1 /*
2  *
3  * This file is part of FFmpeg.
4  *
5  * FFmpeg 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  * FFmpeg 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 FFmpeg; if not, write to the Free Software
17  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18  */
19 
20 #ifndef AVUTIL_FRAME_H
21 #define AVUTIL_FRAME_H
22 
23 #include <stdint.h>
24 
25 #include "libavcodec/version.h"
26 
27 #include "avutil.h"
28 #include "buffer.h"
29 #include "dict.h"
30 #include "rational.h"
31 #include "samplefmt.h"
32 
33 enum AVFrameSideDataType {
34     /**
35      * The data is the AVPanScan struct defined in libavcodec.
36      */
37     AV_FRAME_DATA_PANSCAN,
38 };
39 
40 typedef struct AVFrameSideData {
41     enum AVFrameSideDataType type;
42     uint8_t *data;
43     int      size;
44     AVDictionary *metadata;
45 } AVFrameSideData;
46 
47 /**
48  * This structure describes decoded (raw) audio or video data.
49  *
50  * AVFrame must be allocated using av_frame_alloc(). Note that this only
51  * allocates the AVFrame itself, the buffers for the data must be managed
52  * through other means (see below).
53  * AVFrame must be freed with av_frame_free().
54  *
55  * AVFrame is typically allocated once and then reused multiple times to hold
56  * different data (e.g. a single AVFrame to hold frames received from a
57  * decoder). In such a case, av_frame_unref() will free any references held by
58  * the frame and reset it to its original clean state before it
59  * is reused again.
60  *
61  * The data described by an AVFrame is usually reference counted through the
62  * AVBuffer API. The underlying buffer references are stored in AVFrame.buf /
63  * AVFrame.extended_buf. An AVFrame is considered to be reference counted if at
64  * least one reference is set, i.e. if AVFrame.buf[0] != NULL. In such a case,
65  * every single data plane must be contained in one of the buffers in
66  * AVFrame.buf or AVFrame.extended_buf.
67  * There may be a single buffer for all the data, or one separate buffer for
68  * each plane, or anything in between.
69  *
70  * sizeof(AVFrame) is not a part of the public ABI, so new fields may be added
71  * to the end with a minor bump.
72  * Similarly fields that are marked as to be only accessed by
73  * av_opt_ptr() can be reordered. This allows 2 forks to add fields
74  * without breaking compatibility with each other.
75  */
76 typedef struct AVFrame {
77 #define AV_NUM_DATA_POINTERS 8
78     /**
79      * pointer to the picture/channel planes.
80      * This might be different from the first allocated byte
81      *
82      * Some decoders access areas outside 0,0 - width,height, please
83      * see avcodec_align_dimensions2(). Some filters and swscale can read
84      * up to 16 bytes beyond the planes, if these filters are to be used,
85      * then 16 extra bytes must be allocated.
86      */
87     uint8_t *data[AV_NUM_DATA_POINTERS];
88 
89     /**
90      * For video, size in bytes of each picture line.
91      * For audio, size in bytes of each plane.
92      *
93      * For audio, only linesize[0] may be set. For planar audio, each channel
94      * plane must be the same size.
95      *
96      * For video the linesizes should be multiplies of the CPUs alignment
97      * preference, this is 16 or 32 for modern desktop CPUs.
98      * Some code requires such alignment other code can be slower without
99      * correct alignment, for yet other it makes no difference.
100      */
101     int linesize[AV_NUM_DATA_POINTERS];
102 
103     /**
104      * pointers to the data planes/channels.
105      *
106      * For video, this should simply point to data[].
107      *
108      * For planar audio, each channel has a separate data pointer, and
109      * linesize[0] contains the size of each channel buffer.
110      * For packed audio, there is just one data pointer, and linesize[0]
111      * contains the total size of the buffer for all channels.
112      *
113      * Note: Both data and extended_data should always be set in a valid frame,
114      * but for planar audio with more channels that can fit in data,
115      * extended_data must be used in order to access all channels.
