FFmpeg  1.2.12
utvideodec.c
Go to the documentation of this file.
1 /*
2  * Ut Video decoder
3  * Copyright (c) 2011 Konstantin Shishkov
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
27 #include <stdlib.h>
28 
29 #include "libavutil/intreadwrite.h"
30 #include "avcodec.h"
31 #include "bytestream.h"
32 #include "get_bits.h"
33 #include "dsputil.h"
34 #include "thread.h"
35 #include "utvideo.h"
36 
37 static int build_huff(const uint8_t *src, VLC *vlc, int *fsym)
38 {
39  int i;
40  HuffEntry he[256];
41  int last;
42  uint32_t codes[256];
43  uint8_t bits[256];
44  uint8_t syms[256];
45  uint32_t code;
46 
47  *fsym = -1;
48  for (i = 0; i < 256; i++) {
49  he[i].sym = i;
50  he[i].len = *src++;
51  }
52  qsort(he, 256, sizeof(*he), ff_ut_huff_cmp_len);
53 
54  if (!he[0].len) {
55  *fsym = he[0].sym;
56  return 0;
57  }
58  if (he[0].len > 32)
59  return -1;
60 
61  last = 255;
62  while (he[last].len == 255 && last)
63  last--;
64 
65  code = 1;
66  for (i = last; i >= 0; i--) {
67  codes[i] = code >> (32 - he[i].len);
68  bits[i] = he[i].len;
69  syms[i] = he[i].sym;
70  code += 0x80000000u >> (he[i].len - 1);
71  }
72 
73  return ff_init_vlc_sparse(vlc, FFMIN(he[last].len, 11), last + 1,
74  bits, sizeof(*bits), sizeof(*bits),
75  codes, sizeof(*codes), sizeof(*codes),
76  syms, sizeof(*syms), sizeof(*syms), 0);
77 }
78 
79 static int decode_plane(UtvideoContext *c, int plane_no,
80  uint8_t *dst, int step, int stride,
81  int width, int height,
82  const uint8_t *src, int use_pred)
83 {
84  int i, j, slice, pix;
85  int sstart, send;
86  VLC vlc;
87  GetBitContext gb;
88  int prev, fsym;
89  const int cmask = ~(!plane_no && c->avctx->pix_fmt == AV_PIX_FMT_YUV420P);
90 
91  if (build_huff(src, &vlc, &fsym)) {
92  av_log(c->avctx, AV_LOG_ERROR, "Cannot build Huffman codes\n");
93  return AVERROR_INVALIDDATA;
94  }
95  if (fsym >= 0) { // build_huff reported a symbol to fill slices with
96  send = 0;
97  for (slice = 0; slice < c->slices; slice++) {
98  uint8_t *dest;
99 
100  sstart = send;
101  send = (height * (slice + 1) / c->slices) & cmask;
102  dest = dst + sstart * stride;
103 
104  prev = 0x80;
105  for (j = sstart; j < send; j++) {
106  for (i = 0; i < width * step; i += step) {
107  pix = fsym;
108  if (use_pred) {
109  prev += pix;
110  pix = prev;
111  }
112  dest[i] = pix;
113  }
114  dest += stride;
115  }
116  }
117  return 0;
118  }
119 
120  src += 256;
121 
122  send = 0;
123  for (slice = 0; slice < c->slices; slice++) {
124  uint8_t *dest;
125  int slice_data_start, slice_data_end, slice_size;
126 
127  sstart = send;
128  send = (height * (slice + 1) / c->slices) & cmask;
129  dest = dst + sstart * stride;
130 
131  // slice offset and size validation was done earlier
132  slice_data_start = slice ? AV_RL32(src + slice * 4 - 4) : 0;
133  slice_data_end = AV_RL32(src + slice * 4);
134  slice_size = slice_data_end - slice_data_start;
135 
136  if (!slice_size) {
137  av_log(c->avctx, AV_LOG_ERROR, "Plane has more than one symbol "
138  "yet a slice has a length of zero.\n");
139  goto fail;
