gst/avi/gstavisubtitle.*: Refactor a bit; fix name extraction; don't assume all the data in the chunk is actually sub...

Original commit message from CVS:
* gst/avi/gstavisubtitle.c: (src_template),
(gst_avi_subtitle_extract_utf8_file),
(gst_avi_subtitle_parse_gab2_chunk), (gst_avi_subtitle_chain),
(gst_avi_subtitle_base_init), (gst_avi_subtitle_class_init),
(gst_avi_subtitle_init), (gst_avi_subtitle_change_state):
* gst/avi/gstavisubtitle.h:
Refactor a bit; fix name extraction; don't assume all the data
in the chunk is actually subtitle data, there may be padding at
the end; fix GST_ELEMENT_ERROR usage; store extracted subtitle
file so it's there to send again after a seek (for future use).

1 files changed, 143 insertions, 92 deletions

diff --git a/gst/avi/gstavisubtitle.c b/gst/avi/gstavisubtitle.c
index 2a1ff14b..b5e3eccc 100644
--- a/gst/avi/gstavisubtitle.c
+++ b/gst/avi/gstavisubtitle.c
@@ -1,5 +1,6 @@
-/* GStreamer
+/* GStreamer AVI GAB2 subtitle parser
  * Copyright (C) <2007> Thijs Vermeir <thijsvermeir@gmail.com>
+ * Copyright (C) <2007> Tim-Philipp Müller <tim centricular net>
  *
  * This library is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Library General Public
@@ -17,19 +18,21 @@
  * Boston, MA 02111-1307, USA.
  */
 
- /* example of a subtitle chunk in an avi file
-  * 00000000 (0x8051700): 47 41 42 32 00 02 00 10 00 00 00 45 00 6e 00 67  GAB2.......E.n.g
-  * 00000010 (0x8051710): 00 6c 00 69 00 73 00 68 00 00 00 04 00 8e 00 00  .l.i.s.h........
-  * 00000020 (0x8051720): 00 ef bb bf 31 0d 0a 30 30 3a 30 30 3a 30 30 2c  ....1..00:00:00,
-  * 00000030 (0x8051730): 31 30 30 20 2d 2d 3e 20 30 30 3a 30 30 3a 30 32  100 --> 00:00:02
-  * 00000040 (0x8051740): 2c 30 30 30 0d 0a 3c 62 3e 41 6e 20 55 54 46 38  ,000..<b>An UTF8
-  * 00000050 (0x8051750): 20 53 75 62 74 69 74 6c 65 20 77 69 74 68 20 42   Subtitle with B
-  * 00000060 (0x8051760): 4f 4d 3c 2f 62 3e 0d 0a 0d 0a 32 0d 0a 30 30 3a  OM</b>....2..00:
-  * 00000070 (0x8051770): 30 30 3a 30 32 2c 31 30 30 20 2d 2d 3e 20 30 30  00:02,100 --> 00
-  * 00000080 (0x8051780): 3a 30 30 3a 30 34 2c 30 30 30 0d 0a 53 6f 6d 65  :00:04,000..Some
-  * 00000090 (0x8051790): 74 68 69 6e 67 20 6e 6f 6e 41 53 43 49 49 20 2d  thing nonASCII -
-  * 000000a0 (0x80517a0): 20 c2 b5 c3 b6 c3 a4 c3 bc c3 9f 0d 0a 0d 0a      ..............
-  */
+/* FIXME: BOM detection and format conversion; validate UTF-8; handle seeks */
+
+/* example of a subtitle chunk in an avi file
+ * 00000000: 47 41 42 32 00 02 00 10 00 00 00 45 00 6e 00 67  GAB2.......E.n.g
+ * 00000010: 00 6c 00 69 00 73 00 68 00 00 00 04 00 8e 00 00  .l.i.s.h........
+ * 00000020: 00 ef bb bf 31 0d 0a 30 30 3a 30 30 3a 30 30 2c  ....1..00:00:00,
+ * 00000030: 31 30 30 20 2d 2d 3e 20 30 30 3a 30 30 3a 30 32  100 --> 00:00:02
+ * 00000040: 2c 30 30 30 0d 0a 3c 62 3e 41 6e 20 55 54 46 38  ,000..<b>An UTF8
+ * 00000050: 20 53 75 62 74 69 74 6c 65 20 77 69 74 68 20 42   Subtitle with B
+ * 00000060: 4f 4d 3c 2f 62 3e 0d 0a 0d 0a 32 0d 0a 30 30 3a  OM</b>....2..00:
+ * 00000070: 30 30 3a 30 32 2c 31 30 30 20 2d 2d 3e 20 30 30  00:02,100 --> 00
+ * 00000080: 3a 30 30 3a 30 34 2c 30 30 30 0d 0a 53 6f 6d 65  :00:04,000..Some
+ * 00000090: 74 68 69 6e 67 20 6e 6f 6e 41 53 43 49 49 20 2d  thing nonASCII -
+ * 000000a0: 20 c2 b5 c3 b6 c3 a4 c3 bc c3 9f 0d 0a 0d 0a      ..............
+ */
 