116      */
117     uint8_t **extended_data;
118 
119     /**
120      * width and height of the video frame
121      */
122     int width, height;
123 
124     /**
125      * number of audio samples (per channel) described by this frame
126      */
127     int nb_samples;
128 
129     /**
130      * format of the frame, -1 if unknown or unset
131      * Values correspond to enum AVPixelFormat for video frames,
132      * enum AVSampleFormat for audio)
133      */
134     int format;
135 
136     /**
137      * 1 -> keyframe, 0-> not
138      */
139     int key_frame;
140 
141     /**
142      * Picture type of the frame.
143      */
144     enum AVPictureType pict_type;
145 
146 #if FF_API_AVFRAME_LAVC
147     attribute_deprecated
148     uint8_t *base[AV_NUM_DATA_POINTERS];
149 #endif
150 
151     /**
152      * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified.
153      */
154     AVRational sample_aspect_ratio;
155 
156     /**
157      * Presentation timestamp in time_base units (time when frame should be shown to user).
158      */
159     int64_t pts;
160 
161     /**
162      * PTS copied from the AVPacket that was decoded to produce this frame.
163      */
164     int64_t pkt_pts;
165 
166     /**
167      * DTS copied from the AVPacket that triggered returning this frame. (if frame threading isnt used)
168      * This is also the Presentation time of this AVFrame calculated from
169      * only AVPacket.dts values without pts values.
170      */
171     int64_t pkt_dts;
172 
173     /**
174      * picture number in bitstream order
175      */
176     int coded_picture_number;
177     /**
178      * picture number in display order
179      */
180     int display_picture_number;
181 
182     /**
183      * quality (between 1 (good) and FF_LAMBDA_MAX (bad))
184      */
185     int quality;
186 
187 #if FF_API_AVFRAME_LAVC
188     attribute_deprecated
189     int reference;
190 
191     /**
192      * QP table
193      */
194     attribute_deprecated
195     int8_t *qscale_table;
196     /**
197      * QP store stride
198      */
199     attribute_deprecated
200     int qstride;
201 
202     attribute_deprecated
203     int qscale_type;
204 
205     /**
206      * mbskip_table[mb]>=1 if MB didn't change
207      * stride= mb_width = (width+15)>>4
208      */
209     attribute_deprecated
210     uint8_t *mbskip_table;
211 
212     /**
213      * motion vector table
214      * @code
215      * example:
216      * int mv_sample_log2= 4 - motion_subsample_log2;
217      * int mb_width= (width+15)>>4;
218      * int mv_stride= (mb_width << mv_sample_log2) + 1;
219      * motion_val[direction][x + y*mv_stride][0->mv_x, 1->mv_y];
220      * @endcode
221      */
222     attribute_deprecated
223     int16_t (*motion_val[2])[2];
224 
225     /**
226      * macroblock type table
227      * mb_type_base + mb_width + 2
228      */
229     attribute_deprecated
230     uint32_t *mb_type;
231 
232     /**
233      * DCT coefficients
234      */
235     attribute_deprecated
236     short *dct_coeff;
237 
238     /**
239      * motion reference frame index
240      * the order in which these are stored can depend on the codec.
241      */
242     attribute_deprecated
243     int8_t *ref_index[2];
244 #endif
245 
246     /**
247      * for some private data of the user
248      */
249     void *opaque;
250 
251     /**
252      * error
253      */
254     uint64_t error[AV_NUM_DATA_POINTERS];
255 
256 #if FF_API_AVFRAME_LAVC
257     attribute_deprecated
258     int type;
259 #endif
260 
261     /**
262      * When decoding, this signals how much the picture must be delayed.
263      * extra_delay = repeat_pict / (2*fps)
264      */
265     int repeat_pict;
266 
267     /**
268      * The content of the picture is interlaced.
269      */
270     int interlaced_frame;
271 
272     /**
273      * If the content is interlaced, is top field displayed first.
274      */
275     int top_field_first;
276 
277     /**
278      * Tell user application that palette has changed from previous frame.
