isom.h 11.6 KB
Newer Older
1 2
/*
 * ISO Media common code
3
 * copyright (c) 2001 Fabrice Bellard
4 5 6
 * copyright (c) 2002 Francois Revol <revol@free.fr>
 * copyright (c) 2006 Baptiste Coudurier <baptiste.coudurier@free.fr>
 *
7 8 9
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
10 11
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
12
 * version 2.1 of the License, or (at your option) any later version.
13
 *
14
 * FFmpeg is distributed in the hope that it will be useful,
15 16 17 18 19
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
20
 * License along with FFmpeg; if not, write to the Free Software
21 22 23
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

24 25
#ifndef AVFORMAT_ISOM_H
#define AVFORMAT_ISOM_H
26

27 28 29
#include <stddef.h>
#include <stdint.h>

30
#include "libavutil/encryption_info.h"
31
#include "libavutil/mastering_display_metadata.h"
32 33 34
#include "libavutil/spherical.h"
#include "libavutil/stereo3d.h"

35
#include "avio.h"
36
#include "internal.h"
37
#include "dv.h"
38

39
/* isom.c */
40
extern const AVCodecTag ff_mp4_obj_type[];
41 42
extern const AVCodecTag ff_codec_movvideo_tags[];
extern const AVCodecTag ff_codec_movaudio_tags[];
43
extern const AVCodecTag ff_codec_movsubtitle_tags[];
44

45 46
int ff_mov_iso639_to_lang(const char lang[4], int mp4);
int ff_mov_lang_to_iso639(unsigned code, char to[4]);
47

48 49
struct AVAESCTR;

50 51 52 53 54
/* the QuickTime file format is quite convoluted...
 * it has lots of index tables, each indexing something in another one...
 * Here we just use what is needed to read the chunks
 */

55
typedef struct MOVStts {
56
    unsigned int count;
57
    int duration;
58
} MOVStts;
59

60
typedef struct MOVStsc {
61 62 63 64 65
    int first;
    int count;
    int id;
} MOVStsc;

66 67 68 69 70 71
typedef struct MOVElst {
    int64_t duration;
    int64_t time;
    float rate;
} MOVElst;

72
typedef struct MOVDref {
73 74
    uint32_t type;
    char *path;
75 76 77 78
    char *dir;
    char volume[28];
    char filename[64];
    int16_t nlvl_to, nlvl_from;
79 80
} MOVDref;

81
typedef struct MOVAtom {
82 83 84 85 86 87
    uint32_t type;
    int64_t size; /* total size (excluding the size and type fields) */
} MOVAtom;

struct MOVParseTableEntry;

88
typedef struct MOVFragment {
89 90 91
    unsigned track_id;
    uint64_t base_data_offset;
    uint64_t moof_offset;
92
    uint64_t implicit_offset;
93 94 95 96 97 98
    unsigned stsd_id;
    unsigned duration;
    unsigned size;
    unsigned flags;
} MOVFragment;

99
typedef struct MOVTrackExt {
100 101 102 103 104 105 106
    unsigned track_id;
    unsigned stsd_id;
    unsigned duration;
    unsigned size;
    unsigned flags;
} MOVTrackExt;

107
typedef struct MOVSbgp {
108 109 110 111
    unsigned int count;
    unsigned int index;
} MOVSbgp;

112 113 114 115 116
typedef struct MOVEncryptionIndex {
    // Individual encrypted samples.  If there are no elements, then the default
    // settings will be used.
    unsigned int nb_encrypted_samples;
    AVEncryptionInfo **encrypted_samples;
117 118 119 120

    uint8_t* auxiliary_info_sizes;
    size_t auxiliary_info_sample_count;
    uint8_t auxiliary_info_default_size;
121
    uint64_t* auxiliary_offsets;  ///< Absolute seek position
122
    size_t auxiliary_offsets_count;
123 124
} MOVEncryptionIndex;

