From: Thiago Santos <ts.santos@sisa.samsung.com>
Date: Thu, 13 Feb 2014 15:29:13 +0000 (-0300)
Subject: qtdemux: handle tags in mac encoding
X-Git-Tag: 1.3.1~235
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=7f1d51ba90275a53e00339a75746bc638edded0c;p=platform%2Fupstream%2Fgst-plugins-good.git

qtdemux: handle tags in mac encoding

Check the charset from (C)*** tags and set the charset
to convert from MAC encoding if suitable.

https://bugzilla.gnome.org/show_bug.cgi?id=723166
---

diff --git a/gst/isomp4/qtdemux.c b/gst/isomp4/qtdemux.c
index 2c0010a..fe293ee 100644
--- a/gst/isomp4/qtdemux.c
+++ b/gst/isomp4/qtdemux.c
@@ -9098,6 +9098,7 @@ qtdemux_tag_add_str_full (GstQTDemux * qtdemux, const char *tag,
   guint32 type;
   int offset;
   gboolean ret = TRUE;
+  const gchar *charset = NULL;
 
   data = qtdemux_tree_get_child_by_type (node, FOURCC_data);
   if (data) {
@@ -9119,10 +9120,30 @@ qtdemux_tag_add_str_full (GstQTDemux * qtdemux, const char *tag,
     len = QT_UINT32 (node->data);
     type = QT_UINT32 ((guint8 *) node->data + 4);
     if ((type >> 24) == 0xa9) {
-      /* Type starts with the (C) symbol, so the next 32 bits are
-       * the language code, which we ignore */
+      gint str_len;
+      gint lang_code;
+
+      /* Type starts with the (C) symbol, so the next data is a list
+       * of (string size(16), language code(16), string) */
+
+      str_len = QT_UINT16 ((guint8 *) node->data + 8);
+      lang_code = QT_UINT16 ((guint8 *) node->data + 10);
+
+      /* the string + fourcc + size + 2 16bit fields,
+       * means that there are more tags in this atom */
+      if (len > str_len + 8 + 4) {
+        /* TODO how to represent the same tag in different languages? */
+        GST_WARNING_OBJECT (qtdemux, "Ignoring metadata entry with multiple "
+            "text alternatives, reading only first one");
+      }
+
       offset = 12;
+      len = str_len + 8 + 4;    /* remove trailing strings that we don't use */
       GST_DEBUG_OBJECT (qtdemux, "found international text tag");
+
+      if (lang_code < 0x800) {  /* MAC encoded string */
+        charset = "mac";
+      }
     } else if (len > 14 && qtdemux_is_string_tag_3gp (qtdemux,
             QT_FOURCC ((guint8 *) node->data + 4))) {
       guint32 type = QT_UINT32 ((guint8 *) node->data + 8);
@@ -9144,8 +9165,21 @@ qtdemux_tag_add_str_full (GstQTDemux * qtdemux, const char *tag,
       GST_DEBUG_OBJECT (qtdemux, "found normal text tag");
       ret = FALSE;              /* may have to fallback */
     }
-    s = gst_tag_freeform_string_to_utf8 ((char *) node->data + offset,
-        len - offset, env_vars);
+    if (charset) {
+      GError *err = NULL;
+
+      s = g_convert ((gchar *) node->data + offset, len - offset, "utf8",
+          charset, NULL, NULL, &err);
+      if (err) {
+        GST_DEBUG_OBJECT (qtdemux, "Failed to convert string from charset %s:"
+            " %s(%d): %s", charset, g_quark_to_string (err->domain), err->code,
+            err->message);
+        g_error_free (err);
+      }
+    } else {
+      s = gst_tag_freeform_string_to_utf8 ((char *) node->data + offset,
+          len - offset, env_vars);
+    }
     if (s) {
       GST_DEBUG_OBJECT (qtdemux, "adding tag %s", GST_STR_NULL (s));
       gst_tag_list_add (qtdemux->tag_list, GST_TAG_MERGE_REPLACE, tag, s, NULL);