X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavformat%2Fmatroskadec.c;h=c454713eeb022065b77b24f98bc8b087cd41e67e;hb=f73e3938ac70524826664855210446c3739c4a5e;hp=61a77abdf8ab1661af41383188216d03c9f1a6da;hpb=98a7d560b4f835ba5b99b9974ee1a2c096198bea;p=ffmpeg diff --git a/libavformat/matroskadec.c b/libavformat/matroskadec.c index 61a77abdf8a..c454713eeb0 100644 --- a/libavformat/matroskadec.c +++ b/libavformat/matroskadec.c @@ -22,10 +22,10 @@ /** * @file * Matroska file demuxer - * by Ronald Bultje - * with a little help from Moritz Bunkus - * totally reworked by Aurelien Jacobs - * Specs available on the Matroska project page: http://www.matroska.org/. + * @author Ronald Bultje + * @author with a little help from Moritz Bunkus + * @author totally reworked by Aurelien Jacobs + * @see specs available on the Matroska project page: http://www.matroska.org/ */ #include @@ -38,10 +38,11 @@ #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" +#include "libavutil/dict.h" #if CONFIG_ZLIB #include #endif @@ -127,6 +128,7 @@ typedef struct { int sub_packet_size; int sub_packet_cnt; int pkt_cnt; + uint64_t buf_timecode; uint8_t *buf; } MatroskaTrackAudio; @@ -210,6 +212,11 @@ typedef struct { uint64_t length; } MatroskaLevel; +typedef struct { + uint64_t timecode; + EbmlList blocks; +} MatroskaCluster; + typedef struct { AVFormatContext *ctx; @@ -242,6 +249,16 @@ typedef struct { /* What to skip before effectively reading a packet. */ int skip_to_keyframe; uint64_t skip_to_timecode; + + /* 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 { @@ -251,11 +268,6 @@ 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} }, @@ -509,7 +521,39 @@ static EbmlSyntax matroska_clusters[] = { { 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. @@ -553,8 +597,9 @@ static int ebml_read_num(MatroskaDemuxContext *matroska, AVIOContext *pb, 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 */ @@ -619,9 +664,9 @@ static int ebml_read_float(AVIOContext *pb, int size, double *num) 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; @@ -634,16 +679,19 @@ static int ebml_read_float(AVIOContext *pb, int size, double *num) */ 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; } @@ -736,8 +784,11 @@ static int ebml_parse_id(MatroskaDemuxContext *matroska, EbmlSyntax *syntax, 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); } @@ -796,11 +847,15 @@ static int ebml_parse_elem(MatroskaDemuxContext *matroska, 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++; @@ -898,6 +953,8 @@ static int matroska_probe(AVProbeData *p) * Not fully fool-proof, but good enough. */ for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++) { int probelen = strlen(matroska_doctypes[i]); + if (total < probelen) + continue; for (n = 4+size; n <= 4+size+total-probelen; n++) if (!memcmp(p->buf+n, matroska_doctypes[i], probelen)) return AVPROBE_SCORE_MAX; @@ -928,6 +985,7 @@ static int matroska_decode_buffer(uint8_t** buf, int* buf_size, 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; @@ -957,7 +1015,12 @@ static int matroska_decode_buffer(uint8_t** buf, int* buf_size, 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); @@ -978,7 +1041,12 @@ static int matroska_decode_buffer(uint8_t** buf, int* buf_size, 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); @@ -1033,17 +1101,21 @@ static void matroska_fix_ass_packet(MatroskaDemuxContext *matroska, } } -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, - AVMetadata **metadata, char *prefix) + AVDictionary **metadata, char *prefix) { MatroskaTag *tags = list->elem; char key[1024]; @@ -1059,14 +1131,14 @@ static void matroska_convert_tag(AVFormatContext *s, EbmlList *list, if (prefix) snprintf(key, sizeof(key), "%s/%s", prefix, tags[i].name); else