FFmpeg: libavformat/matroskadec.c Source File

00001 /*
00002  * Matroska file demuxer
00003  * Copyright (c) 2003-2008 The FFmpeg Project
00004  *
00005  * This file is part of FFmpeg.
00006  *
00007  * FFmpeg is free software; you can redistribute it and/or
00008  * modify it under the terms of the GNU Lesser General Public
00009  * License as published by the Free Software Foundation; either
00010  * version 2.1 of the License, or (at your option) any later version.
00011  *
00012  * FFmpeg is distributed in the hope that it will be useful,
00013  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00014  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00015  * Lesser General Public License for more details.
00016  *
00017  * You should have received a copy of the GNU Lesser General Public
00018  * License along with FFmpeg; if not, write to the Free Software
00019  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00020  */
00021 
00031 #include <stdio.h>
00032 #include "avformat.h"
00033 #include "internal.h"
00034 #include "avio_internal.h"
00035 /* For ff_codec_get_id(). */
00036 #include "riff.h"
00037 #include "isom.h"
00038 #include "rm.h"
00039 #include "matroska.h"
00040 #include "libavcodec/mpeg4audio.h"
00041 #include "libavutil/intfloat_readwrite.h"
00042 #include "libavutil/intreadwrite.h"
00043 #include "libavutil/avstring.h"
00044 #include "libavutil/lzo.h"
00045 #if CONFIG_ZLIB
00046 #include <zlib.h>
00047 #endif
00048 #if CONFIG_BZLIB
00049 #include <bzlib.h>
00050 #endif
00051 
00052 typedef enum {
00053     EBML_NONE,
00054     EBML_UINT,
00055     EBML_FLOAT,
00056     EBML_STR,
00057     EBML_UTF8,
00058     EBML_BIN,
00059     EBML_NEST,
00060     EBML_PASS,
00061     EBML_STOP,
00062     EBML_TYPE_COUNT
00063 } EbmlType;
00064 
00065 typedef const struct EbmlSyntax {
00066     uint32_t id;
00067     EbmlType type;
00068     int list_elem_size;
00069     int data_offset;
00070     union {
00071         uint64_t    u;
00072         double      f;
00073         const char *s;
00074         const struct EbmlSyntax *n;
00075     } def;
00076 } EbmlSyntax;
00077 
00078 typedef struct {
00079     int nb_elem;
00080     void *elem;
00081 } EbmlList;
00082 
00083 typedef struct {
00084     int      size;
00085     uint8_t *data;
00086     int64_t  pos;
00087 } EbmlBin;
00088 
00089 typedef struct {
00090     uint64_t version;
00091     uint64_t max_size;
00092     uint64_t id_length;
00093     char    *doctype;
00094     uint64_t doctype_version;
00095 } Ebml;
00096 
00097 typedef struct {
00098     uint64_t algo;
00099     EbmlBin  settings;
00100 } MatroskaTrackCompression;
00101 
00102 typedef struct {
00103     uint64_t scope;
00104     uint64_t type;
00105     MatroskaTrackCompression compression;
00106 } MatroskaTrackEncoding;
00107 
00108 typedef struct {
00109     double   frame_rate;
00110     uint64_t display_width;
00111     uint64_t display_height;
00112     uint64_t pixel_width;
00113     uint64_t pixel_height;
00114     uint64_t fourcc;
00115 } MatroskaTrackVideo;
00116 
00117 typedef struct {
00118     double   samplerate;
00119     double   out_samplerate;
00120     uint64_t bitdepth;
00121     uint64_t channels;
00122 
00123     /* real audio header (extracted from extradata) */
00124     int      coded_framesize;
00125     int      sub_packet_h;
00126     int      frame_size;
00127     int      sub_packet_size;
00128     int      sub_packet_cnt;
00129     int      pkt_cnt;
00130     uint64_t buf_timecode;
00131     uint8_t *buf;
00132 } MatroskaTrackAudio;
00133 
00134 typedef struct {
00135     uint64_t num;
00136     uint64_t uid;
00137     uint64_t type;
00138     char    *name;
00139     char    *codec_id;
00140     EbmlBin  codec_priv;
00141     char    *language;
00142     double time_scale;
00143     uint64_t default_duration;
00144     uint64_t flag_default;
00145     uint64_t flag_forced;
00146     MatroskaTrackVideo video;
00147     MatroskaTrackAudio audio;
00148     EbmlList encodings;
00149 
00150     AVStream *stream;
00151     int64_t end_timecode;
00152     int ms_compat;
00153 } MatroskaTrack;
00154 
00155 typedef struct {
00156     uint64_t uid;
00157     char *filename;
00158     char *mime;
00159     EbmlBin bin;
00160 
00161     AVStream *stream;
00162 } MatroskaAttachement;
00163 
00164 typedef struct {
00165     uint64_t start;
00166     uint64_t end;
00167     uint64_t uid;
00168     char    *title;
00169 
00170     AVChapter *chapter;
00171 } MatroskaChapter;
00172 
00173 typedef struct {
00174     uint64_t track;
00175     uint64_t pos;
00176 } MatroskaIndexPos;
00177 
00178 typedef struct {
00179     uint64_t time;
00180     EbmlList pos;
00181 } MatroskaIndex;
00182 
00183 typedef struct {
00184     char *name;
00185     char *string;
00186     char *lang;
00187     uint64_t def;
00188     EbmlList sub;
00189 } MatroskaTag;
00190 
00191 typedef struct {
00192     char    *type;
00193     uint64_t typevalue;
00194     uint64_t trackuid;
00195     uint64_t chapteruid;
00196     uint64_t attachuid;
00197 } MatroskaTagTarget;
00198 
00199 typedef struct {
00200     MatroskaTagTarget target;
00201     EbmlList tag;
00202 } MatroskaTags;
00203 
00204 typedef struct {
00205     uint64_t id;
00206     uint64_t pos;
00207 } MatroskaSeekhead;
00208 
00209 typedef struct {
00210     uint64_t start;
00211     uint64_t length;
00212 } MatroskaLevel;
00213 
00214 typedef struct {
00215     AVFormatContext *ctx;
00216 
00217     /* EBML stuff */
00218     int num_levels;
00219     MatroskaLevel levels[EBML_MAX_DEPTH];
00220     int level_up;
00221     uint32_t current_id;
00222 
00223     uint64_t time_scale;
00224     double   duration;
00225     char    *title;
00226     EbmlList tracks;
00227     EbmlList attachments;
00228     EbmlList chapters;
00229     EbmlList index;
00230     EbmlList tags;
00231     EbmlList seekhead;
00232 
00233     /* byte position of the segment inside the stream */
00234     int64_t segment_start;
00235 
00236     /* the packet queue */
00237     AVPacket **packets;
00238     int num_packets;
00239     AVPacket *prev_pkt;
00240 
00241     int done;
00242 
00243     /* What to skip before effectively reading a packet. */
00244     int skip_to_keyframe;
00245     uint64_t skip_to_timecode;
00246 } MatroskaDemuxContext;
00247 
00248 typedef struct {
00249     uint64_t duration;
00250     int64_t  reference;
00251     uint64_t non_simple;
00252     EbmlBin  bin;
00253 } MatroskaBlock;
00254 
00255 typedef struct {
00256     uint64_t timecode;
00257     EbmlList blocks;
00258 } MatroskaCluster;
00259 
00260 static EbmlSyntax ebml_header[] = {
00261     { EBML_ID_EBMLREADVERSION,        EBML_UINT, 0, offsetof(Ebml,version), {.u=EBML_VERSION} },
00262     { EBML_ID_EBMLMAXSIZELENGTH,      EBML_UINT, 0, offsetof(Ebml,max_size), {.u=8} },
00263     { EBML_ID_EBMLMAXIDLENGTH,        EBML_UINT, 0, offsetof(Ebml,id_length), {.u=4} },
00264     { EBML_ID_DOCTYPE,                EBML_STR,  0, offsetof(Ebml,doctype), {.s="(none)"} },
00265     { EBML_ID_DOCTYPEREADVERSION,     EBML_UINT, 0, offsetof(Ebml,doctype_version), {.u=1} },
00266     { EBML_ID_EBMLVERSION,            EBML_NONE },
00267     { EBML_ID_DOCTYPEVERSION,         EBML_NONE },
00268     { 0 }
00269 };
00270 
00271 static EbmlSyntax ebml_syntax[] = {
00272     { EBML_ID_HEADER,                 EBML_NEST, 0, 0, {.n=ebml_header} },
00273     { 0 }
00274 };
00275 
00276 static EbmlSyntax matroska_info[] = {
00277     { MATROSKA_ID_TIMECODESCALE,      EBML_UINT,  0, offsetof(MatroskaDemuxContext,time_scale), {.u=1000000} },
00278     { MATROSKA_ID_DURATION,           EBML_FLOAT, 0, offsetof(MatroskaDemuxContext,duration) },
00279     { MATROSKA_ID_TITLE,              EBML_UTF8,  0, offsetof(MatroskaDemuxContext,title) },
00280     { MATROSKA_ID_WRITINGAPP,         EBML_NONE },
00281     { MATROSKA_ID_MUXINGAPP,          EBML_NONE },
00282     { MATROSKA_ID_DATEUTC,            EBML_NONE },
00283     { MATROSKA_ID_SEGMENTUID,         EBML_NONE },
00284     { 0 }
00285 };
00286 
00287 static EbmlSyntax matroska_track_video[] = {
00288     { MATROSKA_ID_VIDEOFRAMERATE,     EBML_FLOAT,0, offsetof(MatroskaTrackVideo,frame_rate) },
00289     { MATROSKA_ID_VIDEODISPLAYWIDTH,  EBML_UINT, 0, offsetof(MatroskaTrackVideo,display_width) },
00290     { MATROSKA_ID_VIDEODISPLAYHEIGHT, EBML_UINT, 0, offsetof(MatroskaTrackVideo,display_height) },
00291     { MATROSKA_ID_VIDEOPIXELWIDTH,    EBML_UINT, 0, offsetof(MatroskaTrackVideo,pixel_width) },
00292     { MATROSKA_ID_VIDEOPIXELHEIGHT,   EBML_UINT, 0, offsetof(MatroskaTrackVideo,pixel_height) },
00293     { MATROSKA_ID_VIDEOCOLORSPACE,    EBML_UINT, 0, offsetof(MatroskaTrackVideo,fourcc) },