279      */
280     int palette_has_changed;
281 
282 #if FF_API_AVFRAME_LAVC
283     attribute_deprecated
284     int buffer_hints;
285 
286     /**
287      * Pan scan.
288      */
289     attribute_deprecated
290     struct AVPanScan *pan_scan;
291 #endif
292 
293     /**
294      * reordered opaque 64bit (generally an integer or a double precision float
295      * PTS but can be anything).
296      * The user sets AVCodecContext.reordered_opaque to represent the input at
297      * that time,
298      * the decoder reorders values as needed and sets AVFrame.reordered_opaque
299      * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque
300      * @deprecated in favor of pkt_pts
301      */
302     int64_t reordered_opaque;
303 
304 #if FF_API_AVFRAME_LAVC
305     /**
306      * @deprecated this field is unused
307      */
308     attribute_deprecated void *hwaccel_picture_private;
309 
310     attribute_deprecated
311     struct AVCodecContext *owner;
312     attribute_deprecated
313     void *thread_opaque;
314 
315     /**
316      * log2 of the size of the block which a single vector in motion_val represents:
317      * (4->16x16, 3->8x8, 2-> 4x4, 1-> 2x2)
318      */
319     attribute_deprecated
320     uint8_t motion_subsample_log2;
321 #endif
322 
323     /**
324      * Sample rate of the audio data.
325      */
326     int sample_rate;
327 
328     /**
329      * Channel layout of the audio data.
330      */
331     uint64_t channel_layout;
332 
333     /**
334      * AVBuffer references backing the data for this frame. If all elements of
335      * this array are NULL, then this frame is not reference counted.
336      *
337      * There may be at most one AVBuffer per data plane, so for video this array
338      * always contains all the references. For planar audio with more than
339      * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in
340      * this array. Then the extra AVBufferRef pointers are stored in the
341      * extended_buf array.
342      */
343     AVBufferRef *buf[AV_NUM_DATA_POINTERS];
344 
345     /**
346      * For planar audio which requires more than AV_NUM_DATA_POINTERS
347      * AVBufferRef pointers, this array will hold all the references which
348      * cannot fit into AVFrame.buf.
349      *
350      * Note that this is different from AVFrame.extended_data, which always
351      * contains all the pointers. This array only contains the extra pointers,
352      * which cannot fit into AVFrame.buf.
353      *
354      * This array is always allocated using av_malloc() by whoever constructs
355      * the frame. It is freed in av_frame_unref().
356      */
357     AVBufferRef **extended_buf;
358     /**
359      * Number of elements in extended_buf.
360      */
361     int        nb_extended_buf;
362 
363     AVFrameSideData **side_data;
364     int            nb_side_data;
365 
366     /**
367      * frame timestamp estimated using various heuristics, in stream time base
368      * Code outside libavcodec should access this field using:
369      * av_frame_get_best_effort_timestamp(frame)
370      * - encoding: unused
371      * - decoding: set by libavcodec, read by user.
372      */
373     int64_t best_effort_timestamp;
374 
375     /**
376      * reordered pos from the last AVPacket that has been input into the decoder
377      * Code outside libavcodec should access this field using:
378      * av_frame_get_pkt_pos(frame)
379      * - encoding: unused
380      * - decoding: Read by user.
381      */
382     int64_t pkt_pos;
383 
384     /**
385      * duration of the corresponding packet, expressed in
386      * AVStream->time_base units, 0 if unknown.
387      * Code outside libavcodec should access this field using:
388      * av_frame_get_pkt_duration(frame)
389      * - encoding: unused
390      * - decoding: Read by user.
391      */
392     int64_t pkt_duration;
393 
394     /**
395      * metadata.
396      * Code outside libavcodec should access this field using:
397      * av_frame_get_metadata(frame)
398      * - encoding: Set by user.
399      * - decoding: Set by libavcodec.
400      */
401     AVDictionary *metadata;
402 
403     /**
404      * decode error flags of the frame, set to a combination of
405      * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there
406      * were errors during the decoding.