125 126 127 128 129 130
typedef struct MOVFragmentStreamInfo {
    int id;
    int64_t sidx_pts;
    int64_t first_tfra_pts;
    int64_t tfdt_dts;
    int index_entry;
131
    MOVEncryptionIndex *encryption_index;
132 133
} MOVFragmentStreamInfo;

134 135
typedef struct MOVFragmentIndexItem {
    int64_t moof_offset;
136
    int headers_read;
137 138 139
    int current;
    int nb_stream_info;
    MOVFragmentStreamInfo * stream_info;
140 141 142
} MOVFragmentIndexItem;

typedef struct MOVFragmentIndex {
143 144 145 146 147
    int allocated_size;
    int complete;
    int current;
    int nb_items;
    MOVFragmentIndexItem * item;
148 149
} MOVFragmentIndex;

150 151 152 153 154
typedef struct MOVIndexRange {
    int64_t start;
    int64_t end;
} MOVIndexRange;

155
typedef struct MOVStreamContext {
156
    AVIOContext *pb;
157
    int pb_is_copied;
158
    int ffindex;          ///< AVStream index
159 160 161 162 163 164
    int next_chunk;
    unsigned int chunk_count;
    int64_t *chunk_offsets;
    unsigned int stts_count;
    MOVStts *stts_data;
    unsigned int ctts_count;
165
    unsigned int ctts_allocated_size;
166 167 168
    MOVStts *ctts_data;
    unsigned int stsc_count;
    MOVStsc *stsc_data;
169
    unsigned int stsc_index;
170
    int stsc_sample;
171 172
    unsigned int stps_count;
    unsigned *stps_data;  ///< partial sync sample for mpeg-2 open gop
173 174
    MOVElst *elst_data;
    unsigned int elst_count;
175 176
    int ctts_index;
    int ctts_sample;
177
    unsigned int sample_size; ///< may contain value calculated from stsd or value from stsz atom
178
    unsigned int stsz_sample_size; ///< always contains sample size from stsz atom
179 180
    unsigned int sample_count;
    int *sample_sizes;
181
    int keyframe_absent;
182 183 184
    unsigned int keyframe_count;
    int *keyframes;
    int time_scale;
185
    int64_t time_offset;  ///< time offset of the edit list entries
186
    int64_t min_corrected_pts;  ///< minimum Composition time shown by the edits excluding empty edits.
187
    int current_sample;
188 189 190
    int64_t current_index;
    MOVIndexRange* index_ranges;
    MOVIndexRange* current_index_range;
191 192 193 194
    unsigned int bytes_per_frame;
    unsigned int samples_per_frame;
    int dv_audio_container;
    int pseudo_stream_id; ///< -1 means demux all ids
195
    int16_t audio_cid;    ///< stsd audio compression id
196 197 198
    unsigned drefs_count;
    MOVDref *drefs;
    int dref_id;
199
    int timecode_track;
200 201 202
    int width;            ///< tkhd width
    int height;           ///< tkhd height
    int dts_shift;        ///< dts shift when ctts is negative
203 204
    uint32_t palette[256];
    int has_palette;
205
    int64_t data_size;
206
    uint32_t tmcd_flags;  ///< tmcd track flags
207
    int64_t track_end;    ///< used for dts generation in fragmented movie files
208
    int start_pad;        ///< amount of samples to skip due to enc-dec delay
209 210
    unsigned int rap_group_count;
    MOVSbgp *rap_group;
211 212 213

    int nb_frames_for_fps;
    int64_t duration_for_fps;
214

215 216 217 218 219 220
    /** extradata array (and size) for multiple stsd */
    uint8_t **extradata;
    int *extradata_size;
    int last_stsd_index;
    int stsd_count;

221
    int32_t *display_matrix;
222 223 224
    AVStereo3D *stereo3d;
    AVSphericalMapping *spherical;
    size_t spherical_size;
225
    AVMasteringDisplayMetadata *mastering;
226 227
    AVContentLightMetadata *coll;
    size_t coll_size;
228

229 230
    uint32_t format;