140  }
141 
142  memcpy(c->slice_bits, src + slice_data_start + c->slices * 4,
143  slice_size);
144  memset(c->slice_bits + slice_size, 0, FF_INPUT_BUFFER_PADDING_SIZE);
145  c->dsp.bswap_buf((uint32_t *) c->slice_bits, (uint32_t *) c->slice_bits,
146  (slice_data_end - slice_data_start + 3) >> 2);
147  init_get_bits(&gb, c->slice_bits, slice_size * 8);
148 
149  prev = 0x80;
150  for (j = sstart; j < send; j++) {
151  for (i = 0; i < width * step; i += step) {
152  if (get_bits_left(&gb) <= 0) {
154  "Slice decoding ran out of bits\n");
155  goto fail;
156  }
157  pix = get_vlc2(&gb, vlc.table, vlc.bits, 4);
158  if (pix < 0) {
159  av_log(c->avctx, AV_LOG_ERROR, "Decoding error\n");
160  goto fail;
161  }
162  if (use_pred) {
163  prev += pix;
164  pix = prev;
165  }
166  dest[i] = pix;
167  }
168  dest += stride;
169  }
170  if (get_bits_left(&gb) > 32)
172  "%d bits left after decoding slice\n", get_bits_left(&gb));
173  }
174 
175  ff_free_vlc(&vlc);
176 
177  return 0;
178 fail:
179  ff_free_vlc(&vlc);
180  return AVERROR_INVALIDDATA;
181 }
182 
183 static void restore_rgb_planes(uint8_t *src, int step, int stride, int width,
184  int height)
185 {
186  int i, j;
187  uint8_t r, g, b;
188 
189  for (j = 0; j < height; j++) {
190  for (i = 0; i < width * step; i += step) {
191  r = src[i];
192  g = src[i + 1];
193  b = src[i + 2];
194  src[i] = r + g - 0x80;
195  src[i + 2] = b + g - 0x80;
196  }
197  src += stride;
198  }
199 }
200 
201 static void restore_median(uint8_t *src, int step, int stride,
202  int width, int height, int slices, int rmode)
203 {
204  int i, j, slice;
205  int A, B, C;
206  uint8_t *bsrc;
207  int slice_start, slice_height;
208  const int cmask = ~rmode;
209 
210  for (slice = 0; slice < slices; slice++) {
211  slice_start = ((slice * height) / slices) & cmask;
212  slice_height = ((((slice + 1) * height) / slices) & cmask) -
213  slice_start;
214 
215  if (!slice_height)
216  continue;
217  bsrc = src + slice_start * stride;
218 
219  // first line - left neighbour prediction
220  bsrc[0] += 0x80;
221  A = bsrc[0];
222  for (i = step; i < width * step; i += step) {
223  bsrc[i] += A;
224  A = bsrc[i];
225  }
226  bsrc += stride;
227  if (slice_height <= 1)
228  continue;
229  // second line - first element has top prediction, the rest uses median
230  C = bsrc[-stride];
231  bsrc[0] += C;
232  A = bsrc[0];
233  for (i = step; i < width * step; i += step) {
234  B = bsrc[i - stride];
235  bsrc[i] += mid_pred(A, B, (uint8_t)(A + B - C));
236  C = B;
237  A = bsrc[i];
238  }
239  bsrc += stride;
240  // the rest of lines use continuous median prediction
241  for (j = 2; j < slice_height; j++) {
242  for (i = 0; i < width * step; i += step) {
243  B = bsrc[i - stride];
244  bsrc[i] += mid_pred(A, B, (uint8_t)(A + B - C));
245  C = B;
246  A = bsrc[i];
247  }
248  bsrc += stride;
249  }
250  }
251 }
252 
253 /* UtVideo interlaced mode treats every two lines as a single one,
254  * so restoring function should take care of possible padding between
255  * two parts of the same "line".
