3 * Copyright (c) 2011 Konstantin Shishkov
5 * This file is part of Libav.
7 * Libav 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.
12 * Libav 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.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
29 #include "libavutil/intreadwrite.h"
31 #include "bytestream.h"
42 typedef struct UtvideoContext {
43 AVCodecContext *avctx;
47 uint32_t frame_info_size, flags, frame_info;
58 typedef struct HuffEntry {
63 static int huff_cmp(const void *a, const void *b)
65 const HuffEntry *aa = a, *bb = b;
66 return (aa->len - bb->len)*256 + aa->sym - bb->sym;
69 static int build_huff(const uint8_t *src, VLC *vlc)
79 for (i = 0; i < 256; i++) {
83 qsort(he, 256, sizeof(*he), huff_cmp);
85 if (!he[0].len || he[0].len > 32)
89 while (he[last].len == 255 && last)
93 for (i = last; i >= 0; i--) {
94 codes[i] = code >> (32 - he[i].len);
97 code += 0x80000000u >> (he[i].len - 1);
100 return init_vlc_sparse(vlc, FFMIN(he[last].len, 9), last + 1,
101 bits, sizeof(*bits), sizeof(*bits),
102 codes, sizeof(*codes), sizeof(*codes),
103 syms, sizeof(*syms), sizeof(*syms), 0);
106 static int decode_plane(UtvideoContext *c, int plane_no,
107 uint8_t *dst, int step, int stride,
108 int width, int height,
109 const uint8_t *src, int src_size, int use_pred)
111 int i, j, slice, pix;
117 if (build_huff(src, &vlc)) {
118 av_log(c->avctx, AV_LOG_ERROR, "Cannot build Huffman codes\n");
119 return AVERROR_INVALIDDATA;
126 for (slice = 0; slice < c->slices; slice++) {
128 int slice_data_start, slice_data_end, slice_size;
131 send = height * (slice + 1) / c->slices;
132 dest = dst + sstart * stride;
134 // slice offset and size validation was done earlier
135 slice_data_start = slice ? AV_RL32(src + slice * 4 - 4) : 0;
136 slice_data_end = AV_RL32(src + slice * 4);
137 slice_size = slice_data_end - slice_data_start;
140 for (j = sstart; j < send; j++) {
141 for (i = 0; i < width * step; i += step)
148 memcpy(c->slice_bits, src + slice_data_start + c->slices * 4, slice_size);
149 memset(c->slice_bits + slice_size, 0, FF_INPUT_BUFFER_PADDING_SIZE);
150 c->dsp.bswap_buf((uint32_t*)c->slice_bits, (uint32_t*)c->slice_bits,
151 (slice_data_end - slice_data_start + 3) >> 2);
152 init_get_bits(&gb, c->slice_bits, slice_size * 8);
155 for (j = sstart; j < send; j++) {
156 for (i = 0; i < width * step; i += step) {
157 if (get_bits_left(&gb) <= 0) {
158 av_log(c->avctx, AV_LOG_ERROR, "Slice decoding ran out of bits\n");
161 pix = get_vlc2(&gb, vlc.table, vlc.bits, 4);
163 av_log(c->avctx, AV_LOG_ERROR, "Decoding error\n");
174 if (get_bits_left(&gb) > 32)
175 av_log(c->avctx, AV_LOG_WARNING, "%d bits left after decoding slice\n",
184 return AVERROR_INVALIDDATA;
187 static const int rgb_order[4] = { 1, 2, 0, 3 };
189 static void restore_rgb_planes(uint8_t *src, int step, int stride, int width, int height)
194 for (j = 0; j < height; j++) {
195 for (i = 0; i < width * step; i += step) {
199 src[i] = r + g - 0x80;
200 src[i + 2] = b + g - 0x80;
206 static void restore_median(uint8_t *src, int step, int stride,
207 int width, int height, int slices)
