blob: 67d78c5ddd2600b76604beb8f62b929d3ba72cf7
1 | /* |
2 | * Apple Intermediate Codec decoder |
3 | * |
4 | * Copyright (c) 2013 Konstantin Shishkov |
5 | * |
6 | * This file is part of FFmpeg. |
7 | * |
8 | * FFmpeg is free software; you can redistribute it and/or |
9 | * modify it under the terms of the GNU Lesser General Public |
10 | * License as published by the Free Software Foundation; either |
11 | * version 2.1 of the License, or (at your option) any later version. |
12 | * |
13 | * FFmpeg is distributed in the hope that it will be useful, |
14 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
15 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
16 | * Lesser General Public License for more details. |
17 | * |
18 | * You should have received a copy of the GNU Lesser General Public |
19 | * License along with FFmpeg; if not, write to the Free Software |
20 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
21 | */ |
22 | |
23 | #include <inttypes.h> |
24 | |
25 | #include "avcodec.h" |
26 | #include "bytestream.h" |
27 | #include "internal.h" |
28 | #include "get_bits.h" |
29 | #include "golomb.h" |
30 | #include "idctdsp.h" |
31 | #include "thread.h" |
32 | #include "unary.h" |
33 | |
34 | #define AIC_HDR_SIZE 24 |
35 | #define AIC_BAND_COEFFS (64 + 32 + 192 + 96) |
36 | |
37 | enum AICBands { |
38 | COEFF_LUMA = 0, |
39 | COEFF_CHROMA, |
40 | COEFF_LUMA_EXT, |
41 | COEFF_CHROMA_EXT, |
42 | NUM_BANDS |
43 | }; |
44 | |
45 | static const int aic_num_band_coeffs[NUM_BANDS] = { 64, 32, 192, 96 }; |
46 | |
47 | static const int aic_band_off[NUM_BANDS] = { 0, 64, 96, 288 }; |
48 | |
49 | static const uint8_t aic_quant_matrix[64] = { |
50 | 8, 16, 19, 22, 22, 26, 26, 27, |
51 | 16, 16, 22, 22, 26, 27, 27, 29, |
52 | 19, 22, 26, 26, 27, 29, 29, 35, |
53 | 22, 24, 27, 27, 29, 32, 34, 38, |
54 | 26, 27, 29, 29, 32, 35, 38, 46, |
55 | 27, 29, 34, 34, 35, 40, 46, 56, |
56 | 29, 34, 34, 37, 40, 48, 56, 69, |
57 | 34, 37, 38, 40, 48, 58, 69, 83, |
58 | }; |
59 | |
60 | static const uint8_t aic_y_scan[64] = { |
61 | 0, 4, 1, 2, 5, 8, 12, 9, |
62 | 6, 3, 7, 10, 13, 14, 11, 15, |
63 | 47, 43, 46, 45, 42, 39, 35, 38, |
64 | 41, 44, 40, 37, 34, 33, 36, 32, |
65 | 16, 20, 17, 18, 21, 24, 28, 25, |
66 | 22, 19, 23, 26, 29, 30, 27, 31, |
67 | 63, 59, 62, 61, 58, 55, 51, 54, |
68 | 57, 60, 56, 53, 50, 49, 52, 48, |
69 | }; |
70 | |
71 | static const uint8_t aic_y_ext_scan[192] = { |
72 | 64, 72, 65, 66, 73, 80, 88, 81, |
73 | 74, 67, 75, 82, 89, 90, 83, 91, |
74 | 0, 4, 1, 2, 5, 8, 12, 9, |
75 | 6, 3, 7, 10, 13, 14, 11, 15, |
76 | 16, 20, 17, 18, 21, 24, 28, 25, |
77 | 22, 19, 23, 26, 29, 30, 27, 31, |
78 | 155, 147, 154, 153, 146, 139, 131, 138, |
79 | 145, 152, 144, 137, 130, 129, 136, 128, |