256  */
257 static void restore_median_il(uint8_t *src, int step, int stride,
258  int width, int height, int slices, int rmode)
259 {
260  int i, j, slice;
261  int A, B, C;
262  uint8_t *bsrc;
263  int slice_start, slice_height;
264  const int cmask = ~(rmode ? 3 : 1);
265  const int stride2 = stride << 1;
266 
267  for (slice = 0; slice < slices; slice++) {
268  slice_start = ((slice * height) / slices) & cmask;
269  slice_height = ((((slice + 1) * height) / slices) & cmask) -
270  slice_start;
271  slice_height >>= 1;
272  if (!slice_height)
273  continue;
274 
275  bsrc = src + slice_start * stride;
276 
277  // first line - left neighbour prediction
278  bsrc[0] += 0x80;
279  A = bsrc[0];
280  for (i = step; i < width * step; i += step) {
281  bsrc[i] += A;
282  A = bsrc[i];
283  }
284  for (i = 0; i < width * step; i += step) {
285  bsrc[stride + i] += A;
286  A = bsrc[stride + i];
287  }
288  bsrc += stride2;
289  if (slice_height <= 1)
290  continue;
291  // second line - first element has top prediction, the rest uses median
292  C = bsrc[-stride2];
293  bsrc[0] += C;
294  A = bsrc[0];
295  for (i = step; i < width * step; i += step) {
296  B = bsrc[i - stride2];
297  bsrc[i] += mid_pred(A, B, (uint8_t)(A + B - C));
298  C = B;
299  A = bsrc[i];
300  }
301  for (i = 0; i < width * step; i += step) {
302  B = bsrc[i - stride];
303  bsrc[stride + i] += mid_pred(A, B, (uint8_t)(A + B - C));
304  C = B;
305  A = bsrc[stride + i];
306  }
307  bsrc += stride2;
308  // the rest of lines use continuous median prediction
309  for (j = 2; j < slice_height; j++) {
310  for (i = 0; i < width * step; i += step) {
311  B = bsrc[i - stride2];
312  bsrc[i] += mid_pred(A, B, (uint8_t)(A + B - C));
313  C = B;
314  A = bsrc[i];
315  }
316  for (i = 0; i < width * step; i += step) {
317  B = bsrc[i - stride];
318  bsrc[i + stride] += mid_pred(A, B, (uint8_t)(A + B - C));
319  C = B;
320  A = bsrc[i + stride];
321  }
322  bsrc += stride2;
323  }
324  }
325 }
326 
327 static int decode_frame(AVCodecContext *avctx, void *data, int *got_frame,
328  AVPacket *avpkt)
329 {
330  const uint8_t *buf = avpkt->data;
331  int buf_size = avpkt->size;
332  UtvideoContext *c = avctx->priv_data;
333  int i, j;
334  const uint8_t *plane_start[5];
335  int plane_size, max_slice_size = 0, slice_start, slice_end, slice_size;
336  int ret;
337  GetByteContext gb;
338 
339  if (c->pic.data[0])
340  ff_thread_release_buffer(avctx, &c->pic);
341 
342  c->pic.reference = 3;
344  if ((ret = ff_thread_get_buffer(avctx, &c->pic)) < 0) {
345  av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
346  return ret;
347  }
348 
349  /* parse plane structure to get frame flags and validate slice offsets */
350  bytestream2_init(&gb, buf, buf_size);
351  for (i = 0; i < c->planes; i++) {
352  plane_start[i] = gb.buffer;
353  if (bytestream2_get_bytes_left(&gb) < 256 + 4 * c->slices) {