212 int slice_start, slice_height;
214 for (slice = 0; slice < slices; slice++) {
215 slice_start = (slice * height) / slices;
216 slice_height = ((slice + 1) * height) / slices - slice_start;
218 bsrc = src + slice_start * stride;
220 // first line - left neighbour prediction
223 for (i = step; i < width * step; i += step) {
228 if (slice_height == 1)
230 // second line - first element has top predition, the rest uses median
234 for (i = step; i < width * step; i += step) {
235 B = bsrc[i - stride];
236 bsrc[i] += mid_pred(A, B, (uint8_t)(A + B - C));
241 // the rest of lines use continuous median prediction
242 for (j = 2; j < slice_height; j++) {
243 for (i = 0; i < width * step; i += step) {
244 B = bsrc[i - stride];
245 bsrc[i] += mid_pred(A, B, (uint8_t)(A + B - C));
254 static int decode_frame(AVCodecContext *avctx, void *data, int *data_size, AVPacket *avpkt)
256 const uint8_t *buf = avpkt->data;
257 int buf_size = avpkt->size;
258 const uint8_t *buf_end = buf + buf_size;
259 UtvideoContext *c = avctx->priv_data;
262 const uint8_t *plane_start[5];
263 int plane_size, max_slice_size = 0, slice_start, slice_end, slice_size;
267 avctx->release_buffer(avctx, &c->pic);
269 c->pic.reference = 1;
270 c->pic.buffer_hints = FF_BUFFER_HINTS_VALID;
271 if ((ret = avctx->get_buffer(avctx, &c->pic)) < 0) {
272 av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
276 /* parse plane structure to retrieve frame flags and validate slice offsets */
278 for (i = 0; i < c->planes; i++) {
279 plane_start[i] = ptr;
280 if (buf_end - ptr < 256 + 4 * c->slices) {
281 av_log(avctx, AV_LOG_ERROR, "Insufficient data for a plane\n");
282 return AVERROR_INVALIDDATA;
287 for (j = 0; j < c->slices; j++) {
288 slice_end = bytestream_get_le32(&ptr);
289 slice_size = slice_end - slice_start;
290 if (slice_size < 0) {
291 av_log(avctx, AV_LOG_ERROR, "Incorrect slice size\n");
292 return AVERROR_INVALIDDATA;
294 slice_start = slice_end;
295 max_slice_size = FFMAX(max_slice_size, slice_size);
297 plane_size = slice_end;
298 if (buf_end - ptr < plane_size) {
299 av_log(avctx, AV_LOG_ERROR, "Plane size is bigger than available data\n");
300 return AVERROR_INVALIDDATA;
304 plane_start[c->planes] = ptr;
305 if (buf_end - ptr < c->frame_info_size) {
306 av_log(avctx, AV_LOG_ERROR, "Not enough data for frame information\n");
307 return AVERROR_INVALIDDATA;
309 c->frame_info = AV_RL32(ptr);
310 av_log(avctx, AV_LOG_DEBUG, "frame information flags %X\n", c->frame_info);
312 c->frame_pred = (c->frame_info >> 8) & 3;
314 if (c->frame_pred == PRED_GRADIENT) {
315 av_log_ask_for_sample(avctx, "Frame uses gradient prediction\n");
316 return AVERROR_PATCHWELCOME;
319 av_fast_malloc(&c->slice_bits, &c->slice_bits_size,
320 max_slice_size + FF_INPUT_BUFFER_PADDING_SIZE);
322 if (!c->slice_bits) {
323 av_log(avctx, AV_LOG_ERROR, "Cannot allocate temporary buffer\n");
324 return AVERROR(ENOMEM);
327 switch (c->avctx->pix_fmt) {
330 for (i = 0; i < c->planes; i++) {
331 ret = decode_plane(c, i, c->pic.data[0] + rgb_order[i], c->planes,
332 c->pic.linesize[0], avctx->width, avctx->height,
