flacdec: we expect framed input now, remove some more code

This commit is contained in:
Tim-Philipp Müller 2011-10-30 18:45:45 +00:00
parent 92361863e6
commit 5ab43cdf91
2 changed files with 32 additions and 120 deletions

View file

@ -247,6 +247,19 @@ static gboolean
gst_flac_dec_set_format (GstAudioDecoder * dec, GstCaps * caps)
{
/* if stream headers are present we could process them here already */
#if 0
///gst_adapter_push (dec->adapter, gst_buffer_ref (buf)); // for all stream headers
/* The first time we get audio data, we know we got all the headers.
* We then loop until all the metadata is processed, then do an extra
* "process_single" step for the audio frame. */
GST_DEBUG_OBJECT (dec,
"First audio frame, ensuring all metadata is processed");
if (!FLAC__stream_decoder_process_until_end_of_metadata (dec->decoder)) {
GST_DEBUG_OBJECT (dec, "process_until_end_of_metadata failed");
}
GST_DEBUG_OBJECT (dec, "All headers and metadata are now processed");
#endif
/* FIXME: refuse caps is there are no stream headers */
GST_LOG_OBJECT (dec, "sink caps: %" GST_PTR_FORMAT, caps);
return TRUE;
}
@ -335,6 +348,8 @@ gst_flac_calculate_crc8 (guint8 * data, guint length)
return crc;
}
/* FIXME: for our purposes it's probably enough to just check for the sync
* marker - we just want to know if it's a header frame or not */
static gboolean
gst_flac_dec_scan_got_frame (GstFlacDec * flacdec, guint8 * data, guint size,
gint64 * last_sample_num)
@ -684,11 +699,7 @@ gst_flac_dec_write (GstFlacDec * flacdec, const FLAC__Frame * frame,
}
gst_buffer_unmap (outbuf, data, size);
GST_DEBUG_OBJECT (flacdec, "pushing %d samples at offset %" G_GINT64_FORMAT
" (%" GST_TIME_FORMAT " + %" GST_TIME_FORMAT ")",
samples, GST_BUFFER_OFFSET (outbuf),
GST_TIME_ARGS (GST_BUFFER_TIMESTAMP (outbuf)),
GST_TIME_ARGS (GST_BUFFER_DURATION (outbuf)));
GST_DEBUG_OBJECT (flacdec, "pushing %d samples", samples);
ret = gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (flacdec), outbuf, 1);
@ -733,58 +744,10 @@ gst_flac_dec_flush (GstAudioDecoder * audio_dec, gboolean hard)
gst_adapter_clear (dec->adapter);
}
static gboolean
gst_flac_dec_chain_parse_headers (GstFlacDec * dec)
{
guint8 marker[4];
guint avail, off;
avail = gst_adapter_available (dec->adapter);
if (avail < 4)
return FALSE;
gst_adapter_copy (dec->adapter, marker, 0, 4);
if (strncmp ((const gchar *) marker, "fLaC", 4) != 0) {
GST_ERROR_OBJECT (dec, "Unexpected header, expected fLaC header");
return TRUE; /* abort header parsing */
}
GST_DEBUG_OBJECT (dec, "fLaC header : len 4 @ %7u", 0);
off = 4;
while (avail > (off + 1 + 3)) {
gboolean is_last;
guint8 mb_hdr[4];
guint len, block_type;
gst_adapter_copy (dec->adapter, mb_hdr, off, 4);
is_last = ((mb_hdr[0] & 0x80) == 0x80);
block_type = mb_hdr[0] & 0x7f;
len = GST_READ_UINT24_BE (mb_hdr + 1);
GST_DEBUG_OBJECT (dec, "Metadata block type %u: len %7u + 4 @ %7u%s",
block_type, len, off, (is_last) ? " (last)" : "");
off += 4 + len;
if (is_last)
break;
if (off >= avail) {
GST_LOG_OBJECT (dec, "Need more data: next offset %u > avail %u", off,
avail);
return FALSE;
}
}
/* want metadata blocks plus at least one frame */
return (off + FLAC__MAX_BLOCK_SIZE >= avail);
}
static GstFlowReturn
gst_flac_dec_handle_frame (GstAudioDecoder * audio_dec, GstBuffer * buf)
{
GstFlacDec *dec;
gboolean got_audio_frame;
dec = GST_FLAC_DEC (audio_dec);
@ -798,13 +761,9 @@ gst_flac_dec_handle_frame (GstAudioDecoder * audio_dec, GstBuffer * buf)
GST_TIME_ARGS (GST_BUFFER_TIMESTAMP (buf)), GST_BUFFER_FLAGS (buf),
gst_buffer_get_size (buf));
