truemotion2.c 25.1 KB
Newer Older
1 2 3 4
/*
 * Duck/ON2 TrueMotion 2 Decoder
 * Copyright (c) 2005 Konstantin Shishkov
 *
5 6 7
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
8 9
 * modify it under the terms of the GNU Lesser General Public
 * 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 14 15 16 17
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * 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

22
/**
23
 * @file
24 25
 * Duck TrueMotion2 decoder.
 */
26

27
#include "avcodec.h"
28
#include "get_bits.h"
29 30 31
#include "dsputil.h"

#define TM2_ESCAPE 0x80000000
Mike Melanson's avatar
Mike Melanson committed
32
#define TM2_DELTAS 64
33 34 35 36 37 38 39 40 41 42 43 44 45
/* Huffman-coded streams of different types of blocks */
enum TM2_STREAMS{ TM2_C_HI = 0, TM2_C_LO, TM2_L_HI, TM2_L_LO,
     TM2_UPD, TM2_MOT, TM2_TYPE, TM2_NUM_STREAMS};
/* Block types */
enum TM2_BLOCKS{ TM2_HI_RES = 0, TM2_MED_RES, TM2_LOW_RES, TM2_NULL_RES,
                 TM2_UPDATE, TM2_STILL, TM2_MOTION};

typedef struct TM2Context{
    AVCodecContext *avctx;
    AVFrame pic;

    GetBitContext gb;
    DSPContext dsp;
46

47 48 49
    uint8_t *buffer;
    int buffer_size;

50 51 52 53 54 55 56 57 58 59
    /* TM2 streams */
    int *tokens[TM2_NUM_STREAMS];
    int tok_lens[TM2_NUM_STREAMS];
    int tok_ptrs[TM2_NUM_STREAMS];
    int deltas[TM2_NUM_STREAMS][TM2_DELTAS];
    /* for blocks decoding */
    int D[4];
    int CD[4];
    int *last;
    int *clast;
60

61 62 63 64 65 66 67 68 69
    /* data for current and previous frame */
    int *Y1, *U1, *V1, *Y2, *U2, *V2;
    int cur;
} TM2Context;

/**
* Huffman codes for each of streams
*/
typedef struct TM2Codes{
70
    VLC vlc; ///< table for FFmpeg bitstream reader
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123
    int bits;
    int *recode; ///< table for converting from code indexes to values
    int length;
} TM2Codes;

/**
* structure for gathering Huffman codes information
*/
typedef struct TM2Huff{
    int val_bits; ///< length of literal
    int max_bits; ///< maximum length of code
    int min_bits; ///< minimum length of code
    int nodes; ///< total number of nodes in tree
    int num; ///< current number filled
    int max_num; ///< total number of codes
    int *nums; ///< literals
    uint32_t *bits; ///< codes
    int *lens; ///< codelengths
} TM2Huff;

static int tm2_read_tree(TM2Context *ctx, uint32_t prefix, int length, TM2Huff *huff)
{
    if(length > huff->max_bits) {
        av_log(ctx->avctx, AV_LOG_ERROR, "Tree exceeded its given depth (%i)\n", huff->max_bits);
        return -1;
    }

    if(!get_bits1(&ctx->gb)) { /* literal */
        if (length == 0) {
            length = 1;
        }
        if(huff->num >= huff->max_num) {
            av_log(ctx->avctx, AV_LOG_DEBUG, "Too many literals\n");
            return -1;
        }
        huff->nums[huff->num] = get_bits_long(&ctx->gb, huff->val_bits);
        huff->bits[huff->num] = prefix;
        huff->lens[huff->num] = length;
        huff->num++;
        return 0;
    } else { /* non-terminal node */
        if(tm2_read_tree(ctx, prefix << 1, length + 1, huff) == -1)
            return -1;
        if(tm2_read_tree(ctx, (prefix << 1) | 1, length + 1, huff) == -1)
            return -1;
    }
    return 0;
}

static int tm2_build_huff_table(TM2Context *ctx, TM2Codes *code)
{
    TM2Huff huff;
    int res = 0;
124

125 126 127 128 129
    huff.val_bits = get_bits(&ctx->gb, 5);
    huff.max_bits = get_bits(&ctx->gb, 5);
    huff.min_bits = get_bits(&ctx->gb, 5);
    huff.nodes = get_bits_long(&ctx->gb, 17);
    huff.num = 0;
130