00294     { MATROSKA_ID_VIDEOPIXELCROPB,    EBML_NONE },
00295     { MATROSKA_ID_VIDEOPIXELCROPT,    EBML_NONE },
00296     { MATROSKA_ID_VIDEOPIXELCROPL,    EBML_NONE },
00297     { MATROSKA_ID_VIDEOPIXELCROPR,    EBML_NONE },
00298     { MATROSKA_ID_VIDEODISPLAYUNIT,   EBML_NONE },
00299     { MATROSKA_ID_VIDEOFLAGINTERLACED,EBML_NONE },
00300     { MATROSKA_ID_VIDEOSTEREOMODE,    EBML_NONE },
00301     { MATROSKA_ID_VIDEOASPECTRATIO,   EBML_NONE },
00302     { 0 }
00303 };
00304 
00305 static EbmlSyntax matroska_track_audio[] = {
00306     { MATROSKA_ID_AUDIOSAMPLINGFREQ,  EBML_FLOAT,0, offsetof(MatroskaTrackAudio,samplerate), {.f=8000.0} },
00307     { MATROSKA_ID_AUDIOOUTSAMPLINGFREQ,EBML_FLOAT,0,offsetof(MatroskaTrackAudio,out_samplerate) },
00308     { MATROSKA_ID_AUDIOBITDEPTH,      EBML_UINT, 0, offsetof(MatroskaTrackAudio,bitdepth) },
00309     { MATROSKA_ID_AUDIOCHANNELS,      EBML_UINT, 0, offsetof(MatroskaTrackAudio,channels), {.u=1} },
00310     { 0 }
00311 };
00312 
00313 static EbmlSyntax matroska_track_encoding_compression[] = {
00314     { MATROSKA_ID_ENCODINGCOMPALGO,   EBML_UINT, 0, offsetof(MatroskaTrackCompression,algo), {.u=0} },
00315     { MATROSKA_ID_ENCODINGCOMPSETTINGS,EBML_BIN, 0, offsetof(MatroskaTrackCompression,settings) },
00316     { 0 }
00317 };
00318 
00319 static EbmlSyntax matroska_track_encoding[] = {
00320     { MATROSKA_ID_ENCODINGSCOPE,      EBML_UINT, 0, offsetof(MatroskaTrackEncoding,scope), {.u=1} },
00321     { MATROSKA_ID_ENCODINGTYPE,       EBML_UINT, 0, offsetof(MatroskaTrackEncoding,type), {.u=0} },
00322     { MATROSKA_ID_ENCODINGCOMPRESSION,EBML_NEST, 0, offsetof(MatroskaTrackEncoding,compression), {.n=matroska_track_encoding_compression} },
00323     { MATROSKA_ID_ENCODINGORDER,      EBML_NONE },
00324     { 0 }
00325 };
00326 
00327 static EbmlSyntax matroska_track_encodings[] = {
00328     { MATROSKA_ID_TRACKCONTENTENCODING, EBML_NEST, sizeof(MatroskaTrackEncoding), offsetof(MatroskaTrack,encodings), {.n=matroska_track_encoding} },
00329     { 0 }
00330 };
00331 
00332 static EbmlSyntax matroska_track[] = {
00333     { MATROSKA_ID_TRACKNUMBER,          EBML_UINT, 0, offsetof(MatroskaTrack,num) },
00334     { MATROSKA_ID_TRACKNAME,            EBML_UTF8, 0, offsetof(MatroskaTrack,name) },
00335     { MATROSKA_ID_TRACKUID,             EBML_UINT, 0, offsetof(MatroskaTrack,uid) },
00336     { MATROSKA_ID_TRACKTYPE,            EBML_UINT, 0, offsetof(MatroskaTrack,type) },
00337     { MATROSKA_ID_CODECID,              EBML_STR,  0, offsetof(MatroskaTrack,codec_id) },
00338     { MATROSKA_ID_CODECPRIVATE,         EBML_BIN,  0, offsetof(MatroskaTrack,codec_priv) },
00339     { MATROSKA_ID_TRACKLANGUAGE,        EBML_UTF8, 0, offsetof(MatroskaTrack,language), {.s="eng"} },
00340     { MATROSKA_ID_TRACKDEFAULTDURATION, EBML_UINT, 0, offsetof(MatroskaTrack,default_duration) },
00341     { MATROSKA_ID_TRACKTIMECODESCALE,   EBML_FLOAT,0, offsetof(MatroskaTrack,time_scale), {.f=1.0} },
00342     { MATROSKA_ID_TRACKFLAGDEFAULT,     EBML_UINT, 0, offsetof(MatroskaTrack,flag_default), {.u=1} },
00343     { MATROSKA_ID_TRACKFLAGFORCED,      EBML_UINT, 0, offsetof(MatroskaTrack,flag_forced), {.u=0} },
00344     { MATROSKA_ID_TRACKVIDEO,           EBML_NEST, 0, offsetof(MatroskaTrack,video), {.n=matroska_track_video} },
00345     { MATROSKA_ID_TRACKAUDIO,           EBML_NEST, 0, offsetof(MatroskaTrack,audio), {.n=matroska_track_audio} },
00346     { MATROSKA_ID_TRACKCONTENTENCODINGS,EBML_NEST, 0, 0, {.n=matroska_track_encodings} },
00347     { MATROSKA_ID_TRACKFLAGENABLED,     EBML_NONE },
00348     { MATROSKA_ID_TRACKFLAGLACING,      EBML_NONE },
00349     { MATROSKA_ID_CODECNAME,            EBML_NONE },
00350     { MATROSKA_ID_CODECDECODEALL,       EBML_NONE },
00351     { MATROSKA_ID_CODECINFOURL,         EBML_NONE },
00352     { MATROSKA_ID_CODECDOWNLOADURL,     EBML_NONE },
00353     { MATROSKA_ID_TRACKMINCACHE,        EBML_NONE },
00354     { MATROSKA_ID_TRACKMAXCACHE,        EBML_NONE },
00355     { MATROSKA_ID_TRACKMAXBLKADDID,     EBML_NONE },
00356     { 0 }
00357 };
00358 
00359 static EbmlSyntax matroska_tracks[] = {
00360     { MATROSKA_ID_TRACKENTRY,         EBML_NEST, sizeof(MatroskaTrack), offsetof(MatroskaDemuxContext,tracks), {.n=matroska_track} },
00361     { 0 }
00362 };
00363 
00364 static EbmlSyntax matroska_attachment[] = {
00365     { MATROSKA_ID_FILEUID,            EBML_UINT, 0, offsetof(MatroskaAttachement,uid) },
00366     { MATROSKA_ID_FILENAME,           EBML_UTF8, 0, offsetof(MatroskaAttachement,filename) },
00367     { MATROSKA_ID_FILEMIMETYPE,       EBML_STR,  0, offsetof(MatroskaAttachement,mime) },
00368     { MATROSKA_ID_FILEDATA,           EBML_BIN,  0, offsetof(MatroskaAttachement,bin) },
00369     { MATROSKA_ID_FILEDESC,           EBML_NONE },
00370     { 0 }
00371 };
00372 
00373 static EbmlSyntax matroska_attachments[] = {
00374     { MATROSKA_ID_ATTACHEDFILE,       EBML_NEST, sizeof(MatroskaAttachement), offsetof(MatroskaDemuxContext,attachments), {.n=matroska_attachment} },
00375     { 0 }
00376 };
00377 
00378 static EbmlSyntax matroska_chapter_display[] = {
00379     { MATROSKA_ID_CHAPSTRING,         EBML_UTF8, 0, offsetof(MatroskaChapter,title) },
00380     { MATROSKA_ID_CHAPLANG,           EBML_NONE },
00381     { 0 }
00382 };
00383 
00384 static EbmlSyntax matroska_chapter_entry[] = {
00385     { MATROSKA_ID_CHAPTERTIMESTART,   EBML_UINT, 0, offsetof(MatroskaChapter,start), {.u=AV_NOPTS_VALUE} },
00386     { MATROSKA_ID_CHAPTERTIMEEND,     EBML_UINT, 0, offsetof(MatroskaChapter,end), {.u=AV_NOPTS_VALUE} },
00387     { MATROSKA_ID_CHAPTERUID,         EBML_UINT, 0, offsetof(MatroskaChapter,uid) },
00388     { MATROSKA_ID_CHAPTERDISPLAY,     EBML_NEST, 0, 0, {.n=matroska_chapter_display} },
00389     { MATROSKA_ID_CHAPTERFLAGHIDDEN,  EBML_NONE },
00390     { MATROSKA_ID_CHAPTERFLAGENABLED, EBML_NONE },
00391     { MATROSKA_ID_CHAPTERPHYSEQUIV,   EBML_NONE },
00392     { MATROSKA_ID_CHAPTERATOM,        EBML_NONE },
00393     { 0 }
00394 };
00395 
00396 static EbmlSyntax matroska_chapter[] = {
00397     { MATROSKA_ID_CHAPTERATOM,        EBML_NEST, sizeof(MatroskaChapter), offsetof(MatroskaDemuxContext,chapters), {.n=matroska_chapter_entry} },
00398     { MATROSKA_ID_EDITIONUID,         EBML_NONE },
00399     { MATROSKA_ID_EDITIONFLAGHIDDEN,  EBML_NONE },
00400     { MATROSKA_ID_EDITIONFLAGDEFAULT, EBML_NONE },
00401     { MATROSKA_ID_EDITIONFLAGORDERED, EBML_NONE },
00402     { 0 }
00403 };
00404 
00405 static EbmlSyntax matroska_chapters[] = {
00406     { MATROSKA_ID_EDITIONENTRY,       EBML_NEST, 0, 0, {.n=matroska_chapter} },
00407     { 0 }
00408 };
00409 
00410 static EbmlSyntax matroska_index_pos[] = {
00411     { MATROSKA_ID_CUETRACK,           EBML_UINT, 0, offsetof(MatroskaIndexPos,track) },
00412     { MATROSKA_ID_CUECLUSTERPOSITION, EBML_UINT, 0, offsetof(MatroskaIndexPos,pos)   },
00413     { MATROSKA_ID_CUEBLOCKNUMBER,     EBML_NONE },
00414     { 0 }
00415 };
00416 
00417 static EbmlSyntax matroska_index_entry[] = {
00418     { MATROSKA_ID_CUETIME,            EBML_UINT, 0, offsetof(MatroskaIndex,time) },
00419     { MATROSKA_ID_CUETRACKPOSITION,   EBML_NEST, sizeof(MatroskaIndexPos), offsetof(MatroskaIndex,pos), {.n=matroska_index_pos} },
00420     { 0 }
00421 };
00422 
00423 static EbmlSyntax matroska_index[] = {
00424     { MATROSKA_ID_POINTENTRY,         EBML_NEST, sizeof(MatroskaIndex), offsetof(MatroskaDemuxContext,index), {.n=matroska_index_entry} },
00425     { 0 }
00426 };
00427 
00428 static EbmlSyntax matroska_simpletag[] = {
00429     { MATROSKA_ID_TAGNAME,            EBML_UTF8, 0, offsetof(MatroskaTag,name) },
00430     { MATROSKA_ID_TAGSTRING,          EBML_UTF8, 0, offsetof(MatroskaTag,string) },
00431     { MATROSKA_ID_TAGLANG,            EBML_STR,  0, offsetof(MatroskaTag,lang), {.s="und"} },
00432     { MATROSKA_ID_TAGDEFAULT,         EBML_UINT, 0, offsetof(MatroskaTag,def) },
00433     { MATROSKA_ID_TAGDEFAULT_BUG,     EBML_UINT, 0, offsetof(MatroskaTag,def) },
00434     { MATROSKA_ID_SIMPLETAG,          EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTag,sub), {.n=matroska_simpletag} },
00435     { 0 }
00436 };
00437 
00438 static EbmlSyntax matroska_tagtargets[] = {
00439     { MATROSKA_ID_TAGTARGETS_TYPE,      EBML_STR,  0, offsetof(MatroskaTagTarget,type) },
00440     { MATROSKA_ID_TAGTARGETS_TYPEVALUE, EBML_UINT, 0, offsetof(MatroskaTagTarget,typevalue), {.u=50} },
00441     { MATROSKA_ID_TAGTARGETS_TRACKUID,  EBML_UINT, 0, offsetof(MatroskaTagTarget,trackuid) },
00442     { MATROSKA_ID_TAGTARGETS_CHAPTERUID,EBML_UINT, 0, offsetof(MatroskaTagTarget,chapteruid) },
