/**
* @file
* Matroska file demuxer
- * by Ronald Bultje <rbultje@ronald.bitfreak.net>
- * with a little help from Moritz Bunkus <moritz@bunkus.org>
- * totally reworked by Aurelien Jacobs <aurel@gnuage.org>
- * Specs available on the Matroska project page: http://www.matroska.org/.
+ * @author Ronald Bultje <rbultje@ronald.bitfreak.net>
+ * @author with a little help from Moritz Bunkus <moritz@bunkus.org>
+ *
@author totally reworked by Aurelien Jacobs <aurel@gnuage.org>
+ * @see specs available on the Matroska project page: http://www.matroska.org/
*/
#include <stdio.h>
#include "rm.h"
#include "matroska.h"
#include "libavcodec/mpeg4audio.h"
-#include "libavutil/intfloat_readwrite.h"
+#include "libavutil/intfloat.h"
#include "libavutil/intreadwrite.h"
#include "libavutil/avstring.h"
#include "libavutil/lzo.h"
uint64_t length;
} MatroskaLevel;
+typedef struct {
+ uint64_t timecode;
+ EbmlList blocks;
+} MatroskaCluster;
+
typedef struct {
AVFormatContext *ctx;
/* File has a CUES element, but we defer parsing until it is needed. */
int cues_parsing_deferred;
+
+ int current_cluster_num_blocks;
+ int64_t current_cluster_pos;
+ MatroskaCluster current_cluster;
+
+ /* File has SSA subtitles which prevent incremental cluster parsing. */
+ int contains_ssa;
} MatroskaDemuxContext;
typedef struct {
EbmlBin bin;
} MatroskaBlock;
-typedef struct {
- uint64_t timecode;
- EbmlList blocks;
-} MatroskaCluster;
-
static EbmlSyntax ebml_header[] = {
{ EBML_ID_EBMLREADVERSION, EBML_UINT, 0, offsetof(Ebml,version), {.u=EBML_VERSION} },
{ EBML_ID_EBMLMAXSIZELENGTH, EBML_UINT, 0, offsetof(Ebml,max_size), {.u=8} },
{ 0 }
};
-static const char *matroska_doctypes[] = { "matroska", "webm" };
+static EbmlSyntax matroska_cluster_incremental_parsing[] = {
+ { MATROSKA_ID_CLUSTERTIMECODE,EBML_UINT,0, offsetof(MatroskaCluster,timecode) },
+ { MATROSKA_ID_BLOCKGROUP, EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
+ { MATROSKA_ID_SIMPLEBLOCK, EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
+ { MATROSKA_ID_CLUSTERPOSITION,EBML_NONE },
+ { MATROSKA_ID_CLUSTERPREVSIZE,EBML_NONE },
+ { MATROSKA_ID_INFO, EBML_NONE },
+ { MATROSKA_ID_CUES, EBML_NONE },
+ { MATROSKA_ID_TAGS, EBML_NONE },
+ { MATROSKA_ID_SEEKHEAD, EBML_NONE },
+ { MATROSKA_ID_CLUSTER, EBML_STOP },
+ { 0 }
+};
+
+static EbmlSyntax matroska_cluster_incremental[] = {
+ { MATROSKA_ID_CLUSTERTIMECODE,EBML_UINT,0, offsetof(MatroskaCluster,timecode) },
+ { MATROSKA_ID_BLOCKGROUP, EBML_STOP },
+ { MATROSKA_ID_SIMPLEBLOCK, EBML_STOP },
+ { MATROSKA_ID_CLUSTERPOSITION,EBML_NONE },
+ { MATROSKA_ID_CLUSTERPREVSIZE,EBML_NONE },
+ { 0 }
+};
+
+static EbmlSyntax matroska_clusters_incremental[] = {
+ { MATROSKA_ID_CLUSTER, EBML_NEST, 0, 0, {.n=matroska_cluster_incremental} },
+ { MATROSKA_ID_INFO, EBML_NONE },
+ { MATROSKA_ID_CUES, EBML_NONE },
+ { MATROSKA_ID_TAGS, EBML_NONE },
+ { MATROSKA_ID_SEEKHEAD, EBML_NONE },
+ { 0 }
+};
+
+static const char *const matroska_doctypes[] = { "matroska", "webm" };
/*
* Return: Whether we reached the end of a level in the hierarchy or not.