80 | 47, 43, 46, 45, 42, 39, 35, 38, |
81 | 41, 44, 40, 37, 34, 33, 36, 32, |
82 | 63, 59, 62, 61, 58, 55, 51, 54, |
83 | 57, 60, 56, 53, 50, 49, 52, 48, |
84 | 96, 104, 97, 98, 105, 112, 120, 113, |
85 | 106, 99, 107, 114, 121, 122, 115, 123, |
86 | 68, 76, 69, 70, 77, 84, 92, 85, |
87 | 78, 71, 79, 86, 93, 94, 87, 95, |
88 | 100, 108, 101, 102, 109, 116, 124, 117, |
89 | 110, 103, 111, 118, 125, 126, 119, 127, |
90 | 187, 179, 186, 185, 178, 171, 163, 170, |
91 | 177, 184, 176, 169, 162, 161, 168, 160, |
92 | 159, 151, 158, 157, 150, 143, 135, 142, |
93 | 149, 156, 148, 141, 134, 133, 140, 132, |
94 | 191, 183, 190, 189, 182, 175, 167, 174, |
95 | 181, 188, 180, 173, 166, 165, 172, 164, |
96 | }; |
97 | |
98 | static const uint8_t aic_c_scan[64] = { |
99 | 0, 4, 1, 2, 5, 8, 12, 9, |
100 | 6, 3, 7, 10, 13, 14, 11, 15, |
101 | 31, 27, 30, 29, 26, 23, 19, 22, |
102 | 25, 28, 24, 21, 18, 17, 20, 16, |
103 | 32, 36, 33, 34, 37, 40, 44, 41, |
104 | 38, 35, 39, 42, 45, 46, 43, 47, |
105 | 63, 59, 62, 61, 58, 55, 51, 54, |
106 | 57, 60, 56, 53, 50, 49, 52, 48, |
107 | }; |
108 | |
109 | static const uint8_t aic_c_ext_scan[192] = { |
110 | 16, 24, 17, 18, 25, 32, 40, 33, |
111 | 26, 19, 27, 34, 41, 42, 35, 43, |
112 | 0, 4, 1, 2, 5, 8, 12, 9, |
113 | 6, 3, 7, 10, 13, 14, 11, 15, |
114 | 20, 28, 21, 22, 29, 36, 44, 37, |
115 | 30, 23, 31, 38, 45, 46, 39, 47, |
116 | 95, 87, 94, 93, 86, 79, 71, 78, |
117 | 85, 92, 84, 77, 70, 69, 76, 68, |
118 | 63, 59, 62, 61, 58, 55, 51, 54, |
119 | 57, 60, 56, 53, 50, 49, 52, 48, |
120 | 91, 83, 90, 89, 82, 75, 67, 74, |
121 | 81, 88, 80, 73, 66, 65, 72, 64, |
122 | 112, 120, 113, 114, 121, 128, 136, 129, |
123 | 122, 115, 123, 130, 137, 138, 131, 139, |
124 | 96, 100, 97, 98, 101, 104, 108, 105, |
125 | 102, 99, 103, 106, 109, 110, 107, 111, |
126 | 116, 124, 117, 118, 125, 132, 140, 133, |
127 | 126, 119, 127, 134, 141, 142, 135, 143, |
128 | 191, 183, 190, 189, 182, 175, 167, 174, |
129 | 181, 188, 180, 173, 166, 165, 172, 164, |
130 | 159, 155, 158, 157, 154, 151, 147, 150, |
131 | 153, 156, 152, 149, 146, 145, 148, 144, |
132 | 187, 179, 186, 185, 178, 171, 163, 170, |
133 | 177, 184, 176, 169, 162, 161, 168, 160, |
134 | }; |
135 | |
136 | static const uint8_t * const aic_scan[NUM_BANDS] = { |
137 | aic_y_scan, aic_c_scan, aic_y_ext_scan, aic_c_ext_scan |
138 | }; |
139 | |
140 | typedef struct AICContext { |
141 | AVCodecContext *avctx; |
142 | AVFrame *frame; |
143 | IDCTDSPContext idsp; |
144 | ScanTable scantable; |
145 | |
146 | int num_x_slices; |
147 | int slice_width; |
148 | int mb_width, mb_height; |
149 | int quant; |
150 | int interlaced; |
151 | |
152 | int16_t *slice_data; |
153 | int16_t *data_ptr[NUM_BANDS]; |
