mjpeg.c

mpeg4 video codec mpeg4 video codec

#define MAX_COMPONENTS 4

typedef struct MJpegDecodeContext {
    AVCodecContext *avctx;
    GetBitContext gb;
    int mpeg_enc_ctx_allocated; /* true if decoding context allocated */

    int start_code; /* current start code */
    int buffer_size;
    uint8_t *buffer;

    int16_t quant_matrixes[4][64];
    VLC vlcs[2][4];
    int qscale[4];      ///< quantizer scale calculated from quant_matrixes

    int org_height;  /* size given at codec init */
    int first_picture;    /* true if decoding first picture */
    int interlaced;     /* true if interlaced */
    int bottom_field;   /* true if bottom field */
    int lossless;
    int ls;
    int rgb;
    int rct;            /* standard rct */  
    int pegasus_rct;    /* pegasus reversible colorspace transform */  
    int bits;           /* bits per component */

    int maxval;
    int near;         ///< near lossless bound (si 0 for lossless)
    int t1,t2,t3;
    int reset;        ///< context halfing intervall ?rename

    int width, height;
    int mb_width, mb_height;
    int nb_components;
    int component_id[MAX_COMPONENTS];
    int h_count[MAX_COMPONENTS]; /* horizontal and vertical count for each component */
    int v_count[MAX_COMPONENTS];
    int comp_index[MAX_COMPONENTS];
    int dc_index[MAX_COMPONENTS];
    int ac_index[MAX_COMPONENTS];
    int nb_blocks[MAX_COMPONENTS];
    int h_scount[MAX_COMPONENTS];
    int v_scount[MAX_COMPONENTS];
    int h_max, v_max; /* maximum h and v counts */
    int quant_index[4];   /* quant table index for each component */
    int last_dc[MAX_COMPONENTS]; /* last DEQUANTIZED dc (XXX: am I right to do that ?) */
    AVFrame picture; /* picture structure */
    int linesize[MAX_COMPONENTS];                   ///< linesize << interlaced
    int8_t *qscale_table;
    __align8(DCTELEM,block[64]);
    ScanTable scantable;
    void (*idct_put)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);

    int restart_interval;
    int restart_count;

    int buggy_avid;
    int cs_itu601;
    int interlace_polarity;

    int mjpb_skiptosod;   
    uint16_t RGBbuffer[32768][4];
} MJpegDecodeContext;

static int mjpeg_decode_dht(MJpegDecodeContext *s);

static int build_vlc(VLC *vlc, const uint8_t *bits_table, const uint8_t *val_table, 
                      int nb_codes, int use_static)
{
    uint8_t huff_size[256];
    uint16_t huff_code[256];

    memset(huff_size, 0, sizeof(huff_size));
    build_huffman_codes(huff_size, huff_code, bits_table, val_table);
    
    return init_vlc(vlc, 9, nb_codes, huff_size, 1, 1, huff_code, 2, 2, use_static);
}

static int mjpeg_decode_init(AVCodecContext *avctx)
{
    MJpegDecodeContext *s = avctx->priv_data;
    MpegEncContext s2;
    memset(s, 0, sizeof(MJpegDecodeContext));

    s->avctx = avctx;

    /* ugly way to get the idct & scantable FIXME */
    memset(&s2, 0, sizeof(MpegEncContext));
    s2.avctx= avctx;
//    s2->out_format = FMT_MJPEG;
    dsputil_init(&s2.dsp, avctx);
    DCT_common_init(&s2);

    s->scantable= s2.intra_scantable;
    s->idct_put= s2.dsp.idct_put;

    s->mpeg_enc_ctx_allocated = 0;
    s->buffer_size = 0;
    s->buffer = NULL;
    s->start_code = -1;
    s->first_picture = 1;
    s->org_height = avctx->coded_height;
    
    build_vlc(&s->vlcs[0][0], bits_dc_luminance, val_dc_luminance, 12, 0);
    build_vlc(&s->vlcs[0][1], bits_dc_chrominance, val_dc_chrominance, 12, 0);
    build_vlc(&s->vlcs[1][0], bits_ac_luminance, val_ac_luminance, 251, 0);
    build_vlc(&s->vlcs[1][1], bits_ac_chrominance, val_ac_chrominance, 251, 0);
    
    if (avctx->flags & CODEC_FLAG_EXTERN_HUFF)
    {
	av_log(avctx, AV_LOG_INFO, "mjpeg: using external huffman table\n");
	init_get_bits(&s->gb, avctx->extradata, avctx->extradata_size*8);
	mjpeg_decode_dht(s);
	/* should check for error - but dunno */
    }