00443     { MATROSKA_ID_TAGTARGETS_ATTACHUID, EBML_UINT, 0, offsetof(MatroskaTagTarget,attachuid) },
00444     { 0 }
00445 };
00446 
00447 static EbmlSyntax matroska_tag[] = {
00448     { MATROSKA_ID_SIMPLETAG,          EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTags,tag), {.n=matroska_simpletag} },
00449     { MATROSKA_ID_TAGTARGETS,         EBML_NEST, 0, offsetof(MatroskaTags,target), {.n=matroska_tagtargets} },
00450     { 0 }
00451 };
00452 
00453 static EbmlSyntax matroska_tags[] = {
00454     { MATROSKA_ID_TAG,                EBML_NEST, sizeof(MatroskaTags), offsetof(MatroskaDemuxContext,tags), {.n=matroska_tag} },
00455     { 0 }
00456 };
00457 
00458 static EbmlSyntax matroska_seekhead_entry[] = {
00459     { MATROSKA_ID_SEEKID,             EBML_UINT, 0, offsetof(MatroskaSeekhead,id) },
00460     { MATROSKA_ID_SEEKPOSITION,       EBML_UINT, 0, offsetof(MatroskaSeekhead,pos), {.u=-1} },
00461     { 0 }
00462 };
00463 
00464 static EbmlSyntax matroska_seekhead[] = {
00465     { MATROSKA_ID_SEEKENTRY,          EBML_NEST, sizeof(MatroskaSeekhead), offsetof(MatroskaDemuxContext,seekhead), {.n=matroska_seekhead_entry} },
00466     { 0 }
00467 };
00468 
00469 static EbmlSyntax matroska_segment[] = {
00470     { MATROSKA_ID_INFO,           EBML_NEST, 0, 0, {.n=matroska_info       } },
00471     { MATROSKA_ID_TRACKS,         EBML_NEST, 0, 0, {.n=matroska_tracks     } },
00472     { MATROSKA_ID_ATTACHMENTS,    EBML_NEST, 0, 0, {.n=matroska_attachments} },
00473     { MATROSKA_ID_CHAPTERS,       EBML_NEST, 0, 0, {.n=matroska_chapters   } },
00474     { MATROSKA_ID_CUES,           EBML_NEST, 0, 0, {.n=matroska_index      } },
00475     { MATROSKA_ID_TAGS,           EBML_NEST, 0, 0, {.n=matroska_tags       } },
00476     { MATROSKA_ID_SEEKHEAD,       EBML_NEST, 0, 0, {.n=matroska_seekhead   } },
00477     { MATROSKA_ID_CLUSTER,        EBML_STOP },
00478     { 0 }
00479 };
00480 
00481 static EbmlSyntax matroska_segments[] = {
00482     { MATROSKA_ID_SEGMENT,        EBML_NEST, 0, 0, {.n=matroska_segment    } },
00483     { 0 }
00484 };
00485 
00486 static EbmlSyntax matroska_blockgroup[] = {
00487     { MATROSKA_ID_BLOCK,          EBML_BIN,  0, offsetof(MatroskaBlock,bin) },
00488     { MATROSKA_ID_SIMPLEBLOCK,    EBML_BIN,  0, offsetof(MatroskaBlock,bin) },
00489     { MATROSKA_ID_BLOCKDURATION,  EBML_UINT, 0, offsetof(MatroskaBlock,duration) },
00490     { MATROSKA_ID_BLOCKREFERENCE, EBML_UINT, 0, offsetof(MatroskaBlock,reference) },
00491     { 1,                          EBML_UINT, 0, offsetof(MatroskaBlock,non_simple), {.u=1} },
00492     { 0 }
00493 };
00494 
00495 static EbmlSyntax matroska_cluster[] = {
00496     { MATROSKA_ID_CLUSTERTIMECODE,EBML_UINT,0, offsetof(MatroskaCluster,timecode) },
00497     { MATROSKA_ID_BLOCKGROUP,     EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
00498     { MATROSKA_ID_SIMPLEBLOCK,    EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
00499     { MATROSKA_ID_CLUSTERPOSITION,EBML_NONE },
00500     { MATROSKA_ID_CLUSTERPREVSIZE,EBML_NONE },
00501     { 0 }
00502 };
00503 
00504 static EbmlSyntax matroska_clusters[] = {
00505     { MATROSKA_ID_CLUSTER,        EBML_NEST, 0, 0, {.n=matroska_cluster} },
00506     { MATROSKA_ID_INFO,           EBML_NONE },
00507     { MATROSKA_ID_CUES,           EBML_NONE },
00508     { MATROSKA_ID_TAGS,           EBML_NONE },
00509     { MATROSKA_ID_SEEKHEAD,       EBML_NONE },
00510     { 0 }
00511 };
00512 
00513 static const char *matroska_doctypes[] = { "matroska", "webm" };
00514 
00515 /*
00516  * Return: Whether we reached the end of a level in the hierarchy or not.
00517  */
00518 static int ebml_level_end(MatroskaDemuxContext *matroska)
00519 {
00520     AVIOContext *pb = matroska->ctx->pb;
00521     int64_t pos = avio_tell(pb);
00522 
00523     if (matroska->num_levels > 0) {
00524         MatroskaLevel *level = &matroska->levels[matroska->num_levels - 1];
00525         if (pos - level->start >= level->length || matroska->current_id) {
00526             matroska->num_levels--;
00527             return 1;
00528         }
00529     }
00530     return 0;
00531 }
00532 
00533 /*
00534  * Read: an "EBML number", which is defined as a variable-length
00535  * array of bytes. The first byte indicates the length by giving a
00536  * number of 0-bits followed by a one. The position of the first
00537  * "one" bit inside the first byte indicates the length of this
00538  * number.
00539  * Returns: number of bytes read, < 0 on error
00540  */
00541 static int ebml_read_num(MatroskaDemuxContext *matroska, AVIOContext *pb,
00542                          int max_size, uint64_t *number)
00543 {
00544     int read = 1, n = 1;
00545     uint64_t total = 0;
00546 
00547     /* The first byte tells us the length in bytes - avio_r8() can normally
00548      * return 0, but since that's not a valid first ebmlID byte, we can
00549      * use it safely here to catch EOS. */
00550     if (!(total = avio_r8(pb))) {
00551         /* we might encounter EOS here */
00552         if (!url_feof(pb)) {
00553             int64_t pos = avio_tell(pb);
00554             av_log(matroska->ctx, AV_LOG_ERROR,
00555                    "Read error at pos. %"PRIu64" (0x%"PRIx64")\n",
00556                    pos, pos);
00557         }
00558         return AVERROR(EIO); /* EOS or actual I/O error */
00559     }
00560 
00561     /* get the length of the EBML number */
00562     read = 8 - ff_log2_tab[total];
00563     if (read > max_size) {
00564         int64_t pos = avio_tell(pb) - 1;
00565         av_log(matroska->ctx, AV_LOG_ERROR,
00566                "Invalid EBML number size tag 0x%02x at pos %"PRIu64" (0x%"PRIx64")\n",
00567                (uint8_t) total, pos, pos);
00568         return AVERROR_INVALIDDATA;
00569     }
00570 
00571     /* read out length */
00572     total ^= 1 << ff_log2_tab[total];
00573     while (n++ < read)
00574         total = (total << 8) | avio_r8(pb);
00575 
00576     *number = total;
00577 
00578     return read;
00579 }
00580 
00586 static int ebml_read_length(MatroskaDemuxContext *matroska, AVIOContext *pb,
00587                             uint64_t *number)
00588 {
00589     int res = ebml_read_num(matroska, pb, 8, number);
00590     if (res > 0 && *number + 1 == 1ULL << (7 * res))
00591         *number = 0xffffffffffffffULL;
00592     return res;
00593 }
00594 
00595 /*
00596  * Read the next element as an unsigned int.
00597  * 0 is success, < 0 is failure.
00598  */
00599 static int ebml_read_uint(AVIOContext *pb, int size, uint64_t *num)
00600 {
00601     int n = 0;
00602 
00603     if (size > 8)
00604         return AVERROR_INVALIDDATA;
00605 
00606     /* big-endian ordering; build up number */
00607     *num = 0;
00608     while (n++ < size)
00609         *num = (*num << 8) | avio_r8(pb);
00610 
00611     return 0;
00612 }
00613 
00614 /*
00615  * Read the next element as a float.
00616  * 0 is success, < 0 is failure.
00617  */
00618 static int ebml_read_float(AVIOContext *pb, int size, double *num)
00619 {
00620     if (size == 0) {
00621         *num = 0;
00622     } else if (size == 4) {
00623         *num= av_int2flt(avio_rb32(pb));
00624     } else if(size==8){
00625         *num= av_int2dbl(avio_rb64(pb));
00626     } else
00627         return AVERROR_INVALIDDATA;
00628 
00629     return 0;
00630 }
00631 
00632 /*
00633  * Read the next element as an ASCII string.
00634  * 0 is success, < 0 is failure.
00635  */
00636 static int ebml_read_ascii(AVIOContext *pb, int size, char **str)
00637 {
00638     av_free(*str);
00639     /* EBML strings are usually not 0-terminated, so we allocate one
00640      * byte more, read the string and NULL-terminate it ourselves. */
00641     if (!(*str = av_malloc(size + 1)))
00642         return AVERROR(ENOMEM);
00643     if (avio_read(pb, (uint8_t *) *str, size) != size) {
00644         av_freep(str);
00645         return AVERROR(EIO);
00646     }
00647     (*str)[size] = '\0';
00648 
00649     return 0;
00650 }
00651 
00652 /*
00653  * Read the next element as binary data.
00654  * 0 is success, < 0 is failure.
00655  */
00656 static int ebml_read_binary(AVIOContext *pb, int length, EbmlBin *bin)
00657 {
00658     av_free(bin->data);
00659     if (!(bin->data = av_malloc(length)))
00660         return AVERROR(ENOMEM);
00661 
00662     bin->size = length;
00663     bin->pos  = avio_tell(pb);
00664     if (avio_read(pb, bin->data, length) != length) {
00665         av_freep(&bin->data);
00666         return AVERROR(EIO);
00667     }
00668 
00669     return 0;
00670 }
00671 
00672 /*
00673  * Read the next element, but only the header. The contents
00674  * are supposed to be sub-elements which can be read separately.
00675  * 0 is success, < 0 is failure.
