_parsed = FALSE;
@@ -2323,12 +2328,23 @@ gst_matroska_demux_move_to_entry (GstMatroskaDemux * demux,
 
   GST_OBJECT_UNLOCK (demux);
 
   return TRUE;
 }
 
+static gint
+gst_matroska_cluster_compare (gint64 * i1, gint64 * i2)
+{
+  if (*i1 < *i2)
+    return -1;
+  else if (*i1 > *i2)
+    return 1;
+  else
+    return 0;
+}
+
 /* searches for a cluster start from @pos,
  * return GST_FLOW_OK and cluster position in @pos if found */
 static GstFlowReturn
 gst_matroska_demux_search_cluster (GstMatroskaDemux * demux, gint64 * pos)
 {
   gint64 newpos = *pos;
@@ -2339,12 +2355,36 @@ gst_matroska_demux_search_cluster (GstMatroskaDemux * demux, gint64 * pos)
   guint64 length;
   guint32 id;
   guint needed;
 
   orig_offset = demux->offset;
 
+  GST_LOG_OBJECT (demux, "searching cluster following offset %" G_GINT64_FORMAT,
+      *pos);
+
+  if (demux->clusters) {
+    gint64 *cpos;
+
+    cpos = gst_util_array_binary_search (demux->clusters->data,
+        demux->clusters->len, sizeof (gint64),
+        (GCompareDataFunc) gst_matroska_cluster_compare,
+        GST_SEARCH_MODE_AFTER, pos, NULL);
+    /* sanity check */
+    if (cpos) {
+      GST_DEBUG_OBJECT (demux,
+          "cluster reported at offset %" G_GINT64_FORMAT, *cpos);
+      demux->offset = *cpos;
+      ret =
+          gst_matroska_demux_peek_id_length_pull (demux, &id, &length, &needed);
+      if (ret == GST_FLOW_OK && id == GST_MATROSKA_ID_CLUSTER) {
+        newpos = *cpos;
+        goto exit;
+      }
+    }
+  }
+
   /* read in at newpos and scan for ebml cluster id */
   while (1) {
     GstByteReader reader;
     gint cluster_pos;
 
     ret = gst_pad_pull_range (demux->sinkpad, newpos, chunk, &buf);
@@ -2407,12 +2447,13 @@ gst_matroska_demux_search_cluster (GstMatroskaDemux * demux, gint64 * pos)
 
   if (buf) {
     gst_buffer_unref (buf);
     buf = NULL;
   }
 
+exit:
   demux->offset = orig_offset;
   *pos = newpos;
   return ret;
 }
 
 /* bisect and scan through file for cluster starting before @time,
@@ -5380,12 +5421,23 @@ gst_matroska_demux_parse_contents_seekentry (GstMatroskaDemux * demux,
     finish:
       /* seek back */
       demux->offset = before_pos;
       break;
     }
 
+    case GST_MATROSKA_ID_CLUSTER:
+    {
+      guint64 pos = seek_pos + demux->ebml_segment_start;
+
+      GST_LOG_OBJECT (demux, "Cluster position");
+      if (G_UNLIKELY (!demux->clusters))
+        demux->clusters = g_array_sized_new (TRUE, TRUE, sizeof (guint64), 100);
+      g_array_append_val (demux->clusters, pos);
+      break;
+    }
+
     default:
       GST_DEBUG_OBJECT (demux, "Ignoring Seek entry for ID=0x%x", seek_id);
       break;
   }
   DEBUG_ELEMENT_STOP (demux, ebml, "Seek", ret);
 
@@ -5426,12 +5478,15 @@ gst_matroska_demux_parse_contents (GstMatroskaDemux * demux, GstEbmlRead * ebml)
         break;
     }
   }
 
   DEBUG_ELEMENT_STOP (demux, ebml, "SeekHead", ret);
 
+  /* Sort clusters by position for easier searching */
+  g_array_sort (demux->clusters, (GCompareFunc) gst_matroska_cluster_compare);
+
   return ret;
 }
 
 #define GST_FLOW_OVERFLOW   GST_FLOW_CUSTOM_ERROR
 
 #define MAX_BLOCK_SIZE (15 * 1024 * 1024)
diff --git a/gst/matroska/matroska-demux.h b/gst/matroska/matroska-demux.h
index b31f79cb6..a35a5935a 100644
--- a/gst/matroska/matroska-demux.h
+++ b/gst/matroska/matroska-demux.h
@@ -89,12 +89,14 @@ typedef struct _GstMatroskaDemux {
 
   /* start-of-segment */
   guint64                  ebml_segment_start;
 
   /* a cue (index) table */
   GArray                  *index;
+  /* cluster positions (optional) */
+  GArray                  *clusters;
 
   /* timescale in the file */
   guint64                  time_scale;
 
   /* keeping track of playback position */
   GstSegment               segment;