aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorAndreas Rheinhardt <andreas.rheinhardt@outlook.com>2021-04-16 20:53:55 +0200
committerAndreas Rheinhardt <andreas.rheinhardt@outlook.com>2021-04-18 02:49:43 +0200
commite1f41b4fdc5d3e8813d8f9d4d81899bd4e6df31b (patch)
tree60579319939655817bf778bb781afc53eb89fcf2
parenta74428921f8bfb33cbe0340bfd810b6945e432d2 (diff)
downloadffmpeg-e1f41b4fdc5d3e8813d8f9d4d81899bd4e6df31b.tar.gz
avformat/wavdec: Fix reading files with id3v2 apic before fmt tag
Up until now the cover images will get the stream index 0 in this case, violating the hardcoded assumption that this is the index of the audio stream. Fix this by creating the audio stream first; this is also in line with the expectations of ff_pcm_read_seek() and ff_spdif_read_packet(). It also simplifies the code to parse the fmt and xma2 tags. Fixes #8540; regression since f5aad350d3695b5b16e7d135154a4c61e4dce9d8. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
-rw-r--r--libavformat/wavdec.c78
1 files changed, 39 insertions, 39 deletions
diff --git a/libavformat/wavdec.c b/libavformat/wavdec.c
index aca9a8382b..e11535b6ae 100644
--- a/libavformat/wavdec.c
+++ b/libavformat/wavdec.c
@@ -49,6 +49,7 @@ typedef struct WAVDemuxContext {
const AVClass *class;
int64_t data_end;
int w64;
+ AVStream *vst;
int64_t smv_data_ofs;
int smv_block_size;
int smv_frames_per_jpeg;
@@ -170,30 +171,26 @@ static void handle_stream_probing(AVStream *st)
}
}
-static int wav_parse_fmt_tag(AVFormatContext *s, int64_t size, AVStream **st)
+static int wav_parse_fmt_tag(AVFormatContext *s, int64_t size, AVStream *st)
{
AVIOContext *pb = s->pb;
WAVDemuxContext *wav = s->priv_data;
int ret;
/* parse fmt header */
- *st = avformat_new_stream(s, NULL);
- if (!*st)
- return AVERROR(ENOMEM);
-
- ret = ff_get_wav_header(s, pb, (*st)->codecpar, size, wav->rifx);
+ ret = ff_get_wav_header(s, pb, st->codecpar, size, wav->rifx);
if (ret < 0)
return ret;
- handle_stream_probing(*st);
+ handle_stream_probing(st);
- (*st)->need_parsing = AVSTREAM_PARSE_FULL_RAW;
+ st->need_parsing = AVSTREAM_PARSE_FULL_RAW;
- avpriv_set_pts_info(*st, 64, 1, (*st)->codecpar->sample_rate);
+ avpriv_set_pts_info(st, 64, 1, st->codecpar->sample_rate);
return 0;
}
-static int wav_parse_xma2_tag(AVFormatContext *s, int64_t size, AVStream **st)
+static int wav_parse_xma2_tag(AVFormatContext *s, int64_t size, AVStream *st)
{
AVIOContext *pb = s->pb;
int version, num_streams, i, channels = 0, ret;
@@ -201,13 +198,9 @@ static int wav_parse_xma2_tag(AVFormatContext *s, int64_t size, AVStream **st)
if (size < 36)
return AVERROR_INVALIDDATA;
- *st = avformat_new_stream(s, NULL);
- if (!*st)
- return AVERROR(ENOMEM);
-
- (*st)->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
- (*st)->codecpar->codec_id = AV_CODEC_ID_XMA2;
- (*st)->need_parsing = AVSTREAM_PARSE_FULL_RAW;
+ st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
+ st->codecpar->codec_id = AV_CODEC_ID_XMA2;
+ st->need_parsing = AVSTREAM_PARSE_FULL_RAW;
version = avio_r8(pb);
if (version != 3 && version != 4)
@@ -216,26 +209,26 @@ static int wav_parse_xma2_tag(AVFormatContext *s, int64_t size, AVStream **st)
if (size != (32 + ((version==3)?0:8) + 4*num_streams))
return AVERROR_INVALIDDATA;
avio_skip(pb, 10);
- (*st)->codecpar->sample_rate = avio_rb32(pb);
+ st->codecpar->sample_rate = avio_rb32(pb);
if (version == 4)
avio_skip(pb, 8);
avio_skip(pb, 4);
- (*st)->duration = avio_rb32(pb);
+ st->duration = avio_rb32(pb);
avio_skip(pb, 8);
for (i = 0; i < num_streams; i++) {
channels += avio_r8(pb);
avio_skip(pb, 3);
}
- (*st)->codecpar->channels = channels;
+ st->codecpar->channels = channels;
- if ((*st)->codecpar->channels <= 0 || (*st)->codecpar->sample_rate <= 0)
+ if (st->codecpar->channels <= 0 || st->codecpar->sample_rate <= 0)
return AVERROR_INVALIDDATA;
- avpriv_set_pts_info(*st, 64, 1, (*st)->codecpar->sample_rate);
+ avpriv_set_pts_info(st, 64, 1, st->codecpar->sample_rate);
avio_seek(pb, -size, SEEK_CUR);
- if ((ret = ff_get_extradata(s, (*st)->codecpar, pb, size)) < 0)