00676  */
00677 static int ebml_read_master(MatroskaDemuxContext *matroska, uint64_t length)
00678 {
00679     AVIOContext *pb = matroska->ctx->pb;
00680     MatroskaLevel *level;
00681 
00682     if (matroska->num_levels >= EBML_MAX_DEPTH) {
00683         av_log(matroska->ctx, AV_LOG_ERROR,
00684                "File moves beyond max. allowed depth (%d)\n", EBML_MAX_DEPTH);
00685         return AVERROR(ENOSYS);
00686     }
00687 
00688     level = &matroska->levels[matroska->num_levels++];
00689     level->start = avio_tell(pb);
00690     level->length = length;
00691 
00692     return 0;
00693 }
00694 
00695 /*
00696  * Read signed/unsigned "EBML" numbers.
00697  * Return: number of bytes processed, < 0 on error
00698  */
00699 static int matroska_ebmlnum_uint(MatroskaDemuxContext *matroska,
00700                                  uint8_t *data, uint32_t size, uint64_t *num)
00701 {
00702     AVIOContext pb;
00703     ffio_init_context(&pb, data, size, 0, NULL, NULL, NULL, NULL);
00704     return ebml_read_num(matroska, &pb, FFMIN(size, 8), num);
00705 }
00706 
00707 /*
00708  * Same as above, but signed.
00709  */
00710 static int matroska_ebmlnum_sint(MatroskaDemuxContext *matroska,
00711                                  uint8_t *data, uint32_t size, int64_t *num)
00712 {
00713     uint64_t unum;
00714     int res;
00715 
00716     /* read as unsigned number first */
00717     if ((res = matroska_ebmlnum_uint(matroska, data, size, &unum)) < 0)
00718         return res;
00719 
00720     /* make signed (weird way) */
00721     *num = unum - ((1LL << (7*res - 1)) - 1);
00722 
00723     return res;
00724 }
00725 
00726 static int ebml_parse_elem(MatroskaDemuxContext *matroska,
00727                            EbmlSyntax *syntax, void *data);
00728 
00729 static int ebml_parse_id(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
00730                          uint32_t id, void *data)
00731 {
00732     int i;
00733     for (i=0; syntax[i].id; i++)
00734         if (id == syntax[i].id)
00735             break;
00736     if (!syntax[i].id && id == MATROSKA_ID_CLUSTER &&
00737         matroska->num_levels > 0 &&
00738         matroska->levels[matroska->num_levels-1].length == 0xffffffffffffff)
00739         return 0;  // we reached the end of an unknown size cluster
00740     if (!syntax[i].id && id != EBML_ID_VOID && id != EBML_ID_CRC32)
00741         av_log(matroska->ctx, AV_LOG_INFO, "Unknown entry 0x%X\n", id);
00742     return ebml_parse_elem(matroska, &syntax[i], data);
00743 }
00744 
00745 static int ebml_parse(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
00746                       void *data)
00747 {
00748     if (!matroska->current_id) {
00749         uint64_t id;
00750         int res = ebml_read_num(matroska, matroska->ctx->pb, 4, &id);
00751         if (res < 0)
00752             return res;
00753         matroska->current_id = id | 1 << 7*res;
00754     }
00755     return ebml_parse_id(matroska, syntax, matroska->current_id, data);
00756 }
00757 
00758 static int ebml_parse_nest(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
00759                            void *data)
00760 {
00761     int i, res = 0;
00762 
00763     for (i=0; syntax[i].id; i++)
00764         switch (syntax[i].type) {
00765         case EBML_UINT:
00766             *(uint64_t *)((char *)data+syntax[i].data_offset) = syntax[i].def.u;
00767             break;
00768         case EBML_FLOAT:
00769             *(double   *)((char *)data+syntax[i].data_offset) = syntax[i].def.f;
00770             break;
00771         case EBML_STR:
00772         case EBML_UTF8:
00773             *(char    **)((char *)data+syntax[i].data_offset) = av_strdup(syntax[i].def.s);
00774             break;
00775         }
00776 
00777     while (!res && !ebml_level_end(matroska))
00778         res = ebml_parse(matroska, syntax, data);
00779 
00780     return res;
00781 }
00782 
00783 static int ebml_parse_elem(MatroskaDemuxContext *matroska,
00784                            EbmlSyntax *syntax, void *data)
00785 {
00786     static const uint64_t max_lengths[EBML_TYPE_COUNT] = {
00787         [EBML_UINT]  = 8,
00788         [EBML_FLOAT] = 8,
00789         // max. 16 MB for strings
00790         [EBML_STR]   = 0x1000000,
00791         [EBML_UTF8]  = 0x1000000,
00792         // max. 256 MB for binary data
00793         [EBML_BIN]   = 0x10000000,
00794         // no limits for anything else
00795     };
00796     AVIOContext *pb = matroska->ctx->pb;
00797     uint32_t id = syntax->id;
00798     uint64_t length;
00799     int res;
00800 
00801     data = (char *)data + syntax->data_offset;
00802     if (syntax->list_elem_size) {
00803         EbmlList *list = data;
00804         list->elem = av_realloc(list->elem, (list->nb_elem+1)*syntax->list_elem_size);
00805         data = (char*)list->elem + list->nb_elem*syntax->list_elem_size;
00806         memset(data, 0, syntax->list_elem_size);
00807         list->nb_elem++;
00808     }
00809 
00810     if (syntax->type != EBML_PASS && syntax->type != EBML_STOP) {
00811         matroska->current_id = 0;
00812         if ((res = ebml_read_length(matroska, pb, &length)) < 0)
00813             return res;
00814         if (max_lengths[syntax->type] && length > max_lengths[syntax->type]) {
00815             av_log(matroska->ctx, AV_LOG_ERROR,
00816                    "Invalid length 0x%"PRIx64" > 0x%"PRIx64" for syntax element %i\n",
00817                    length, max_lengths[syntax->type], syntax->type);
00818             return AVERROR_INVALIDDATA;
00819         }
00820     }
00821 
00822     switch (syntax->type) {
00823     case EBML_UINT:  res = ebml_read_uint  (pb, length, data);  break;
00824     case EBML_FLOAT: res = ebml_read_float (pb, length, data);  break;
00825     case EBML_STR:
00826     case EBML_UTF8:  res = ebml_read_ascii (pb, length, data);  break;
00827     case EBML_BIN:   res = ebml_read_binary(pb, length, data);  break;
00828     case EBML_NEST:  if ((res=ebml_read_master(matroska, length)) < 0)
00829                          return res;
00830                      if (id == MATROSKA_ID_SEGMENT)
00831                          matroska->segment_start = avio_tell(matroska->ctx->pb);
00832                      return ebml_parse_nest(matroska, syntax->def.n, data);
00833     case EBML_PASS:  return ebml_parse_id(matroska, syntax->def.n, id, data);
00834     case EBML_STOP:  return 1;
00835     default:         return avio_skip(pb,length)<0 ? AVERROR(EIO) : 0;
00836     }
00837     if (res == AVERROR_INVALIDDATA)
00838         av_log(matroska->ctx, AV_LOG_ERROR, "Invalid element\n");
00839     else if (res == AVERROR(EIO))
00840         av_log(matroska->ctx, AV_LOG_ERROR, "Read error\n");
00841     return res;
00842 }
00843 
00844 static void ebml_free(EbmlSyntax *syntax, void *data)
00845 {
00846     int i, j;
00847     for (i=0; syntax[i].id; i++) {
00848         void *data_off = (char *)data + syntax[i].data_offset;
00849         switch (syntax[i].type) {
00850         case EBML_STR:
00851         case EBML_UTF8:  av_freep(data_off);                      break;
00852         case EBML_BIN:   av_freep(&((EbmlBin *)data_off)->data);  break;
00853         case EBML_NEST:
00854             if (syntax[i].list_elem_size) {
00855                 EbmlList *list = data_off;
00856                 char *ptr = list->elem;
00857                 for (j=0; j<list->nb_elem; j++, ptr+=syntax[i].list_elem_size)
00858                     ebml_free(syntax[i].def.n, ptr);
00859                 av_free(list->elem);
00860             } else
00861                 ebml_free(syntax[i].def.n, data_off);
00862         default:  break;
00863         }
00864     }
00865 }
00866 
00867 
00868 /*
00869  * Autodetecting...
00870  */
00871 static int matroska_probe(AVProbeData *p)
00872 {
00873     uint64_t total = 0;
00874     int len_mask = 0x80, size = 1, n = 1, i;
00875 
00876     /* EBML header? */
00877     if (AV_RB32(p->buf) != EBML_ID_HEADER)
00878         return 0;
00879 
00880     /* length of header */
00881     total = p->buf[4];
00882     while (size <= 8 && !(total & len_mask)) {
00883         size++;
00884         len_mask >>= 1;
00885     }
00886     if (size > 8)
00887       return 0;
00888     total &= (len_mask - 1);
00889     while (n < size)
00890         total = (total << 8) | p->buf[4 + n++];
00891 
00892     /* Does the probe data contain the whole header? */
00893     if (p->buf_size < 4 + size + total)
00894       return 0;
00895 
00896     /* The header should contain a known document type. For now,
00897      * we don't parse the whole header but simply check for the
00898      * availability of that array of characters inside the header.
