added proposed API for parse_only mode to extract compressed frames from compressed...
[libav.git] / libavcodec / avcodec.h
1 #ifndef AVCODEC_H
2 #define AVCODEC_H
3
4 #include "common.h"
5
6 #define LIBAVCODEC_VERSION_INT 0x000406
7 #define LIBAVCODEC_VERSION "0.4.6"
8 #define LIBAVCODEC_BUILD 4618
9 #define LIBAVCODEC_BUILD_STR "4618"
10
11 enum CodecID {
12 CODEC_ID_NONE,
13 CODEC_ID_MPEG1VIDEO,
14 CODEC_ID_H263,
15 CODEC_ID_RV10,
16 CODEC_ID_MP2,
17 CODEC_ID_MP3LAME,
18 CODEC_ID_AC3,
19 CODEC_ID_MJPEG,
20 CODEC_ID_MPEG4,
21 CODEC_ID_RAWVIDEO,
22 CODEC_ID_MSMPEG4V1,
23 CODEC_ID_MSMPEG4V2,
24 CODEC_ID_MSMPEG4V3,
25 CODEC_ID_WMV1,
26 CODEC_ID_WMV2,
27 CODEC_ID_H263P,
28 CODEC_ID_H263I,
29 CODEC_ID_SVQ1,
30
31 /* various pcm "codecs" */
32 CODEC_ID_PCM_S16LE,
33 CODEC_ID_PCM_S16BE,
34 CODEC_ID_PCM_U16LE,
35 CODEC_ID_PCM_U16BE,
36 CODEC_ID_PCM_S8,
37 CODEC_ID_PCM_U8,
38 CODEC_ID_PCM_MULAW,
39 CODEC_ID_PCM_ALAW,
40
41 /* various adpcm codecs */
42 CODEC_ID_ADPCM_IMA_QT,
43 CODEC_ID_ADPCM_IMA_WAV,
44 CODEC_ID_ADPCM_MS,
45 };
46 #define CODEC_ID_MSMPEG4 CODEC_ID_MSMPEG4V3
47
48 enum CodecType {
49 CODEC_TYPE_UNKNOWN = -1,
50 CODEC_TYPE_VIDEO,
51 CODEC_TYPE_AUDIO,
52 };
53
54 enum PixelFormat {
55 PIX_FMT_ANY = -1,
56 PIX_FMT_YUV420P,
57 PIX_FMT_YUV422,
58 PIX_FMT_RGB24,
59 PIX_FMT_BGR24,
60 PIX_FMT_YUV422P,
61 PIX_FMT_YUV444P,
62 PIX_FMT_YUV410P
63 };
64
65 /* currently unused, may be used if 24/32 bits samples ever supported */
66 enum SampleFormat {
67 SAMPLE_FMT_S16 = 0, /* signed 16 bits */
68 };
69
70 /* in bytes */
71 #define AVCODEC_MAX_AUDIO_FRAME_SIZE 18432
72
73 /* motion estimation type, EPZS by default */
74 enum Motion_Est_ID {
75 ME_ZERO = 1,
76 ME_FULL,
77 ME_LOG,
78 ME_PHODS,
79 ME_EPZS,
80 ME_X1
81 };
82
83 /* only for ME compatiblity with old apps */
84 extern int motion_estimation_method;
85
86 /* ME algos sorted by quality */
87 static const int Motion_Est_QTab[] = { ME_ZERO, ME_PHODS, ME_LOG,
88 ME_X1, ME_EPZS, ME_FULL };
89
90 #define FF_MAX_B_FRAMES 4
91
92 /* encoding support */
93 /* note not everything is supported yet */
94
95 #define CODEC_FLAG_HQ 0x0001 /* high quality (non real time) encoding */
96 #define CODEC_FLAG_QSCALE 0x0002 /* use fixed qscale */
97 #define CODEC_FLAG_4MV 0x0004 /* 4 MV per MB allowed */
98 #define CODEC_FLAG_QPEL 0x0010 /* use qpel MC */
99 #define CODEC_FLAG_GMC 0x0020 /* use GMC */
100 #define CODEC_FLAG_TYPE 0x0040 /* fixed I/P frame type, from avctx->key_frame */
101 #define CODEC_FLAG_PART 0x0080 /* use data partitioning */
102 /* parent program gurantees that the input for b-frame containing streams is not written to
103 for at least s->max_b_frames+1 frames, if this is not set than the input will be copied */
104 #define CODEC_FLAG_INPUT_PRESERVED 0x0100
105 #define CODEC_FLAG_PASS1 0x0200 /* use internal 2pass ratecontrol in first pass mode */
106 #define CODEC_FLAG_PASS2 0x0400 /* use internal 2pass ratecontrol in second pass mode */
107 #define CODEC_FLAG_EXTERN_HUFF 0x1000 /* use external huffman table (for mjpeg) */
108 #define CODEC_FLAG_GRAY 0x2000 /* only decode/encode grayscale */
109 #define CODEC_FLAG_EMU_EDGE 0x4000/* dont draw edges */
110 #define CODEC_FLAG_DR1 0x8000 /* dr1 */
111 /* codec capabilities */
112
113 /* decoder can use draw_horiz_band callback */