av_log(matroska->ctx, AV_LOG_ERROR,
"Read error at pos. %"PRIu64" (0x%"PRIx64")\n",
pos, pos);
+ return pb->error ? pb->error : AVERROR(EIO);
}
- return AVERROR(EIO); /* EOS or actual I/O error */
+ return AVERROR_EOF;
}
/* get the length of the EBML number */
if (size == 0) {
*num = 0;
} else if (size == 4) {
- *num= av_int2flt(avio_rb32(pb));
- } else if(size==8){
- *num= av_int2dbl(avio_rb64(pb));
+ *num = av_int2float(avio_rb32(pb));
+ } else if (size == 8){
+ *num = av_int2double(avio_rb64(pb));
} else
return AVERROR_INVALIDDATA;
*/
static int ebml_read_ascii(AVIOContext *pb, int size, char **str)
{
- av_free(*str);
+ char *res;
+
/* EBML strings are usually not 0-terminated, so we allocate one
* byte more, read the string and NULL-terminate it ourselves. */
- if (!(*str = av_malloc(size + 1)))
+ if (!(res = av_malloc(size + 1)))
return AVERROR(ENOMEM);
- if (avio_read(pb, (uint8_t *) *str, size) != size) {
- av_freep(str);
+ if (avio_read(pb, (uint8_t *) res, size) != size) {
+ av_free(res);
return AVERROR(EIO);
}
- (*str)[size] = '\0';
+ (res)[size] = '\0';
+ av_free(*str);
+ *str = res;
return 0;
}
matroska->num_levels > 0 &&
matroska->levels[matroska->num_levels-1].length == 0xffffffffffffff)
return 0; // we reached the end of an unknown size cluster
- if (!syntax[i].id && id != EBML_ID_VOID && id != EBML_ID_CRC32)
+ if (!syntax[i].id && id != EBML_ID_VOID && id != EBML_ID_CRC32) {
av_log(matroska->ctx, AV_LOG_INFO, "Unknown entry 0x%X\n", id);
+ if (matroska->ctx->error_recognition & AV_EF_EXPLODE)
+ return AVERROR_INVALIDDATA;
+ }
return ebml_parse_elem(matroska, &syntax[i], data);
}
uint32_t id = syntax->id;
uint64_t length;
int res;
+ void *newelem;
data = (char *)data + syntax->data_offset;
if (syntax->list_elem_size) {
EbmlList *list = data;
- list->elem = av_realloc(list->elem, (list->nb_elem+1)*syntax->list_elem_size);
+ newelem = av_realloc(list->elem, (list->nb_elem+1)*syntax->list_elem_size);
+ if (!newelem)
+ return AVERROR(ENOMEM);
+ list->elem = newelem;
data = (char*)list->elem + list->nb_elem*syntax->list_elem_size;
memset(data, 0, syntax->list_elem_size);
list->nb_elem++;
uint8_t* data = *buf;
int isize = *buf_size;
uint8_t* pkt_data = NULL;
+ uint8_t av_unused *newpktdata;
int pkt_size = isize;
int result = 0;
int olen;
zstream.avail_in = isize;
do {
pkt_size *= 3;
- pkt_data = av_realloc(pkt_data, pkt_size);
+ newpktdata = av_realloc(pkt_data, pkt_size);
+ if (!newpktdata) {
+ inflateEnd(&zstream);
+ goto failed;
+ }
+ pkt_data = newpktdata;
zstream.avail_out = pkt_size - zstream.total_out;
zstream.next_out = pkt_data + zstream.total_out;
result = inflate(&zstream, Z_NO_FLUSH);
bzstream.avail_in = isize;
do {
pkt_size *= 3;
- pkt_data = av_realloc(pkt_data, pkt_size);
+ newpktdata = av_realloc(pkt_data, pkt_size);
+ if (!newpktdata) {
+ BZ2_bzDecompressEnd(&bzstream);
+ goto failed;
+ }
+ pkt_data = newpktdata;
bzstream.avail_out = pkt_size - bzstream.total_out_lo32;
bzstream.next_out = pkt_data + bzstream.total_out_lo32;
result = BZ2_bzDecompress(&bzstream);
}
}
-static void matroska_merge_packets(AVPacket *out, AVPacket *in)
+static int matroska_merge_packets(AVPacket *out, AVPacket *in)
{
- out->data = av_realloc(out->data, out->size+in->size);
+ void *newdata = av_realloc(out->data, out->size+in->size);
+ if (!newdata)
+ return AVERROR(ENOMEM);
+ out->data = newdata;
memcpy(out->data+out->size, in->data, in->size);
out->size += in->size;
av_destruct_packet(in);
av_free(in);