00899      * Not fully fool-proof, but good enough. */
00900     for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++) {
00901         int probelen = strlen(matroska_doctypes[i]);
00902         for (n = 4+size; n <= 4+size+total-probelen; n++)
00903             if (!memcmp(p->buf+n, matroska_doctypes[i], probelen))
00904                 return AVPROBE_SCORE_MAX;
00905     }
00906 
00907     // probably valid EBML header but no recognized doctype
00908     return AVPROBE_SCORE_MAX/2;
00909 }
00910 
00911 static MatroskaTrack *matroska_find_track_by_num(MatroskaDemuxContext *matroska,
00912                                                  int num)
00913 {
00914     MatroskaTrack *tracks = matroska->tracks.elem;
00915     int i;
00916 
00917     for (i=0; i < matroska->tracks.nb_elem; i++)
00918         if (tracks[i].num == num)
00919             return &tracks[i];
00920 
00921     av_log(matroska->ctx, AV_LOG_ERROR, "Invalid track number %d\n", num);
00922     return NULL;
00923 }
00924 
00925 static int matroska_decode_buffer(uint8_t** buf, int* buf_size,
00926                                   MatroskaTrack *track)
00927 {
00928     MatroskaTrackEncoding *encodings = track->encodings.elem;
00929     uint8_t* data = *buf;
00930     int isize = *buf_size;
00931     uint8_t* pkt_data = NULL;
00932     int pkt_size = isize;
00933     int result = 0;
00934     int olen;
00935 
00936     if (pkt_size >= 10000000)
00937         return -1;
00938 
00939     switch (encodings[0].compression.algo) {
00940     case MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP:
00941         return encodings[0].compression.settings.size;
00942     case MATROSKA_TRACK_ENCODING_COMP_LZO:
00943         do {
00944             olen = pkt_size *= 3;
00945             pkt_data = av_realloc(pkt_data, pkt_size+AV_LZO_OUTPUT_PADDING);
00946             result = av_lzo1x_decode(pkt_data, &olen, data, &isize);
00947         } while (result==AV_LZO_OUTPUT_FULL && pkt_size<10000000);
00948         if (result)
00949             goto failed;
00950         pkt_size -= olen;
00951         break;
00952 #if CONFIG_ZLIB
00953     case MATROSKA_TRACK_ENCODING_COMP_ZLIB: {
00954         z_stream zstream = {0};
00955         if (inflateInit(&zstream) != Z_OK)
00956             return -1;
00957         zstream.next_in = data;
00958         zstream.avail_in = isize;
00959         do {
00960             pkt_size *= 3;
00961             pkt_data = av_realloc(pkt_data, pkt_size);
00962             zstream.avail_out = pkt_size - zstream.total_out;
00963             zstream.next_out = pkt_data + zstream.total_out;
00964             result = inflate(&zstream, Z_NO_FLUSH);
00965         } while (result==Z_OK && pkt_size<10000000);
00966         pkt_size = zstream.total_out;
00967         inflateEnd(&zstream);
00968         if (result != Z_STREAM_END)
00969             goto failed;
00970         break;
00971     }
00972 #endif
00973 #if CONFIG_BZLIB
00974     case MATROSKA_TRACK_ENCODING_COMP_BZLIB: {
00975         bz_stream bzstream = {0};
00976         if (BZ2_bzDecompressInit(&bzstream, 0, 0) != BZ_OK)
00977             return -1;
00978         bzstream.next_in = data;
00979         bzstream.avail_in = isize;
00980         do {
00981             pkt_size *= 3;
00982             pkt_data = av_realloc(pkt_data, pkt_size);
00983             bzstream.avail_out = pkt_size - bzstream.total_out_lo32;
00984             bzstream.next_out = pkt_data + bzstream.total_out_lo32;
00985             result = BZ2_bzDecompress(&bzstream);
00986         } while (result==BZ_OK && pkt_size<10000000);
00987         pkt_size = bzstream.total_out_lo32;
00988         BZ2_bzDecompressEnd(&bzstream);
00989         if (result != BZ_STREAM_END)
00990             goto failed;
00991         break;
00992     }
00993 #endif
00994     default:
00995         return -1;
00996     }
00997 
00998     *buf = pkt_data;
00999     *buf_size = pkt_size;
01000     return 0;
01001  failed:
01002     av_free(pkt_data);
01003     return -1;
01004 }
01005 
01006 static void matroska_fix_ass_packet(MatroskaDemuxContext *matroska,
01007                                     AVPacket *pkt, uint64_t display_duration)
01008 {
01009     char *line, *layer, *ptr = pkt->data, *end = ptr+pkt->size;
01010     for (; *ptr!=',' && ptr<end-1; ptr++);
01011     if (*ptr == ',')
01012         layer = ++ptr;
01013     for (; *ptr!=',' && ptr<end-1; ptr++);
01014     if (*ptr == ',') {
01015         int64_t end_pts = pkt->pts + display_duration;
01016         int sc = matroska->time_scale * pkt->pts / 10000000;
01017         int ec = matroska->time_scale * end_pts  / 10000000;
01018         int sh, sm, ss, eh, em, es, len;
01019         sh = sc/360000;  sc -= 360000*sh;
01020         sm = sc/  6000;  sc -=   6000*sm;
01021         ss = sc/   100;  sc -=    100*ss;
01022         eh = ec/360000;  ec -= 360000*eh;
01023         em = ec/  6000;  ec -=   6000*em;
01024         es = ec/   100;  ec -=    100*es;
01025         *ptr++ = '\0';
01026         len = 50 + end-ptr + FF_INPUT_BUFFER_PADDING_SIZE;
01027         if (!(line = av_malloc(len)))
01028             return;
01029         snprintf(line,len,"Dialogue: %s,%d:%02d:%02d.%02d,%d:%02d:%02d.%02d,%s\r\n",
01030                  layer, sh, sm, ss, sc, eh, em, es, ec, ptr);
01031         av_free(pkt->data);
01032         pkt->data = line;
01033         pkt->size = strlen(line);
01034     }
01035 }
01036 
01037 static void matroska_merge_packets(AVPacket *out, AVPacket *in)
01038 {
01039     out->data = av_realloc(out->data, out->size+in->size);
01040     memcpy(out->data+out->size, in->data, in->size);
01041     out->size += in->size;
01042     av_destruct_packet(in);
01043     av_free(in);
01044 }
01045 
01046 static void matroska_convert_tag(AVFormatContext *s, EbmlList *list,
01047                                  AVMetadata **metadata, char *prefix)
01048 {
01049     MatroskaTag *tags = list->elem;
01050     char key[1024];
01051     int i;
01052 
01053     for (i=0; i < list->nb_elem; i++) {
01054         const char *lang = strcmp(tags[i].lang, "und") ? tags[i].lang : NULL;
01055 
01056         if (!tags[i].name) {
01057             av_log(s, AV_LOG_WARNING, "Skipping invalid tag with no TagName.\n");
01058             continue;
01059         }
01060         if (prefix)  snprintf(key, sizeof(key), "%s/%s", prefix, tags[i].name);
01061         else         av_strlcpy(key, tags[i].name, sizeof(key));
01062         if (tags[i].def || !lang) {
01063         av_metadata_set2(metadata, key, tags[i].string, 0);
01064         if (tags[i].sub.nb_elem)
01065             matroska_convert_tag(s, &tags[i].sub, metadata, key);
01066         }
01067         if (lang) {
01068             av_strlcat(key, "-", sizeof(key));
01069             av_strlcat(key, lang, sizeof(key));
01070             av_metadata_set2(metadata, key, tags[i].string, 0);
01071             if (tags[i].sub.nb_elem)
01072                 matroska_convert_tag(s, &tags[i].sub, metadata, key);
01073         }
01074     }
01075     ff_metadata_conv(metadata, NULL, ff_mkv_metadata_conv);
01076 }
01077 
01078 static void matroska_convert_tags(AVFormatContext *s)
01079 {
01080     MatroskaDemuxContext *matroska = s->priv_data;
01081     MatroskaTags *tags = matroska->tags.elem;
01082     int i, j;
01083 
01084     for (i=0; i < matroska->tags.nb_elem; i++) {
01085         if (tags[i].target.attachuid) {
01086             MatroskaAttachement *attachment = matroska->attachments.elem;
01087             for (j=0; j<matroska->attachments.nb_elem; j++)
01088                 if (attachment[j].uid == tags[i].target.attachuid)
01089                     matroska_convert_tag(s, &tags[i].tag,
01090                                          &attachment[j].stream->metadata, NULL);
01091         } else if (tags[i].target.chapteruid) {
01092             MatroskaChapter *chapter = matroska->chapters.elem;
01093             for (j=0; j<matroska->chapters.nb_elem; j++)
01094                 if (chapter[j].uid == tags[i].target.chapteruid)
01095                     matroska_convert_tag(s, &tags[i].tag,
01096                                          &chapter[j].chapter->metadata, NULL);
01097         } else if (tags[i].target.trackuid) {
01098             MatroskaTrack *track = matroska->tracks.elem;
01099             for (j=0; j<matroska->tracks.nb_elem; j++)
01100                 if (track[j].uid == tags[i].target.trackuid)
01101                     matroska_convert_tag(s, &tags[i].tag,
01102                                          &track[j].stream->metadata, NULL);
01103         } else {
01104             matroska_convert_tag(s, &tags[i].tag, &s->metadata,
01105                                  tags[i].target.type);
01106         }
01107     }
01108 }
01109 
01110 static void matroska_execute_seekhead(MatroskaDemuxContext *matroska)
01111 {
01112     EbmlList *seekhead_list = &matroska->seekhead;
01113     MatroskaSeekhead *seekhead = seekhead_list->elem;
01114     uint32_t level_up = matroska->level_up;
01115     int64_t before_pos = avio_tell(matroska->ctx->pb);
01116     uint32_t saved_id = matroska->current_id;
01117     MatroskaLevel level;
01118     int i;
01119 
01120     // we should not do any seeking in the streaming case
01121     if (url_is_streamed(matroska->ctx->pb) ||
01122         (matroska->ctx->flags & AVFMT_FLAG_IGNIDX))
01123         return;
01124 
01125     for (i=0; i<seekhead_list->nb_elem; i++) {
01126         int64_t offset = seekhead[i].pos + matroska->segment_start;
01127 
01128         if (seekhead[i].pos <= before_pos
01129             || seekhead[i].id == MATROSKA_ID_SEEKHEAD
01130             || seekhead[i].id == MATROSKA_ID_CLUSTER)
01131             continue;
01132 
01133         /* seek */
01134         if (avio_seek(matroska->ctx->pb, offset, SEEK_SET) != offset)
01135             continue;
01136 
01137         /* We don't want to lose our seekhead level, so we add
01138          * a dummy. This is a crude hack. */
01139         if (matroska->num_levels == EBML_MAX_DEPTH) {
01140             av_log(matroska->ctx, AV_LOG_INFO,
01141                    "Max EBML element depth (%d) reached, "
01142                    "cannot parse further.\n", EBML_MAX_DEPTH);
01143             break;
01144         }
01145 
01146         level.start = 0;
01147         level.length = (uint64_t)-1;
01148         matroska->levels[matroska->num_levels] = level;
01149         matroska->num_levels++;
01150         matroska->current_id = 0;
01151 
01152         ebml_parse(matroska, matroska_segment, matroska);
01153 
01154         /* remove dummy level */
01155         while (matroska->num_levels) {
01156             uint64_t length = matroska->levels[--matroska->num_levels].length;
01157             if (length == (uint64_t)-1)
01158                 break;
01159         }
01160     }
01161 
01162     /* seek back */
01163     avio_seek(matroska->ctx->pb, before_pos, SEEK_SET);
01164     matroska->level_up = level_up;
01165     matroska->current_id = saved_id;
01166 }
01167 
01168 static int matroska_aac_profile(char *codec_id)
01169 {
01170     static const char * const aac_profiles[] = { "MAIN", "LC", "SSR" };
01171     int profile;
01172 
01173     for (profile=0; profile<FF_ARRAY_ELEMS(aac_profiles); profile++)
01174         if (strstr(codec_id, aac_profiles[profile]))
01175             break;
01176     return profile + 1;
01177 }
01178 
01179 static int matroska_aac_sri(int samplerate)
01180 {
01181     int sri;
01182 
01183     for (sri=0; sri<FF_ARRAY_ELEMS(ff_mpeg4audio_sample_rates); sri++)
01184         if (ff_mpeg4audio_sample_rates[sri] == samplerate)
01185             break;
01186     return sri;
01187 }
01188 
01189 static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap)
01190 {
01191     MatroskaDemuxContext *matroska = s->priv_data;
01192     EbmlList *attachements_list = &matroska->attachments;
01193     MatroskaAttachement *attachements;
01194     EbmlList *chapters_list = &matroska->chapters;