114 #define CODEC_CAP_DRAW_HORIZ_BAND 0x0001
115 #define CODEC_CAP_DR1 0x0002 /* direct rendering method 1 */
116 /* if 'parse_only' field is true, then avcodec_parse_frame() can be
117 used */
118 #define CODEC_CAP_PARSE_ONLY 0x0004
119
120 #define FRAME_RATE_BASE 10000
121
122 typedef struct AVCodecContext {
123 int bit_rate;
124 int bit_rate_tolerance; /* amount of +- bits (>0)*/
125 int flags;
126 int sub_id; /* some codecs needs additionnal format info. It is
127 stored there */
128
129 int me_method; /* ME algorithm used for video coding */
130
131 /* extra data from parent application to codec, e.g. huffman table
132 for mjpeg */
133 /* the parent should allocate and free this buffer */
134 void *extradata;
135 int extradata_size;
136
137 /* video only */
138 int frame_rate; /* frames per sec multiplied by FRAME_RATE_BASE */
139 int width, height;
140 int aspect_ratio_info;
141 #define FF_ASPECT_SQUARE 1
142 #define FF_ASPECT_4_3_625 2
143 #define FF_ASPECT_4_3_525 3
144 #define FF_ASPECT_16_9_625 4
145 #define FF_ASPECT_16_9_525 5
146 int gop_size; /* 0 = intra only */
147 enum PixelFormat pix_fmt; /* pixel format, see PIX_FMT_xxx */
148 int repeat_pict; /* when decoding, this signal how much the picture */
149 /* must be delayed. */
150 /* extra_delay = (repeat_pict / 2) * (1/fps) */
151 /* if non NULL, 'draw_horiz_band' is called by the libavcodec
152 decoder to draw an horizontal band. It improve cache usage. Not
153 all codecs can do that. You must check the codec capabilities
154 before */
155 void (*draw_horiz_band)(struct AVCodecContext *s,
156 UINT8 **src_ptr, int linesize,
157 int y, int width, int height);
158
159 /* audio only */
160 int sample_rate; /* samples per sec */
161 int channels;
162 int sample_fmt; /* sample format, currenly unused */
163
164 /* the following data should not be initialized */
165 int frame_size; /* in samples, initialized when calling 'init' */
166 int frame_number; /* audio or video frame number */
167 int real_pict_num; /* returns the real picture number of
168 previous encoded frame */
169 int key_frame; /* true if the previous compressed frame was
170 a key frame (intra, or seekable) */
171 int pict_type; /* picture type of the previous
172 encoded frame */
173 /* FIXME: these should have FF_ */
174 #define I_TYPE 1 // Intra
175 #define P_TYPE 2 // Predicted
176 #define B_TYPE 3 // Bi-dir predicted
177 #define S_TYPE 4 // S(GMC)-VOP MPEG4
178
179 int delay; /* number of frames the decoded output
180 will be delayed relative to the encoded input */
181 uint8_t *mbskip_table; /* =1 if MB didnt change, is only valid for I/P frames
182 stride= mb_width = (width+15)>>4 */
183
184 /* encoding parameters */
185 int quality; /* quality of the previous encoded frame
186 (between 1 (good) and 31 (bad))
187 this is allso used to set the quality in vbr mode
188 and the per frame quality in CODEC_FLAG_TYPE (second pass mode) */
189 float qcompress; /* amount of qscale change between easy & hard scenes (0.0-1.0)*/
190 float qblur; /* amount of qscale smoothing over time (0.0-1.0) */
191 int qmin; /* min qscale */
192 int qmax; /* max qscale */
193 int max_qdiff; /* max qscale difference between frames */