av_strlcpy(key, tags[i].name, sizeof(key)); if (tags[i].def || !lang) { - av_metadata_set2(metadata, key, tags[i].string, 0); + av_dict_set(metadata, key, tags[i].string, 0); if (tags[i].sub.nb_elem) matroska_convert_tag(s, &tags[i].sub, metadata, key); } if (lang) { av_strlcat(key, "-", sizeof(key)); av_strlcat(key, lang, sizeof(key)); - av_metadata_set2(metadata, key, tags[i].string, 0); + av_dict_set(metadata, key, tags[i].string, 0); if (tags[i].sub.nb_elem) matroska_convert_tag(s, &tags[i].sub, metadata, key); } @@ -1084,19 +1156,21 @@ static void matroska_convert_tags(AVFormatContext *s) if (tags[i].target.attachuid) { MatroskaAttachement *attachment = matroska->attachments.elem; for (j=0; jattachments.nb_elem; j++) - if (attachment[j].uid == tags[i].target.attachuid) + if (attachment[j].uid == tags[i].target.attachuid + && attachment[j].stream) matroska_convert_tag(s, &tags[i].tag, &attachment[j].stream->metadata, NULL); } else if (tags[i].target.chapteruid) { MatroskaChapter *chapter = matroska->chapters.elem; for (j=0; jchapters.nb_elem; j++) - if (chapter[j].uid == tags[i].target.chapteruid) + if (chapter[j].uid == tags[i].target.chapteruid + && chapter[j].chapter) matroska_convert_tag(s, &tags[i].tag, &chapter[j].chapter->metadata, NULL); } else if (tags[i].target.trackuid) { MatroskaTrack *track = matroska->tracks.elem; for (j=0; jtracks.nb_elem; j++) - if (track[j].uid == tags[i].target.trackuid) + if (track[j].uid == tags[i].target.trackuid && track[j].stream) matroska_convert_tag(s, &tags[i].tag, &track[j].stream->metadata, NULL); } else { @@ -1106,7 +1180,7 @@ static void matroska_convert_tags(AVFormatContext *s) } } -static void matroska_execute_seekhead(MatroskaDemuxContext *matroska) +static int matroska_parse_seekhead_entry(MatroskaDemuxContext *matroska, int idx) { EbmlList *seekhead_list = &matroska->seekhead; MatroskaSeekhead *seekhead = seekhead_list->elem; @@ -1114,54 +1188,110 @@ static void matroska_execute_seekhead(MatroskaDemuxContext *matroska) int64_t before_pos = avio_tell(matroska->ctx->pb); uint32_t saved_id = matroska->current_id; MatroskaLevel level; + int64_t offset; + int ret = 0; + + if (idx >= seekhead_list->nb_elem + || seekhead[idx].id == MATROSKA_ID_SEEKHEAD + || seekhead[idx].id == MATROSKA_ID_CLUSTER) + return 0; + + /* seek */ + offset = seekhead[idx].pos + matroska->segment_start; + if (avio_seek(matroska->ctx->pb, offset, SEEK_SET) == offset) { + /* We don't want to lose our seekhead level, so we add + * a dummy. This is a crude hack. */ + if (matroska->num_levels == EBML_MAX_DEPTH) { + av_log(matroska->ctx, AV_LOG_INFO, + "Max EBML element depth (%d) reached, " + "cannot parse further.\n", EBML_MAX_DEPTH); + ret = AVERROR_INVALIDDATA; + } else { + level.start = 0; + level.length = (uint64_t)-1; + matroska->levels[matroska->num_levels] = level; + matroska->num_levels++; + matroska->current_id = 0; + + ret = ebml_parse(matroska, matroska_segment, matroska); + + /* remove dummy level */ + while (matroska->num_levels) { + uint64_t length = matroska->levels[--matroska->num_levels].length; + if (length == (uint64_t)-1) + break; + } + } + } + /* seek back */ + avio_seek(matroska->ctx->pb, before_pos, SEEK_SET); + matroska->level_up = level_up; + matroska->current_id = saved_id; + + return ret; +} + +static void matroska_execute_seekhead(MatroskaDemuxContext *matroska) +{ + EbmlList *seekhead_list = &matroska->seekhead; + int64_t before_pos = avio_tell(matroska->ctx->pb); int i; // we should not do any seeking in the streaming case - if (url_is_streamed(matroska->ctx->pb) || + if (!matroska->ctx->pb->seekable || (matroska->ctx->flags & AVFMT_FLAG_IGNIDX)) return; - for (i=0; inb_elem; i++) { - int64_t