01195     MatroskaChapter *chapters;
01196     MatroskaTrack *tracks;
01197     EbmlList *index_list;
01198     MatroskaIndex *index;
01199     int index_scale = 1;
01200     uint64_t max_start = 0;
01201     Ebml ebml = { 0 };
01202     AVStream *st;
01203     int i, j, res;
01204 
01205     matroska->ctx = s;
01206 
01207     /* First read the EBML header. */
01208     if (ebml_parse(matroska, ebml_syntax, &ebml)
01209         || ebml.version > EBML_VERSION       || ebml.max_size > sizeof(uint64_t)
01210         || ebml.id_length > sizeof(uint32_t) || ebml.doctype_version > 2) {
01211         av_log(matroska->ctx, AV_LOG_ERROR,
01212                "EBML header using unsupported features\n"
01213                "(EBML version %"PRIu64", doctype %s, doc version %"PRIu64")\n",
01214                ebml.version, ebml.doctype, ebml.doctype_version);
01215         ebml_free(ebml_syntax, &ebml);
01216         return AVERROR_PATCHWELCOME;
01217     }
01218     for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++)
01219         if (!strcmp(ebml.doctype, matroska_doctypes[i]))
01220             break;
01221     if (i >= FF_ARRAY_ELEMS(matroska_doctypes)) {
01222         av_log(s, AV_LOG_WARNING, "Unknown EBML doctype '%s'\n", ebml.doctype);
01223     }
01224     ebml_free(ebml_syntax, &ebml);
01225 
01226     /* The next thing is a segment. */
01227     if ((res = ebml_parse(matroska, matroska_segments, matroska)) < 0)
01228         return res;
01229     matroska_execute_seekhead(matroska);
01230 
01231     if (!matroska->time_scale)
01232         matroska->time_scale = 1000000;
01233     if (matroska->duration)
01234         matroska->ctx->duration = matroska->duration * matroska->time_scale
01235                                   * 1000 / AV_TIME_BASE;
01236     av_metadata_set2(&s->metadata, "title", matroska->title, 0);
01237 
01238     tracks = matroska->tracks.elem;
01239     for (i=0; i < matroska->tracks.nb_elem; i++) {
01240         MatroskaTrack *track = &tracks[i];
01241         enum CodecID codec_id = CODEC_ID_NONE;
01242         EbmlList *encodings_list = &tracks->encodings;
01243         MatroskaTrackEncoding *encodings = encodings_list->elem;
01244         uint8_t *extradata = NULL;
01245         int extradata_size = 0;
01246         int extradata_offset = 0;
01247         AVIOContext b;
01248 
01249         /* Apply some sanity checks. */
01250         if (track->type != MATROSKA_TRACK_TYPE_VIDEO &&
01251             track->type != MATROSKA_TRACK_TYPE_AUDIO &&
01252             track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
01253             av_log(matroska->ctx, AV_LOG_INFO,
01254                    "Unknown or unsupported track type %"PRIu64"\n",
01255                    track->type);
01256             continue;
01257         }
01258         if (track->codec_id == NULL)
01259             continue;
01260 
01261         if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
01262             if (!track->default_duration)
01263                 track->default_duration = 1000000000/track->video.frame_rate;
01264             if (!track->video.display_width)
01265                 track->video.display_width = track->video.pixel_width;
01266             if (!track->video.display_height)
01267                 track->video.display_height = track->video.pixel_height;
01268         } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
01269             if (!track->audio.out_samplerate)
01270                 track->audio.out_samplerate = track->audio.samplerate;
01271         }
01272         if (encodings_list->nb_elem > 1) {
01273             av_log(matroska->ctx, AV_LOG_ERROR,
01274                    "Multiple combined encodings no supported");
01275         } else if (encodings_list->nb_elem == 1) {
01276             if (encodings[0].type ||
01277                 (encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP &&
01278 #if CONFIG_ZLIB
01279                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_ZLIB &&
01280 #endif
01281 #if CONFIG_BZLIB
01282                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_BZLIB &&
01283 #endif
01284                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_LZO)) {
01285                 encodings[0].scope = 0;
01286                 av_log(matroska->ctx, AV_LOG_ERROR,
01287                        "Unsupported encoding type");
01288             } else if (track->codec_priv.size && encodings[0].scope&2) {
01289                 uint8_t *codec_priv = track->codec_priv.data;
01290                 int offset = matroska_decode_buffer(&track->codec_priv.data,
01291                                                     &track->codec_priv.size,
01292                                                     track);
01293                 if (offset < 0) {
01294                     track->codec_priv.data = NULL;
01295                     track->codec_priv.size = 0;
01296                     av_log(matroska->ctx, AV_LOG_ERROR,
01297                            "Failed to decode codec private data\n");
01298                 } else if (offset > 0) {
01299                     track->codec_priv.data = av_malloc(track->codec_priv.size + offset);
01300                     memcpy(track->codec_priv.data,
01301                            encodings[0].compression.settings.data, offset);
01302                     memcpy(track->codec_priv.data+offset, codec_priv,
01303                            track->codec_priv.size);
01304                     track->codec_priv.size += offset;
01305                 }
01306                 if (codec_priv != track->codec_priv.data)
01307                     av_free(codec_priv);
01308             }
01309         }
01310 
01311         for(j=0; ff_mkv_codec_tags[j].id != CODEC_ID_NONE; j++){
01312             if(!strncmp(ff_mkv_codec_tags[j].str, track->codec_id,
01313                         strlen(ff_mkv_codec_tags[j].str))){
01314                 codec_id= ff_mkv_codec_tags[j].id;
01315                 break;
01316             }
01317         }
01318 
01319         st = track->stream = av_new_stream(s, 0);
01320         if (st == NULL)
01321             return AVERROR(ENOMEM);
01322 
01323         if (!strcmp(track->codec_id, "V_MS/VFW/FOURCC")
01324             && track->codec_priv.size >= 40
01325             && track->codec_priv.data != NULL) {
01326             track->ms_compat = 1;
01327             track->video.fourcc = AV_RL32(track->codec_priv.data + 16);
01328             codec_id = ff_codec_get_id(ff_codec_bmp_tags, track->video.fourcc);
01329             extradata_offset = 40;
01330         } else if (!strcmp(track->codec_id, "A_MS/ACM")
01331                    && track->codec_priv.size >= 14
01332                    && track->codec_priv.data != NULL) {
01333             ffio_init_context(&b, track->codec_priv.data, track->codec_priv.size,
01334                           URL_RDONLY, NULL, NULL, NULL, NULL);
01335             ff_get_wav_header(&b, st->codec, track->codec_priv.size);
01336             codec_id = st->codec->codec_id;
01337             extradata_offset = FFMIN(track->codec_priv.size, 18);
01338         } else if (!strcmp(track->codec_id, "V_QUICKTIME")
01339                    && (track->codec_priv.size >= 86)
01340                    && (track->codec_priv.data != NULL)) {
01341             track->video.fourcc = AV_RL32(track->codec_priv.data);
01342             codec_id=ff_codec_get_id(codec_movvideo_tags, track->video.fourcc);
01343         } else if (codec_id == CODEC_ID_PCM_S16BE) {
01344             switch (track->audio.bitdepth) {
01345             case  8:  codec_id = CODEC_ID_PCM_U8;     break;
01346             case 24:  codec_id = CODEC_ID_PCM_S24BE;  break;
01347             case 32:  codec_id = CODEC_ID_PCM_S32BE;  break;
01348             }
01349         } else if (codec_id == CODEC_ID_PCM_S16LE) {
01350             switch (track->audio.bitdepth) {
01351             case  8:  codec_id = CODEC_ID_PCM_U8;     break;
01352             case 24:  codec_id = CODEC_ID_PCM_S24LE;  break;
01353             case 32:  codec_id = CODEC_ID_PCM_S32LE;  break;
01354             }
01355         } else if (codec_id==CODEC_ID_PCM_F32LE && track->audio.bitdepth==64) {
01356             codec_id = CODEC_ID_PCM_F64LE;
01357         } else if (codec_id == CODEC_ID_AAC && !track->codec_priv.size) {
01358             int profile = matroska_aac_profile(track->codec_id);
01359             int sri = matroska_aac_sri(track->audio.samplerate);
01360             extradata = av_malloc(5);
01361             if (extradata == NULL)
01362                 return AVERROR(ENOMEM);
01363             extradata[0] = (profile << 3) | ((sri&0x0E) >> 1);
01364             extradata[1] = ((sri&0x01) << 7) | (track->audio.channels<<3);
01365             if (strstr(track->codec_id, "SBR")) {
01366                 sri = matroska_aac_sri(track->audio.out_samplerate);
01367                 extradata[2] = 0x56;
01368                 extradata[3] = 0xE5;
01369                 extradata[4] = 0x80 | (sri<<3);
01370                 extradata_size = 5;
01371             } else
01372                 extradata_size = 2;
01373         } else if (codec_id == CODEC_ID_TTA) {
01374             extradata_size = 30;
01375             extradata = av_mallocz(extradata_size);
01376             if (extradata == NULL)
01377                 return AVERROR(ENOMEM);
01378             ffio_init_context(&b, extradata, extradata_size, 1,
01379                           NULL, NULL, NULL, NULL);
01380             avio_write(&b, "TTA1", 4);
01381             avio_wl16(&b, 1);
01382             avio_wl16(&b, track->audio.channels);
01383             avio_wl16(&b, track->audio.bitdepth);
01384             avio_wl32(&b, track->audio.out_samplerate);
01385             avio_wl32(&b, matroska->ctx->duration * track->audio.out_samplerate);
01386         } else if (codec_id == CODEC_ID_RV10 || codec_id == CODEC_ID_RV20 ||
01387                    codec_id == CODEC_ID_RV30 || codec_id == CODEC_ID_RV40) {
01388             extradata_offset = 26;
01389         } else if (codec_id == CODEC_ID_RA_144) {
01390             track->audio.out_samplerate = 8000;
01391             track->audio.channels = 1;
01392         } else if (codec_id == CODEC_ID_RA_288 || codec_id == CODEC_ID_COOK ||
01393                    codec_id == CODEC_ID_ATRAC3 || codec_id == CODEC_ID_SIPR) {
01394             int flavor;
01395             ffio_init_context(&b, track->codec_priv.data,track->codec_priv.size,
01396                           0, NULL, NULL, NULL, NULL);
01397             avio_skip(&b, 22);
01398             flavor                       = avio_rb16(&b);
01399             track->audio.coded_framesize = avio_rb32(&b);
01400             avio_skip(&b, 12);
01401             track->audio.sub_packet_h    = avio_rb16(&b);
01402             track->audio.frame_size      = avio_rb16(&b);
01403             track->audio.sub_packet_size = avio_rb16(&b);
01404             track->audio.buf = av_malloc(track->audio.frame_size * track->audio.sub_packet_h);
01405             if (codec_id == CODEC_ID_RA_288) {
01406                 st->codec->block_align = track->audio.coded_framesize;
01407                 track->codec_priv.size = 0;
01408             } else {
01409                 if (codec_id == CODEC_ID_SIPR && flavor < 4) {
01410                     const int sipr_bit_rate[4] = { 6504, 8496, 5000, 16000 };
01411                     track->audio.sub_packet_size = ff_sipr_subpk_size[flavor];
01412                     st->codec->bit_rate = sipr_bit_rate[flavor];
01413                 }
01414                 st->codec->block_align = track->audio.sub_packet_size;
01415                 extradata_offset = 78;
01416             }
01417         }
01418         track->codec_priv.size -= extradata_offset;
01419 
01420         if (codec_id == CODEC_ID_NONE)
01421             av_log(matroska->ctx, AV_LOG_INFO,
01422                    "Unknown/unsupported CodecID %s.\n", track->codec_id);