154 | |
155 | DECLARE_ALIGNED(16, int16_t, block)[64]; |
156 | DECLARE_ALIGNED(16, uint8_t, quant_matrix)[64]; |
157 | } AICContext; |
158 | |
159 | static int aic_decode_header(AICContext *ctx, const uint8_t *src, int size) |
160 | { |
161 | uint32_t frame_size; |
162 | int width, height; |
163 | |
164 | if (src[0] != 1) { |
165 | av_log(ctx->avctx, AV_LOG_ERROR, "Invalid version %d\n", src[0]); |
166 | return AVERROR_INVALIDDATA; |
167 | } |
168 | if (src[1] != AIC_HDR_SIZE - 2) { |
169 | av_log(ctx->avctx, AV_LOG_ERROR, "Invalid header size %d\n", src[1]); |
170 | return AVERROR_INVALIDDATA; |
171 | } |
172 | frame_size = AV_RB32(src + 2); |
173 | width = AV_RB16(src + 6); |
174 | height = AV_RB16(src + 8); |
175 | if (frame_size > size) { |
176 | av_log(ctx->avctx, AV_LOG_ERROR, "Frame size should be %"PRIu32" got %d\n", |
177 | frame_size, size); |
178 | return AVERROR_INVALIDDATA; |
179 | } |
180 | if (width != ctx->avctx->width || height != ctx->avctx->height) { |
181 | av_log(ctx->avctx, AV_LOG_ERROR, |
182 | "Picture dimension changed: old: %d x %d, new: %d x %d\n", |
183 | ctx->avctx->width, ctx->avctx->height, width, height); |
184 | return AVERROR_INVALIDDATA; |
185 | } |
186 | ctx->quant = src[15]; |
187 | ctx->interlaced = ((src[16] >> 4) == 3); |
188 | |
189 | return 0; |
190 | } |
191 | |
192 | #define GET_CODE(val, type, add_bits) \ |
193 | do { \ |
194 | if (type) \ |
195 | val = get_ue_golomb(gb); \ |
196 | else \ |
197 | val = get_unary(gb, 1, 31); \ |
198 | if (add_bits) \ |
199 | val = (val << add_bits) + get_bits(gb, add_bits); \ |
200 | } while (0) |
201 | |
202 | static int aic_decode_coeffs(GetBitContext *gb, int16_t *dst, |
203 | int band, int slice_width, int force_chroma) |
204 | { |
205 | int has_skips, coeff_type, coeff_bits, skip_type, skip_bits; |
206 | const int num_coeffs = aic_num_band_coeffs[band]; |
207 | const uint8_t *scan = aic_scan[band | force_chroma]; |
208 | int mb, idx; |
209 | unsigned val; |
210 | |
211 | has_skips = get_bits1(gb); |
212 | coeff_type = get_bits1(gb); |
213 | coeff_bits = get_bits(gb, 3); |
214 | |
215 | if (has_skips) { |
216 | skip_type = get_bits1(gb); |
217 | skip_bits = get_bits(gb, 3); |
218 | |
219 | for (mb = 0; mb < slice_width; mb++) { |
220 | idx = -1; |
221 | do { |
222 | GET_CODE(val, skip_type, skip_bits); |
223 | if (val >= 0x10000) |
224 | return AVERROR_INVALIDDATA; |
225 | idx += val + 1; |
226 | if (idx >= num_coeffs) |
227 | break; |
228 | GET_CODE(val, coeff_type, coeff_bits); |
229 | val++; |
230 | if (val >= 0x10000) |
231 | return AVERROR_INVALIDDATA; |
232 | dst[scan[idx]] = val; |
233 | } while (idx < num_coeffs - 1); |
234 | dst += num_coeffs; |
235 | } |
236 | } else { |