131 132 133 134 135 136 137
    /* check for correct codes parameters */
    if((huff.val_bits < 1) || (huff.val_bits > 32) ||
       (huff.max_bits < 0) || (huff.max_bits > 32)) {
        av_log(ctx->avctx, AV_LOG_ERROR, "Incorrect tree parameters - literal length: %i, max code length: %i\n",
               huff.val_bits, huff.max_bits);
        return -1;
    }
138
    if((huff.nodes <= 0) || (huff.nodes > 0x10000)) {
139 140 141 142 143 144
        av_log(ctx->avctx, AV_LOG_ERROR, "Incorrect number of Huffman tree nodes: %i\n", huff.nodes);
        return -1;
    }
    /* one-node tree */
    if(huff.max_bits == 0)
        huff.max_bits = 1;
145

146 147 148 149 150
    /* allocate space for codes - it is exactly ceil(nodes / 2) entries */
    huff.max_num = (huff.nodes + 1) >> 1;
    huff.nums = av_mallocz(huff.max_num * sizeof(int));
    huff.bits = av_mallocz(huff.max_num * sizeof(uint32_t));
    huff.lens = av_mallocz(huff.max_num * sizeof(int));
151

152 153
    if(tm2_read_tree(ctx, 0, 0, &huff) == -1)
        res = -1;
154

155 156 157 158 159
    if(huff.num != huff.max_num) {
        av_log(ctx->avctx, AV_LOG_ERROR, "Got less codes than expected: %i of %i\n",
               huff.num, huff.max_num);
        res = -1;
    }
160

161 162 163
    /* convert codes to vlc_table */
    if(res != -1) {
        int i;
164

165 166 167 168 169 170
        res = init_vlc(&code->vlc, huff.max_bits, huff.max_num,
                    huff.lens, sizeof(int), sizeof(int),
                    huff.bits, sizeof(uint32_t), sizeof(uint32_t), 0);
        if(res < 0) {
            av_log(ctx->avctx, AV_LOG_ERROR, "Cannot build VLC table\n");
            res = -1;
171
        } else
172 173 174 175 176 177 178 179 180 181 182 183 184
            res = 0;
        if(res != -1) {
            code->bits = huff.max_bits;
            code->length = huff.max_num;
            code->recode = av_malloc(code->length * sizeof(int));
            for(i = 0; i < code->length; i++)
                code->recode[i] = huff.nums[i];
        }
    }
    /* free allocated memory */
    av_free(huff.nums);
    av_free(huff.bits);
    av_free(huff.lens);
185

186 187 188 189 190
    return res;
}

static void tm2_free_codes(TM2Codes *code)
{
191
    av_free(code->recode);
192
    if(code->vlc.table)
193
        ff_free_vlc(&code->vlc);
194 195 196 197 198 199 200 201 202
}

static inline int tm2_get_token(GetBitContext *gb, TM2Codes *code)
{
    int val;
    val = get_vlc2(gb, code->vlc.table, code->bits, 1);
    return code->recode[val];
}

Michael Niedermayer's avatar
Michael Niedermayer committed
203
static inline int tm2_read_header(TM2Context *ctx, const uint8_t *buf)
204 205
{
    uint32_t magic;
Michael Niedermayer's avatar
Michael Niedermayer committed
206
    const uint8_t *obuf;
207

208
    obuf = buf;
209

210
    magic = AV_RL32(buf);
211
    buf += 4;
212

213 214 215 216 217 218 219 220 221
    if(magic == 0x00000100) { /* old header */
/*      av_log (ctx->avctx, AV_LOG_ERROR, "TM2 old header: not implemented (yet)\n"); */
        return 40;
    } else if(magic == 0x00000101) { /* new header */
        return 40;
    } else {
        av_log (ctx->avctx, AV_LOG_ERROR, "Not a TM2 header: 0x%08X\n", magic);
        return -1;
    }
222

223
    return buf - obuf;
224 225 226 227 228
}

static int tm2_read_deltas(TM2Context *ctx, int stream_id) {
    int d, mb;
    int i, v;
229

230 231
    d = get_bits(&ctx->gb, 9);
    mb = get_bits(&ctx->gb, 5);
232

233 234 235 236
    if((d < 1) || (d > TM2_DELTAS) || (mb < 1) || (mb > 32)) {
        av_log(ctx->avctx, AV_LOG_ERROR, "Incorrect delta table: %i deltas x %i bits\n", d, mb);
        return -1;
    }
237

238 239 240 241 242 243 244 245 246
    for(i = 0; i < d; i++) {
        v = get_bits_long(&ctx->gb, mb);
        if(v & (1 << (mb - 1)))
            ctx->deltas[stream_id][i] = v - (1 << mb);
        else
            ctx->deltas[stream_id][i] = v;
    }
    for(; i < TM2_DELTAS; i++)
        ctx->deltas[stream_id][i] = 0;
247