354  av_log(avctx, AV_LOG_ERROR, "Insufficient data for a plane\n");
355  return AVERROR_INVALIDDATA;
356  }
357  bytestream2_skipu(&gb, 256);
358  slice_start = 0;
359  slice_end = 0;
360  for (j = 0; j < c->slices; j++) {
361  slice_end = bytestream2_get_le32u(&gb);
362  slice_size = slice_end - slice_start;
363  if (slice_end < 0 || slice_size < 0 ||
364  bytestream2_get_bytes_left(&gb) < slice_end) {
365  av_log(avctx, AV_LOG_ERROR, "Incorrect slice size\n");
366  return AVERROR_INVALIDDATA;
367  }
368  slice_start = slice_end;
369  max_slice_size = FFMAX(max_slice_size, slice_size);
370  }
371  plane_size = slice_end;
372  bytestream2_skipu(&gb, plane_size);
373  }
374  plane_start[c->planes] = gb.buffer;
376  av_log(avctx, AV_LOG_ERROR, "Not enough data for frame information\n");
377  return AVERROR_INVALIDDATA;
378  }
379  c->frame_info = bytestream2_get_le32u(&gb);
380  av_log(avctx, AV_LOG_DEBUG, "frame information flags %X\n", c->frame_info);
381 
382  c->frame_pred = (c->frame_info >> 8) & 3;
383 
384  if (c->frame_pred == PRED_GRADIENT) {
385  av_log_ask_for_sample(avctx, "Frame uses gradient prediction\n");
386  return AVERROR_PATCHWELCOME;
387  }
388 
390  max_slice_size + FF_INPUT_BUFFER_PADDING_SIZE);
391 
392  if (!c->slice_bits) {
393  av_log(avctx, AV_LOG_ERROR, "Cannot allocate temporary buffer\n");
394  return AVERROR(ENOMEM);
395  }
396 
397  switch (c->avctx->pix_fmt) {
398  case AV_PIX_FMT_RGB24:
399  case AV_PIX_FMT_RGBA:
400  for (i = 0; i < c->planes; i++) {
401  ret = decode_plane(c, i, c->pic.data[0] + ff_ut_rgb_order[i],
402  c->planes, c->pic.linesize[0], avctx->width,
403  avctx->height, plane_start[i],
404  c->frame_pred == PRED_LEFT);
405  if (ret)
406  return ret;
407  if (c->frame_pred == PRED_MEDIAN) {
408  if (!c->interlaced) {
410  c->planes, c->pic.linesize[0], avctx->width,
411  avctx->height, c->slices, 0);
412  } else {
414  c->planes, c->pic.linesize[0],
415  avctx->width, avctx->height, c->slices,
416  0);
417  }
418  }
419  }
420  restore_rgb_planes(c->pic.data[0], c->planes, c->pic.linesize[0],
421  avctx->width, avctx->height);
422  break;
423  case AV_PIX_FMT_YUV420P:
424  for (i = 0; i < 3; i++) {
425  ret = decode_plane(c, i, c->pic.data[i], 1, c->pic.linesize[i],
426  avctx->width >> !!i, avctx->height >> !!i,
427  plane_start[i], c->frame_pred == PRED_LEFT);
428  if (ret)
429  return ret;
430  if (c->frame_pred == PRED_MEDIAN) {
431  if (!c->interlaced) {
432  restore_median(c->pic.data[i], 1, c->pic.linesize[i],
433  avctx->width >> !!i, avctx->height >> !!i,
434  c->slices, !i);
435  } else {
436  restore_median_il(c->pic.data[i], 1, c->pic.linesize[i],
437  avctx->width >> !!i,
438  avctx->height >> !!i,
439  c->slices, !i);
440  }
441  }
442  }
443  break;
444  case AV_PIX_FMT_YUV422P:
445  for (i = 0; i < 3; i++) {
446  ret = decode_plane(c, i, c->pic.data[i], 1, c->pic.linesize[i],
447  avctx->width >> !!i, avctx->height,