/* drop any in-stream headers, we've processed those in set_format already */
if (G_UNLIKELY (!dec->got_headers)) {
// FIXME
}
/* FIXME: should always be framed */
dec->framed = TRUE;
if (dec->framed) {
gboolean got_audio_frame;
gint64 unused;
guint8 *data;
gsize size;
@ -813,8 +772,14 @@ gst_flac_dec_handle_frame (GstAudioDecoder * audio_dec, GstBuffer * buf)
data = gst_buffer_map (buf, &size, NULL, GST_MAP_READ);
got_audio_frame = gst_flac_dec_scan_got_frame (dec, data, size, &unused);
gst_buffer_unmap (buf, data, size);
} else {
got_audio_frame = TRUE;
if (!got_audio_frame) {
GST_INFO_OBJECT (dec, "dropping in-stream header, %d bytes", size);
return GST_FLOW_OK;
}
GST_INFO_OBJECT (dec, "first audio frame, got all in-stream headers now");
dec->got_headers = TRUE;
}
gst_adapter_push (dec->adapter, gst_buffer_ref (buf));
@ -822,65 +787,13 @@ gst_flac_dec_handle_frame (GstAudioDecoder * audio_dec, GstBuffer * buf)
dec->last_flow = GST_FLOW_OK;
if (!dec->framed) {
if (G_UNLIKELY (!dec->got_headers)) {
if (!gst_flac_dec_chain_parse_headers (dec)) {
GST_LOG_OBJECT (dec, "don't have metadata blocks yet, need more data");
goto out;
}
GST_INFO_OBJECT (dec, "have all metadata blocks now");
dec->got_headers = TRUE;
}
/* wait until we have at least 64kB because libflac's StreamDecoder
* interface is a bit dumb it seems (if we don't have as much data as
* it wants it will call our read callback repeatedly and the only
* way to stop that is to error out or EOS, which will affect the
* decoder state). And the decoder seems to always ask for MAX_BLOCK_SIZE
* bytes rather than the max. block size from the header). Requiring
* MAX_BLOCK_SIZE bytes here should make sure it always gets enough data
* to decode at least one block */
while (gst_adapter_available (dec->adapter) >= FLAC__MAX_BLOCK_SIZE &&
dec->last_flow == GST_FLOW_OK) {
GST_LOG_OBJECT (dec, "%u bytes available",
gst_adapter_available (dec->adapter));
if (!FLAC__stream_decoder_process_single (dec->decoder)) {
GST_DEBUG_OBJECT (dec, "process_single failed");
break;
}
if (FLAC__stream_decoder_get_state (dec->decoder) ==
FLAC__STREAM_DECODER_ABORTED) {
GST_WARNING_OBJECT (dec, "Read callback caused internal abort");
dec->last_flow = GST_FLOW_ERROR;
break;
}
}
} else if (dec->framed && got_audio_frame) {
/* framed - there should always be enough data to decode something */
GST_LOG_OBJECT (dec, "%u bytes available",
gst_adapter_available (dec->adapter));
if (G_UNLIKELY (!dec->got_headers)) {
/* The first time we get audio data, we know we got all the headers.
* We then loop until all the metadata is processed, then do an extra
* "process_single" step for the audio frame. */
GST_DEBUG_OBJECT (dec,
"First audio frame, ensuring all metadata is processed");
if (!FLAC__stream_decoder_process_until_end_of_metadata (dec->decoder)) {
GST_DEBUG_OBJECT (dec, "process_until_end_of_metadata failed");
}
GST_DEBUG_OBJECT (dec,
"All metadata is now processed, reading to process audio data");
dec->got_headers = TRUE;
}
if (!FLAC__stream_decoder_process_single (dec->decoder)) {
GST_DEBUG_OBJECT (dec, "process_single failed");
}
} else {
GST_DEBUG_OBJECT (dec, "don't have all headers yet");
}
out:
if (!FLAC__stream_decoder_process_single (dec->decoder)) {
GST_INFO_OBJECT (dec, "process_single failed");
}
return dec->last_flow;
}

View file

@ -44,9 +44,8 @@ struct _GstFlacDec {
/*< private >*/
FLAC__StreamDecoder *decoder;
GstAdapter *adapter;
gboolean framed; // FIXME
gboolean got_headers; /* if we've parsed the headers */
gboolean got_headers; /* have we received all the header buffers yet? */
GstTagList *tags;