194 int max_b_frames; /* maximum b frames, the output will be delayed by max_b_frames+1 relative to the input */
195 float b_quant_factor;/* qscale factor between ips and b frames */
196 int rc_strategy;
197 int b_frame_strategy;
198
199 int hurry_up; /* when set to 1 during decoding, b frames will be skiped
200 when set to 2 idct/dequant will be skipped too */
201
202 struct AVCodec *codec;
203 void *priv_data;
204
205 /* The following data is for RTP friendly coding */
206 /* By now only H.263/H.263+/MPEG4 coder honours this */
207 int rtp_mode; /* 1 for activate RTP friendly-mode */
208 /* highers numbers represent more error-prone */
209 /* enviroments, by now just "1" exist */
210
211 int rtp_payload_size; /* The size of the RTP payload, the coder will */
212 /* do it's best to deliver a chunk with size */
213 /* below rtp_payload_size, the chunk will start */
214 /* with a start code on some codecs like H.263 */
215 /* This doesn't take account of any particular */
216 /* headers inside the transmited RTP payload */
217
218
219 /* The RTP callcack: This function is called */
220 /* every time the encoder as a packet to send */
221 /* Depends on the encoder if the data starts */
222 /* with a Start Code (it should) H.263 does */
223 void (*rtp_callback)(void *data, int size, int packet_number);
224
225 /* These are for PSNR calculation, if you set get_psnr to 1 */
226 /* after encoding you will have the PSNR on psnr_y/cb/cr */
227 int get_psnr;
228 float psnr_y;
229 float psnr_cb;
230 float psnr_cr;
231
232 /* statistics, used for 2-pass encoding */
233 int mv_bits;
234 int header_bits;
235 int i_tex_bits;
236 int p_tex_bits;
237 int i_count;
238 int p_count;
239 int skip_count;
240 int misc_bits; // cbp, mb_type
241 int frame_bits;
242
243 /* the following fields are ignored */
244 void *opaque; /* can be used to carry app specific stuff */
245 char codec_name[32];
246 enum CodecType codec_type; /* see CODEC_TYPE_xxx */
247 enum CodecID codec_id; /* see CODEC_ID_xxx */
248 unsigned int codec_tag; /* codec tag, only used if unknown codec */
249
250 int workaround_bugs; /* workaround bugs in encoders which cannot be detected automatically */
251 int luma_elim_threshold;
252 int chroma_elim_threshold;
253 int strict_std_compliance; /* strictly follow the std (MPEG4, ...) */
254 float b_quant_offset;/* qscale offset between ips and b frames, not implemented yet */
255 int error_resilience;
256
257 #ifndef MBC
258 #define MBC 128
259 #define MBR 96
260 #endif
261 #define QP_TYPE int //FIXME note xxx this might be changed to int8_t
262
263 QP_TYPE *quant_store; /* field for communicating with external postprocessing */
264
265 unsigned qstride;
266 uint8_t *dr_buffer[3];
267 int dr_stride;
268 void *dr_opaque_frame;
269 void (*get_buffer_callback)(struct AVCodecContext *c, int width, int height, int pict_type);
270
271 int has_b_frames; // is 1 if the decoded stream contains b frames
272 int dr_uvstride;
273 int dr_ip_buffer_count;
274 int block_align; /* currently only for adpcm codec in wav/avi */
275
276 int parse_only; /* decoding only: if true, only parsing is done
277 (function avcodec_parse_frame()). The frame
278 data is returned. Only MPEG codecs support this now. */
279
280 //FIXME this should be reordered after kabis API is finished ...