248 249 250
    return 0;
}

251 252
static int tm2_read_stream(TM2Context *ctx, const uint8_t *buf, int stream_id, int buf_size)
{
253 254 255 256 257
    int i;
    int cur = 0;
    int skip = 0;
    int len, toks;
    TM2Codes codes;
258

259 260 261 262 263
    if (buf_size < 4) {
        av_log(ctx->avctx, AV_LOG_ERROR, "not enough space for len left\n");
        return -1;
    }

264
    /* get stream length in dwords */
265
    len = AV_RB32(buf); buf += 4; cur += 4;
266
    skip = len * 4 + 4;
267

268 269
    if(len == 0)
        return 4;
270

271 272 273 274 275
    if (len >= INT_MAX/4-1 || len < 0 || len > buf_size) {
        av_log(ctx->avctx, AV_LOG_ERROR, "Error, invalid stream size.\n");
        return -1;
    }

276
    toks = AV_RB32(buf); buf += 4; cur += 4;
277
    if(toks & 1) {
278
        len = AV_RB32(buf); buf += 4; cur += 4;
279
        if(len == TM2_ESCAPE) {
280
            len = AV_RB32(buf); buf += 4; cur += 4;
281 282
        }
        if(len > 0) {
283 284
            if (skip <= cur)
                return -1;
Mike Melanson's avatar
Mike Melanson committed
285
            init_get_bits(&ctx->gb, buf, (skip - cur) * 8);
286 287 288 289 290 291 292
            if(tm2_read_deltas(ctx, stream_id) == -1)
                return -1;
            buf += ((get_bits_count(&ctx->gb) + 31) >> 5) << 2;
            cur += ((get_bits_count(&ctx->gb) + 31) >> 5) << 2;
        }
    }
    /* skip unused fields */
293
    if(AV_RB32(buf) == TM2_ESCAPE) {
294 295 296 297
        buf += 4; cur += 4; /* some unknown length - could be escaped too */
    }
    buf += 4; cur += 4;
    buf += 4; cur += 4; /* unused by decoder */
298

299
    if (skip <= cur)
300
        return -1;
Mike Melanson's avatar
Mike Melanson committed
301
    init_get_bits(&ctx->gb, buf, (skip - cur) * 8);
302 303 304 305
    if(tm2_build_huff_table(ctx, &codes) == -1)
        return -1;
    buf += ((get_bits_count(&ctx->gb) + 31) >> 5) << 2;
    cur += ((get_bits_count(&ctx->gb) + 31) >> 5) << 2;
306

307 308 309 310 311 312 313 314 315
    toks >>= 1;
    /* check if we have sane number of tokens */
    if((toks < 0) || (toks > 0xFFFFFF)){
        av_log(ctx->avctx, AV_LOG_ERROR, "Incorrect number of tokens: %i\n", toks);
        tm2_free_codes(&codes);
        return -1;
    }
    ctx->tokens[stream_id] = av_realloc(ctx->tokens[stream_id], toks * sizeof(int));
    ctx->tok_lens[stream_id] = toks;
316
    len = AV_RB32(buf); buf += 4; cur += 4;
317
    if(len > 0) {
318 319
        if (skip <= cur)
            return -1;
Mike Melanson's avatar
Mike Melanson committed
320
        init_get_bits(&ctx->gb, buf, (skip - cur) * 8);
321 322 323 324 325
        for(i = 0; i < toks; i++) {
            if (get_bits_left(&ctx->gb) <= 0) {
                av_log(ctx->avctx, AV_LOG_ERROR, "Incorrect number of tokens: %i\n", toks);
                return -1;
            }
326
            ctx->tokens[stream_id][i] = tm2_get_token(&ctx->gb, &codes);
327
        }
328
    } else {
Mike Melanson's avatar
Mike Melanson committed
329 330
        for(i = 0; i < toks; i++)
            ctx->tokens[stream_id][i] = codes.recode[0];
331 332
    }
    tm2_free_codes(&codes);
333

334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349
    return skip;
}

static inline int GET_TOK(TM2Context *ctx,int type) {
    if(ctx->tok_ptrs[type] >= ctx->tok_lens[type]) {
        av_log(ctx->avctx, AV_LOG_ERROR, "Read token from stream %i out of bounds (%i>=%i)\n", type, ctx->tok_ptrs[type], ctx->tok_lens[type]);
        return 0;
    }
    if(type <= TM2_MOT)
        return ctx->deltas[type][ctx->tokens[type][ctx->tok_ptrs[type]++]];
    return ctx->tokens[type][ctx->tok_ptrs[type]++];
}