231
    int has_sidx;  // If there is an sidx entry for this stream.
232 233
    struct {
        struct AVAESCTR* aes_ctr;
234 235 236
        unsigned int per_sample_iv_size;  // Either 0, 8, or 16.
        AVEncryptionInfo *default_encrypted_sample;
        MOVEncryptionIndex *encryption_index;
237
    } cenc;
238 239 240
} MOVStreamContext;

typedef struct MOVContext {
241
    const AVClass *class; ///< class for private options
242 243
    AVFormatContext *fc;
    int time_scale;
244 245 246
    int64_t duration;     ///< duration of the longest track
    int found_moov;       ///< 'moov' atom has been found
    int found_mdat;       ///< 'mdat' atom has been found
247
    int found_hdlr_mdta;  ///< 'hdlr' atom with type 'mdta' has been found
248
    int trak_index;       ///< Index of the current 'trak'
249 250
    char **meta_keys;
    unsigned meta_keys_count;
251 252
    DVDemuxContext *dv_demux;
    AVFormatContext *dv_fctx;
253
    int isom;             ///< 1 if file is ISO Media (mp4/3gp)
254 255 256
    MOVFragment fragment; ///< current fragment in moof atom
    MOVTrackExt *trex_data;
    unsigned trex_count;
257
    int itunes_metadata;  ///< metadata are itunes style
258
    int handbrake_version;
259 260
    int *chapter_tracks;
    unsigned int nb_chapter_tracks;
261
    int use_absolute_path;
262
    int ignore_editlist;
263
    int advanced_editlist;
264
    int ignore_chapters;
265
    int seek_individually;
266
    int64_t next_root_atom; ///< offset of the next root atom
267
    int export_all;
268
    int export_xmp;
269 270
    int *bitrates;          ///< bitrates read before streams creation
    int bitrates_count;
271
    int moov_retry;
272 273
    int use_mfra_for;
    int has_looked_for_mfra;
274
    MOVFragmentIndex frag_index;
275
    int atom_depth;
276 277 278 279 280 281 282 283
    unsigned int aax_mode;  ///< 'aax' file has been detected
    uint8_t file_key[20];
    uint8_t file_iv[20];
    void *activation_bytes;
    int activation_bytes_size;
    void *audible_fixed_key;
    int audible_fixed_key_size;
    struct AVAES *aes_decrypt;
284 285
    uint8_t *decryption_key;
    int decryption_key_len;
286
    int enable_drefs;
287
    int32_t movie_display_matrix[3][3]; ///< display matrix from mvhd
288 289
} MOVContext;

290 291 292
int ff_mp4_read_descr_len(AVIOContext *pb);
int ff_mp4_read_descr(AVFormatContext *fc, AVIOContext *pb, int *tag);
int ff_mp4_read_dec_config_descr(AVFormatContext *fc, AVStream *st, AVIOContext *pb);
293
void ff_mp4_parse_es_descr(AVIOContext *pb, int *es_id);
294

Alex Converse's avatar
Alex Converse committed
295
#define MP4ODescrTag                    0x01
296 297 298 299
#define MP4IODescrTag                   0x02
#define MP4ESDescrTag                   0x03
#define MP4DecConfigDescrTag            0x04
#define MP4DecSpecificDescrTag          0x05
Alex Converse's avatar
Alex Converse committed
300
#define MP4SLDescrTag                   0x06
301

302 303 304 305 306 307
#define MOV_TFHD_BASE_DATA_OFFSET       0x01
#define MOV_TFHD_STSD_ID                0x02
#define MOV_TFHD_DEFAULT_DURATION       0x08
#define MOV_TFHD_DEFAULT_SIZE           0x10
#define MOV_TFHD_DEFAULT_FLAGS          0x20
#define MOV_TFHD_DURATION_IS_EMPTY  0x010000
308
#define MOV_TFHD_DEFAULT_BASE_IS_MOOF 0x020000
309