+ return 0;
}
static void matroska_convert_tag(AVFormatContext *s, EbmlList *list,
matroska->num_levels++;
matroska->current_id = 0;
- ebml_parse(matroska, matroska_segment, matroska);
+ ret = ebml_parse(matroska, matroska_segment, matroska);
/* remove dummy level */
while (matroska->num_levels) {
static void matroska_execute_seekhead(MatroskaDemuxContext *matroska)
{
EbmlList *seekhead_list = &matroska->seekhead;
- MatroskaSeekhead *seekhead = seekhead_list->elem;
int64_t before_pos = avio_tell(matroska->ctx->pb);
int i;
return;
for (i = 0; i < seekhead_list->nb_elem; i++) {
+ MatroskaSeekhead *seekhead = seekhead_list->elem;
if (seekhead[i].pos <= before_pos)
continue;
{
int sri;
- for (sri=0; sri<FF_ARRAY_ELEMS(ff_mpeg4audio_sample_rates); sri++)
- if (ff_mpeg4audio_sample_rates[sri] == samplerate)
+ for (sri=0; sri<FF_ARRAY_ELEMS(avpriv_mpeg4audio_sample_rates); sri++)
+ if (avpriv_mpeg4audio_sample_rates[sri] == samplerate)
break;
return sri;
}
-static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap)
+static int matroska_read_header(AVFormatContext *s)
{
MatroskaDemuxContext *matroska = s->priv_data;
EbmlList *attachements_list = &matroska->attachments;
continue;
if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
- if (!track->default_duration)
+ if (!track->default_duration && track->video.frame_rate > 0)
track->default_duration = 1000000000/track->video.frame_rate;
if (!track->video.display_width)
track->video.display_width = track->video.pixel_width;
}
if (encodings_list->nb_elem > 1) {
av_log(matroska->ctx, AV_LOG_ERROR,
- "Multiple combined encodings no supported");
+ "Multiple combined encodings not supported");
} else if (encodings_list->nb_elem == 1) {
if (encodings[0].type ||
(encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP &&
}
}
- st = track->stream = av_new_stream(s, 0);
+ st = track->stream = avformat_new_stream(s, NULL);
if (st == NULL)
return AVERROR(ENOMEM);
&& (track->codec_priv.size >= 86)
&& (track->codec_priv.data != NULL)) {
track->video.fourcc = AV_RL32(track->codec_priv.data);
- codec_id=ff_codec_get_id(codec_movvideo_tags, track->video.fourcc);
+ codec_id=ff_codec_get_id(ff_codec_movvideo_tags, track->video.fourcc);
} else if (codec_id == CODEC_ID_PCM_S16BE) {
switch (track->audio.bitdepth) {
case 8: codec_id = CODEC_ID_PCM_U8; break;
} else if (codec_id == CODEC_ID_AAC && !track->codec_priv.size) {
int profile = matroska_aac_profile(track->codec_id);
int sri = matroska_aac_sri(track->audio.samplerate);
- extradata = av_malloc(5);
+ extradata = av_mallocz(5 + FF_INPUT_BUFFER_PADDING_SIZE);
if (extradata == NULL)
return AVERROR(ENOMEM);
extradata[0] = (profile << 3) | ((sri&0x0E) >> 1);
if (track->time_scale < 0.01)
track->time_scale = 1.0;
- av_set_pts_info(st, 64, matroska->time_scale*track->time_scale, 1000*1000*1000); /* 64 bit pts in ns */
+ avpriv_set_pts_info(st, 64, matroska->time_scale*track->time_scale, 1000*1000*1000); /* 64 bit pts in ns */
st->codec->codec_id = codec_id;
st->start_time = 0;
if (track->flag_forced)
st->disposition |= AV_DISPOSITION_FORCED;
- if (track->default_duration)
- av_reduce(&st->codec->time_base.num, &st->codec->time_base.den,
- track->default_duration, 1000000000, 30000);
-
if (!st->codec->extradata) {
if(extradata){
st->codec->extradata = extradata;
255);
if (st->codec->codec_id != CODEC_ID_H264)
st->need_parsing = AVSTREAM_PARSE_HEADERS;
- if (track->default_duration)
- st->avg_frame_rate = av_d2q(1000000000.0/track->default_duration, INT_MAX);