01423 
01424         if (track->time_scale < 0.01)
01425             track->time_scale = 1.0;
01426         av_set_pts_info(st, 64, matroska->time_scale*track->time_scale, 1000*1000*1000); /* 64 bit pts in ns */
01427 
01428         st->codec->codec_id = codec_id;
01429         st->start_time = 0;
01430         if (strcmp(track->language, "und"))
01431             av_metadata_set2(&st->metadata, "language", track->language, 0);
01432         av_metadata_set2(&st->metadata, "title", track->name, 0);
01433 
01434         if (track->flag_default)
01435             st->disposition |= AV_DISPOSITION_DEFAULT;
01436         if (track->flag_forced)
01437             st->disposition |= AV_DISPOSITION_FORCED;
01438 
01439         if (track->default_duration)
01440             av_reduce(&st->codec->time_base.num, &st->codec->time_base.den,
01441                       track->default_duration, 1000000000, 30000);
01442 
01443         if (!st->codec->extradata) {
01444             if(extradata){
01445                 st->codec->extradata = extradata;
01446                 st->codec->extradata_size = extradata_size;
01447             } else if(track->codec_priv.data && track->codec_priv.size > 0){
01448                 st->codec->extradata = av_mallocz(track->codec_priv.size +
01449                                                   FF_INPUT_BUFFER_PADDING_SIZE);
01450                 if(st->codec->extradata == NULL)
01451                     return AVERROR(ENOMEM);
01452                 st->codec->extradata_size = track->codec_priv.size;
01453                 memcpy(st->codec->extradata,
01454                        track->codec_priv.data + extradata_offset,
01455                        track->codec_priv.size);
01456             }
01457         }
01458 
01459         if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
01460             st->codec->codec_type = AVMEDIA_TYPE_VIDEO;
01461             st->codec->codec_tag  = track->video.fourcc;
01462             st->codec->width  = track->video.pixel_width;
01463             st->codec->height = track->video.pixel_height;
01464             av_reduce(&st->sample_aspect_ratio.num,
01465                       &st->sample_aspect_ratio.den,
01466                       st->codec->height * track->video.display_width,
01467                       st->codec-> width * track->video.display_height,
01468                       255);
01469             if (st->codec->codec_id != CODEC_ID_H264)
01470             st->need_parsing = AVSTREAM_PARSE_HEADERS;
01471             if (track->default_duration)
01472                 st->avg_frame_rate = av_d2q(1000000000.0/track->default_duration, INT_MAX);
01473         } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
01474             st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
01475             st->codec->sample_rate = track->audio.out_samplerate;
01476             st->codec->channels = track->audio.channels;
01477             if (st->codec->codec_id != CODEC_ID_AAC)
01478             st->need_parsing = AVSTREAM_PARSE_HEADERS;
01479         } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
01480             st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE;
01481         }
01482     }
01483 
01484     attachements = attachements_list->elem;
01485     for (j=0; j<attachements_list->nb_elem; j++) {
01486         if (!(attachements[j].filename && attachements[j].mime &&
01487               attachements[j].bin.data && attachements[j].bin.size > 0)) {
01488             av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
01489         } else {
01490             AVStream *st = av_new_stream(s, 0);
01491             if (st == NULL)
01492                 break;
01493             av_metadata_set2(&st->metadata, "filename",attachements[j].filename, 0);
01494             st->codec->codec_id = CODEC_ID_NONE;
01495             st->codec->codec_type = AVMEDIA_TYPE_ATTACHMENT;
01496             st->codec->extradata  = av_malloc(attachements[j].bin.size);
01497             if(st->codec->extradata == NULL)
01498                 break;
01499             st->codec->extradata_size = attachements[j].bin.size;
01500             memcpy(st->codec->extradata, attachements[j].bin.data, attachements[j].bin.size);
01501 
01502             for (i=0; ff_mkv_mime_tags[i].id != CODEC_ID_NONE; i++) {
01503                 if (!strncmp(ff_mkv_mime_tags[i].str, attachements[j].mime,
01504                              strlen(ff_mkv_mime_tags[i].str))) {
01505                     st->codec->codec_id = ff_mkv_mime_tags[i].id;
01506                     break;
01507                 }
01508             }
01509             attachements[j].stream = st;
01510         }
01511     }
01512 
01513     chapters = chapters_list->elem;
01514     for (i=0; i<chapters_list->nb_elem; i++)
01515         if (chapters[i].start != AV_NOPTS_VALUE && chapters[i].uid
01516             && (max_start==0 || chapters[i].start > max_start)) {
01517             chapters[i].chapter =
01518             ff_new_chapter(s, chapters[i].uid, (AVRational){1, 1000000000},
01519                            chapters[i].start, chapters[i].end,
01520                            chapters[i].title);
01521             av_metadata_set2(&chapters[i].chapter->metadata,
01522                              "title", chapters[i].title, 0);
01523             max_start = chapters[i].start;
01524         }
01525 
01526     index_list = &matroska->index;
01527     index = index_list->elem;
01528     if (index_list->nb_elem
01529         && index[0].time > 100000000000000/matroska->time_scale) {
01530         av_log(matroska->ctx, AV_LOG_WARNING, "Working around broken index.\n");
01531         index_scale = matroska->time_scale;
01532     }
01533     for (i=0; i<index_list->nb_elem; i++) {
01534         EbmlList *pos_list = &index[i].pos;
01535         MatroskaIndexPos *pos = pos_list->elem;
01536         for (j=0; j<pos_list->nb_elem; j++) {
01537             MatroskaTrack *track = matroska_find_track_by_num(matroska,
01538                                                               pos[j].track);
01539             if (track && track->stream)
01540                 av_add_index_entry(track->stream,
01541                                    pos[j].pos + matroska->segment_start,
01542                                    index[i].time/index_scale, 0, 0,
01543                                    AVINDEX_KEYFRAME);
01544         }
01545     }
01546 
01547     matroska_convert_tags(s);
01548 
01549     return 0;
01550 }
01551 
01552 /*
01553  * Put one packet in an application-supplied AVPacket struct.
01554  * Returns 0 on success or -1 on failure.
01555  */
01556 static int matroska_deliver_packet(MatroskaDemuxContext *matroska,
01557                                    AVPacket *pkt)
01558 {
01559     if (matroska->num_packets > 0) {
01560         memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
01561         av_free(matroska->packets[0]);
01562         if (matroska->num_packets > 1) {
01563             memmove(&matroska->packets[0], &matroska->packets[1],
01564                     (matroska->num_packets - 1) * sizeof(AVPacket *));
01565             matroska->packets =
01566                 av_realloc(matroska->packets, (matroska->num_packets - 1) *
01567                            sizeof(AVPacket *));
01568         } else {
01569             av_freep(&matroska->packets);
01570         }
01571         matroska->num_packets--;
01572         return 0;
01573     }
01574 
01575     return -1;
01576 }
01577 
01578 /*
01579  * Free all packets in our internal queue.
01580  */
01581 static void matroska_clear_queue(MatroskaDemuxContext *matroska)
01582 {
01583     if (matroska->packets) {
01584         int n;
01585         for (n = 0; n < matroska->num_packets; n++) {
01586             av_free_packet(matroska->packets[n]);
01587             av_free(matroska->packets[n]);
01588         }
01589         av_freep(&matroska->packets);
01590         matroska->num_packets = 0;
01591     }
01592 }
01593 
01594 static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data,
01595                                 int size, int64_t pos, uint64_t cluster_time,
01596                                 uint64_t duration, int is_keyframe,
01597                                 int64_t cluster_pos)
01598 {
01599     uint64_t timecode = AV_NOPTS_VALUE;
01600     MatroskaTrack *track;
01601     int res = 0;
01602     AVStream *st;
01603     AVPacket *pkt;
01604     int16_t block_time;
01605     uint32_t *lace_size = NULL;
01606     int n, flags, laces = 0;
01607     uint64_t num;
01608 
01609     if ((n = matroska_ebmlnum_uint(matroska, data, size, &num)) < 0) {
01610         av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
01611         return res;
01612     }
01613     data += n;
01614     size -= n;
01615 
01616     track = matroska_find_track_by_num(matroska, num);
01617     if (size <= 3 || !track || !track->stream) {
01618         av_log(matroska->ctx, AV_LOG_INFO,
01619                "Invalid stream %"PRIu64" or size %u\n", num, size);
01620         return res;
01621     }
01622     st = track->stream;
01623     if (st->discard >= AVDISCARD_ALL)
01624         return res;
01625     if (!duration)
01626         duration = track->default_duration / matroska->time_scale;
01627 
01628     block_time = AV_RB16(data);
01629     data += 2;
01630     flags = *data++;
01631     size -= 3;
01632     if (is_keyframe == -1)
01633         is_keyframe = flags & 0x80 ? AV_PKT_FLAG_KEY : 0;
01634 
01635     if (cluster_time != (uint64_t)-1
01636         && (block_time >= 0 || cluster_time >= -block_time)) {
01637         timecode = cluster_time + block_time;
01638         if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE
01639             && timecode < track->end_timecode)
01640             is_keyframe = 0;  /* overlapping subtitles are not key frame */
01641         if (is_keyframe)
01642             av_add_index_entry(st, cluster_pos, timecode, 0,0,AVINDEX_KEYFRAME);
01643         track->end_timecode = FFMAX(track->end_timecode, timecode+duration);
01644     }
01645 
01646     if (matroska->skip_to_keyframe && track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
01647         if (!is_keyframe || timecode < matroska->skip_to_timecode)
01648             return res;
01649         matroska->skip_to_keyframe = 0;
01650     }
01651 
01652     switch ((flags & 0x06) >> 1) {
01653         case 0x0: /* no lacing */
01654             laces = 1;
01655             lace_size = av_mallocz(sizeof(int));
01656             lace_size[0] = size;
01657             break;
01658 
01659         case 0x1: /* Xiph lacing */
01660         case 0x2: /* fixed-size lacing */
01661         case 0x3: /* EBML lacing */
01662             assert(size>0); // size <=3 is checked before size-=3 above
01663             laces = (*data) + 1;
01664             data += 1;
01665             size -= 1;
01666             lace_size = av_mallocz(laces * sizeof(int));
01667 
01668             switch ((flags & 0x06) >> 1) {
01669                 case 0x1: /* Xiph lacing */ {
01670                     uint8_t temp;
01671                     uint32_t total = 0;
01672                     for (n = 0; res == 0 && n < laces - 1; n++) {
01673                         while (1) {
01674                             if (size == 0) {
01675                                 res = -1;
01676                                 break;
01677                             }
01678                             temp = *data;
01679                             lace_size[n] += temp;
01680                             data += 1;
01681                             size -= 1;
01682                             if (temp != 0xff)
01683                                 break;
01684                         }
01685                         total += lace_size[n];
01686                     }
01687                     lace_size[n] = size - total;