/* blocks decoding routines */

/* common Y, U, V pointers initialisation */
350
#define TM2_INIT_POINTERS() \
351 352 353 354 355 356 357 358 359 360 361 362 363
    int *last, *clast; \
    int *Y, *U, *V;\
    int Ystride, Ustride, Vstride;\
\
    Ystride = ctx->avctx->width;\
    Vstride = (ctx->avctx->width + 1) >> 1;\
    Ustride = (ctx->avctx->width + 1) >> 1;\
    Y = (ctx->cur?ctx->Y2:ctx->Y1) + by * 4 * Ystride + bx * 4;\
    V = (ctx->cur?ctx->V2:ctx->V1) + by * 2 * Vstride + bx * 2;\
    U = (ctx->cur?ctx->U2:ctx->U1) + by * 2 * Ustride + bx * 2;\
    last = ctx->last + bx * 4;\
    clast = ctx->clast + bx * 4;

364
#define TM2_INIT_POINTERS_2() \
365 366 367 368 369 370 371 372 373 374 375 376 377
    int *Yo, *Uo, *Vo;\
    int oYstride, oUstride, oVstride;\
\
    TM2_INIT_POINTERS();\
    oYstride = Ystride;\
    oVstride = Vstride;\
    oUstride = Ustride;\
    Yo = (ctx->cur?ctx->Y1:ctx->Y2) + by * 4 * oYstride + bx * 4;\
    Vo = (ctx->cur?ctx->V1:ctx->V2) + by * 2 * oVstride + bx * 2;\
    Uo = (ctx->cur?ctx->U1:ctx->U2) + by * 2 * oUstride + bx * 2;

/* recalculate last and delta values for next blocks */
#define TM2_RECALC_BLOCK(CHR, stride, last, CD) {\
378
    CD[0] = CHR[1] - last[1];\
379
    CD[1] = (int)CHR[stride + 1] - (int)CHR[1];\
380 381
    last[0] = (int)CHR[stride + 0];\
    last[1] = (int)CHR[stride + 1];}
382 383 384 385 386 387

/* common operations - add deltas to 4x4 block of luma or 2x2 blocks of chroma */
static inline void tm2_apply_deltas(TM2Context *ctx, int* Y, int stride, int *deltas, int *last)
{
    int ct, d;
    int i, j;
388

389 390 391 392 393 394
    for(j = 0; j < 4; j++){
        ct = ctx->D[j];
        for(i = 0; i < 4; i++){
            d = deltas[i + j * 4];
            ct += d;
            last[i] += ct;
395
            Y[i] = av_clip_uint8(last[i]);
396 397 398 399 400 401 402 403 404 405 406 407 408
        }
        Y += stride;
        ctx->D[j] = ct;
    }
}

static inline void tm2_high_chroma(int *data, int stride, int *last, int *CD, int *deltas)
{
    int i, j;
    for(j = 0; j < 2; j++){
        for(i = 0; i < 2; i++){
            CD[j] += deltas[i + j * 2];
            last[i] += CD[j];
409
            data[i] = last[i];
410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425
        }
        data += stride;
    }
}

static inline void tm2_low_chroma(int *data, int stride, int *clast, int *CD, int *deltas, int bx)
{
    int t;
    int l;
    int prev;

    if(bx > 0)
        prev = clast[-3];
    else
        prev = 0;
    t = (CD[0] + CD[1]) >> 1;
426
    l = (prev - CD[0] - CD[1] + clast[1]) >> 1;
427 428 429
    CD[1] = CD[0] + CD[1] - t;
    CD[0] = t;
    clast[0] = l;
430

431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446
    tm2_high_chroma(data, stride, clast, CD, deltas);
}

static inline void tm2_hi_res_block(TM2Context *ctx, AVFrame *pic, int bx, int by)
{
    int i;
    int deltas[16];
    TM2_INIT_POINTERS();

    /* hi-res chroma */
    for(i = 0; i < 4; i++) {
        deltas[i] = GET_TOK(ctx, TM2_C_HI);
        deltas[i + 4] = GET_TOK(ctx, TM2_C_HI);
    }
    tm2_high_chroma(U, Ustride, clast, ctx->CD, deltas);
    tm2_high_chroma(V, Vstride, clast + 2, ctx->CD + 2, deltas + 4);
447

448 449 450
    /* hi-res luma */
    for(i = 0; i < 16; i++)
        deltas[i] = GET_TOK(ctx, TM2_L_HI);
451