310 311 312 313 314 315 316
#define MOV_TRUN_DATA_OFFSET            0x01
#define MOV_TRUN_FIRST_SAMPLE_FLAGS     0x04
#define MOV_TRUN_SAMPLE_DURATION       0x100
#define MOV_TRUN_SAMPLE_SIZE           0x200
#define MOV_TRUN_SAMPLE_FLAGS          0x400
#define MOV_TRUN_SAMPLE_CTS            0x800

317 318 319 320 321 322 323 324 325 326
#define MOV_FRAG_SAMPLE_FLAG_DEGRADATION_PRIORITY_MASK 0x0000ffff
#define MOV_FRAG_SAMPLE_FLAG_IS_NON_SYNC               0x00010000
#define MOV_FRAG_SAMPLE_FLAG_PADDING_MASK              0x000e0000
#define MOV_FRAG_SAMPLE_FLAG_REDUNDANCY_MASK           0x00300000
#define MOV_FRAG_SAMPLE_FLAG_DEPENDED_MASK             0x00c00000
#define MOV_FRAG_SAMPLE_FLAG_DEPENDS_MASK              0x03000000

#define MOV_FRAG_SAMPLE_FLAG_DEPENDS_NO                0x02000000
#define MOV_FRAG_SAMPLE_FLAG_DEPENDS_YES               0x01000000

327 328 329 330 331
#define MOV_TKHD_FLAG_ENABLED       0x0001
#define MOV_TKHD_FLAG_IN_MOVIE      0x0002
#define MOV_TKHD_FLAG_IN_PREVIEW    0x0004
#define MOV_TKHD_FLAG_IN_POSTER     0x0008

332 333 334 335 336
#define MOV_SAMPLE_DEPENDENCY_UNKNOWN 0x0
#define MOV_SAMPLE_DEPENDENCY_YES     0x1
#define MOV_SAMPLE_DEPENDENCY_NO      0x2


337 338 339 340 341 342 343 344 345 346 347 348 349
#define TAG_IS_AVCI(tag)                    \
    ((tag) == MKTAG('a', 'i', '5', 'p') ||  \
     (tag) == MKTAG('a', 'i', '5', 'q') ||  \
     (tag) == MKTAG('a', 'i', '5', '2') ||  \
     (tag) == MKTAG('a', 'i', '5', '3') ||  \
     (tag) == MKTAG('a', 'i', '5', '5') ||  \
     (tag) == MKTAG('a', 'i', '5', '6') ||  \
     (tag) == MKTAG('a', 'i', '1', 'p') ||  \
     (tag) == MKTAG('a', 'i', '1', 'q') ||  \
     (tag) == MKTAG('a', 'i', '1', '2') ||  \
     (tag) == MKTAG('a', 'i', '1', '3') ||  \
     (tag) == MKTAG('a', 'i', '1', '5') ||  \
     (tag) == MKTAG('a', 'i', '1', '6') ||  \
350
     (tag) == MKTAG('a', 'i', 'v', 'x') ||  \
351 352 353
     (tag) == MKTAG('A', 'V', 'i', 'n'))


354
int ff_mov_read_esds(AVFormatContext *fc, AVIOContext *pb);
355

356
int ff_mov_read_stsd_entries(MOVContext *c, AVIOContext *pb, int entries);
357
void ff_mov_write_chan(AVIOContext *pb, int64_t channel_layout);
358

359
#define FF_MOV_FLAG_MFRA_AUTO -1
360 361 362
#define FF_MOV_FLAG_MFRA_DTS 1
#define FF_MOV_FLAG_MFRA_PTS 2

363 364 365 366 367 368 369 370 371 372 373 374 375 376
/**
 * Compute codec id for 'lpcm' tag.
 * See CoreAudioTypes and AudioStreamBasicDescription at Apple.
 */
static inline enum AVCodecID ff_mov_get_lpcm_codec_id(int bps, int flags)
{
    /* lpcm flags:
     * 0x1 = float
     * 0x2 = big-endian
     * 0x4 = signed
     */
    return ff_get_pcm_codec_id(bps, flags & 1, flags & 2, flags & 4 ? -1 : 0);
}

377
#endif /* AVFORMAT_ISOM_H */