Consolidate and improve audio decoding test for all decoders.

media/filters/audio_file_reader.cc

Index: media/filters/audio_file_reader.cc
diff --git a/media/filters/audio_file_reader.cc b/media/filters/audio_file_reader.cc
index b20fd8d9ce35795b7ed94e5cea8177d80f0f25f0..0aa4a6a787ca1f15779296c76f0224a5dccc621d 100644
--- a/media/filters/audio_file_reader.cc
+++ b/media/filters/audio_file_reader.cc
@@ -257,4 +257,11 @@ bool AudioFileReader::ReadPacket(AVPacket* output_packet) {
   return false;
 }
 
+bool AudioFileReader::SeekForTesting(base::TimeDelta seek_time) {

[wolenetz, 2014/06/06 21:00:05]

How ingrained is embedding alternate timebase times in a base::TimeDelta? It
seems strange / confusing versus just int64. Does it give us any advantage like
non-finite or invalid value representation that is meaningful for wherever we
put a non-microsecond time value as a microseconds in a TimeDelta? Especially
strange to me if this is ingrained in our non-test code.

[DaleCurtis, 2014/06/06 21:17:00]

See my comment about skipping timebase conversion when calling this function.

+  return av_seek_frame(glue_->format_context(),
+                       stream_index_,
+                       seek_time.InMicroseconds(),
+                       AVSEEK_FLAG_BACKWARD) >= 0;
+}
+
 }  // namespace media