448  plane_start[i], c->frame_pred == PRED_LEFT);
449  if (ret)
450  return ret;
451  if (c->frame_pred == PRED_MEDIAN) {
452  if (!c->interlaced) {
453  restore_median(c->pic.data[i], 1, c->pic.linesize[i],
454  avctx->width >> !!i, avctx->height,
455  c->slices, 0);
456  } else {
457  restore_median_il(c->pic.data[i], 1, c->pic.linesize[i],
458  avctx->width >> !!i, avctx->height,
459  c->slices, 0);
460  }
461  }
462  }
463  break;
464  }
465 
466  c->pic.key_frame = 1;
468  c->pic.interlaced_frame = !!c->interlaced;
469 
470  *got_frame = 1;
471  *(AVFrame*)data = c->pic;
472 
473  /* always report that the buffer was completely consumed */
474  return buf_size;
475 }
476 
478 {
479  UtvideoContext * const c = avctx->priv_data;
480 
481  c->avctx = avctx;
482 
483  ff_dsputil_init(&c->dsp, avctx);
484 
485  if (avctx->extradata_size < 16) {
486  av_log(avctx, AV_LOG_ERROR,
487  "Insufficient extradata size %d, should be at least 16\n",
488  avctx->extradata_size);
489  return AVERROR_INVALIDDATA;
490  }
491 
492  av_log(avctx, AV_LOG_DEBUG, "Encoder version %d.%d.%d.%d\n",
493  avctx->extradata[3], avctx->extradata[2],
494  avctx->extradata[1], avctx->extradata[0]);
495  av_log(avctx, AV_LOG_DEBUG, "Original format %X\n",
496  AV_RB32(avctx->extradata + 4));
497  c->frame_info_size = AV_RL32(avctx->extradata + 8);
498  c->flags = AV_RL32(avctx->extradata + 12);
499 
500  if (c->frame_info_size != 4)
501  av_log_ask_for_sample(avctx, "Frame info is not 4 bytes\n");
502  av_log(avctx, AV_LOG_DEBUG, "Encoding parameters %08X\n", c->flags);
503  c->slices = (c->flags >> 24) + 1;
504  c->compression = c->flags & 1;
505  c->interlaced = c->flags & 0x800;
506 
507  c->slice_bits_size = 0;
508 
509  switch (avctx->codec_tag) {
510  case MKTAG('U', 'L', 'R', 'G'):
511  c->planes = 3;
512  avctx->pix_fmt = AV_PIX_FMT_RGB24;
513  break;
514  case MKTAG('U', 'L', 'R', 'A'):
515  c->planes = 4;
516  avctx->pix_fmt = AV_PIX_FMT_RGBA;
517  break;
518  case MKTAG('U', 'L', 'Y', '0'):
519  c->planes = 3;
520  avctx->pix_fmt = AV_PIX_FMT_YUV420P;
521  break;
522  case MKTAG('U', 'L', 'Y', '2'):
523  c->planes = 3;
524  avctx->pix_fmt = AV_PIX_FMT_YUV422P;
525  break;
526  default:
527  av_log(avctx, AV_LOG_ERROR, "Unknown Ut Video FOURCC provided (%08X)\n",
528  avctx->codec_tag);
529  return AVERROR_INVALIDDATA;
530  }
531 
532  return 0;
533 }
534 
536 {
537  UtvideoContext * const c = avctx->priv_data;
538 
539  if (c->pic.data[0])
540  ff_thread_release_buffer(avctx, &c->pic);
541 
542  av_freep(&c->slice_bits);
543 
544  return 0;
545 }
546 
548  .name = "utvideo",
549  .type = AVMEDIA_TYPE_VIDEO,
550  .id = AV_CODEC_ID_UTVIDEO,
551  .priv_data_size = sizeof(UtvideoContext),
552  .init = decode_init,
553  .close = decode_end,
554  .decode = decode_frame,
555  .capabilities = CODEC_CAP_DR1 | CODEC_CAP_FRAME_THREADS,
556  .long_name = NULL_IF_CONFIG_SMALL("Ut Video"),
557 };