452 453 454 455 456 457 458 459
    tm2_apply_deltas(ctx, Y, Ystride, deltas, last);
}

static inline void tm2_med_res_block(TM2Context *ctx, AVFrame *pic, int bx, int by)
{
    int i;
    int deltas[16];
    TM2_INIT_POINTERS();
460

461 462 463 464 465 466 467 468 469 470 471 472
    /* low-res chroma */
    deltas[0] = GET_TOK(ctx, TM2_C_LO);
    deltas[1] = deltas[2] = deltas[3] = 0;
    tm2_low_chroma(U, Ustride, clast, ctx->CD, deltas, bx);

    deltas[0] = GET_TOK(ctx, TM2_C_LO);
    deltas[1] = deltas[2] = deltas[3] = 0;
    tm2_low_chroma(V, Vstride, clast + 2, ctx->CD + 2, deltas, bx);

    /* hi-res luma */
    for(i = 0; i < 16; i++)
        deltas[i] = GET_TOK(ctx, TM2_L_HI);
473

474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495
    tm2_apply_deltas(ctx, Y, Ystride, deltas, last);
}

static inline void tm2_low_res_block(TM2Context *ctx, AVFrame *pic, int bx, int by)
{
    int i;
    int t1, t2;
    int deltas[16];
    TM2_INIT_POINTERS();

    /* low-res chroma */
    deltas[0] = GET_TOK(ctx, TM2_C_LO);
    deltas[1] = deltas[2] = deltas[3] = 0;
    tm2_low_chroma(U, Ustride, clast, ctx->CD, deltas, bx);

    deltas[0] = GET_TOK(ctx, TM2_C_LO);
    deltas[1] = deltas[2] = deltas[3] = 0;
    tm2_low_chroma(V, Vstride, clast + 2, ctx->CD + 2, deltas, bx);

    /* low-res luma */
    for(i = 0; i < 16; i++)
        deltas[i] = 0;
496

497 498 499 500
    deltas[ 0] = GET_TOK(ctx, TM2_L_LO);
    deltas[ 2] = GET_TOK(ctx, TM2_L_LO);
    deltas[ 8] = GET_TOK(ctx, TM2_L_LO);
    deltas[10] = GET_TOK(ctx, TM2_L_LO);
501

502 503 504 505 506 507 508 509 510 511 512 513
    if(bx > 0)
        last[0] = (last[-1] - ctx->D[0] - ctx->D[1] - ctx->D[2] - ctx->D[3] + last[1]) >> 1;
    else
        last[0] = (last[1]  - ctx->D[0] - ctx->D[1] - ctx->D[2] - ctx->D[3])>> 1;
    last[2] = (last[1] + last[3]) >> 1;

    t1 = ctx->D[0] + ctx->D[1];
    ctx->D[0] = t1 >> 1;
    ctx->D[1] = t1 - (t1 >> 1);
    t2 = ctx->D[2] + ctx->D[3];
    ctx->D[2] = t2 >> 1;
    ctx->D[3] = t2 - (t2 >> 1);
514

515 516 517 518 519 520 521 522 523 524
    tm2_apply_deltas(ctx, Y, Ystride, deltas, last);
}

static inline void tm2_null_res_block(TM2Context *ctx, AVFrame *pic, int bx, int by)
{
    int i;
    int ct;
    int left, right, diff;
    int deltas[16];
    TM2_INIT_POINTERS();
525

526 527 528 529 530 531
    /* null chroma */
    deltas[0] = deltas[1] = deltas[2] = deltas[3] = 0;
    tm2_low_chroma(U, Ustride, clast, ctx->CD, deltas, bx);

    deltas[0] = deltas[1] = deltas[2] = deltas[3] = 0;
    tm2_low_chroma(V, Vstride, clast + 2, ctx->CD + 2, deltas, bx);
532

533 534 535 536 537
    /* null luma */
    for(i = 0; i < 16; i++)
        deltas[i] = 0;

    ct = ctx->D[0] + ctx->D[1] + ctx->D[2] + ctx->D[3];
538

539 540 541 542
    if(bx > 0)
        left = last[-1] - ct;
    else
        left = 0;
543

544 545 546 547 548 549 550 551
    right = last[3];
    diff = right - left;
    last[0] = left + (diff >> 2);
    last[1] = left + (diff >> 1);
    last[2] = right - (diff >> 2);
    last[3] = right;
    {
        int tp = left;
552