+ if ((ret = ff_get_extradata(s, st->codecpar, pb, size)) < 0)
return ret;
return 0;
@@ -407,6 +400,11 @@ static int wav_read_header(AVFormatContext *s)
}
+ /* Create the audio stream now so that its index is always zero */
+ st = avformat_new_stream(s, NULL);
+ if (!st)
+ return AVERROR(ENOMEM);
+
for (;;) {
AVStream *vst;
size = next_tag(pb, &tag, wav->rifx);
@@ -418,7 +416,7 @@ static int wav_read_header(AVFormatContext *s)
switch (tag) {
case MKTAG('f', 'm', 't', ' '):
/* only parse the first 'fmt ' tag found */
- if (!got_xma2 && !got_fmt && (ret = wav_parse_fmt_tag(s, size, &st)) < 0) {
+ if (!got_xma2 && !got_fmt && (ret = wav_parse_fmt_tag(s, size, st)) < 0) {
return ret;
} else if (got_fmt)
av_log(s, AV_LOG_WARNING, "found more than one 'fmt ' tag\n");
@@ -427,7 +425,7 @@ static int wav_read_header(AVFormatContext *s)
break;
case MKTAG('X', 'M', 'A', '2'):
/* only parse the first 'XMA2' tag found */
- if (!got_fmt && !got_xma2 && (ret = wav_parse_xma2_tag(s, size, &st)) < 0) {
+ if (!got_fmt && !got_xma2 && (ret = wav_parse_xma2_tag(s, size, st)) < 0) {
return ret;
} else if (got_xma2)
av_log(s, AV_LOG_WARNING, "found more than one 'XMA2' tag\n");
@@ -484,6 +482,7 @@ static int wav_read_header(AVFormatContext *s)
vst = avformat_new_stream(s, NULL);
if (!vst)
return AVERROR(ENOMEM);
+ wav->vst = vst;
avio_r8(pb);
vst->id = 1;
vst->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
@@ -693,23 +692,24 @@ static int wav_read_packet(AVFormatContext *s, AVPacket *pkt)
{
int ret, size;
int64_t left;
- AVStream *st;
WAVDemuxContext *wav = s->priv_data;
+ AVStream *st = s->streams[0];
if (CONFIG_SPDIF_DEMUXER && wav->spdif == 1)
return ff_spdif_read_packet(s, pkt);
if (wav->smv_data_ofs > 0) {
int64_t audio_dts, video_dts;
+ AVStream *vst = wav->vst;
smv_retry:
- audio_dts = (int32_t)s->streams[0]->cur_dts;
- video_dts = (int32_t)s->streams[1]->cur_dts;
+ audio_dts = (int32_t)st->cur_dts;
+ video_dts = (int32_t)vst->cur_dts;
if (audio_dts != AV_NOPTS_VALUE && video_dts != AV_NOPTS_VALUE) {
/*We always return a video frame first to get the pixel format first*/
wav->smv_last_stream = wav->smv_given_first ?
- av_compare_ts(video_dts, s->streams[1]->time_base,
- audio_dts, s->streams[0]->time_base) > 0 : 0;
+ av_compare_ts(video_dts, vst->time_base,
+ audio_dts, st->time_base) > 0 : 0;
wav->smv_given_first = 1;
}
wav->smv_last_stream = !wav->smv_last_stream;
@@ -732,7 +732,7 @@ smv_retry:
pkt->duration = wav->smv_frames_per_jpeg;
wav->smv_block++;
- pkt->stream_index = 1;
+ pkt->stream_index = vst->index;
smv_out:
avio_seek(s->pb, old_pos, SEEK_SET);
if (ret == AVERROR_EOF) {
@@ -743,8 +743,6 @@ smv_out:
}
}
- st = s->streams[0];
-
left = wav->data_end - avio_tell(s->pb);
if (wav->ignore_length)
left = INT_MAX;
@@ -781,22 +779,24 @@ static int wav_read_seek(AVFormatContext *s,
int stream_index, int64_t timestamp, int flags)
{
WAVDemuxContext *wav = s->priv_data;
- AVStream *st;
+ AVStream *ast = s->streams[0], *vst = wav->vst;
wav->smv_eof = 0;
wav->audio_eof = 0;
+
+ if (stream_index != 0 && (!vst || stream_index != vst->index))
+ return AVERROR(EINVAL);
if (wav->smv_data_ofs > 0) {
int64_t smv_timestamp = timestamp;
if (stream_index == 0)
- smv_timestamp = av_rescale_q(timestamp, s->streams[0]->time_base, s->streams[1]->time_base);
+ smv_timestamp = av_rescale_q(timestamp, ast->time_base, vst->time_base);
else
- timestamp = av_rescale_q(smv_timestamp, s->streams[1]->time_base, s->streams[0]->time_base);
+ timestamp = av_rescale_q(smv_timestamp, vst->time_base, ast->time_base);
if (wav->smv_frames_per_jpeg > 0) {
wav->smv_block = smv_timestamp / wav->smv_frames_per_jpeg;
}
}
- st = s->streams[0];
- switch (st->codecpar->codec_id) {
+ switch (ast->codecpar->codec_id) {
case AV_CODEC_ID_MP2:
case AV_CODEC_ID_MP3:
case AV_CODEC_ID_AC3:
@@ -807,7 +807,7 @@ static int wav_read_seek(AVFormatContext *s,
default:
break;
}
- return ff_pcm_read_seek(s, stream_index, timestamp, flags);
+ return ff_pcm_read_seek(s, 0, timestamp, flags);
}
static const AVClass wav_demuxer_class = {