281 /*
282 Note: Below are located reserved fields for further usage
283 It requires for ABI !!!
284 If you'll perform some changes then borrow new space from these fields
285 (void * can be safety replaced with struct * ;)
286 P L E A S E ! ! !
287 IMPORTANT: Never change order of already declared fields!!!
288 */
289 unsigned long long int
290 ull_res0,ull_res1,ull_res2,ull_res3,ull_res4,ull_res5,
291 ull_res6,ull_res7,ull_res8,ull_res9,ull_res10,ull_res11,ull_res12;
292 float
293 flt_res0,flt_res1,flt_res2,flt_res3,flt_res4,flt_res5,
294 flt_res6,flt_res7,flt_res8,flt_res9,flt_res10,flt_res11;
295 void
296 *ptr_res0,*ptr_res1,*ptr_res2,*ptr_res3,*ptr_res4,*ptr_res5,
297 *ptr_res6;
298 unsigned long int
299 ul_res0,ul_res1,ul_res2,ul_res3,ul_res4,ul_res5,
300 ul_res6,ul_res7,ul_res8,ul_res9,ul_res10,ul_res11,ul_res12;
301 unsigned int
302 ui_res0;
303 unsigned short int
304 us_res0,us_res1,us_res2,us_res3,us_res4,us_res5,
305 us_res6,us_res7,us_res8,us_res9,us_res10,us_res11,us_res12;
306 unsigned char
307 uc_res0,uc_res1,uc_res2,uc_res3,uc_res4,uc_res5,
308 uc_res6,uc_res7,uc_res8,uc_res9,uc_res10,uc_res11,uc_res12;
309 } AVCodecContext;
310
311 typedef struct AVCodec {
312 char *name;
313 int type;
314 int id;
315 int priv_data_size;
316 int (*init)(AVCodecContext *);
317 int (*encode)(AVCodecContext *, UINT8 *buf, int buf_size, void *data);
318 int (*close)(AVCodecContext *);
319 int (*decode)(AVCodecContext *, void *outdata, int *outdata_size,
320 UINT8 *buf, int buf_size);
321 int capabilities;
322 struct AVCodec *next;
323 /*
324 Note: Below are located reserved fields for further usage
325 It requires for ABI !!!
326 If you'll perform some changes then borrow new space from these fields
327 (void * can be safety replaced with struct * ;)
328 P L E A S E ! ! !
329 IMPORTANT: Never change order of already declared fields!!!
330 */
331 unsigned long long int
332 ull_res0,ull_res1,ull_res2,ull_res3,ull_res4,ull_res5,
333 ull_res6,ull_res7,ull_res8,ull_res9,ull_res10,ull_res11,ull_res12;
334 float
335 flt_res0,flt_res1,flt_res2,flt_res3,flt_res4,flt_res5,
336 flt_res6,flt_res7,flt_res8,flt_res9,flt_res10,flt_res11,flt_res12;
337 void
338 *ptr_res0,*ptr_res1,*ptr_res2,*ptr_res3,*ptr_res4,*ptr_res5,
339 *ptr_res6,*ptr_res7,*ptr_res8,*ptr_res9,*ptr_res10,*ptr_res11,*ptr_res12;
340 } AVCodec;
341
342 /* three components are given, that's all */
343 typedef struct AVPicture {
344 UINT8 *data[3];
345 int linesize[3];
346 } AVPicture;
347
348 extern AVCodec ac3_encoder;
349 extern AVCodec mp2_encoder;
350 extern AVCodec mp3lame_encoder;
351 extern AVCodec mpeg1video_encoder;
352 extern AVCodec h263_encoder;
353 extern AVCodec h263p_encoder;
354 extern AVCodec rv10_encoder;
355 extern AVCodec mjpeg_encoder;
356 extern AVCodec mpeg4_encoder;
357 extern AVCodec msmpeg4v1_encoder;
358 extern AVCodec msmpeg4v2_encoder;
359 extern AVCodec msmpeg4v3_encoder;
360 extern AVCodec wmv1_encoder;
361 extern AVCodec wmv2_encoder;
362
363 extern AVCodec h263_decoder;
364 extern AVCodec mpeg4_decoder;
365 extern AVCodec msmpeg4v1_decoder;
366 extern AVCodec msmpeg4v2_decoder;