offset = seekhead[i].pos + matroska->segment_start; - - if (seekhead[i].pos <= before_pos - || seekhead[i].id == MATROSKA_ID_SEEKHEAD - || seekhead[i].id == MATROSKA_ID_CLUSTER) + for (i = 0; i < seekhead_list->nb_elem; i++) { + MatroskaSeekhead *seekhead = seekhead_list->elem; + if (seekhead[i].pos <= before_pos) continue; - /* seek */ - if (avio_seek(matroska->ctx->pb, offset, SEEK_SET) != offset) + // defer cues parsing until we actually need cue data. + if (seekhead[i].id == MATROSKA_ID_CUES) { + matroska->cues_parsing_deferred = 1; continue; + } - /* We don't want to lose our seekhead level, so we add - * a dummy. This is a crude hack. */ - if (matroska->num_levels == EBML_MAX_DEPTH) { - av_log(matroska->ctx, AV_LOG_INFO, - "Max EBML element depth (%d) reached, " - "cannot parse further.\n", EBML_MAX_DEPTH); + if (matroska_parse_seekhead_entry(matroska, i) < 0) break; - } + } +} - level.start = 0; - level.length = (uint64_t)-1; - matroska->levels[matroska->num_levels] = level; - matroska->num_levels++; - matroska->current_id = 0; +static void matroska_parse_cues(MatroskaDemuxContext *matroska) { + EbmlList *seekhead_list = &matroska->seekhead; + MatroskaSeekhead *seekhead = seekhead_list->elem; + EbmlList *index_list; + MatroskaIndex *index; + int index_scale = 1; + int i, j; - ebml_parse(matroska, matroska_segment, matroska); + for (i = 0; i < seekhead_list->nb_elem; i++) + if (seekhead[i].id == MATROSKA_ID_CUES) + break; + assert(i <= seekhead_list->nb_elem); - /* remove dummy level */ - while (matroska->num_levels) { - uint64_t length = matroska->levels[--matroska->num_levels].length; - if (length == (uint64_t)-1) - break; + matroska_parse_seekhead_entry(matroska, i); + + index_list = &matroska->index; + index = index_list->elem; + if (index_list->nb_elem + && index[0].time > 1E14/matroska->time_scale) { + av_log(matroska->ctx, AV_LOG_WARNING, "Working around broken index.\n"); + index_scale = matroska->time_scale; + } + for (i = 0; i < index_list->nb_elem; i++) { + EbmlList *pos_list = &index[i].pos; + MatroskaIndexPos *pos = pos_list->elem; + for (j = 0; j < pos_list->nb_elem; j++) { + MatroskaTrack *track = matroska_find_track_by_num(matroska, pos[j].track); + if (track && track->stream) + av_add_index_entry(track->stream, + pos[j].pos + matroska->segment_start, + index[i].time/index_scale, 0, 0, + AVINDEX_KEYFRAME); } } - - /* seek back */ - avio_seek(matroska->ctx->pb, before_pos, SEEK_SET); - matroska->level_up = level_up; - matroska->current_id = saved_id; } static int matroska_aac_profile(char *codec_id) @@ -1179,13 +1309,13 @@ static int matroska_aac_sri(int samplerate) { int sri; - for (sri=0; sripriv_data; EbmlList *attachements_list = &matroska->attachments; @@ -1193,9 +1323,6 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) EbmlList *chapters_list = &matroska->chapters; MatroskaChapter *chapters; MatroskaTrack *tracks; - EbmlList *index_list; - MatroskaIndex *index; - int index_scale = 1; uint64_t max_start = 0; Ebml ebml = { 0 }; AVStream *st; @@ -1232,7 +1359,7 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) if (matroska->duration) matroska->ctx->duration = matroska->duration * matroska->time_scale * 1000 / AV_TIME_BASE; - av_metadata_set2(&s->metadata, "title", matroska->title, 0); + av_dict_set(&s->metadata, "title", matroska->title, 0); tracks = matroska->tracks.elem; for (i=0; i < matroska->tracks.nb_elem; i++) { @@ -1258,7 +1385,7 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) 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; @@ -1270,7 +1397,7 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) } 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 && @@ -1315,7 +1442,7 