2 * Sierra VMD Audio & Video Decoders
3 * Copyright (C) 2004 the ffmpeg project
5 * This file is part of FFmpeg.
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.
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.
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
24 * Sierra VMD audio & video decoders
25 * by Vladimir "VAG" Gneushev (vagsoft at mail.ru)
26 * for more information on the Sierra VMD format, visit:
27 * http://www.pcisys.net/~melanson/codecs/
29 * The video decoder outputs PAL8 colorspace data. The decoder expects
30 * a 0x330-byte VMD file header to be transmitted via extradata during
31 * codec initialization. Each encoded frame that is sent to this decoder
32 * is expected to be prepended with the appropriate 16-byte frame
33 * information record from the VMD file.
35 * The audio decoder, like the video decoder, expects each encoded data
36 * chunk to be prepended with the appropriate 16-byte frame information
37 * record from the VMD file. It does not require the 0x330-byte VMD file
38 * header, but it does need the audio setup parameters passed in through
39 * normal libavcodec API means.
46 #include "libavutil/intreadwrite.h"
49 #define VMD_HEADER_SIZE 0x330
50 #define PALETTE_COUNT 256
56 typedef struct VmdVideoContext {
58 AVCodecContext *avctx;
62 const unsigned char *buf;
65 unsigned char palette[PALETTE_COUNT * 4];
66 unsigned char *unpack_buffer;
67 int unpack_buffer_size;
72 #define QUEUE_SIZE 0x1000
73 #define QUEUE_MASK 0x0FFF
75 static void lz_unpack(const unsigned char *src, int src_len,
76 unsigned char *dest, int dest_len)
78 const unsigned char *s;
79 const unsigned char *s_end;
82 unsigned char queue[QUEUE_SIZE];
84 unsigned int dataleft;
85 unsigned int chainofs;
86 unsigned int chainlen;
92 s_end = src + src_len;
98 dataleft = AV_RL32(s);
100 memset(queue, 0x20, QUEUE_SIZE);
101 if (AV_RL32(s) == 0x56781234) {
107 speclen = 100; /* no speclen */
110 while (s_end - s > 0 && dataleft > 0) {
112 if ((tag == 0xFF) && (dataleft > 8)) {
113 if (d_end - d < 8 || s_end - s < 8)
115 for (i = 0; i < 8; i++) {
116 queue[qpos++] = *d++ = *s++;
121 for (i = 0; i < 8; i++) {
125 if (d_end - d < 1 || s_end - s < 1)
127 queue[qpos++] = *d++ = *s++;
134 chainofs |= ((*s & 0xF0) << 4);
135 chainlen = (*s++ & 0x0F) + 3;
136 if (chainlen == speclen) {
139 chainlen = *s++ + 0xF + 3;
141 if (d_end - d < chainlen)
143 for (j = 0; j < chainlen; j++) {
144 *d = queue[chainofs++ & QUEUE_MASK];
145 queue[qpos++] = *d++;
148 dataleft -= chainlen;
156 static int rle_unpack(const unsigned char *src, int src_len, int src_count,
157 unsigned char *dest, int dest_len)
159 const unsigned char *ps;
160 const unsigned char *ps_end;
163 unsigned char *dest_end = dest + dest_len;
166 ps_end = src + src_len;
182 if (dest_end - pd < l || ps_end - ps < l)
188 if (dest_end - pd < i || ps_end - ps < 2)
190 for (i = 0; i < l; i++) {
197 } while (i < src_count);
202 static void vmd_decode(VmdVideoContext *s)
205 unsigned int *palette32;
206 unsigned char r, g, b;
208 /* point to the start of the encoded data */
209 const unsigned char *p = s->buf + 16;
210 const unsigned char *p_end = s->buf + s->size;
212 const unsigned char *pb;
213 const unsigned char *pb_end;
215 unsigned char *dp; /* pointer to current frame */
216 unsigned char *pp; /* pointer to previous frame */
