frame: add a function for removing side data from a frame
[libav.git] / libavutil / frame.h
1 /*
2 *
3 * This file is part of Libav.
4 *
5 * Libav is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Lesser General Public
7 * License as published by the Free Software Foundation; either
8 * version 2.1 of the License, or (at your option) any later version.
9 *
10 * Libav is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * Lesser General Public License for more details.
14 *
15 * You should have received a copy of the GNU Lesser General Public
16 * License along with Libav; if not, write to the Free Software
17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18 */
19
20 /**
21 * @file
22 * @ingroup lavu_frame
23 * reference-counted frame API
24 */
25
26 #ifndef AVUTIL_FRAME_H
27 #define AVUTIL_FRAME_H
28
29 #include <stdint.h>
30
31 #include "avutil.h"
32 #include "buffer.h"
33 #include "dict.h"
34 #include "rational.h"
35 #include "samplefmt.h"
36 #include "version.h"
37
38
39 /**
40 * @defgroup lavu_frame AVFrame
41 * @ingroup lavu_data
42 *
43 * @{
44 * AVFrame is an abstraction for reference-counted raw multimedia data.
45 */
46
47 enum AVFrameSideDataType {
48 /**
49 * The data is the AVPanScan struct defined in libavcodec.
50 */
51 AV_FRAME_DATA_PANSCAN,
52 /**
53 * ATSC A53 Part 4 Closed Captions.
54 * A53 CC bitstream is stored as uint8_t in AVFrameSideData.data.
55 * The number of bytes of CC data is AVFrameSideData.size.
56 */
57 AV_FRAME_DATA_A53_CC,
58 /**
59 * Stereoscopic 3d metadata.
60 * The data is the AVStereo3D struct defined in libavutil/stereo3d.h.
61 */
62 AV_FRAME_DATA_STEREO3D,
63 /**
64 * The data is the AVMatrixEncoding enum defined in libavutil/channel_layout.h.
65 */
66 AV_FRAME_DATA_MATRIXENCODING,
67 /**
68 * Metadata relevant to a downmix procedure.
69 * The data is the AVDownmixInfo struct defined in libavutil/downmix_info.h.
70 */
71 AV_FRAME_DATA_DOWNMIX_INFO,
72 /**
73 * ReplayGain information in the form of the AVReplayGain struct.
74 */
75 AV_FRAME_DATA_REPLAYGAIN,
76 };
77
78 typedef struct AVFrameSideData {
79 enum AVFrameSideDataType type;
80 uint8_t *data;
81 int size;
82 AVDictionary *metadata;
83 } AVFrameSideData;
84
85 /**
86 * This structure describes decoded (raw) audio or video data.
87 *
88 * AVFrame must be allocated using av_frame_alloc(). Note that this only
89 * allocates the AVFrame itself, the buffers for the data must be managed
90 * through other means (see below).
91 * AVFrame must be freed with av_frame_free().
92 *
93 * AVFrame is typically allocated once and then reused multiple times to hold
94 * different data (e.g. a single AVFrame to hold frames received from a
95 * decoder). In such a case, av_frame_unref() will free any references held by
96 * the frame and reset it to its original clean state before it
97 * is reused again.
98 *
99 * The data described by an AVFrame is usually reference counted through the
100 * AVBuffer API. The underlying buffer references are stored in AVFrame.buf /
101 * AVFrame.extended_buf. An AVFrame is considered to be reference counted if at
102 * least one reference is set, i.e. if AVFrame.buf[0] != NULL. In such a case,
103 * every single data plane must be contained in one of the buffers in
104 * AVFrame.buf or AVFrame.extended_buf.
105 * There may be a single buffer for all the data, or one separate buffer for
106 * each plane, or anything in between.
107 *
108 * sizeof(AVFrame) is not a part of the public ABI, so new fields may be added
109 * to the end with a minor bump.
110 */
111 typedef struct AVFrame {
112 #define AV_NUM_DATA_POINTERS 8
113 /**
114 * pointer to the picture/channel planes.
115 * This might be different from the first allocated byte
116 */
117 uint8_t *data[AV_NUM_DATA_POINTERS];
118
119 /**
120 * For video, size in bytes of each picture line.
121 * For audio, size in bytes of each plane.
122 *
123 * For audio, only linesize[0] may be set. For planar audio, each channel
124 * plane must be the same size.
125 *
126 * @note The linesize may be larger than the size of usable data -- there
127 * may be extra padding present for performance reasons.