367 extern AVCodec msmpeg4v3_decoder;
368 extern AVCodec wmv1_decoder;
369 extern AVCodec wmv2_decoder;
370 extern AVCodec mpeg_decoder;
371 extern AVCodec h263i_decoder;
372 extern AVCodec rv10_decoder;
373 extern AVCodec svq1_decoder;
374 extern AVCodec mjpeg_decoder;
375 extern AVCodec mp2_decoder;
376 extern AVCodec mp3_decoder;
377
378 /* pcm codecs */
379 #define PCM_CODEC(id, name) \
380 extern AVCodec name ## _decoder; \
381 extern AVCodec name ## _encoder;
382
383 PCM_CODEC(CODEC_ID_PCM_S16LE, pcm_s16le);
384 PCM_CODEC(CODEC_ID_PCM_S16BE, pcm_s16be);
385 PCM_CODEC(CODEC_ID_PCM_U16LE, pcm_u16le);
386 PCM_CODEC(CODEC_ID_PCM_U16BE, pcm_u16be);
387 PCM_CODEC(CODEC_ID_PCM_S8, pcm_s8);
388 PCM_CODEC(CODEC_ID_PCM_U8, pcm_u8);
389 PCM_CODEC(CODEC_ID_PCM_ALAW, pcm_alaw);
390 PCM_CODEC(CODEC_ID_PCM_MULAW, pcm_mulaw);
391
392 /* adpcm codecs */
393
394 PCM_CODEC(CODEC_ID_ADPCM_IMA_QT, adpcm_ima_qt);
395 PCM_CODEC(CODEC_ID_ADPCM_IMA_WAV, adpcm_ima_wav);
396 PCM_CODEC(CODEC_ID_ADPCM_MS, adpcm_ms);
397
398 #undef PCM_CODEC
399
400 /* dummy raw video codec */
401 extern AVCodec rawvideo_codec;
402
403 /* the following codecs use external GPL libs */
404 extern AVCodec ac3_decoder;
405
406 /* resample.c */
407
408 struct ReSampleContext;
409
410 typedef struct ReSampleContext ReSampleContext;
411
412 ReSampleContext *audio_resample_init(int output_channels, int input_channels,
413 int output_rate, int input_rate);
414 int audio_resample(ReSampleContext *s, short *output, short *input, int nb_samples);
415 void audio_resample_close(ReSampleContext *s);
416
417 /* YUV420 format is assumed ! */
418
419 struct ImgReSampleContext;
420
421 typedef struct ImgReSampleContext ImgReSampleContext;
422
423 ImgReSampleContext *img_resample_init(int output_width, int output_height,
424 int input_width, int input_height);
425 void img_resample(ImgReSampleContext *s,
426 AVPicture *output, AVPicture *input);
427
428 void img_resample_close(ImgReSampleContext *s);
429
430 void avpicture_fill(AVPicture *picture, UINT8 *ptr,
431 int pix_fmt, int width, int height);
432 int avpicture_get_size(int pix_fmt, int width, int height);
433
434 /* convert among pixel formats */
435 int img_convert(AVPicture *dst, int dst_pix_fmt,
436 AVPicture *src, int pix_fmt,
437 int width, int height);
438
439 /* deinterlace a picture */
440 int avpicture_deinterlace(AVPicture *dst, AVPicture *src,
441 int pix_fmt, int width, int height);
442
443 /* external high level API */
444
445 extern AVCodec *first_avcodec;
446
447 /* returns LIBAVCODEC_VERSION_INT constant */
448 unsigned avcodec_version(void);
449 /* returns LIBAVCODEC_BUILD constant */
450 unsigned avcodec_build(void);
451 void avcodec_init(void);
452
453 void avcodec_set_bit_exact(void);
454
455 void register_avcodec(AVCodec *format);
456 AVCodec *avcodec_find_encoder(enum CodecID id);
457 AVCodec *avcodec_find_encoder_by_name(const char *name);
458 AVCodec *avcodec_find_decoder(enum CodecID id);
459 AVCodec *avcodec_find_decoder_by_name(const char *name);
460 void avcodec_string(char *buf, int buf_size, AVCodecContext *enc, int encode);
461
