summaryrefslogtreecommitdiff
path: root/libavcodec/wmavoice.c
diff options
context:
space:
mode:
authorMichael Niedermayer <michaelni@gmx.at>2011-10-29 02:08:54 +0200
committerMichael Niedermayer <michaelni@gmx.at>2011-10-29 02:08:54 +0200
commit6faf0a21e18f314c48a886864145abe715be6572 (patch)
treef67c3e543a8b2c3283875881536d0a69da515e5e /libavcodec/wmavoice.c
parented1aa8921749a1c70d4453326da7f7b5a6f6f6e7 (diff)
parent61856d06eb30955290911140e6745bad93a25323 (diff)
downloadffmpeg-6faf0a21e18f314c48a886864145abe715be6572.tar.gz
Merge remote-tracking branch 'qatar/master'
* qatar/master: (53 commits) probe: Restore identification of files with very large id3 tags and no extension. probe: Remove id3 tag presence as a criteria to do file extension checking. mpegts: MP4 SL support mpegts: MP4 OD support mpegts: Add support for Sections in PMT mpegts: Replace the MP4 descriptor parser with a recursive parser. mpegts: Add support for multiple mp4 descriptors mpegts: Parse mpeg2 SL descriptors. isom: Add MPEG4SYSTEMS dummy object type indication. aacdec: allow output reconfiguration on channel changes nellymoserenc: take float input samples instead of int16 nellymoserdec: use dsp functions for overlap and windowing nellymoserdec: do not fail if there is extra data in the packet nellymoserdec: fail if output buffer is too small nellymoserdec: remove pointless buffer size check. lavf: add init_put_byte() to the list of visible symbols. seek-test: free options dictionary after use snow: do not draw_edge if emu_edge is set tools/pktdumper: update to recent avformat api seek-test: update to recent avformat api ... Conflicts: doc/APIchanges libavcodec/mpegaudiodec.c libavcodec/nellymoserdec.c libavcodec/snow.c libavcodec/version.h libavcodec/wmadec.c libavformat/avformat.h libavformat/mpegts.c libavformat/mxfdec.c libavformat/utils.c libavformat/wtv.c Merged-by: Michael Niedermayer <michaelni@gmx.at>
Diffstat (limited to 'libavcodec/wmavoice.c')
-rw-r--r--libavcodec/wmavoice.c33
1 files changed, 20 insertions, 13 deletions
diff --git a/libavcodec/wmavoice.c b/libavcodec/wmavoice.c
index b7a6f88a5b..ca7b368f63 100644
--- a/libavcodec/wmavoice.c
+++ b/libavcodec/wmavoice.c
@@ -1730,7 +1730,7 @@ static int synth_superframe(AVCodecContext *ctx,
{
WMAVoiceContext *s = ctx->priv_data;
GetBitContext *gb = &s->gb, s_gb;
- int n, res, n_samples = 480;
+ int n, res, out_size, n_samples = 480;
double lsps[MAX_FRAMES][MAX_LSPS];
const double *mean_lsf = s->lsps == 16 ?
wmavoice_mean_lsf16[s->lsp_def_mode] : wmavoice_mean_lsf10[s->lsp_def_mode];
@@ -1748,7 +1748,10 @@ static int synth_superframe(AVCodecContext *ctx,
s->sframe_cache_size = 0;
}
- if ((res = check_bits_for_superframe(gb, s)) == 1) return 1;
+ if ((res = check_bits_for_superframe(gb, s)) == 1) {
+ *data_size = 0;
+ return 1;
+ }
/* First bit is speech/music bit, it differentiates between WMAVoice
* speech samples (the actual codec) and WMAVoice music samples, which
@@ -1789,6 +1792,14 @@ static int synth_superframe(AVCodecContext *ctx,
stabilize_lsps(lsps[n], s->lsps);
}
+ out_size = n_samples * av_get_bytes_per_sample(ctx->sample_fmt);
+ if (*data_size < out_size) {
+ av_log(ctx, AV_LOG_ERROR,
+ "Output buffer too small (%d given - %zu needed)\n",
+ *data_size, out_size);
+ return -1;
+ }
+
/* Parse frames, optionally preceeded by per-frame (independent) LSPs. */
for (n = 0; n < 3; n++) {
if (!s->has_residual_lsps) {
@@ -1808,8 +1819,10 @@ static int synth_superframe(AVCodecContext *ctx,
&samples[n * MAX_FRAMESIZE],
lsps[n], n == 0 ? s->prev_lsps : lsps[n - 1],
&excitation[s->history_nsamples + n * MAX_FRAMESIZE],
- &synth[s->lsps + n * MAX_FRAMESIZE])))
+ &synth[s->lsps + n * MAX_FRAMESIZE]))) {
+ *data_size = 0;
return res;
+ }
}
/* Statistics? FIXME - we don't check for length, a slight overrun
@@ -1821,7 +1834,7 @@ static int synth_superframe(AVCodecContext *ctx,
}
/* Specify nr. of output samples */
- *data_size = n_samples * sizeof(float);
+ *data_size = out_size;
/* Update history */
memcpy(s->prev_lsps, lsps[2],
@@ -1915,22 +1928,16 @@ static int wmavoice_decode_packet(AVCodecContext *ctx, void *data,
GetBitContext *gb = &s->gb;
int size, res, pos;
- if (*data_size < 480 * sizeof(float)) {
- av_log(ctx, AV_LOG_ERROR,
- "Output buffer too small (%d given - %zu needed)\n",
- *data_size, 480 * sizeof(float));
- return -1;
- }
- *data_size = 0;
-
/* Packets are sometimes a multiple of ctx->block_align, with a packet
* header at each ctx->block_align bytes. However, FFmpeg's ASF demuxer
* feeds us ASF packets, which may concatenate multiple "codec" packets
* in a single "muxer" packet, so we artificially emulate that by
* capping the packet size at ctx->block_align. */
for (size = avpkt->size; size > ctx->block_align; size -= ctx->block_align);
- if (!size)
+ if (!size) {
+ *data_size = 0;
return 0;
+ }
init_get_bits(&s->gb, avpkt->data, size << 3);
/* size == ctx->block_align is used to indicate whether we are dealing with