128 */
129 int linesize[AV_NUM_DATA_POINTERS];
130
131 /**
132 * pointers to the data planes/channels.
133 *
134 * For video, this should simply point to data[].
135 *
136 * For planar audio, each channel has a separate data pointer, and
137 * linesize[0] contains the size of each channel buffer.
138 * For packed audio, there is just one data pointer, and linesize[0]
139 * contains the total size of the buffer for all channels.
140 *
141 * Note: Both data and extended_data should always be set in a valid frame,
142 * but for planar audio with more channels that can fit in data,
143 * extended_data must be used in order to access all channels.
144 */
145 uint8_t **extended_data;
146
147 /**
148 * width and height of the video frame
149 */
150 int width, height;
151
152 /**
153 * number of audio samples (per channel) described by this frame
154 */
155 int nb_samples;
156
157 /**
158 * format of the frame, -1 if unknown or unset
159 * Values correspond to enum AVPixelFormat for video frames,
160 * enum AVSampleFormat for audio)
161 */
162 int format;
163
164 /**
165 * 1 -> keyframe, 0-> not
166 */
167 int key_frame;
168
169 /**
170 * Picture type of the frame.
171 */
172 enum AVPictureType pict_type;
173
174 #if FF_API_AVFRAME_LAVC
175 attribute_deprecated
176 uint8_t *base[AV_NUM_DATA_POINTERS];
177 #endif
178
179 /**
180 * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified.
181 */
182 AVRational sample_aspect_ratio;
183
184 /**
185 * Presentation timestamp in time_base units (time when frame should be shown to user).
186 */
187 int64_t pts;
188
189 /**
190 * PTS copied from the AVPacket that was decoded to produce this frame.
191 */
192 int64_t pkt_pts;
193
194 /**
195 * DTS copied from the AVPacket that triggered returning this frame.
196 */
197 int64_t pkt_dts;
198
199 /**
200 * picture number in bitstream order
201 */
202 int coded_picture_number;
203 /**
204 * picture number in display order
205 */
206 int display_picture_number;
207
208 /**
209 * quality (between 1 (good) and FF_LAMBDA_MAX (bad))
210 */
211 int quality;
212
213 #if FF_API_AVFRAME_LAVC
214 attribute_deprecated
215 int reference;
216
217 /**
218 * QP table
219 */
220 attribute_deprecated
221 int8_t *qscale_table;
222 /**
223 * QP store stride
224 */
225 attribute_deprecated
226 int qstride;
227
228 attribute_deprecated
229 int qscale_type;
230
231 /**
232 * mbskip_table[mb]>=1 if MB didn't change
233 * stride= mb_width = (width+15)>>4
234 */
235 attribute_deprecated
236 uint8_t *mbskip_table;
237
238 /**
239 * motion vector table
240 * @code
241 * example:
242 * int mv_sample_log2= 4 - motion_subsample_log2;
243 * int mb_width= (width+15)>>4;
244 * int mv_stride= (mb_width << mv_sample_log2) + 1;
245 * motion_val[direction][x + y*mv_stride][0->mv_x, 1->mv_y];
246 * @endcode
247 */
248 attribute_deprecated
249 int16_t (*motion_val[2])[2];
250
251 /**
252 * macroblock type table
253 * mb_type_base + mb_width + 2
254 */
255 attribute_deprecated
256 uint32_t *mb_type;
257
258 /**
259 * DCT coefficients
260 */
261 attribute_deprecated
262 short *dct_coeff;
263
264 /**
265 * motion reference frame index
266 * the order in which these are stored can depend on the codec.
267 */
268 attribute_deprecated
269 int8_t *ref_index[2];
270 #endif
271
272 /**
273 * for some private data of the user
274 */
275 void *opaque;
276
277 /**
278 * error
279 */
280 uint64_t error[AV_NUM_DATA_POINTERS];
281
282 #if FF_API_AVFRAME_LAVC
283 attribute_deprecated
284 int type;
285 #endif
286
287 /**
288 * When decoding, this signals how much the picture must be delayed.
289 * extra_delay = repeat_pict / (2*fps)
290 */
291 int repeat_pict;
292
293 /**
294 * The content of the picture is interlaced.
295 */
296 int interlaced_frame;
297
298 /**
299 * If the content is interlaced, is top field displayed first.
300 */
301 int top_field_first;
302
303 /**
304 * Tell user application that palette has changed from previous frame.