 #ifdef HAVE_CONFIG_H
 #include "config.h"
@@ -41,7 +44,6 @@
 
 GST_DEBUG_CATEGORY_STATIC (avisubtitle_debug);
 #define GST_CAT_DEFAULT avisubtitle_debug
-GST_DEBUG_CATEGORY_EXTERN (GST_CAT_EVENT);
 
 static GstStaticPadTemplate sink_template = GST_STATIC_PAD_TEMPLATE ("sink",
     GST_PAD_SINK,
@@ -56,19 +58,20 @@ static GstStaticPadTemplate src_template = GST_STATIC_PAD_TEMPLATE ("src",
     );
 
 static GstFlowReturn gst_avi_subtitle_chain (GstPad * pad, GstBuffer * buffer);
+static GstStateChangeReturn gst_avi_subtitle_change_state (GstElement * element,
+    GstStateChange transition);
 
 GST_BOILERPLATE (GstAviSubtitle, gst_avi_subtitle, GstElement,
     GST_TYPE_ELEMENT);
 
 static GstBuffer *
-gst_avi_subtitle_utf8_file (GstBuffer * buffer, guint offset)
+gst_avi_subtitle_extract_utf8_file (GstBuffer * buffer, guint offset, guint len)
 {
   guint8 *file = GST_BUFFER_DATA (buffer) + offset;
 
   if (file[0] == 0xEF && file[1] == 0xBB && file[2] == 0xBF) {
     /* UTF-8 */
-    return gst_buffer_create_sub (buffer, offset + 3,
-        GST_BUFFER_SIZE (buffer) - offset - 3);
+    return gst_buffer_create_sub (buffer, offset + 3, len - 3);
   }
   /* TODO Check for:
    * 00 00 FE FF    UTF-32, big-endian
@@ -78,121 +81,137 @@ gst_avi_subtitle_utf8_file (GstBuffer * buffer, guint offset)
    */
 
   /* No BOM detected assuming UTF-8 */
-  return gst_buffer_create_sub (buffer, offset,
-      GST_BUFFER_SIZE (buffer) - offset);
+  return gst_buffer_create_sub (buffer, offset, len);
 }
 
 static GstFlowReturn
-gst_avi_subtitle_chain (GstPad * pad, GstBuffer * buffer)
+gst_avi_subtitle_parse_gab2_chunk (GstAviSubtitle * sub, GstBuffer * buf)
 {
-  guint name_length, file_length;
-  gunichar2 *name;
-
-  // gchar* name_utf8;
-  GstFlowReturn ret;
-  GstAviSubtitle *avisubtitle = GST_AVI_SUBTITLE (GST_PAD_PARENT (pad));
+  const guint8 *data;
+  gchar *name_utf8;
+  guint name_length;
+  guint file_length;
+  guint size;
 
-  /* we expext only one buffer packet with the whole srt/ssa file in it */
+  data = GST_BUFFER_DATA (buf);
+  size = GST_BUFFER_SIZE (buf);
 
-  /* check the magic word "GAB2\0" */
-  if (GST_BUFFER_SIZE (buffer) <= 11
-      || memcmp (GST_BUFFER_DATA (buffer), "GAB2\0", 5) != 0)
+  /* check the magic word "GAB2\0", and the next word must be 2 */
+  if (size < 12 || memcmp (data, "GAB2\0\2\0", 5 + 2) != 0)
     goto wrong_magic_word;
 
-  /* next word must be 2 */
-  if (GST_READ_UINT16_LE (GST_BUFFER_DATA (buffer) + 5) != 0x2)
-    goto wrong_fixed_word_1;
+  /* read 'name' of subtitle */
+  name_length = GST_READ_UINT32_LE (data + 5 + 2);
+  GST_LOG_OBJECT (sub, "length of name: %u", name_length);
+  if (size <= 17 + name_length)
+    goto wrong_name_length;
 
