qtdemux: Detect and expose CEA 608/708 Closed Caption tracks

https://bugzilla.gnome.org/show_bug.cgi?id=606643

3 files changed, 156 insertions, 2 deletions

diff --git a/gst/isomp4/fourcc.h b/gst/isomp4/fourcc.h
index be5c7c04e..6f467ffbb 100644
--- a/gst/isomp4/fourcc.h
+++ b/gst/isomp4/fourcc.h
@@ -95,6 +95,12 @@ G_BEGIN_DECLS
 #define FOURCC_avc1     GST_MAKE_FOURCC('a','v','c','1')
 #define FOURCC_avc3     GST_MAKE_FOURCC('a','v','c','3')
 #define FOURCC_avcC     GST_MAKE_FOURCC('a','v','c','C')
+#define FOURCC_c608     GST_MAKE_FOURCC('c','6','0','8')
+#define FOURCC_c708     GST_MAKE_FOURCC('c','7','0','8')
+#define FOURCC_ccdp     GST_MAKE_FOURCC('c','c','d','p')
+#define FOURCC_cdat     GST_MAKE_FOURCC('c','d','a','t')
+#define FOURCC_cdt2     GST_MAKE_FOURCC('c','d','t','2')
+#define FOURCC_clcp     GST_MAKE_FOURCC('c','l','c','p')
 #define FOURCC_clip     GST_MAKE_FOURCC('c','l','i','p')
 #define FOURCC_cmov     GST_MAKE_FOURCC('c','m','o','v')
 #define FOURCC_cmvd     GST_MAKE_FOURCC('c','m','v','d')
diff --git a/gst/isomp4/qtdemux.c b/gst/isomp4/qtdemux.c
index 5449ff9d2..1a965d7df 100644
--- a/gst/isomp4/qtdemux.c
+++ b/gst/isomp4/qtdemux.c
@@ -5363,6 +5363,120 @@ gst_qtdemux_align_buffer (GstQTDemux * demux,
   return buffer;
 }
 
+static guint8 *
+convert_to_ccdata (const guint8 * ccpair, guint8 ccpair_size, guint field,
+    gsize * res)
+{
+  guint8 *storage;
+  gsize i;
+
+  /* We are converting from pairs to triplets */
+  *res = ccpair_size / 2 * 3;
+  storage = g_malloc (*res);
+  for (i = 0; i * 2 < ccpair_size; i += 1) {
+    if (field == 1)
+      storage[i * 3] = 0xfc;
+    else
+      storage[i * 3] = 0xfd;
+    storage[i * 3 + 1] = ccpair[i * 2];
+    storage[i * 3 + 2] = ccpair[i * 2 + 1];
+  }
+
+  return storage;
+}
+
+static guint8 *
+extract_cc_from_data (QtDemuxStream * stream, const guint8 * data, gsize size,
+    gsize * cclen)
+{
+  guint8 *res = NULL;
+  guint32 atom_length, fourcc;
+  QtDemuxStreamStsdEntry *stsd_entry;
+
+  GST_MEMDUMP ("caption atom", data, size);
+
+  /* There might be multiple atoms */
+
+  *cclen = 0;
+  if (size < 8)
+    goto invalid_cdat;
+  atom_length = QT_UINT32 (data);
+  fourcc = QT_FOURCC (data + 4);
+  if (G_UNLIKELY (atom_length > size || atom_length == 8))
+    goto invalid_cdat;
+
+  GST_DEBUG_OBJECT (stream->pad, "here");
+
+  /* Check if we have somethig compatible */
+  stsd_entry = CUR_STREAM (stream);
+  switch (stsd_entry->fourcc) {
+    case FOURCC_c608:{
+      guint8 *cdat = NULL, *cdt2 = NULL;
+      gsize cdat_size = 0, cdt2_size = 0;
+      /* Should be cdat or cdt2 */
+      if (fourcc != FOURCC_cdat && fourcc != FOURCC_cdt2) {
+        GST_WARNING_OBJECT (stream->pad,
+            "Unknown data atom (%" GST_FOURCC_FORMAT ") for CEA608",
+            GST_FOURCC_ARGS (fourcc));
+        goto invalid_cdat;
+      }
+      /* Convert to cc_data triplet */
+      if (fourcc == FOURCC_cdat)
+        cdat = convert_to_ccdata (data + 8, atom_length - 8, 1, &cdat_size);
+      else
+        cdt2 = convert_to_ccdata (data + 8, atom_length - 8, 2, &cdt2_size);
+      GST_DEBUG_OBJECT (stream->pad, "size:%" G_GSIZE_FORMAT " atom_length:%u",
+          size, atom_length);
+      /* Check for another atom ? */
+      if (size > atom_length + 8) {
+        guint32 new_atom_length = QT_UINT32 (data + atom_length);
+        if (size <= atom_length + new_atom_length) {
+          fourcc = QT_FOURCC (data + atom_length + 4);
+          if (fourcc == FOURCC_cdat)
+            cdat =
+                convert_to_ccdata (data + atom_length + 8, new_atom_length - 8,
+                1, &cdat_size);
+          else
+            cdt2 =
+                convert_to_ccdata (data + atom_length + 8, new_atom_length - 8,
+                2, &cdt2_size);
+        }
+      }
+      *cclen = cdat_size + cdt2_size;
+      res = g_malloc (*cclen);
+      if (cdat_size)
+        memcpy (res, cdat, cdat_size);
+      if (cdt2_size)
+        memcpy (res + cdat_size, cdt2, cdt2_size);
+      g_free (cdat);
+      g_free (cdt2);
+    }
+      break;
+    case FOURCC_c708:
+      if (fourcc != FOURCC_ccdp) {
+        GST_WARNING_OBJECT (stream->pad,
+            "Unknown data atom (%" GST_FOURCC_FORMAT ") for CEA708",
+            GST_FOURCC_ARGS (fourcc));
+        goto invalid_cdat;
+      }
+      *cclen = atom_length - 8;
+      res = g_memdup (data + 8, *cclen);
+      break;
+    default:
+      /* Keep this here in case other closed caption formats are added */
+      g_assert_not_reached ();
+      break;
+  }
+
+  GST_MEMDUMP ("Output", res, *cclen);
+  return res;
+
+  /* Errors */
+invalid_cdat:
+  GST_WARNING ("[cdat] atom is too small or invalid");
+  return NULL;
+}
+
 /* the input buffer metadata must be writable,
  * but time/duration etc not yet set and need not be preserved */
 static GstBuffer *
@@ -5383,7 +5497,7 @@ gst_qtdemux_process_buffer (GstQTDemux * qtdemux, QtDemuxStream * stream,
 