333 plane_start[i], plane_start[i + 1] - plane_start[i],
334 c->frame_pred == PRED_LEFT);
337 if (c->frame_pred == PRED_MEDIAN)
338 restore_median(c->pic.data[0] + rgb_order[i], c->planes,
339 c->pic.linesize[0], avctx->width, avctx->height,
342 restore_rgb_planes(c->pic.data[0], c->planes, c->pic.linesize[0],
343 avctx->width, avctx->height);
345 case PIX_FMT_YUV420P:
346 for (i = 0; i < 3; i++) {
347 ret = decode_plane(c, i, c->pic.data[i], 1,
348 c->pic.linesize[i], avctx->width >> !!i, avctx->height >> !!i,
349 plane_start[i], plane_start[i + 1] - plane_start[i],
350 c->frame_pred == PRED_LEFT);
353 if (c->frame_pred == PRED_MEDIAN)
354 restore_median(c->pic.data[i], 1, c->pic.linesize[i],
355 avctx->width >> !!i, avctx->height >> !!i,
359 case PIX_FMT_YUV422P:
360 for (i = 0; i < 3; i++) {
361 ret = decode_plane(c, i, c->pic.data[i], 1,
362 c->pic.linesize[i], avctx->width >> !!i, avctx->height,
363 plane_start[i], plane_start[i + 1] - plane_start[i],
364 c->frame_pred == PRED_LEFT);
367 if (c->frame_pred == PRED_MEDIAN)
368 restore_median(c->pic.data[i], 1, c->pic.linesize[i],
369 avctx->width >> !!i, avctx->height, c->slices);
374 *data_size = sizeof(AVFrame);
375 *(AVFrame*)data = c->pic;
377 /* always report that the buffer was completely consumed */
381 static av_cold int decode_init(AVCodecContext *avctx)
383 UtvideoContext * const c = avctx->priv_data;
387 dsputil_init(&c->dsp, avctx);
389 if (avctx->extradata_size < 16) {
390 av_log(avctx, AV_LOG_ERROR, "Insufficient extradata size %d, should be at least 16\n",
391 avctx->extradata_size);
392 return AVERROR_INVALIDDATA;
395 av_log(avctx, AV_LOG_DEBUG, "Encoder version %d.%d.%d.%d\n",
396 avctx->extradata[3], avctx->extradata[2],
397 avctx->extradata[1], avctx->extradata[0]);
398 av_log(avctx, AV_LOG_DEBUG, "Original format %X\n", AV_RB32(avctx->extradata + 4));
399 c->frame_info_size = AV_RL32(avctx->extradata + 8);
400 c->flags = AV_RL32(avctx->extradata + 12);
402 if (c->frame_info_size != 4)
403 av_log_ask_for_sample(avctx, "Frame info is not 4 bytes\n");
404 av_log(avctx, AV_LOG_DEBUG, "Encoding parameters %08X\n", c->flags);
405 c->slices = (c->flags >> 24) + 1;
406 c->compression = c->flags & 1;
407 c->interlaced = c->flags & 0x800;
409 c->slice_bits_size = 0;
411 switch (avctx->codec_tag) {
412 case MKTAG('U', 'L', 'R', 'G'):
414 avctx->pix_fmt = PIX_FMT_RGB24;
416 case MKTAG('U', 'L', 'R', 'A'):
418 avctx->pix_fmt = PIX_FMT_RGBA;
420 case MKTAG('U', 'L', 'Y', '0'):
422 avctx->pix_fmt = PIX_FMT_YUV420P;
424 case MKTAG('U', 'L', 'Y', '2'):
426 avctx->pix_fmt = PIX_FMT_YUV422P;
429 av_log(avctx, AV_LOG_ERROR, "Unknown Ut Video FOURCC provided (%08X)\n",
431 return AVERROR_INVALIDDATA;
437 static av_cold int decode_end(AVCodecContext *avctx)
439 UtvideoContext * const c = avctx->priv_data;
442 avctx->release_buffer(avctx, &c->pic);
444 av_freep(&c->slice_bits);
449 AVCodec ff_utvideo_decoder = {
451 .type = AVMEDIA_TYPE_VIDEO,
452 .id = CODEC_ID_UTVIDEO,
453 .priv_data_size = sizeof(UtvideoContext),
456 .decode = decode_frame,
457 .capabilities = CODEC_CAP_DR1,
458 .long_name = NULL_IF_CONFIG_SMALL("Ut Video"),