462 int avcodec_open(AVCodecContext *avctx, AVCodec *codec);
463 int avcodec_decode_audio(AVCodecContext *avctx, INT16 *samples,
464 int *frame_size_ptr,
465 UINT8 *buf, int buf_size);
466 int avcodec_decode_video(AVCodecContext *avctx, AVPicture *picture,
467 int *got_picture_ptr,
468 UINT8 *buf, int buf_size);
469 int avcodec_parse_frame(AVCodecContext *avctx, UINT8 **pdata,
470 int *data_size_ptr,
471 UINT8 *buf, int buf_size);
472 int avcodec_encode_audio(AVCodecContext *avctx, UINT8 *buf, int buf_size,
473 const short *samples);
474 int avcodec_encode_video(AVCodecContext *avctx, UINT8 *buf, int buf_size,
475 const AVPicture *pict);
476
477 int avcodec_close(AVCodecContext *avctx);
478
479 void avcodec_register_all(void);
480
481 void avcodec_flush_buffers(AVCodecContext *avctx);
482
483 #ifdef FF_POSTPROCESS
484 extern int quant_store[MBR+1][MBC+1]; // [Review]
485 #endif
486
487
488 /**
489 * Interface for 0.5.0 version
490 *
491 * do not even think about it's usage for this moment
492 */
493
494 typedef struct {
495 // compressed size used from given memory buffer
496 int size;
497 /// I/P/B frame type
498 int frame_type;
499 } avc_enc_result_t;
500
501 /**
502 * Commands
503 * order can't be changed - once it was defined
504 */
505 typedef enum {
506 // general commands
507 AVC_OPEN_BY_NAME = 0xACA000,
508 AVC_OPEN_BY_CODEC_ID,
509 AVC_OPEN_BY_FOURCC,
510 AVC_CLOSE,
511
512 AVC_FLUSH,
513 // pin - struct { uint8_t* src, uint_t src_size }
514 // pout - struct { AVPicture* img, consumed_bytes,
515 AVC_DECODE,
516 // pin - struct { AVPicture* img, uint8_t* dest, uint_t dest_size }
517 // pout - uint_t used_from_dest_size
518 AVC_ENCODE,
519
520 // query/get video commands
521 AVC_GET_VERSION = 0xACB000,
522 AVC_GET_WIDTH,
523 AVC_GET_HEIGHT,
524 AVC_GET_DELAY,
525 AVC_GET_QUANT_TABLE,
526 // ...
527
528 // query/get audio commands
529 AVC_GET_FRAME_SIZE = 0xABC000,
530
531 // maybe define some simple structure which
532 // might be passed to the user - but they can't
533 // contain any codec specific parts and these
534 // calls are usualy necessary only few times
535
536 // set video commands
537 AVC_SET_WIDTH = 0xACD000,
538 AVC_SET_HEIGHT,
539
540 // set video encoding commands
541 AVC_SET_FRAME_RATE = 0xACD800,
542 AVC_SET_QUALITY,
543 AVC_SET_HURRY_UP,
544
545 // set audio commands
546 AVC_SET_SAMPLE_RATE = 0xACE000,
547 AVC_SET_CHANNELS,
548
549 } avc_cmd_t;
550
551 /**
552 * \param handle allocated private structure by libavcodec
553 * for initialization pass NULL - will be returned pout
554 * user is supposed to know nothing about its structure
555 * \param cmd type of operation to be performed
556 * \param pint input parameter
557 * \param pout output parameter
558 *
559 * \returns command status - eventually for query command it might return
560 * integer resulting value
561 */
562 int avcodec(void* handle, avc_cmd_t cmd, void* pin, void* pout);
563
564 /* memory */
565 void *av_malloc(int size);
566 void *av_mallocz(int size);
567 void av_free(void *ptr);
568 void __av_freep(void **ptr);
569 #define av_freep(p) __av_freep((void **)(p))
570
571 #endif /* AVCODEC_H */