220 int frame_x, frame_y;
221 int frame_width, frame_height;
223 frame_x = AV_RL16(&s->buf[6]);
224 frame_y = AV_RL16(&s->buf[8]);
225 frame_width = AV_RL16(&s->buf[10]) - frame_x + 1;
226 frame_height = AV_RL16(&s->buf[12]) - frame_y + 1;
227 if (frame_x < 0 || frame_width < 0 ||
228 frame_x >= s->avctx->width ||
229 frame_width > s->avctx->width ||
230 frame_x + frame_width > s->avctx->width)
232 if (frame_y < 0 || frame_height < 0 ||
233 frame_y >= s->avctx->height ||
234 frame_height > s->avctx->height ||
235 frame_y + frame_height > s->avctx->height)
238 if ((frame_width == s->avctx->width && frame_height == s->avctx->height) &&
239 (frame_x || frame_y)) {
247 /* if only a certain region will be updated, copy the entire previous
248 * frame before the decode */
249 if (s->prev_frame.data[0] &&
250 (frame_x || frame_y || (frame_width != s->avctx->width) ||
251 (frame_height != s->avctx->height))) {
253 memcpy(s->frame.data[0], s->prev_frame.data[0],
254 s->avctx->height * s->frame.linesize[0]);
257 /* check if there is a new palette */
258 if (s->buf[15] & 0x02) {
259 if (p_end - p < 2 + 3 * PALETTE_COUNT)
262 palette32 = (unsigned int *)s->palette;
263 for (i = 0; i < PALETTE_COUNT; i++) {
267 palette32[i] = (r << 16) | (g << 8) | (b);
271 /* originally UnpackFrame in VAG's code */
276 lz_unpack(pb, p_end - pb, s->unpack_buffer, s->unpack_buffer_size);
278 pb = s->unpack_buffer;
279 pb_end = s->unpack_buffer + s->unpack_buffer_size;
282 dp = &s->frame.data[0][frame_y * s->frame.linesize[0] + frame_x];
283 pp = &s->prev_frame.data[0][frame_y * s->prev_frame.linesize[0] + frame_x];
286 for (i = 0; i < frame_height; i++) {
293 len = (len & 0x7F) + 1;
294 if (ofs + len > frame_width || pb_end - pb < len)
296 memcpy(&dp[ofs], pb, len);
300 /* interframe pixel copy */
301 if (ofs + len + 1 > frame_width || !s->prev_frame.data[0])
303 memcpy(&dp[ofs], &pp[ofs], len + 1);
306 } while (ofs < frame_width);
307 if (ofs > frame_width) {
308 av_log(s->avctx, AV_LOG_ERROR, "VMD video: offset > width (%d > %d)\n",
312 dp += s->frame.linesize[0];
313 pp += s->prev_frame.linesize[0];
318 for (i = 0; i < frame_height; i++) {
319 if (pb_end -pb < frame_width)
321 memcpy(dp, pb, frame_width);
323 dp += s->frame.linesize[0];
324 pp += s->prev_frame.linesize[0];
329 for (i = 0; i < frame_height; i++) {
336 len = (len & 0x7F) + 1;
340 len = rle_unpack(pb, pb_end - pb, len, &dp[ofs], frame_width - ofs);
342 if (pb_end - pb < len)
344 memcpy(&dp[ofs], pb, len);
349 /* interframe pixel copy */
350 if (ofs + len + 1 > frame_width || !s->prev_frame.data[0])
352 memcpy(&dp[ofs], &pp[ofs], len + 1);
355 } while (ofs < frame_width);
356 if (ofs > frame_width) {
357 av_log(s->avctx, AV_LOG_ERROR, "VMD video: offset > width (%d > %d)\n",
360 dp += s->frame.linesize[0];
361 pp += s->prev_frame.linesize[0];
368 static av_cold int vmdvideo_decode_init(AVCodecContext *avctx)
370 VmdVideoContext *s = avctx->priv_data;
372 unsigned int *palette32;
373 int palette_index = 0;
374 unsigned char r, g, b;
375 unsigned char *vmd_header;
376 unsigned char *raw_palette;
379 avctx->pix_fmt = PIX_FMT_PAL8;
381 /* make sure the VMD header made it */
382 if (s->avctx->extradata_size != VMD_HEADER_SIZE) {
383 av_log(s->avctx, AV_LOG_ERROR, "VMD video: expected extradata size of %d\n",