-  name_length = GST_READ_UINT32_LE (GST_BUFFER_DATA (buffer) + 7);
-  GST_LOG ("length of name: %d", name_length);
-  if (GST_BUFFER_SIZE (buffer) <= 17 + name_length)
-    goto wrong_length_1;
+  name_utf8 = g_convert ((gchar *) data + 11, name_length, "UTF-8", "UTF-16LE",
+      NULL, NULL, NULL);
 
-  name = (gunichar2 *) & (GST_BUFFER_DATA (buffer)[11]);
-  // FIXME Take care for endianess in UTF-16
-  // name_utf8 = g_utf16_to_utf8( name, name_length, NULL, NULL, NULL);
-  // GST_LOG("avi subtitle name: %s", name_utf8);
-  // g_free (name_utf8);
+  if (name_utf8) {
+    /* FIXME: put in a taglist */
+    GST_LOG_OBJECT (sub, "subtitle name: %s", name_utf8);
+    g_free (name_utf8);
+  }
 
   /* next word must be 4 */
-  if (GST_READ_UINT16_LE (GST_BUFFER_DATA (buffer) + 11 + name_length) != 0x4)
+  if (GST_READ_UINT16_LE (data + 11 + name_length) != 0x4)
     goto wrong_fixed_word_2;
 
-  file_length =
-      GST_READ_UINT32_LE (GST_BUFFER_DATA (buffer) + 13 + name_length);
-  GST_LOG ("length srt/ssa file: %d", file_length);
+  file_length = GST_READ_UINT32_LE (data + 13 + name_length);
+  GST_LOG_OBJECT (sub, "length srt/ssa file: %u", file_length);
 
-  if (GST_BUFFER_SIZE (buffer) != 17 + name_length + file_length)
+  if (size < (17 + name_length + file_length))
     goto wrong_total_length;
 
-  /* push the file over the src pad */
-  ret =
-      gst_pad_push (avisubtitle->src, gst_avi_subtitle_utf8_file (buffer,
-          17 + name_length));
-  gst_buffer_unref (buffer);
+  /* store this, so we can send it again after a seek; note that we shouldn't
+   * assume all the remaining data in the chunk is subtitle data, there may
+   * be padding at the end for some reason, so only parse file_length bytes */
+  sub->subfile =
+      gst_avi_subtitle_extract_utf8_file (buf, 17 + name_length, file_length);
 
-  return ret;
+  return GST_FLOW_OK;
 
-  /* all the errors */
+  /* ERRORS */
 wrong_magic_word:
-  GST_ELEMENT_ERROR (avisubtitle, STREAM, DECODE, NULL, ("Wrong magic word"));
-  gst_buffer_unref (buffer);
-  return GST_FLOW_ERROR;
+  {
+    GST_ELEMENT_ERROR (sub, STREAM, DECODE, (NULL), ("Wrong magic word"));
+    return GST_FLOW_ERROR;
+  }
+wrong_name_length:
+  {
+    GST_ELEMENT_ERROR (sub, STREAM, DECODE, (NULL),
+        ("name doesn't fit in buffer (%d < %d)", size, 17 + name_length));
+    return GST_FLOW_ERROR;
+  }
+wrong_fixed_word_2:
+  {
+    GST_ELEMENT_ERROR (sub, STREAM, DECODE, (NULL),
+        ("wrong fixed word: expected %u, got %u", 4,
+            GST_READ_UINT16_LE (data + 11 + name_length)));
+    return GST_FLOW_ERROR;
+  }
+wrong_total_length:
+  {
+    GST_ELEMENT_ERROR (sub, STREAM, DECODE, (NULL),
+        ("buffer size is wrong: need %d bytes, have %d bytes",
+            17 + name_length + file_length, size));
+    return GST_FLOW_ERROR;
+  }
+}
 
-wrong_fixed_word_1:
-  GST_ELEMENT_ERROR (avisubtitle, STREAM, DECODE, NULL,
-      ("wrong fixed word: expected %d found %d", 2,
-          GST_READ_UINT16_LE (GST_BUFFER_DATA (buffer) + 5)));
-  gst_buffer_unref (buffer);
-  return GST_FLOW_ERROR;
+static GstFlowReturn
+gst_avi_subtitle_chain (GstPad * pad, GstBuffer * buffer)
+{
+  GstAviSubtitle *sub = GST_AVI_SUBTITLE (GST_PAD_PARENT (pad));
+  GstFlowReturn ret;
 