+ if (track->default_duration) {
+ av_reduce(&st->r_frame_rate.num, &st->r_frame_rate.den,
+ 1000000000, track->default_duration, 30000);
+ st->avg_frame_rate = st->r_frame_rate;
+ }
} else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
st->codec->sample_rate = track->audio.out_samplerate;
st->need_parsing = AVSTREAM_PARSE_HEADERS;
} else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE;
+ if (st->codec->codec_id == CODEC_ID_SSA)
+ matroska->contains_ssa = 1;
}
}
attachements[j].bin.data && attachements[j].bin.size > 0)) {
av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
} else {
- AVStream *st = av_new_stream(s, 0);
+ AVStream *st = avformat_new_stream(s, NULL);
if (st == NULL)
break;
av_dict_set(&st->metadata, "filename",attachements[j].filename, 0);
+ av_dict_set(&st->metadata, "mimetype", attachements[j].mime, 0);
st->codec->codec_id = CODEC_ID_NONE;
st->codec->codec_type = AVMEDIA_TYPE_ATTACHMENT;
st->codec->extradata = av_malloc(attachements[j].bin.size);
if (chapters[i].start != AV_NOPTS_VALUE && chapters[i].uid
&& (max_start==0 || chapters[i].start > max_start)) {
chapters[i].chapter =
- ff_new_chapter(s, chapters[i].uid, (AVRational){1, 1000000000},
+ avpriv_new_chapter(s, chapters[i].uid, (AVRational){1, 1000000000},
chapters[i].start, chapters[i].end,
chapters[i].title);
av_dict_set(&chapters[i].chapter->metadata,
memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
av_free(matroska->packets[0]);
if (matroska->num_packets > 1) {
+ void *newpackets;
memmove(&matroska->packets[0], &matroska->packets[1],
(matroska->num_packets - 1) * sizeof(AVPacket *));
- matroska->packets =
- av_realloc(matroska->packets, (matroska->num_packets - 1) *
- sizeof(AVPacket *));
+ newpackets = av_realloc(matroska->packets,
+ (matroska->num_packets - 1) * sizeof(AVPacket *));
+ if (newpackets)
+ matroska->packets = newpackets;
} else {
av_freep(&matroska->packets);
+ matroska->prev_pkt = NULL;
}
matroska->num_packets--;
return 0;
if ((n = matroska_ebmlnum_uint(matroska, data, size, &num)) < 0) {
av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
- return res;
+ return n;
}
data += n;
size -= n;
track = matroska_find_track_by_num(matroska, num);
- if (size <= 3 || !track || !track->stream) {
+ if (!track || !track->stream) {
av_log(matroska->ctx, AV_LOG_INFO,
"Invalid stream %"PRIu64" or size %u\n", num, size);
return AVERROR_INVALIDDATA;
- }
+ } else if (size <= 3)
+ return 0;
st = track->stream;
if (st->discard >= AVDISCARD_ALL)
return res;
lace_size[n] = lace_size[n - 1] + snum;
total += lace_size[n];
}
- lace_size[n] = size - total;
+ lace_size[laces - 1] = size - total;
break;
}
}
if (!track->audio.pkt_cnt) {
if (track->audio.sub_packet_cnt == 0)
track->audio.buf_timecode = timecode;
- if (st->codec->codec_id == CODEC_ID_RA_288)
+ if (st->codec->codec_id == CODEC_ID_RA_288) {
+ if (size < cfs * h / 2) {
+ av_log(matroska->ctx, AV_LOG_ERROR,
+ "Corrupt int4 RM-style audio packet size\n");
+ res = AVERROR_INVALIDDATA;
+ goto end;
+ }
for (x=0; x<h/2; x++)
memcpy(track->audio.buf+x*2*w+y*cfs,
data+x*cfs, cfs);
- else if (st->codec->codec_id == CODEC_ID_SIPR)
+ } else if (st->codec->codec_id == CODEC_ID_SIPR) {
+ if (size < w) {
+ av_log(matroska->ctx, AV_LOG_ERROR,
+ "Corrupt sipr RM-style audio packet size\n");
+ res = AVERROR_INVALIDDATA;
+ goto end;
+ }
memcpy(track->audio.buf + y*w, data, w);
- else
+ } else {
+ if (size < sps * w / sps) {