387 vmd_header = (unsigned char *)avctx->extradata;
389 s->unpack_buffer_size = AV_RL32(&vmd_header[800]);
390 s->unpack_buffer = av_malloc(s->unpack_buffer_size);
391 if (!s->unpack_buffer)
394 /* load up the initial palette */
395 raw_palette = &vmd_header[28];
396 palette32 = (unsigned int *)s->palette;
397 for (i = 0; i < PALETTE_COUNT; i++) {
398 r = raw_palette[palette_index++] * 4;
399 g = raw_palette[palette_index++] * 4;
400 b = raw_palette[palette_index++] * 4;
401 palette32[i] = (r << 16) | (g << 8) | (b);
404 avcodec_get_frame_defaults(&s->frame);
405 avcodec_get_frame_defaults(&s->prev_frame);
410 static int vmdvideo_decode_frame(AVCodecContext *avctx,
411 void *data, int *data_size,
414 const uint8_t *buf = avpkt->data;
415 int buf_size = avpkt->size;
416 VmdVideoContext *s = avctx->priv_data;
424 s->frame.reference = 3;
425 if (avctx->get_buffer(avctx, &s->frame)) {
426 av_log(s->avctx, AV_LOG_ERROR, "VMD Video: get_buffer() failed\n");
432 /* make the palette available on the way out */
433 memcpy(s->frame.data[1], s->palette, PALETTE_COUNT * 4);
436 FFSWAP(AVFrame, s->frame, s->prev_frame);
437 if (s->frame.data[0])
438 avctx->release_buffer(avctx, &s->frame);
440 *data_size = sizeof(AVFrame);
441 *(AVFrame*)data = s->prev_frame;
443 /* report that the buffer was completely consumed */
447 static av_cold int vmdvideo_decode_end(AVCodecContext *avctx)
449 VmdVideoContext *s = avctx->priv_data;
451 if (s->prev_frame.data[0])
452 avctx->release_buffer(avctx, &s->prev_frame);
453 av_free(s->unpack_buffer);
463 #define BLOCK_TYPE_AUDIO 1
464 #define BLOCK_TYPE_INITIAL 2
465 #define BLOCK_TYPE_SILENCE 3
467 typedef struct VmdAudioContext {
472 static const uint16_t vmdaudio_table[128] = {
473 0x000, 0x008, 0x010, 0x020, 0x030, 0x040, 0x050, 0x060, 0x070, 0x080,
474 0x090, 0x0A0, 0x0B0, 0x0C0, 0x0D0, 0x0E0, 0x0F0, 0x100, 0x110, 0x120,
475 0x130, 0x140, 0x150, 0x160, 0x170, 0x180, 0x190, 0x1A0, 0x1B0, 0x1C0,
476 0x1D0, 0x1E0, 0x1F0, 0x200, 0x208, 0x210, 0x218, 0x220, 0x228, 0x230,
477 0x238, 0x240, 0x248, 0x250, 0x258, 0x260, 0x268, 0x270, 0x278, 0x280,
478 0x288, 0x290, 0x298, 0x2A0, 0x2A8, 0x2B0, 0x2B8, 0x2C0, 0x2C8, 0x2D0,
479 0x2D8, 0x2E0, 0x2E8, 0x2F0, 0x2F8, 0x300, 0x308, 0x310, 0x318, 0x320,
480 0x328, 0x330, 0x338, 0x340, 0x348, 0x350, 0x358, 0x360, 0x368, 0x370,
481 0x378, 0x380, 0x388, 0x390, 0x398, 0x3A0, 0x3A8, 0x3B0, 0x3B8, 0x3C0,
482 0x3C8, 0x3D0, 0x3D8, 0x3E0, 0x3E8, 0x3F0, 0x3F8, 0x400, 0x440, 0x480,
483 0x4C0, 0x500, 0x540, 0x580, 0x5C0, 0x600, 0x640, 0x680, 0x6C0, 0x700,
484 0x740, 0x780, 0x7C0, 0x800, 0x900, 0xA00, 0xB00, 0xC00, 0xD00, 0xE00,
485 0xF00, 0x1000, 0x1400, 0x1800, 0x1C00, 0x2000, 0x3000, 0x4000
488 static av_cold int vmdaudio_decode_init(AVCodecContext *avctx)
490 VmdAudioContext *s = avctx->priv_data;
492 if (avctx->channels < 1 || avctx->channels > 2) {
493 av_log(avctx, AV_LOG_ERROR, "invalid number of channels\n");
494 return AVERROR(EINVAL);
496 if (avctx->block_align < 1) {
497 av_log(avctx, AV_LOG_ERROR, "invalid block align\n");
498 return AVERROR(EINVAL);
501 if (avctx->bits_per_coded_sample == 16)
502 avctx->sample_fmt = AV_SAMPLE_FMT_S16;
504 avctx->sample_fmt = AV_SAMPLE_FMT_U8;
505 s->out_bps = av_get_bytes_per_sample(avctx->sample_fmt);
507 s->chunk_size = avctx->block_align + avctx->channels * (s->out_bps == 2);
