From 2f177a7616d080937752fe19f15d633fd9e32da3 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tim-Philipp=20M=C3=BCller?= <tim@centricular.net>
Date: Sun, 16 Dec 2012 12:05:02 +0000
Subject: [PATCH] typefinding: detect stand-alone SSA/ASS subtitle files

https://bugzilla.gnome.org/show_bug.cgi?id=625113
---
 gst/typefind/gsttypefindfunctions.c | 90 +++++++++++++++++++++++++++++
 1 file changed, 90 insertions(+)

diff --git a/gst/typefind/gsttypefindfunctions.c b/gst/typefind/gsttypefindfunctions.c
index 9203bccb90..258ec24f27 100644
--- a/gst/typefind/gsttypefindfunctions.c
+++ b/gst/typefind/gsttypefindfunctions.c
@@ -4699,6 +4699,93 @@ dvdiso_type_find (GstTypeFind * tf, gpointer private)
       "application/octet-stream", NULL);
 }
 
+/* SSA/ASS subtitles
+ *
+ * http://en.wikipedia.org/wiki/SubStation_Alpha
+ * http://matroska.org/technical/specs/subtitles/ssa.html
+ */
+static void
+ssa_type_find (GstTypeFind * tf, gpointer private)
+{
+  const gchar *start, *end, *ver_str, *media_type = NULL;
+  const guint8 *data;
+  gchar *str, *script_type, *p = NULL;
+  gint64 len;
+
+  data = gst_type_find_peek (tf, 0, 32);
+
+  if (data == NULL)
+    return;
+
+  /* there might be a BOM at the beginning */
+  if (memcmp (data, "[Script Info]", 13) != 0 &&
+      memcmp (data + 2, "[Script Info]", 13) != 0 &&
+      memcmp (data + 3, "[Script Info]", 13) != 0 &&
+      memcmp (data + 4, "[Script Info]", 13) != 0) {
+    return;
+  }
+
+  /* now check if we have SSA or ASS */
+  len = gst_type_find_get_length (tf);
+  if (len > 8192)
+    len = 8192;
+
+  data = gst_type_find_peek (tf, 0, len);
+  if (data == NULL)
+    return;
+
+  /* skip BOM */
+  start = (gchar *) memchr (data, '[', 5);
+  g_assert (start);
+  len -= (start - (gchar *) data);
+
+  /* ignore anything non-UTF8 for now, in future we might at least allow
+   * other UTF variants that are clearly prefixed with the appropriate BOM */
+  if (!g_utf8_validate (start, len, &end) && (len - (end - start)) > 6) {
+    GST_FIXME ("non-UTF8 SSA/ASS file");
+    return;
+  }
+
+  /* something at start,  but not a UTF-8 BOM? */
+  if (data[0] != '[' && (data[0] != 0xEF || data[1] != 0xBB || data[2] != 0xBF))
+    return;
+
+  /* ignore any partial UTF-8 characters at the end */
+  len = end - start;
+
+  /* create a NUL-terminated string so it's easier to process it safely */
+  str = g_strndup (start, len - 1);
+  script_type = strstr (str, "ScriptType:");
+  if (script_type != NULL) {
+    gdouble version;
+
+    ver_str = script_type + 11;
+    while (*ver_str == ' ' || *ver_str == 'v' || *ver_str == 'V')
+      ++ver_str;
+    version = g_ascii_strtod (ver_str, &p);
+    if (version == 4.0 && p != NULL && *p == '+')
+      media_type = "application/x-ass";
+    else if (version >= 1.0 && version <= 4.0)
+      media_type = "application/x-ssa";
+  }
+
+  if (media_type == NULL) {
+    if (strstr (str, "[v4+ Styles]") || strstr (str, "[V4+ Styles]"))
+      media_type = "application/x-ass";
+    else if (strstr (str, "[v4 Styles]") || strstr (str, "[V4 Styles]"))
+      media_type = "application/x-ssa";
+  }
+
+  if (media_type != NULL) {
+    gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
+        media_type, "parsed", G_TYPE_BOOLEAN, FALSE, NULL);
+  } else {
+    GST_WARNING ("could not detect SSA/ASS variant");
+  }
+
+  g_free (str);
+}
+
 /*** generic typefind for streams that have some data at a specific position***/
 typedef struct
 {
@@ -5073,6 +5160,9 @@ plugin_init (GstPlugin * plugin)
   TYPE_FIND_REGISTER (plugin, "application/octet-stream", GST_RANK_MARGINAL,
       dvdiso_type_find, NULL, NULL, NULL, NULL);
 
+  TYPE_FIND_REGISTER (plugin, "application/x-ssa", GST_RANK_SECONDARY,
+      ssa_type_find, "ssa,ass", NULL, NULL, NULL);
+
   return TRUE;
 }