553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603
        ctx->D[0] = (tp + (ct >> 2)) - left;
        left += ctx->D[0];
        ctx->D[1] = (tp + (ct >> 1)) - left;
        left += ctx->D[1];
        ctx->D[2] = ((tp + ct) - (ct >> 2)) - left;
        left += ctx->D[2];
        ctx->D[3] = (tp + ct) - left;
    }
    tm2_apply_deltas(ctx, Y, Ystride, deltas, last);
}

static inline void tm2_still_block(TM2Context *ctx, AVFrame *pic, int bx, int by)
{
    int i, j;
    TM2_INIT_POINTERS_2();

    /* update chroma */
    for(j = 0; j < 2; j++){
        for(i = 0; i < 2; i++){
            U[i] = Uo[i];
            V[i] = Vo[i];
        }
        U += Ustride; V += Vstride;
        Uo += oUstride; Vo += oVstride;
    }
    U -= Ustride * 2;
    V -= Vstride * 2;
    TM2_RECALC_BLOCK(U, Ustride, clast, ctx->CD);
    TM2_RECALC_BLOCK(V, Vstride, (clast + 2), (ctx->CD + 2));

    /* update deltas */
    ctx->D[0] = Yo[3] - last[3];
    ctx->D[1] = Yo[3 + oYstride] - Yo[3];
    ctx->D[2] = Yo[3 + oYstride * 2] - Yo[3 + oYstride];
    ctx->D[3] = Yo[3 + oYstride * 3] - Yo[3 + oYstride * 2];

    for(j = 0; j < 4; j++){
        for(i = 0; i < 4; i++){
            Y[i] = Yo[i];
            last[i] = Yo[i];
        }
        Y += Ystride;
        Yo += oYstride;
    }
}

static inline void tm2_update_block(TM2Context *ctx, AVFrame *pic, int bx, int by)
{
    int i, j;
    int d;
    TM2_INIT_POINTERS_2();
604

605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644
    /* update chroma */
    for(j = 0; j < 2; j++){
        for(i = 0; i < 2; i++){
            U[i] = Uo[i] + GET_TOK(ctx, TM2_UPD);
            V[i] = Vo[i] + GET_TOK(ctx, TM2_UPD);
        }
        U += Ustride; V += Vstride;
        Uo += oUstride; Vo += oVstride;
    }
    U -= Ustride * 2;
    V -= Vstride * 2;
    TM2_RECALC_BLOCK(U, Ustride, clast, ctx->CD);
    TM2_RECALC_BLOCK(V, Vstride, (clast + 2), (ctx->CD + 2));

    /* update deltas */
    ctx->D[0] = Yo[3] - last[3];
    ctx->D[1] = Yo[3 + oYstride] - Yo[3];
    ctx->D[2] = Yo[3 + oYstride * 2] - Yo[3 + oYstride];
    ctx->D[3] = Yo[3 + oYstride * 3] - Yo[3 + oYstride * 2];

    for(j = 0; j < 4; j++){
        d = last[3];
        for(i = 0; i < 4; i++){
            Y[i] = Yo[i] + GET_TOK(ctx, TM2_UPD);
            last[i] = Y[i];
        }
        ctx->D[j] = last[3] - d;
        Y += Ystride;
        Yo += oYstride;
    }
}

static inline void tm2_motion_block(TM2Context *ctx, AVFrame *pic, int bx, int by)
{
    int i, j;
    int mx, my;
    TM2_INIT_POINTERS_2();

    mx = GET_TOK(ctx, TM2_MOT);
    my = GET_TOK(ctx, TM2_MOT);
645

646 647 648 649 650
    if (4*bx+mx<0 || 4*by+my<0 || 4*bx+mx+4 > ctx->avctx->width || 4*by+my+4 > ctx->avctx->height) {
        av_log(0,0, "MV out of picture\n");
        return;
    }

651 652 653
    Yo += my * oYstride + mx;
    Uo += (my >> 1) * oUstride + (mx >> 1);
    Vo += (my >> 1) * oVstride + (mx >> 1);
654

655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692
    /* copy chroma */
    for(j = 0; j < 2; j++){
        for(i = 0; i < 2; i++){
            U[i] = Uo[i];
            V[i] = Vo[i];
        }
        U += Ustride; V += Vstride;
        Uo += oUstride; Vo += oVstride;
    }
    U -= Ustride * 2;
    V -= Vstride * 2;
    TM2_RECALC_BLOCK(U, Ustride, clast, ctx->CD);
    TM2_RECALC_BLOCK(V, Vstride, (clast + 2), (ctx->CD + 2));