509 av_log(avctx, AV_LOG_DEBUG, "%d channels, %d bits/sample, "
510 "block align = %d, sample rate = %d\n",
511 avctx->channels, avctx->bits_per_coded_sample, avctx->block_align,
517 static void decode_audio_s16(int16_t *out, const uint8_t *buf, int buf_size,
521 const uint8_t *buf_end = buf + buf_size;
523 int st = channels - 1;
525 /* decode initial raw sample */
526 for (ch = 0; ch < channels; ch++) {
527 predictor[ch] = (int16_t)AV_RL16(buf);
529 *out++ = predictor[ch];
532 /* decode DPCM samples */
534 while (buf < buf_end) {
537 predictor[ch] -= vmdaudio_table[b & 0x7F];
539 predictor[ch] += vmdaudio_table[b];
540 predictor[ch] = av_clip_int16(predictor[ch]);
541 *out++ = predictor[ch];
546 static int vmdaudio_decode_frame(AVCodecContext *avctx,
547 void *data, int *data_size,
550 const uint8_t *buf = avpkt->data;
551 const uint8_t *buf_end;
552 int buf_size = avpkt->size;
553 VmdAudioContext *s = avctx->priv_data;
554 int block_type, silent_chunks, audio_chunks;
555 int nb_samples, out_size;
556 uint8_t *output_samples_u8 = data;
557 int16_t *output_samples_s16 = data;
560 av_log(avctx, AV_LOG_WARNING, "skipping small junk packet\n");
566 if (block_type < BLOCK_TYPE_AUDIO || block_type > BLOCK_TYPE_SILENCE) {
567 av_log(avctx, AV_LOG_ERROR, "unknown block type: %d\n", block_type);
568 return AVERROR(EINVAL);
573 /* get number of silent chunks */
575 if (block_type == BLOCK_TYPE_INITIAL) {
578 av_log(avctx, AV_LOG_ERROR, "packet is too small\n");
579 return AVERROR(EINVAL);
581 flags = AV_RB32(buf);
582 silent_chunks = av_popcount(flags);
585 } else if (block_type == BLOCK_TYPE_SILENCE) {
587 buf_size = 0; // should already be zero but set it just to be sure
590 /* ensure output buffer is large enough */
591 audio_chunks = buf_size / s->chunk_size;
592 nb_samples = ((silent_chunks + audio_chunks) * avctx->block_align) / avctx->channels;
593 out_size = nb_samples * avctx->channels * s->out_bps;
594 if (*data_size < out_size)
597 /* decode silent chunks */
598 if (silent_chunks > 0) {
599 int silent_size = avctx->block_align * silent_chunks;
600 if (s->out_bps == 2) {
601 memset(output_samples_s16, 0x00, silent_size * 2);
602 output_samples_s16 += silent_size;
604 memset(output_samples_u8, 0x80, silent_size);
605 output_samples_u8 += silent_size;
609 /* decode audio chunks */
610 if (audio_chunks > 0) {
611 buf_end = buf + buf_size;
612 while (buf < buf_end) {
613 if (s->out_bps == 2) {
614 decode_audio_s16(output_samples_s16, buf, s->chunk_size,
616 output_samples_s16 += avctx->block_align;
618 memcpy(output_samples_u8, buf, s->chunk_size);
619 output_samples_u8 += avctx->block_align;
621 buf += s->chunk_size;
625 *data_size = out_size;
631 * Public Data Structures
634 AVCodec ff_vmdvideo_decoder = {
636 .type = AVMEDIA_TYPE_VIDEO,
637 .id = CODEC_ID_VMDVIDEO,
638 .priv_data_size = sizeof(VmdVideoContext),
639 .init = vmdvideo_decode_init,
640 .close = vmdvideo_decode_end,
641 .decode = vmdvideo_decode_frame,
642 .capabilities = CODEC_CAP_DR1,
643 .long_name = NULL_IF_CONFIG_SMALL("Sierra VMD video"),
646 AVCodec ff_vmdaudio_decoder = {
648 .type = AVMEDIA_TYPE_AUDIO,
649 .id = CODEC_ID_VMDAUDIO,
650 .priv_data_size = sizeof(VmdAudioContext),
651 .init = vmdaudio_decode_init,
652 .decode = vmdaudio_decode_frame,
653 .long_name = NULL_IF_CONFIG_SMALL("Sierra VMD audio"),