01688                     break;
01689                 }
01690 
01691                 case 0x2: /* fixed-size lacing */
01692                     for (n = 0; n < laces; n++)
01693                         lace_size[n] = size / laces;
01694                     break;
01695 
01696                 case 0x3: /* EBML lacing */ {
01697                     uint32_t total;
01698                     n = matroska_ebmlnum_uint(matroska, data, size, &num);
01699                     if (n < 0) {
01700                         av_log(matroska->ctx, AV_LOG_INFO,
01701                                "EBML block data error\n");
01702                         break;
01703                     }
01704                     data += n;
01705                     size -= n;
01706                     total = lace_size[0] = num;
01707                     for (n = 1; res == 0 && n < laces - 1; n++) {
01708                         int64_t snum;
01709                         int r;
01710                         r = matroska_ebmlnum_sint(matroska, data, size, &snum);
01711                         if (r < 0) {
01712                             av_log(matroska->ctx, AV_LOG_INFO,
01713                                    "EBML block data error\n");
01714                             break;
01715                         }
01716                         data += r;
01717                         size -= r;
01718                         lace_size[n] = lace_size[n - 1] + snum;
01719                         total += lace_size[n];
01720                     }
01721                     lace_size[n] = size - total;
01722                     break;
01723                 }
01724             }
01725             break;
01726     }
01727 
01728     if (res == 0) {
01729         for (n = 0; n < laces; n++) {
01730             if ((st->codec->codec_id == CODEC_ID_RA_288 ||
01731                  st->codec->codec_id == CODEC_ID_COOK ||
01732                  st->codec->codec_id == CODEC_ID_SIPR ||
01733                  st->codec->codec_id == CODEC_ID_ATRAC3) &&
01734                  st->codec->block_align && track->audio.sub_packet_size) {
01735                 int a = st->codec->block_align;
01736                 int sps = track->audio.sub_packet_size;
01737                 int cfs = track->audio.coded_framesize;
01738                 int h = track->audio.sub_packet_h;
01739                 int y = track->audio.sub_packet_cnt;
01740                 int w = track->audio.frame_size;
01741                 int x;
01742 
01743                 if (!track->audio.pkt_cnt) {
01744                     if (track->audio.sub_packet_cnt == 0)
01745                         track->audio.buf_timecode = timecode;
01746                     if (st->codec->codec_id == CODEC_ID_RA_288)
01747                         for (x=0; x<h/2; x++)
01748                             memcpy(track->audio.buf+x*2*w+y*cfs,
01749                                    data+x*cfs, cfs);
01750                     else if (st->codec->codec_id == CODEC_ID_SIPR)
01751                         memcpy(track->audio.buf + y*w, data, w);
01752                     else
01753                         for (x=0; x<w/sps; x++)
01754                             memcpy(track->audio.buf+sps*(h*x+((h+1)/2)*(y&1)+(y>>1)), data+x*sps, sps);
01755 
01756                     if (++track->audio.sub_packet_cnt >= h) {
01757                         if (st->codec->codec_id == CODEC_ID_SIPR)
01758                             ff_rm_reorder_sipr_data(track->audio.buf, h, w);
01759                         track->audio.sub_packet_cnt = 0;
01760                         track->audio.pkt_cnt = h*w / a;
01761                     }
01762                 }
01763                 while (track->audio.pkt_cnt) {
01764                     pkt = av_mallocz(sizeof(AVPacket));
01765                     av_new_packet(pkt, a);
01766                     memcpy(pkt->data, track->audio.buf
01767                            + a * (h*w / a - track->audio.pkt_cnt--), a);
01768                     pkt->pts = track->audio.buf_timecode;
01769                     track->audio.buf_timecode = AV_NOPTS_VALUE;
01770                     pkt->pos = pos;
01771                     pkt->stream_index = st->index;
01772                     dynarray_add(&matroska->packets,&matroska->num_packets,pkt);
01773                 }
01774             } else {
01775                 MatroskaTrackEncoding *encodings = track->encodings.elem;
01776                 int offset = 0, pkt_size = lace_size[n];
01777                 uint8_t *pkt_data = data;
01778 
01779                 if (pkt_size > size) {
01780                     av_log(matroska->ctx, AV_LOG_ERROR, "Invalid packet size\n");
01781                     break;
01782                 }
01783 
01784                 if (encodings && encodings->scope & 1) {
01785                     offset = matroska_decode_buffer(&pkt_data,&pkt_size, track);
01786                     if (offset < 0)
01787                         continue;
01788                 }
01789 
01790                 pkt = av_mallocz(sizeof(AVPacket));
01791                 /* XXX: prevent data copy... */
01792                 if (av_new_packet(pkt, pkt_size+offset) < 0) {
01793                     av_free(pkt);
01794                     res = AVERROR(ENOMEM);
01795                     break;
01796                 }
01797                 if (offset)
01798                     memcpy (pkt->data, encodings->compression.settings.data, offset);
01799                 memcpy (pkt->data+offset, pkt_data, pkt_size);
01800 
01801                 if (pkt_data != data)
01802                     av_free(pkt_data);
01803 
01804                 if (n == 0)
01805                     pkt->flags = is_keyframe;
01806                 pkt->stream_index = st->index;
01807 
01808                 if (track->ms_compat)
01809                     pkt->dts = timecode;
01810                 else
01811                     pkt->pts = timecode;
01812                 pkt->pos = pos;
01813                 if (st->codec->codec_id == CODEC_ID_TEXT)
01814                     pkt->convergence_duration = duration;
01815                 else if (track->type != MATROSKA_TRACK_TYPE_SUBTITLE)
01816                     pkt->duration = duration;
01817 
01818                 if (st->codec->codec_id == CODEC_ID_SSA)
01819                     matroska_fix_ass_packet(matroska, pkt, duration);
01820 
01821                 if (matroska->prev_pkt &&
01822                     timecode != AV_NOPTS_VALUE &&
01823                     matroska->prev_pkt->pts == timecode &&
01824                     matroska->prev_pkt->stream_index == st->index &&
01825                     st->codec->codec_id == CODEC_ID_SSA)
01826                     matroska_merge_packets(matroska->prev_pkt, pkt);
01827                 else {
01828                     dynarray_add(&matroska->packets,&matroska->num_packets,pkt);
01829                     matroska->prev_pkt = pkt;
01830                 }
01831             }
01832 
01833             if (timecode != AV_NOPTS_VALUE)
01834                 timecode = duration ? timecode + duration : AV_NOPTS_VALUE;
01835             data += lace_size[n];
01836             size -= lace_size[n];
01837         }
01838     }
01839 
01840     av_free(lace_size);
01841     return res;
01842 }
01843 
01844 static int matroska_parse_cluster(MatroskaDemuxContext *matroska)
01845 {
01846     MatroskaCluster cluster = { 0 };
01847     EbmlList *blocks_list;
01848     MatroskaBlock *blocks;
01849     int i, res;
01850     int64_t pos = avio_tell(matroska->ctx->pb);
01851     matroska->prev_pkt = NULL;
01852     if (matroska->current_id)
01853         pos -= 4;  /* sizeof the ID which was already read */
01854     res = ebml_parse(matroska, matroska_clusters, &cluster);
01855     blocks_list = &cluster.blocks;
01856     blocks = blocks_list->elem;
01857     for (i=0; i<blocks_list->nb_elem; i++)
01858         if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
01859             int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
01860             res=matroska_parse_block(matroska,
01861                                      blocks[i].bin.data, blocks[i].bin.size,
01862                                      blocks[i].bin.pos,  cluster.timecode,
01863                                      blocks[i].duration, is_keyframe,
01864                                      pos);
01865         }
01866     ebml_free(matroska_cluster, &cluster);
01867     if (res < 0)  matroska->done = 1;
01868     return res;
01869 }
01870 
01871 static int matroska_read_packet(AVFormatContext *s, AVPacket *pkt)
01872 {
01873     MatroskaDemuxContext *matroska = s->priv_data;
01874 
01875     while (matroska_deliver_packet(matroska, pkt)) {
01876         if (matroska->done)
01877             return AVERROR_EOF;
01878         matroska_parse_cluster(matroska);
01879     }
01880 
01881     return 0;
01882 }
01883 
01884 static int matroska_read_seek(AVFormatContext *s, int stream_index,
01885                               int64_t timestamp, int flags)
01886 {
01887     MatroskaDemuxContext *matroska = s->priv_data;
01888     MatroskaTrack *tracks = matroska->tracks.elem;
01889     AVStream *st = s->streams[stream_index];
01890     int i, index, index_sub, index_min;
01891 
01892     if (!st->nb_index_entries)
01893         return 0;
01894     timestamp = FFMAX(timestamp, st->index_entries[0].timestamp);
01895 
01896     if ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
01897         avio_seek(s->pb, st->index_entries[st->nb_index_entries-1].pos, SEEK_SET);
01898         while ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
01899             matroska_clear_queue(matroska);
01900             if (matroska_parse_cluster(matroska) < 0)
01901                 break;
01902         }
01903     }
01904 
01905     matroska_clear_queue(matroska);
01906     if (index < 0)
01907         return 0;
01908 
01909     index_min = index;
01910     for (i=0; i < matroska->tracks.nb_elem; i++) {
01911         tracks[i].audio.pkt_cnt = 0;
01912         tracks[i].audio.sub_packet_cnt = 0;
01913         tracks[i].audio.buf_timecode = AV_NOPTS_VALUE;
01914         tracks[i].end_timecode = 0;
01915         if (tracks[i].type == MATROSKA_TRACK_TYPE_SUBTITLE
01916             && !tracks[i].stream->discard != AVDISCARD_ALL) {
01917             index_sub = av_index_search_timestamp(tracks[i].stream, st->index_entries[index].timestamp, AVSEEK_FLAG_BACKWARD);
01918             if (index_sub >= 0
01919                 && st->index_entries[index_sub].pos < st->index_entries[index_min].pos
01920                 && st->index_entries[index].timestamp - st->index_entries[index_sub].timestamp < 30000000000/matroska->time_scale)
01921                 index_min = index_sub;
01922         }
01923     }
01924 
01925     avio_seek(s->pb, st->index_entries[index_min].pos, SEEK_SET);
01926     matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
01927     matroska->skip_to_timecode = st->index_entries[index].timestamp;
01928     matroska->done = 0;
01929     av_update_cur_dts(s, st, st->index_entries[index].timestamp);
01930     return 0;
01931 }
01932 
01933 static int matroska_read_close(AVFormatContext *s)
01934 {
01935     MatroskaDemuxContext *matroska = s->priv_data;
01936     MatroskaTrack *tracks = matroska->tracks.elem;
01937     int n;
01938 
01939     matroska_clear_queue(matroska);
01940 
01941     for (n=0; n < matroska->tracks.nb_elem; n++)
01942         if (tracks[n].type == MATROSKA_TRACK_TYPE_AUDIO)
01943             av_free(tracks[n].audio.buf);
01944     ebml_free(matroska_segment, matroska);
01945 
01946     return 0;
01947 }
01948 
01949 AVInputFormat ff_matroska_demuxer = {
01950     "matroska,webm",
01951     NULL_IF_CONFIG_SMALL("Matroska/WebM file format"),
01952     sizeof(MatroskaDemuxContext),
01953     matroska_probe,
01954     matroska_read_header,
01955     matroska_read_packet,
01956     matroska_read_close,
01957     matroska_read_seek,
01958 };