237 | for (mb = 0; mb < slice_width; mb++) { |
238 | for (idx = 0; idx < num_coeffs; idx++) { |
239 | GET_CODE(val, coeff_type, coeff_bits); |
240 | if (val >= 0x10000) |
241 | return AVERROR_INVALIDDATA; |
242 | dst[scan[idx]] = val; |
243 | } |
244 | dst += num_coeffs; |
245 | } |
246 | } |
247 | return 0; |
248 | } |
249 | |
250 | static void recombine_block(int16_t *dst, const uint8_t *scan, |
251 | int16_t **base, int16_t **ext) |
252 | { |
253 | int i, j; |
254 | |
255 | for (i = 0; i < 4; i++) { |
256 | for (j = 0; j < 4; j++) |
257 | dst[scan[i * 8 + j]] = (*base)[j]; |
258 | for (j = 0; j < 4; j++) |
259 | dst[scan[i * 8 + j + 4]] = (*ext)[j]; |
260 | *base += 4; |
261 | *ext += 4; |
262 | } |
263 | for (; i < 8; i++) { |
264 | for (j = 0; j < 8; j++) |
265 | dst[scan[i * 8 + j]] = (*ext)[j]; |
266 | *ext += 8; |
267 | } |
268 | } |
269 | |
270 | static void recombine_block_il(int16_t *dst, const uint8_t *scan, |
271 | int16_t **base, int16_t **ext, |
272 | int block_no) |
273 | { |
274 | int i, j; |
275 | |
276 | if (block_no < 2) { |
277 | for (i = 0; i < 8; i++) { |
278 | for (j = 0; j < 4; j++) |
279 | dst[scan[i * 8 + j]] = (*base)[j]; |
280 | for (j = 0; j < 4; j++) |
281 | dst[scan[i * 8 + j + 4]] = (*ext)[j]; |
282 | *base += 4; |
283 | *ext += 4; |
284 | } |
285 | } else { |
286 | for (i = 0; i < 64; i++) |
287 | dst[scan[i]] = (*ext)[i]; |
288 | *ext += 64; |
289 | } |
290 | } |
291 | |
292 | static void unquant_block(int16_t *block, int q, uint8_t *quant_matrix) |
293 | { |
294 | int i; |
295 | |
296 | for (i = 0; i < 64; i++) { |
297 | int val = (uint16_t)block[i]; |
298 | int sign = val & 1; |
299 | |
300 | block[i] = (((val >> 1) ^ -sign) * q * quant_matrix[i] >> 4) |
301 | + sign; |
302 | } |
303 | } |
304 | |
305 | static int aic_decode_slice(AICContext *ctx, int mb_x, int mb_y, |
306 | const uint8_t *src, int src_size) |
307 | { |
308 | GetBitContext gb; |
309 | int ret, i, mb, blk; |
310 | int slice_width = FFMIN(ctx->slice_width, ctx->mb_width - mb_x); |
311 | uint8_t *Y, *C[2]; |
312 | uint8_t *dst; |
313 | int16_t *base_y = ctx->data_ptr[COEFF_LUMA]; |
314 | int16_t *base_c = ctx->data_ptr[COEFF_CHROMA]; |
315 | int16_t *ext_y = ctx->data_ptr[COEFF_LUMA_EXT]; |
316 | int16_t *ext_c = ctx->data_ptr[COEFF_CHROMA_EXT]; |
317 | const int ystride = ctx->frame->linesize[0]; |
318 | |
319 | Y = ctx->frame->data[0] + mb_x * 16 + mb_y * 16 * ystride; |
320 | for (i = 0; i < 2; i++) |
321 | C[i] = ctx->frame->data[i + 1] + mb_x * 8 |
322 | + mb_y * 8 * ctx->frame->linesize[i + 1]; |
323 | init_get_bits(&gb, src, src_size * 8); |
324 | |
325 | memset(ctx->slice_data, 0, |
326 | sizeof(*ctx->slice_data) * slice_width * AIC_BAND_COEFFS); |
327 | for (i = 0; i < NUM_BANDS; i++) |