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) } } - st = track->stream = av_new_stream(s, 0); + st = track->stream = avformat_new_stream(s, NULL); if (st == NULL) return AVERROR(ENOMEM); @@ -1329,16 +1456,19 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) } else if (!strcmp(track->codec_id, "A_MS/ACM") && track->codec_priv.size >= 14 && track->codec_priv.data != NULL) { + int ret; ffio_init_context(&b, track->codec_priv.data, track->codec_priv.size, - URL_RDONLY, NULL, NULL, NULL, NULL); - ff_get_wav_header(&b, st->codec, track->codec_priv.size); + AVIO_FLAG_READ, NULL, NULL, NULL, NULL); + ret = ff_get_wav_header(&b, st->codec, track->codec_priv.size); + if (ret < 0) + return ret; codec_id = st->codec->codec_id; extradata_offset = FFMIN(track->codec_priv.size, 18); } else if (!strcmp(track->codec_id, "V_QUICKTIME") && (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; @@ -1356,7 +1486,7 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) } 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); @@ -1422,23 +1552,19 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) 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 (strcmp(track->language, "und")) - av_metadata_set2(&st->metadata, "language", track->language, 0); - av_metadata_set2(&st->metadata, "title", track->name, 0); + av_dict_set(&st->metadata, "language", track->language, 0); + av_dict_set(&st->metadata, "title", track->name, 0); if (track->flag_default) st->disposition |= AV_DISPOSITION_DEFAULT; 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; @@ -1467,8 +1593,11 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) 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; @@ -1477,6 +1606,8 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) 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; } } @@ -1486,10 +1617,11 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) 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_metadata_set2(&st->metadata, "filename",attachements[j].filename, 0); + 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); @@ -1514,35 +1646,14 @@ static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap) 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_metadata_set2(&chapters[i].chapter->metadata, + av_dict_set(&chapters[i].chapter->metadata, "title", chapters[i].title, 0); max_start = chapters[i].start; } - index_list = &matroska->index; - index = index_list->elem; - if (index_list->nb_elem - && index[0].time > 100000000000000/matroska->time_scale) { - av_log(matroska->ctx, AV_LOG_WARNING, "Working around broken index.\n"); - index_scale = matroska->time_scale; - } - for (i=0; inb_elem; i++) { - EbmlList *pos_list = &index[i].pos; - MatroskaIndexPos *pos = pos_list->elem; - for (j=0; jnb_elem; j++) { - MatroskaTrack *track = matroska_find_track_by_num(matroska, - pos[j].track); - if (track && track->stream) - av_add_index_entry(track->stream, - pos[j].pos + matroska->segment_start, - index[i].time/index_scale, 0, 0, - AVINDEX_KEYFRAME); - } - } - matroska_convert_tags(s); return 0; @@ -1559,13 +1670,16 @@ static int matroska_deliver_packet(MatroskaDemuxContext *matroska, 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; @@ -1607,17 +1721,18 @@ static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, 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 res; - } + return AVERROR_INVALIDDATA; + } else if (size <= 3) + return 0; st = track->stream; if (st->discard >= AVDISCARD_ALL) return res; @@ -1717,7 +1832,7 @@ static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, lace_size[n] = lace_size[n - 1] + snum; total += lace_size[n]; } - lace_size[n] = size - total; + lace_size[laces - 1] = size - total; break; } } @@ -1740,15 +1855,36 @@ static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, int x; if (!track->audio.pkt_cnt) { - if (st->codec->codec_id == CODEC_ID_RA_288) + if (track->audio.sub_packet_cnt == 0) + track->audio.buf_timecode = timecode; + 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; xaudio.