    return 0;
}


/**
 * finds the end of the current frame in the bitstream.
 * @return the position of the first byte of the next frame, or -1
 */
static int find_frame_end(ParseContext *pc, const uint8_t *buf, int buf_size){
    int vop_found, i;
    uint16_t state;
    
    vop_found= pc->frame_start_found;
    state= pc->state;
    
    i=0;
    if(!vop_found){
        for(i=0; i<buf_size; i++){
            state= (state<<8) | buf[i];
            if(state == 0xFFD8){
                i++;
                vop_found=1;
                break;
            }
        }
    }

    if(vop_found){
        /* EOF considered as end of frame */
        if (buf_size == 0)
            return 0;
        for(; i<buf_size; i++){
            state= (state<<8) | buf[i];
            if(state == 0xFFD8){
                pc->frame_start_found=0;
                pc->state=0; 
                return i-1;
            }
        }
    }
    pc->frame_start_found= vop_found;
    pc->state= state;
    return END_NOT_FOUND;
}

static int jpeg_parse(AVCodecParserContext *s,
                           AVCodecContext *avctx,
                           uint8_t **poutbuf, int *poutbuf_size, 
                           const uint8_t *buf, int buf_size)
{
    ParseContext *pc = s->priv_data;
    int next;
    
    next= find_frame_end(pc, buf, buf_size);

    if (ff_combine_frame(pc, next, (uint8_t **)&buf, &buf_size) < 0) {
        *poutbuf = NULL;
        *poutbuf_size = 0;
        return buf_size;
    }

    *poutbuf = (uint8_t *)buf;
    *poutbuf_size = buf_size;
    return next;
}

/* quantize tables */
static int mjpeg_decode_dqt(MJpegDecodeContext *s)
{
    int len, index, i, j;
    
    len = get_bits(&s->gb, 16) - 2;

    while (len >= 65) {
        /* only 8 bit precision handled */
        if (get_bits(&s->gb, 4) != 0)
	{
	    dprintf("dqt: 16bit precision\n");
            return -1;
	}
        index = get_bits(&s->gb, 4);
        if (index >= 4)
            return -1;
        dprintf("index=%d\n", index);
        /* read quant table */
        for(i=0;i<64;i++) {
            j = s->scantable.permutated[i];
	    s->quant_matrixes[index][j] = get_bits(&s->gb, 8);
        }

        //XXX FIXME finetune, and perhaps add dc too
        s->qscale[index]= FFMAX(
            s->quant_matrixes[index][s->scantable.permutated[1]],
            s->quant_matrixes[index][s->scantable.permutated[8]]) >> 1;
	dprintf("qscale[%d]: %d\n", index, s->qscale[index]);
        len -= 65;
    }
    
    return 0;
}

/* decode huffman tables and build VLC decoders */
static int mjpeg_decode_dht(MJpegDecodeContext *s)
{
    int len, index, i, class, n, v, code_max;
    uint8_t bits_table[17];
    uint8_t val_table[256];
    
    len = get_bits(&s->gb, 16) - 2;

    while (len > 0) {
        if (len < 17)
            return -1;
        class = get_bits(&s->gb, 4);
        if (class >= 2)
            return -1;
        index = get_bits(&s->gb, 4);
        if (index >= 4)
            return -1;
        n = 0;
        for(i=1;i<=16;i++) {
            bits_table[i] = get_bits(&s->gb, 8);
            n += bits_table[i];
        }
        len -= 17;
        if (len < n || n > 256)
            return -1;

        code_max = 0;
        for(i=0;i<n;i++) {
            v = get_bits(&s->gb, 8);
            if (v > code_max)
                code_max = v;
            val_table[i] = v;
        }
        len -= n;

        /* build VLC and flush previous vlc if present */
        free_vlc(&s->vlcs[class][index]);
        dprintf("class=%d index=%d nb_codes=%d\n",
               class, index, code_max + 1);
        if(build_vlc(&s->vlcs[class][index], bits_table, val_table, code_max + 1, 0) < 0){
            return -1;
        }
    }
    return 0;
}

static int mjpeg_decode_sof(MJpegDecodeContext *s)
{
    int len, nb_components, i, width, height;