-wrong_length_1:
-  GST_ELEMENT_ERROR (avisubtitle, STREAM, DECODE, NULL,
-      ("length of the buffer is too small (%d < %d)", GST_BUFFER_SIZE (buffer),
-          17 + name_length));
-  gst_buffer_unref (buffer);
-  return GST_FLOW_ERROR;
+  if (sub->subfile != NULL) {
+    GST_WARNING_OBJECT (sub, "Got more buffers than expected, dropping");
+    ret = GST_FLOW_UNEXPECTED;
+    goto done;
+  }
 
-wrong_fixed_word_2:
-  GST_ELEMENT_ERROR (avisubtitle, STREAM, DECODE, NULL,
-      ("wrong fixed word: expected %d found %d", 4,
-          GST_READ_UINT16_LE (GST_BUFFER_DATA (buffer) + 11 + name_length)));
-  gst_buffer_unref (buffer);
-  return GST_FLOW_ERROR;
+  /* we expect exactly one buffer with the whole srt/ssa file in it */
+  ret = gst_avi_subtitle_parse_gab2_chunk (sub, buffer);
+  if (ret != GST_FLOW_OK)
+    goto done;
+
+  /* now push the subtitle data downstream */
+  ret = gst_pad_push (sub->src, gst_buffer_ref (sub->subfile));
+
+done:
 
-wrong_total_length:
-  GST_ELEMENT_ERROR (avisubtitle, STREAM, DECODE, NULL,
-      ("buffer size is wrong: need %d bytes, have %d bytes",
-          17 + name_length + file_length, GST_BUFFER_SIZE (buffer)));
   gst_buffer_unref (buffer);
-  return GST_FLOW_ERROR;
+  return ret;
 }
 
 static void
 gst_avi_subtitle_base_init (gpointer klass)
 {
-  static const GstElementDetails gst_avi_demux_details =
-      GST_ELEMENT_DETAILS ("Avi subtitle parser",
-      "Codec/Demuxer",
-      "Parse avi subtitle stream",
-      "Thijs Vermeir <thijsvermeir@gmail.com>");
   GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
 
-  /* add the pad templates to the element */
+  GST_DEBUG_CATEGORY_INIT (avisubtitle_debug, "avisubtitle", 0,
+      "parse avi subtitle stream");
+
   gst_element_class_add_pad_template (element_class,
       gst_static_pad_template_get (&sink_template));
   gst_element_class_add_pad_template (element_class,
       gst_static_pad_template_get (&src_template));
 
-  /* set the element details */
-  gst_element_class_set_details (element_class, &gst_avi_demux_details);
+  gst_element_class_set_details_simple (element_class,
+      "Avi subtitle parser", "Codec/Demuxer", "Parse avi subtitle stream",
+      "Thijs Vermeir <thijsvermeir@gmail.com>");
 }
 
 static void
 gst_avi_subtitle_class_init (GstAviSubtitleClass * klass)
 {
-  GST_DEBUG_CATEGORY_INIT (avisubtitle_debug, "avisubtitle", 0,
-      "parse avi subtitle stream");
+  GstElementClass *gstelement_class = (GstElementClass *) klass;
+
+  gstelement_class->change_state =
+      GST_DEBUG_FUNCPTR (gst_avi_subtitle_change_state);
 }
 
 static void
@@ -202,6 +221,38 @@ gst_avi_subtitle_init (GstAviSubtitle * self, GstAviSubtitleClass * klass)
   gst_element_add_pad (GST_ELEMENT (self), self->src);
 
   self->sink = gst_pad_new_from_static_template (&sink_template, "sink");
+  gst_pad_set_chain_function (self->sink,
+      GST_DEBUG_FUNCPTR (gst_avi_subtitle_chain));
   gst_element_add_pad (GST_ELEMENT (self), self->sink);
-  gst_pad_set_chain_function (self->sink, gst_avi_subtitle_chain);
+}
+
+static GstStateChangeReturn
+gst_avi_subtitle_change_state (GstElement * element, GstStateChange transition)
+{
+  GstStateChangeReturn ret;
+  GstAviSubtitle *sub = GST_AVI_SUBTITLE (element);
+
+  switch (transition) {
+    case GST_STATE_CHANGE_NULL_TO_READY:
+    case GST_STATE_CHANGE_READY_TO_PAUSED:
+    default:
+      break;
+  }
+
+  ret = GST_ELEMENT_CLASS (parent_class)->change_state (element, transition);
+  if (ret == GST_STATE_CHANGE_FAILURE)
+    return ret;
+
+  switch (transition) {
+    case GST_STATE_CHANGE_PAUSED_TO_READY:
+      if (sub->subfile) {
+        gst_buffer_unref (sub->subfile);
+        sub->subfile = NULL;
+      }
+      break;
+    default:
+      break;
+  }
+
+  return ret;
 }