305 */
306 int palette_has_changed;
307
308 #if FF_API_AVFRAME_LAVC
309 attribute_deprecated
310 int buffer_hints;
311
312 /**
313 * Pan scan.
314 */
315 attribute_deprecated
316 struct AVPanScan *pan_scan;
317 #endif
318
319 /**
320 * reordered opaque 64bit (generally an integer or a double precision float
321 * PTS but can be anything).
322 * The user sets AVCodecContext.reordered_opaque to represent the input at
323 * that time,
324 * the decoder reorders values as needed and sets AVFrame.reordered_opaque
325 * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque
326 * @deprecated in favor of pkt_pts
327 */
328 int64_t reordered_opaque;
329
330 #if FF_API_AVFRAME_LAVC
331 /**
332 * @deprecated this field is unused
333 */
334 attribute_deprecated void *hwaccel_picture_private;
335
336 attribute_deprecated
337 struct AVCodecContext *owner;
338 attribute_deprecated
339 void *thread_opaque;
340
341 /**
342 * log2 of the size of the block which a single vector in motion_val represents:
343 * (4->16x16, 3->8x8, 2-> 4x4, 1-> 2x2)
344 */
345 attribute_deprecated
346 uint8_t motion_subsample_log2;
347 #endif
348
349 /**
350 * Sample rate of the audio data.
351 */
352 int sample_rate;
353
354 /**
355 * Channel layout of the audio data.
356 */
357 uint64_t channel_layout;
358
359 /**
360 * AVBuffer references backing the data for this frame. If all elements of
361 * this array are NULL, then this frame is not reference counted.
362 *
363 * There may be at most one AVBuffer per data plane, so for video this array
364 * always contains all the references. For planar audio with more than
365 * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in
366 * this array. Then the extra AVBufferRef pointers are stored in the
367 * extended_buf array.
368 */
369 AVBufferRef *buf[AV_NUM_DATA_POINTERS];
370
371 /**
372 * For planar audio which requires more than AV_NUM_DATA_POINTERS
373 * AVBufferRef pointers, this array will hold all the references which
374 * cannot fit into AVFrame.buf.
375 *
376 * Note that this is different from AVFrame.extended_data, which always
377 * contains all the pointers. This array only contains the extra pointers,
378 * which cannot fit into AVFrame.buf.
379 *
380 * This array is always allocated using av_malloc() by whoever constructs
381 * the frame. It is freed in av_frame_unref().
382 */
383 AVBufferRef **extended_buf;
384 /**
385 * Number of elements in extended_buf.
386 */
387 int nb_extended_buf;
388
389 AVFrameSideData **side_data;
390 int nb_side_data;
391
392 /**
393 * @defgroup lavu_frame_flags AV_FRAME_FLAGS
394 * Flags describing additional frame properties.
395 *
396 * @{
397 */
398
399 /**
400 * The frame data may be corrupted, e.g. due to decoding errors.
401 */
402 #define AV_FRAME_FLAG_CORRUPT (1 << 0)
403 /**
404 * @}
405 */
406
407 /**
408 * Frame flags, a combination of @ref lavu_frame_flags
409 */
410 int flags;
411 } AVFrame;
412
413 /**
414 * Allocate an AVFrame and set its fields to default values. The resulting
415 * struct must be freed using av_frame_free().
416 *
417 * @return An AVFrame filled with default values or NULL on failure.
418 *
419 * @note this only allocates the AVFrame itself, not the data buffers. Those
420 * must be allocated through other means, e.g. with av_frame_get_buffer() or
421 * manually.
422 */
423 AVFrame *av_frame_alloc(void);
424
425 /**
426 * Free the frame and any dynamically allocated objects in it,
427 * e.g. extended_data. If the frame is reference counted, it will be
428 * unreferenced first.
429 *
430 * @param frame frame to be freed. The pointer will be set to NULL.
431 */
432 void av_frame_free(AVFrame **frame);
433
434 /**
435 * Set up a new reference to the data described by the source frame.
436 *
437 * Copy frame properties from src to dst and create a new reference for each
438 * AVBufferRef from src.
439 *
440 * If src is not reference counted, new buffers are allocated and the data is
441 * copied.
442 *
443 * @return 0 on success, a negative AVERROR on error
444 */
445 int av_frame_ref(AVFrame *dst, const AVFrame *src);
446
447 /**
448 * Create a new frame that references the same data as src.