407      * Code outside libavcodec should access this field using:
408      * av_frame_get_decode_error_flags(frame)
409      * - encoding: unused
410      * - decoding: set by libavcodec, read by user.
411      */
412     int decode_error_flags;
413 #define FF_DECODE_ERROR_INVALID_BITSTREAM   1
414 #define FF_DECODE_ERROR_MISSING_REFERENCE   2
415 
416     /**
417      * number of audio channels, only used for audio.
418      * Code outside libavcodec should access this field using:
419      * av_frame_get_channels(frame)
420      * - encoding: unused
421      * - decoding: Read by user.
422      */
423     int channels;
424 
425     /**
426      * size of the corresponding packet containing the compressed
427      * frame. It must be accessed using av_frame_get_pkt_size() and
428      * av_frame_set_pkt_size().
429      * It is set to a negative value if unknown.
430      * - encoding: unused
431      * - decoding: set by libavcodec, read by user.
432      */
433     int pkt_size;
434 
435     /**
436      * Not to be accessed directly from outside libavutil
437      */
438     AVBufferRef *qp_table_buf;
439 } AVFrame;
440 
441 /**
442  * Accessors for some AVFrame fields.
443  * The position of these field in the structure is not part of the ABI,
444  * they should not be accessed directly outside libavcodec.
445  */
446 int64_t av_frame_get_best_effort_timestamp(const AVFrame *frame);
447 void    av_frame_set_best_effort_timestamp(AVFrame *frame, int64_t val);
448 int64_t av_frame_get_pkt_duration         (const AVFrame *frame);
449 void    av_frame_set_pkt_duration         (AVFrame *frame, int64_t val);
450 int64_t av_frame_get_pkt_pos              (const AVFrame *frame);
451 void    av_frame_set_pkt_pos              (AVFrame *frame, int64_t val);
452 int64_t av_frame_get_channel_layout       (const AVFrame *frame);
453 void    av_frame_set_channel_layout       (AVFrame *frame, int64_t val);
454 int     av_frame_get_channels             (const AVFrame *frame);
455 void    av_frame_set_channels             (AVFrame *frame, int     val);
456 int     av_frame_get_sample_rate          (const AVFrame *frame);
457 void    av_frame_set_sample_rate          (AVFrame *frame, int     val);
458 AVDictionary *av_frame_get_metadata       (const AVFrame *frame);
459 void          av_frame_set_metadata       (AVFrame *frame, AVDictionary *val);
460 int     av_frame_get_decode_error_flags   (const AVFrame *frame);
461 void    av_frame_set_decode_error_flags   (AVFrame *frame, int     val);
462 int     av_frame_get_pkt_size(const AVFrame *frame);
463 void    av_frame_set_pkt_size(AVFrame *frame, int val);
464 AVDictionary **avpriv_frame_get_metadatap(AVFrame *frame);
465 int8_t *av_frame_get_qp_table(AVFrame *f, int *stride, int *type);
466 int av_frame_set_qp_table(AVFrame *f, AVBufferRef *buf, int stride, int type);
467 
468 /**
469  * Allocate an AVFrame and set its fields to default values.  The resulting
470  * struct must be freed using av_frame_free().
471  *
472  * @return An AVFrame filled with default values or NULL on failure.
473  *
474  * @note this only allocates the AVFrame itself, not the data buffers. Those
475  * must be allocated through other means, e.g. with av_frame_get_buffer() or
476  * manually.
477  */
478 AVFrame *av_frame_alloc(void);
479 
480 /**
481  * Free the frame and any dynamically allocated objects in it,
482  * e.g. extended_data. If the frame is reference counted, it will be
483  * unreferenced first.
484  *
485  * @param frame frame to be freed. The pointer will be set to NULL.
486  */
487 void av_frame_free(AVFrame **frame);
488 
489 /**
490  * Setup a new reference to the data described by an given frame.
491  *
492  * Copy frame properties from src to dst and create a new reference for each
493  * AVBufferRef from src.
494  *
495  * If src is not reference counted, new buffers are allocated and the data is
496  * copied.