    /* XXX: verify len field validity */
    len = get_bits(&s->gb, 16);
    s->bits= get_bits(&s->gb, 8);
    
    if(s->pegasus_rct) s->bits=9;  
    if(s->bits==9 && !s->pegasus_rct) s->rct=1;    //FIXME ugly

    if (s->bits != 8 && !s->lossless){
        av_log(s->avctx, AV_LOG_ERROR, "only 8 bits/component accepted\n");
        return -1;
    }
    height = get_bits(&s->gb, 16);
    width = get_bits(&s->gb, 16);
    
    dprintf("sof0: picture: %dx%d\n", width, height);
    if(avcodec_check_dimensions(s->avctx, width, height))
        return -1;

    nb_components = get_bits(&s->gb, 8);
    if (nb_components <= 0 ||
        nb_components > MAX_COMPONENTS)
        return -1;
    s->nb_components = nb_components;
    s->h_max = 1;
    s->v_max = 1;
    for(i=0;i<nb_components;i++) {
        /* component id */
        s->component_id[i] = get_bits(&s->gb, 8) - 1;
        s->h_count[i] = get_bits(&s->gb, 4);
        s->v_count[i] = get_bits(&s->gb, 4);
        /* compute hmax and vmax (only used in interleaved case) */
        if (s->h_count[i] > s->h_max)
            s->h_max = s->h_count[i];
        if (s->v_count[i] > s->v_max)
            s->v_max = s->v_count[i];
        s->quant_index[i] = get_bits(&s->gb, 8);
        if (s->quant_index[i] >= 4)
            return -1;
        dprintf("component %d %d:%d id: %d quant:%d\n", i, s->h_count[i],
	    s->v_count[i], s->component_id[i], s->quant_index[i]);
    }

    if(s->v_max==1 && s->h_max==1 && s->lossless==1) s->rgb=1;

    /* if different size, realloc/alloc picture */
    /* XXX: also check h_count and v_count */
    if (width != s->width || height != s->height) {
        av_freep(&s->qscale_table);
            
        s->width = width;
        s->height = height;

        /* test interlaced mode */
        if (s->first_picture &&
            s->org_height != 0 &&
            s->height < ((s->org_height * 3) / 4)) {
            s->interlaced = 1;
//	    s->bottom_field = (s->interlace_polarity) ? 1 : 0;
	    s->bottom_field = 0;
            height *= 2;
        }

        avcodec_set_dimensions(s->avctx, width, height);

        s->qscale_table= av_mallocz((s->width+15)/16);

        s->first_picture = 0;
    }
    
    if(s->interlaced && s->bottom_field)
        return 0;
 
    /* XXX: not complete test ! */
    switch((s->h_count[0] << 4) | s->v_count[0]) {
    case 0x11:
        if(s->rgb){
            s->avctx->pix_fmt = PIX_FMT_RGBA32;
        }else if(s->nb_components==3)
            s->avctx->pix_fmt = s->cs_itu601 ? PIX_FMT_YUV444P : PIX_FMT_YUVJ444P;
        else
            s->avctx->pix_fmt = PIX_FMT_GRAY8;
        break;
    case 0x21:
        s->avctx->pix_fmt = s->cs_itu601 ? PIX_FMT_YUV422P : PIX_FMT_YUVJ422P;
        break;
    default:
    case 0x22:
        s->avctx->pix_fmt = s->cs_itu601 ? PIX_FMT_YUV420P : PIX_FMT_YUVJ420P;
        break;
    }

    if(s->picture.data[0])
        s->avctx->release_buffer(s->avctx, &s->picture);

    s->picture.reference= 0;
    if(s->avctx->get_buffer(s->avctx, &s->picture) < 0){
        av_log(s->avctx, AV_LOG_ERROR, "get_buffer() failed\n");
		return -1;
	    }
    s->picture.pict_type= I_TYPE;
    s->picture.key_frame= 1;
    
    for(i=0; i<3; i++){
        s->linesize[i]= s->picture.linesize[i] << s->interlaced;
        }

//    printf("%d %d %d %d %d %d\n", s->width, s->height, s->linesize[0], s->linesize[1], s->interlaced, s->avctx->height);

    if (len != (8+(3*nb_components)))
    {
	dprintf("decode_sof0: error, len(%d) mismatch\n", len);
    }
    
    return 0;
}

static inline int mjpeg_decode_dc(MJpegDecodeContext *s, int dc_index)
{
    int code;
    code = get_vlc2(&s->gb, s->vlcs[0][dc_index].table, 9, 2);
    if (code < 0)
    {
	dprintf("mjpeg_decode_dc: bad vlc: %d:%d (%p)\n", 0, dc_index,
                &s->vlcs[0][dc_index]);
        return 0xffff;
    }

    if(code)
        return get_xbits(&s->gb, code);
    else
        return 0;
}

/* decode block and dequantize */
static int decode_block(MJpegDecodeContext *s, DCTELEM *block, 
                        int component, int dc_index, int ac_index, int quant_index)
{
    int code, i, j, level, val;
    VLC *ac_vlc;
    int16_t *quant_matrix;

    /* DC coef */
    val = mjpeg_decode_dc(s, dc_index);
    if (val == 0xffff) {
        dprintf("error dc\n");