449 *
450 * This is a shortcut for av_frame_alloc()+av_frame_ref().
451 *
452 * @return newly created AVFrame on success, NULL on error.
453 */
454 AVFrame *av_frame_clone(const AVFrame *src);
455
456 /**
457 * Unreference all the buffers referenced by frame and reset the frame fields.
458 */
459 void av_frame_unref(AVFrame *frame);
460
461 /**
462 * Move everythnig contained in src to dst and reset src.
463 */
464 void av_frame_move_ref(AVFrame *dst, AVFrame *src);
465
466 /**
467 * Allocate new buffer(s) for audio or video data.
468 *
469 * The following fields must be set on frame before calling this function:
470 * - format (pixel format for video, sample format for audio)
471 * - width and height for video
472 * - nb_samples and channel_layout for audio
473 *
474 * This function will fill AVFrame.data and AVFrame.buf arrays and, if
475 * necessary, allocate and fill AVFrame.extended_data and AVFrame.extended_buf.
476 * For planar formats, one buffer will be allocated for each plane.
477 *
478 * @param frame frame in which to store the new buffers.
479 * @param align required buffer size alignment
480 *
481 * @return 0 on success, a negative AVERROR on error.
482 */
483 int av_frame_get_buffer(AVFrame *frame, int align);
484
485 /**
486 * Check if the frame data is writable.
487 *
488 * @return A positive value if the frame data is writable (which is true if and
489 * only if each of the underlying buffers has only one reference, namely the one
490 * stored in this frame). Return 0 otherwise.
491 *
492 * If 1 is returned the answer is valid until av_buffer_ref() is called on any
493 * of the underlying AVBufferRefs (e.g. through av_frame_ref() or directly).
494 *
495 * @see av_frame_make_writable(), av_buffer_is_writable()
496 */
497 int av_frame_is_writable(AVFrame *frame);
498
499 /**
500 * Ensure that the frame data is writable, avoiding data copy if possible.
501 *
502 * Do nothing if the frame is writable, allocate new buffers and copy the data
503 * if it is not.
504 *
505 * @return 0 on success, a negative AVERROR on error.
506 *
507 * @see av_frame_is_writable(), av_buffer_is_writable(),
508 * av_buffer_make_writable()
509 */
510 int av_frame_make_writable(AVFrame *frame);
511
512 /**
513 * Copy the frame data from src to dst.
514 *
515 * This function does not allocate anything, dst must be already initialized and
516 * allocated with the same parameters as src.
517 *
518 * This function only copies the frame data (i.e. the contents of the data /
519 * extended data arrays), not any other properties.
520 *
521 * @return >= 0 on success, a negative AVERROR on error.
522 */
523 int av_frame_copy(AVFrame *dst, const AVFrame *src);
524
525 /**
526 * Copy only "metadata" fields from src to dst.
527 *
528 * Metadata for the purpose of this function are those fields that do not affect
529 * the data layout in the buffers. E.g. pts, sample rate (for audio) or sample
530 * aspect ratio (for video), but not width/height or channel layout.
531 * Side data is also copied.
532 */
533 int av_frame_copy_props(AVFrame *dst, const AVFrame *src);
534
535 /**
536 * Get the buffer reference a given data plane is stored in.
537 *
538 * @param plane index of the data plane of interest in frame->extended_data.
539 *
540 * @return the buffer reference that contains the plane or NULL if the input
541 * frame is not valid.
542 */
543 AVBufferRef *av_frame_get_plane_buffer(AVFrame *frame, int plane);
544
545 /**
546 * Add a new side data to a frame.
547 *
548 * @param frame a frame to which the side data should be added
549 * @param type type of the added side data
550 * @param size size of the side data
551 *
552 * @return newly added side data on success, NULL on error
553 */
554 AVFrameSideData *av_frame_new_side_data(AVFrame *frame,
555 enum AVFrameSideDataType type,
556 int size);
557
558 /**
559 * @return a pointer to the side data of a given type on success, NULL if there
560 * is no side data with such type in this frame.
561 */
562 AVFrameSideData *av_frame_get_side_data(const AVFrame *frame,
563 enum AVFrameSideDataType type);
564
565 /**
566 * If side data of the supplied type exists in the frame, free it and remove it
567 * from the frame.
568 */
569 void av_frame_remove_side_data(AVFrame *frame, enum AVFrameSideDataType type);
570
571 /**
572 * @}
573 */
574
575 #endif /* AVUTIL_FRAME_H */