497  *
498  * @return 0 on success, a negative AVERROR on error
499  */
500 int av_frame_ref(AVFrame *dst, AVFrame *src);
501 
502 /**
503  * Create a new frame that references the same data as src.
504  *
505  * This is a shortcut for av_frame_alloc()+av_frame_ref().
506  *
507  * @return newly created AVFrame on success, NULL on error.
508  */
509 AVFrame *av_frame_clone(AVFrame *src);
510 
511 /**
512  * Unreference all the buffers referenced by frame and reset the frame fields.
513  */
514 void av_frame_unref(AVFrame *frame);
515 
516 /**
517  * Move everythnig contained in src to dst and reset src.
518  */
519 void av_frame_move_ref(AVFrame *dst, AVFrame *src);
520 
521 /**
522  * Allocate new buffer(s) for audio or video data.
523  *
524  * The following fields must be set on frame before calling this function:
525  * - format (pixel format for video, sample format for audio)
526  * - width and height for video
527  * - nb_samples and channel_layout for audio
528  *
529  * This function will fill AVFrame.data and AVFrame.buf arrays and, if
530  * necessary, allocate and fill AVFrame.extended_data and AVFrame.extended_buf.
531  * For planar formats, one buffer will be allocated for each plane.
532  *
533  * @param frame frame in which to store the new buffers.
534  * @param align required buffer size alignment
535  *
536  * @return 0 on success, a negative AVERROR on error.
537  */
538 int av_frame_get_buffer(AVFrame *frame, int align);
539 
540 /**
541  * Check if the frame data is writable.
542  *
543  * @return A positive value if the frame data is writable (which is true if and
544  * only if each of the underlying buffers has only one reference, namely the one
545  * stored in this frame). Return 0 otherwise.
546  *
547  * If 1 is returned the answer is valid until av_buffer_ref() is called on any
548  * of the underlying AVBufferRefs (e.g. through av_frame_ref() or directly).
549  *
550  * @see av_frame_make_writable(), av_buffer_is_writable()
551  */
552 int av_frame_is_writable(AVFrame *frame);
553 
554 /**
555  * Ensure that the frame data is writable, avoiding data copy if possible.
556  *
557  * Do nothing if the frame is writable, allocate new buffers and copy the data
558  * if it is not.
559  *
560  * @return 0 on success, a negative AVERROR on error.
561  *
562  * @see av_frame_is_writable(), av_buffer_is_writable(),
563  * av_buffer_make_writable()
564  */
565 int av_frame_make_writable(AVFrame *frame);
566 
567 /**
568  * Copy only "metadata" fields from src to dst.
569  *
570  * Metadata for the purpose of this function are those fields that do not affect
571  * the data layout in the buffers.  E.g. pts, sample rate (for audio) or sample
572  * aspect ratio (for video), but not width/height or channel layout.
573  * Side data is also copied.
574  */
575 int av_frame_copy_props(AVFrame *dst, const AVFrame *src);
576 
577 /**
578  * Get the buffer reference a given data plane is stored in.
579  *
580  * @param plane index of the data plane of interest in frame->extended_data.
581  *
582  * @return the buffer reference that contains the plane or NULL if the input
583  * frame is not valid.
584  */
585 AVBufferRef *av_frame_get_plane_buffer(AVFrame *frame, int plane);
586 
587 /**
588  * Add a new side data to a frame.
589  *
590  * @param frame a frame to which the side data should be added
591  * @param type type of the added side data
592  * @param size size of the side data
593  *
594  * @return newly added side data on success, NULL on error
595  */
596 AVFrameSideData *av_frame_new_side_data(AVFrame *frame,
597                                         enum AVFrameSideDataType type,
598                                         int size);
599 
600 /**
601  * @return a pointer to the side data of a given type on success, NULL if there
602  * is no side data with such type in this frame.
603  */
604 AVFrameSideData *av_frame_get_side_data(AVFrame *frame,
605                                         enum AVFrameSideDataType type);
606 
607 #endif /* AVUTIL_FRAME_H */
608