    /* copy luma */
    for(j = 0; j < 4; j++){
        for(i = 0; i < 4; i++){
            Y[i] = Yo[i];
        }
        Y += Ystride;
        Yo += oYstride;
    }
    /* calculate deltas */
    Y -= Ystride * 4;
    ctx->D[0] = Y[3] - last[3];
    ctx->D[1] = Y[3 + Ystride] - Y[3];
    ctx->D[2] = Y[3 + Ystride * 2] - Y[3 + Ystride];
    ctx->D[3] = Y[3 + Ystride * 3] - Y[3 + Ystride * 2];
    for(i = 0; i < 4; i++)
        last[i] = Y[i + Ystride * 3];
}

static int tm2_decode_blocks(TM2Context *ctx, AVFrame *p)
{
    int i, j;
    int bw, bh;
    int type;
    int keyframe = 1;
693 694
    int *Y, *U, *V;
    uint8_t *dst;
695

696 697 698 699 700
    bw = ctx->avctx->width >> 2;
    bh = ctx->avctx->height >> 2;

    for(i = 0; i < TM2_NUM_STREAMS; i++)
        ctx->tok_ptrs[i] = 0;
701

702 703 704 705
    if (ctx->tok_lens[TM2_TYPE]<bw*bh){
        av_log(ctx->avctx,AV_LOG_ERROR,"Got %i tokens for %i blocks\n",ctx->tok_lens[TM2_TYPE],bw*bh);
        return -1;
    }
706

707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744
    memset(ctx->last, 0, 4 * bw * sizeof(int));
    memset(ctx->clast, 0, 4 * bw * sizeof(int));

    for(j = 0; j < bh; j++) {
        memset(ctx->D, 0, 4 * sizeof(int));
        memset(ctx->CD, 0, 4 * sizeof(int));
        for(i = 0; i < bw; i++) {
            type = GET_TOK(ctx, TM2_TYPE);
            switch(type) {
            case TM2_HI_RES:
                tm2_hi_res_block(ctx, p, i, j);
                break;
            case TM2_MED_RES:
                tm2_med_res_block(ctx, p, i, j);
                break;
            case TM2_LOW_RES:
                tm2_low_res_block(ctx, p, i, j);
                break;
            case TM2_NULL_RES:
                tm2_null_res_block(ctx, p, i, j);
                break;
            case TM2_UPDATE:
                tm2_update_block(ctx, p, i, j);
                keyframe = 0;
                break;
            case TM2_STILL:
                tm2_still_block(ctx, p, i, j);
                keyframe = 0;
                break;
            case TM2_MOTION:
                tm2_motion_block(ctx, p, i, j);
                keyframe = 0;
                break;
            default:
                av_log(ctx->avctx, AV_LOG_ERROR, "Skipping unknown block type %i\n", type);
            }
        }
    }
745 746

    /* copy data from our buffer to AVFrame */
747 748 749 750
    Y = (ctx->cur?ctx->Y2:ctx->Y1);
    U = (ctx->cur?ctx->U2:ctx->U1);
    V = (ctx->cur?ctx->V2:ctx->V1);
    dst = p->data[0];
751 752
    for(j = 0; j < ctx->avctx->height; j++){
        for(i = 0; i < ctx->avctx->width; i++){
753 754 755 756
            int y = Y[i], u = U[i >> 1], v = V[i >> 1];
            dst[3*i+0] = av_clip_uint8(y + v);
            dst[3*i+1] = av_clip_uint8(y);
            dst[3*i+2] = av_clip_uint8(y + u);
757
        }
758 759 760 761
        Y += ctx->avctx->width;
        if (j & 1) {
            U += ctx->avctx->width >> 1;
            V += ctx->avctx->width >> 1;
762
        }
763
        dst += p->linesize[0];
764
    }
765

766 767 768
    return keyframe;
}

769 770 771 772
static const int tm2_stream_order[TM2_NUM_STREAMS] = {
    TM2_C_HI, TM2_C_LO, TM2_L_HI, TM2_L_LO, TM2_UPD, TM2_MOT, TM2_TYPE
};

773
static int decode_frame(AVCodecContext *avctx,
774
                        void *data, int *data_size,
775
                        AVPacket *avpkt)
776
{
777 778
    const uint8_t *buf = avpkt->data;
    int buf_size = avpkt->size;
779
    TM2Context * const l = avctx->priv_data;
780
    AVFrame * const p = &l->pic;
781
    int i, skip, t;
782