+ av_log(matroska->ctx, AV_LOG_ERROR,
+ "Corrupt generic RM-style audio packet size\n");
+ res = AVERROR_INVALIDDATA;
+ goto end;
+ }
for (x=0; x<w/sps; x++)
memcpy(track->audio.buf+sps*(h*x+((h+1)/2)*(y&1)+(y>>1)), data+x*sps, sps);
+ }
if (++track->audio.sub_packet_cnt >= h) {
if (st->codec->codec_id == CODEC_ID_SIPR)
}
}
+end:
av_free(lace_size);
return res;
}
+static int matroska_parse_cluster_incremental(MatroskaDemuxContext *matroska)
+{
+ EbmlList *blocks_list;
+ MatroskaBlock *blocks;
+ int i, res;
+ res = ebml_parse(matroska,
+ matroska_cluster_incremental_parsing,
+ &matroska->current_cluster);
+ if (res == 1) {
+ /* New Cluster */
+ if (matroska->current_cluster_pos)
+ ebml_level_end(matroska);
+ ebml_free(matroska_cluster, &matroska->current_cluster);
+ memset(&matroska->current_cluster, 0, sizeof(MatroskaCluster));
+ matroska->current_cluster_num_blocks = 0;
+ matroska->current_cluster_pos = avio_tell(matroska->ctx->pb);
+ matroska->prev_pkt = NULL;
+ /* sizeof the ID which was already read */
+ if (matroska->current_id)
+ matroska->current_cluster_pos -= 4;
+ res = ebml_parse(matroska,
+ matroska_clusters_incremental,
+ &matroska->current_cluster);
+ /* Try parsing the block again. */
+ if (res == 1)
+ res = ebml_parse(matroska,
+ matroska_cluster_incremental_parsing,
+ &matroska->current_cluster);
+ }
+
+ if (!res &&
+ matroska->current_cluster_num_blocks <
+ matroska->current_cluster.blocks.nb_elem) {
+ blocks_list = &matroska->current_cluster.blocks;
+ blocks = blocks_list->elem;
+
+ matroska->current_cluster_num_blocks = blocks_list->nb_elem;
+ i = blocks_list->nb_elem - 1;
+ if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
+ int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
+ if (!blocks[i].non_simple)
+ blocks[i].duration = AV_NOPTS_VALUE;
+ res = matroska_parse_block(matroska,
+ blocks[i].bin.data, blocks[i].bin.size,
+ blocks[i].bin.pos,
+ matroska->current_cluster.timecode,
+ blocks[i].duration, is_keyframe,
+ matroska->current_cluster_pos);
+ }
+ }
+
+ if (res < 0) matroska->done = 1;
+ return res;
+}
+
static int matroska_parse_cluster(MatroskaDemuxContext *matroska)
{
MatroskaCluster cluster = { 0 };
EbmlList *blocks_list;
MatroskaBlock *blocks;
int i, res;
- int64_t pos = avio_tell(matroska->ctx->pb);
+ int64_t pos;
+ if (!matroska->contains_ssa)
+ return matroska_parse_cluster_incremental(matroska);
+ pos = avio_tell(matroska->ctx->pb);
matroska->prev_pkt = NULL;
if (matroska->current_id)
pos -= 4; /* sizeof the ID which was already read */
ret = matroska_parse_cluster(matroska);
}
+ if (ret == AVERROR_INVALIDDATA) {
+ pkt->flags |= AV_PKT_FLAG_CORRUPT;
+ return 0;
+ }
+
return ret;
}
avio_seek(s->pb, st->index_entries[st->nb_index_entries-1].pos, SEEK_SET);
matroska->current_id = 0;
while ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
+ matroska->prev_pkt = NULL;
matroska_clear_queue(matroska);
if (matroska_parse_cluster(matroska) < 0)
break;
matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
matroska->skip_to_timecode = st->index_entries[index].timestamp;
matroska->done = 0;
- av_update_cur_dts(s, st, st->index_entries[index].timestamp);
+ ff_update_cur_dts(s, st, st->index_entries[index].timestamp);
return 0;
}
for (n=0; n < matroska->tracks.nb_elem; n++)
if (tracks[n].type == MATROSKA_TRACK_TYPE_AUDIO)
av_free(tracks[n].audio.buf);
+ ebml_free(matroska_cluster, &matroska->current_cluster);
ebml_free(matroska_segment, matroska);
return 0;
projects / ffmpeg / blobdiff