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; xaudio.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) @@ -1762,6 +1898,8 @@ static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, av_new_packet(pkt, a); memcpy(pkt->data, track->audio.buf + a * (h*w / a - track->audio.pkt_cnt--), a); + pkt->pts = track->audio.buf_timecode; + track->audio.buf_timecode = AV_NOPTS_VALUE; pkt->pos = pos; pkt->stream_index = st->index; dynarray_add(&matroska->packets,&matroska->num_packets,pkt); @@ -1832,24 +1970,83 @@ static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, } } +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 */ res = ebml_parse(matroska, matroska_clusters, &cluster); blocks_list = &cluster.blocks; blocks = blocks_list->elem; - for (i=0; inb_elem; i++) + for (i=0; inb_elem && !res; i++) 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) @@ -1868,14 +2065,20 @@ static int matroska_parse_cluster(MatroskaDemuxContext *matroska) static int matroska_read_packet(AVFormatContext *s, AVPacket *pkt) { MatroskaDemuxContext *matroska = s->priv_data; + int ret = 0; - while (matroska_deliver_packet(matroska, pkt)) { + while (!ret && matroska_deliver_packet(matroska, pkt)) { if (matroska->done) return AVERROR_EOF; - matroska_parse_cluster(matroska); + ret = matroska_parse_cluster(matroska); } - return 0; + if (ret == AVERROR_INVALIDDATA) { + pkt->flags |= AV_PKT_FLAG_CORRUPT; + return 0; + } + + return ret; } static int matroska_read_seek(AVFormatContext *s, int stream_index, @@ -1886,13 +2089,21 @@ static int matroska_read_seek(AVFormatContext *s, int stream_index, AVStream *st = s->streams[stream_index]; int i, index, index_sub, index_min; + /* Parse the CUES now since we need the index data to seek. */ + if (matroska->cues_parsing_deferred) { + matroska_parse_cues(matroska); + matroska->cues_parsing_deferred = 0; + } + if (!st->nb_index_entries) return 0; timestamp = FFMAX(timestamp, st->index_entries[0].timestamp); if ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) { 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; @@ -1905,6 +2116,9 @@ static int matroska_read_seek(AVFormatContext *s, int stream_index, index_min = index; for (i=0; i < matroska->tracks.nb_elem; i++) { + tracks[i].audio.pkt_cnt = 0; + tracks[i].audio.sub_packet_cnt = 0; + tracks[i].audio.buf_timecode = AV_NOPTS_VALUE; tracks[i].end_timecode = 0; if (tracks[i].type == MATROSKA_TRACK_TYPE_SUBTITLE && !tracks[i].stream->discard != AVDISCARD_ALL) { @@ -1917,10 +2131,11 @@ static int matroska_read_seek(AVFormatContext *s, int stream_index, } avio_seek(s->pb, st->index_entries[index_min].pos, SEEK_SET); + matroska->current_id = 0; 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; } @@ -1935,18 +2150,19 @@ static int matroska_read_close(AVFormatContext *s) 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; } AVInputFormat ff_matroska_demuxer = { - "matroska,webm", - NULL_IF_CONFIG_SMALL("Matroska/WebM file format"), - sizeof(MatroskaDemuxContext), - matroska_probe, - matroska_read_header, - matroska_read_packet, - matroska_read_close, - matroska_read_seek, + .name = "matroska,webm", + .long_name = NULL_IF_CONFIG_SMALL("Matroska/WebM file format"), + .priv_data_size = sizeof(MatroskaDemuxContext), + .read_probe = matroska_probe, + .read_header = matroska_read_header, + .read_packet = matroska_read_packet, + .read_close = matroska_read_close, + .read_seek = matroska_read_seek, };