783 784
    av_fast_padded_malloc(&l->buffer, &l->buffer_size, buf_size);
    if(!l->buffer){
785 786 787
        av_log(avctx, AV_LOG_ERROR, "Cannot allocate temporary buffer\n");
        return -1;
    }
788
    p->reference = 3;
Mike Melanson's avatar
Mike Melanson committed
789 790
    p->buffer_hints = FF_BUFFER_HINTS_VALID | FF_BUFFER_HINTS_PRESERVE | FF_BUFFER_HINTS_REUSABLE;
    if(avctx->reget_buffer(avctx, p) < 0){
791 792 793 794
        av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
        return -1;
    }

795 796
    l->dsp.bswap_buf((uint32_t*)l->buffer, (const uint32_t*)buf, buf_size >> 2);
    skip = tm2_read_header(l, l->buffer);
797

798
    if(skip == -1){
799
        return -1;
800
    }
801

802
    for(i = 0; i < TM2_NUM_STREAMS; i++){
803
        t = tm2_read_stream(l, l->buffer + skip, tm2_stream_order[i], buf_size - skip);
804 805 806 807 808
        if(t == -1){
            return -1;
        }
        skip += t;
    }
809 810
    p->key_frame = tm2_decode_blocks(l, p);
    if(p->key_frame)
811
        p->pict_type = AV_PICTURE_TYPE_I;
812
    else
813
        p->pict_type = AV_PICTURE_TYPE_P;
814

815 816 817
    l->cur = !l->cur;
    *data_size = sizeof(AVFrame);
    *(AVFrame*)data = l->pic;
818

819 820 821
    return buf_size;
}

822
static av_cold int decode_init(AVCodecContext *avctx){
823 824 825 826 827 828 829
    TM2Context * const l = avctx->priv_data;
    int i;

    if((avctx->width & 3) || (avctx->height & 3)){
        av_log(avctx, AV_LOG_ERROR, "Width and height must be multiple of 4\n");
        return -1;
    }
830

831 832
    l->avctx = avctx;
    l->pic.data[0]=NULL;
833
    avctx->pix_fmt = PIX_FMT_BGR24;
834
    avcodec_get_frame_defaults(&l->pic);
835

836
    ff_dsputil_init(&l->dsp, avctx);
837

838 839
    l->last = av_malloc(4 * sizeof(int) * (avctx->width >> 2));
    l->clast = av_malloc(4 * sizeof(int) * (avctx->width >> 2));
840

841 842 843 844
    for(i = 0; i < TM2_NUM_STREAMS; i++) {
        l->tokens[i] = NULL;
        l->tok_lens[i] = 0;
    }
845

846 847 848 849 850 851 852
    l->Y1 = av_malloc(sizeof(int) * avctx->width * avctx->height);
    l->U1 = av_malloc(sizeof(int) * ((avctx->width + 1) >> 1) * ((avctx->height + 1) >> 1));
    l->V1 = av_malloc(sizeof(int) * ((avctx->width + 1) >> 1) * ((avctx->height + 1) >> 1));
    l->Y2 = av_malloc(sizeof(int) * avctx->width * avctx->height);
    l->U2 = av_malloc(sizeof(int) * ((avctx->width + 1) >> 1) * ((avctx->height + 1) >> 1));
    l->V2 = av_malloc(sizeof(int) * ((avctx->width + 1) >> 1) * ((avctx->height + 1) >> 1));
    l->cur = 0;
853

854 855 856
    return 0;
}

857
static av_cold int decode_end(AVCodecContext *avctx){
858
    TM2Context * const l = avctx->priv_data;
859
    AVFrame *pic = &l->pic;
860 861
    int i;

862 863
    av_free(l->last);
    av_free(l->clast);
864
    for(i = 0; i < TM2_NUM_STREAMS; i++)
865
        av_free(l->tokens[i]);
866 867 868 869 870 871 872 873
    if(l->Y1){
        av_free(l->Y1);
        av_free(l->U1);
        av_free(l->V1);
        av_free(l->Y2);
        av_free(l->U2);
        av_free(l->V2);
    }
874 875
    av_freep(&l->buffer);
    l->buffer_size = 0;
876 877 878 879

    if (pic->data[0])
        avctx->release_buffer(avctx, pic);

880 881 882
    return 0;
}

883
AVCodec ff_truemotion2_decoder = {
884 885 886 887 888 889 890 891
    .name           = "truemotion2",
    .type           = AVMEDIA_TYPE_VIDEO,
    .id             = CODEC_ID_TRUEMOTION2,
    .priv_data_size = sizeof(TM2Context),
    .init           = decode_init,
    .close          = decode_end,
    .decode         = decode_frame,
    .capabilities   = CODEC_CAP_DR1,
892
    .long_name = NULL_IF_CONFIG_SMALL("Duck TrueMotion 2.0"),
893
};