328 | if ((ret = aic_decode_coeffs(&gb, ctx->data_ptr[i], |
329 | i, slice_width, |
330 | !ctx->interlaced)) < 0) |
331 | return ret; |
332 | |
333 | for (mb = 0; mb < slice_width; mb++) { |
334 | for (blk = 0; blk < 4; blk++) { |
335 | if (!ctx->interlaced) |
336 | recombine_block(ctx->block, ctx->scantable.permutated, |
337 | &base_y, &ext_y); |
338 | else |
339 | recombine_block_il(ctx->block, ctx->scantable.permutated, |
340 | &base_y, &ext_y, blk); |
341 | unquant_block(ctx->block, ctx->quant, ctx->quant_matrix); |
342 | ctx->idsp.idct(ctx->block); |
343 | |
344 | if (!ctx->interlaced) { |
345 | dst = Y + (blk >> 1) * 8 * ystride + (blk & 1) * 8; |
346 | ctx->idsp.put_signed_pixels_clamped(ctx->block, dst, ystride); |
347 | } else { |
348 | dst = Y + (blk & 1) * 8 + (blk >> 1) * ystride; |
349 | ctx->idsp.put_signed_pixels_clamped(ctx->block, dst, |
350 | ystride * 2); |
351 | } |
352 | } |
353 | Y += 16; |
354 | |
355 | for (blk = 0; blk < 2; blk++) { |
356 | recombine_block(ctx->block, ctx->scantable.permutated, |
357 | &base_c, &ext_c); |
358 | unquant_block(ctx->block, ctx->quant, ctx->quant_matrix); |
359 | ctx->idsp.idct(ctx->block); |
360 | ctx->idsp.put_signed_pixels_clamped(ctx->block, C[blk], |
361 | ctx->frame->linesize[blk + 1]); |
362 | C[blk] += 8; |
363 | } |
364 | } |
365 | |
366 | return 0; |
367 | } |
368 | |
369 | static int aic_decode_frame(AVCodecContext *avctx, void *data, int *got_frame, |
370 | AVPacket *avpkt) |
371 | { |
372 | AICContext *ctx = avctx->priv_data; |
373 | const uint8_t *buf = avpkt->data; |
374 | int buf_size = avpkt->size; |
375 | GetByteContext gb; |
376 | uint32_t off; |
377 | int x, y, ret; |
378 | int slice_size; |
379 | ThreadFrame frame = { .f = data }; |
380 | |
381 | ctx->frame = data; |
382 | ctx->frame->pict_type = AV_PICTURE_TYPE_I; |
383 | ctx->frame->key_frame = 1; |
384 | |
385 | off = FFALIGN(AIC_HDR_SIZE + ctx->num_x_slices * ctx->mb_height * 2, 4); |
386 | |
387 | if (buf_size < off) { |
388 | av_log(avctx, AV_LOG_ERROR, "Too small frame\n"); |
389 | return AVERROR_INVALIDDATA; |
390 | } |
391 | |
392 | ret = aic_decode_header(ctx, buf, buf_size); |
393 | if (ret < 0) { |
394 | av_log(avctx, AV_LOG_ERROR, "Invalid header\n"); |
395 | return ret; |
396 | } |
397 | |
398 | if ((ret = ff_thread_get_buffer(avctx, &frame, 0)) < 0) |
399 | return ret; |
400 | |
401 | bytestream2_init(&gb, buf + AIC_HDR_SIZE, |
402 | ctx->num_x_slices * ctx->mb_height * 2); |
403 | |
404 | for (y = 0; y < ctx->mb_height; y++) { |
405 | for (x = 0; x < ctx->mb_width; x += ctx->slice_width) { |
406 | slice_size = bytestream2_get_le16(&gb) * 4; |
407 | if (slice_size + off > buf_size || !slice_size) { |