   if (G_UNLIKELY (stream->subtype != FOURCC_text
           && stream->subtype != FOURCC_sbtl &&
-          stream->subtype != FOURCC_subp)) {
+          stream->subtype != FOURCC_subp && stream->subtype != FOURCC_clcp)) {
     return buf;
   }
 
@@ -5401,6 +5515,23 @@ gst_qtdemux_process_buffer (GstQTDemux * qtdemux, QtDemuxStream * stream,
     return buf;
   }
 
+  if (stream->subtype == FOURCC_clcp) {
+    guint8 *cc;
+    gsize cclen = 0;
+    /* For closed caption, we need to extract the information from the
+     * [cdat],[cdt2] or [ccdp] atom */
+    cc = extract_cc_from_data (stream, map.data, map.size, &cclen);
+    gst_buffer_unmap (buf, &map);
+    gst_buffer_unref (buf);
+    if (cc) {
+      buf = _gst_buffer_new_wrapped (cc, cclen, g_free);
+    } else {
+      /* Conversion failed or there's nothing */
+      buf = NULL;
+    }
+    return buf;
+  }
+
   nsize = GST_READ_UINT16_BE (map.data);
   nsize = MIN (nsize, map.size - 2);
 
@@ -11550,7 +11681,8 @@ qtdemux_parse_trak (GstQTDemux * qtdemux, GNode * trak)
       }
       entry->sampled = TRUE;
     } else if (stream->subtype == FOURCC_subp || stream->subtype == FOURCC_text
-        || stream->subtype == FOURCC_sbtl || stream->subtype == FOURCC_subt) {
+        || stream->subtype == FOURCC_sbtl || stream->subtype == FOURCC_subt
+        || stream->subtype == FOURCC_clcp) {
 
       entry->sampled = TRUE;
       entry->sparse = TRUE;
@@ -14541,6 +14673,21 @@ qtdemux_sub_caps (GstQTDemux * qtdemux, QtDemuxStream * stream,
       _codec ("XML subtitles");
       caps = gst_caps_new_empty_simple ("application/ttml+xml");
       break;
+    case FOURCC_c608:
+      _codec ("CEA 608 Closed Caption");
+      caps =
+          gst_caps_new_simple ("closedcaption/x-cea-608", "format",
+          G_TYPE_STRING, "cc_data", NULL);
+      stream->need_process = TRUE;
+      break;
+    case FOURCC_c708:
+      _codec ("CEA 708 Closed Caption");
+      caps =
+          gst_caps_new_simple ("closedcaption/x-cea-708", "format",
+          G_TYPE_STRING, "cdp", NULL);
+      stream->need_process = TRUE;
+      break;
+
     default:
     {
       caps = _get_unknown_codec_name ("text", fourcc);
diff --git a/gst/isomp4/qtdemux_types.c b/gst/isomp4/qtdemux_types.c
index 88db8c2ce..1d5840394 100644
--- a/gst/isomp4/qtdemux_types.c
+++ b/gst/isomp4/qtdemux_types.c
@@ -213,6 +213,7 @@ static const QtNodeType qt_node_types[] = {
   {FOURCC_pssh, "protection system specific header", 0},
   {FOURCC_tenc, "track encryption", 0},
   {FOURCC_stpp, "XML subtitle sample entry", 0},
+  {FOURCC_clcp, "Closed Caption", 0},
   {0, "unknown", 0,},
 };