408 | av_log(avctx, AV_LOG_ERROR, |
409 | "Incorrect slice size %d at %d.%d\n", slice_size, x, y); |
410 | return AVERROR_INVALIDDATA; |
411 | } |
412 | |
413 | ret = aic_decode_slice(ctx, x, y, buf + off, slice_size); |
414 | if (ret < 0) { |
415 | av_log(avctx, AV_LOG_ERROR, |
416 | "Error decoding slice at %d.%d\n", x, y); |
417 | return ret; |
418 | } |
419 | |
420 | off += slice_size; |
421 | } |
422 | } |
423 | |
424 | *got_frame = 1; |
425 | |
426 | return avpkt->size; |
427 | } |
428 | |
429 | static av_cold int aic_decode_init(AVCodecContext *avctx) |
430 | { |
431 | AICContext *ctx = avctx->priv_data; |
432 | int i; |
433 | uint8_t scan[64]; |
434 | |
435 | ctx->avctx = avctx; |
436 | |
437 | avctx->pix_fmt = AV_PIX_FMT_YUV420P; |
438 | |
439 | ff_idctdsp_init(&ctx->idsp, avctx); |
440 | |
441 | for (i = 0; i < 64; i++) |
442 | scan[i] = i; |
443 | ff_init_scantable(ctx->idsp.idct_permutation, &ctx->scantable, scan); |
444 | for (i = 0; i < 64; i++) |
445 | ctx->quant_matrix[ctx->idsp.idct_permutation[i]] = aic_quant_matrix[i]; |
446 | |
447 | ctx->mb_width = FFALIGN(avctx->width, 16) >> 4; |
448 | ctx->mb_height = FFALIGN(avctx->height, 16) >> 4; |
449 | |
450 | ctx->num_x_slices = (ctx->mb_width + 15) >> 4; |
451 | ctx->slice_width = 16; |
452 | for (i = 1; i < ctx->mb_width; i++) { |
453 | if (!(ctx->mb_width % i) && (ctx->mb_width / i <= 32)) { |
454 | ctx->slice_width = ctx->mb_width / i; |
455 | ctx->num_x_slices = i; |
456 | break; |
457 | } |
458 | } |
459 | |
460 | ctx->slice_data = av_malloc_array(ctx->slice_width, AIC_BAND_COEFFS |
461 | * sizeof(*ctx->slice_data)); |
462 | if (!ctx->slice_data) { |
463 | av_log(avctx, AV_LOG_ERROR, "Error allocating slice buffer\n"); |
464 | |
465 | return AVERROR(ENOMEM); |
466 | } |
467 | |
468 | for (i = 0; i < NUM_BANDS; i++) |
469 | ctx->data_ptr[i] = ctx->slice_data + ctx->slice_width |
470 | * aic_band_off[i]; |
471 | |
472 | return 0; |
473 | } |
474 | |
475 | static av_cold int aic_decode_close(AVCodecContext *avctx) |
476 | { |
477 | AICContext *ctx = avctx->priv_data; |
478 | |
479 | av_freep(&ctx->slice_data); |
480 | |
481 | return 0; |
482 | } |
483 | |
484 | AVCodec ff_aic_decoder = { |
485 | .name = "aic", |
486 | .long_name = NULL_IF_CONFIG_SMALL("Apple Intermediate Codec"), |
487 | .type = AVMEDIA_TYPE_VIDEO, |
488 | .id = AV_CODEC_ID_AIC, |
489 | .priv_data_size = sizeof(AICContext), |
490 | .init = aic_decode_init, |
491 | .close = aic_decode_close, |
492 | .decode = aic_decode_frame, |
493 | .capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_FRAME_THREADS, |
494 | .init_thread_copy = ONLY_IF_THREADS_ENABLED(aic_decode_init), |
495 | .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE, |
496 | }; |
497 |