Lagarith decoder by Nathan Caldwell, saintdev at gmail

Originally committed as revision 26270 to svn://svn.ffmpeg.org/ffmpeg/trunk

6 files changed, 696 insertions, 1 deletions

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index ee16b91510..90f07d6247 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -185,6 +185,7 @@ OBJS-$(CONFIG_JPEGLS_DECODER)          += jpeglsdec.o jpegls.o \
 OBJS-$(CONFIG_JPEGLS_ENCODER)          += jpeglsenc.o jpegls.o
 OBJS-$(CONFIG_KGV1_DECODER)            += kgv1dec.o
 OBJS-$(CONFIG_KMVC_DECODER)            += kmvc.o
+OBJS-$(CONFIG_LAGARITH_DECODER)        += lagarith.o lagarithrac.o
 OBJS-$(CONFIG_LJPEG_ENCODER)           += ljpegenc.o mjpegenc.o mjpeg.o \
                                           mpegvideo_enc.o motion_est.o  \
                                           ratecontrol.o mpeg12data.o    \
diff --git a/libavcodec/allcodecs.c b/libavcodec/allcodecs.c
index 5ba199861e..94da5c04c1 100644
--- a/libavcodec/allcodecs.c
+++ b/libavcodec/allcodecs.c
@@ -128,6 +128,7 @@ void avcodec_register_all(void)
     REGISTER_ENCDEC  (JPEGLS, jpegls);
     REGISTER_DECODER (KGV1, kgv1);
     REGISTER_DECODER (KMVC, kmvc);
+    REGISTER_DECODER (LAGARITH, lagarith);
     REGISTER_ENCODER (LJPEG, ljpeg);
     REGISTER_DECODER (LOCO, loco);
     REGISTER_DECODER (MDEC, mdec);
diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h
index 2355a6eb81..601f8ed6bd 100644
--- a/libavcodec/avcodec.h
+++ b/libavcodec/avcodec.h
@@ -32,7 +32,7 @@
 #include "libavutil/cpu.h"
 
 #define LIBAVCODEC_VERSION_MAJOR 52
-#define LIBAVCODEC_VERSION_MINOR 107
+#define LIBAVCODEC_VERSION_MINOR 108
 #define LIBAVCODEC_VERSION_MICRO  0
 
 #define LIBAVCODEC_VERSION_INT  AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \
@@ -259,6 +259,7 @@ enum CodecID {
     CODEC_ID_A64_MULTI5,
     CODEC_ID_R10K,
     CODEC_ID_MXPEG,
+    CODEC_ID_LAGARITH,
 
     /* various PCM "codecs" */
     CODEC_ID_PCM_S16LE= 0x10000,
diff --git a/libavcodec/lagarith.c b/libavcodec/lagarith.c
new file mode 100644
index 0000000000..95179df729
--- /dev/null
+++ b/libavcodec/lagarith.c
@@ -0,0 +1,518 @@
+/*
+ * Lagarith lossless decoder
+ * Copyright (c) 2009 Nathan Caldwell <saintdev (at) gmail.com>
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * 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
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/**
+ * @file libavcodec/lagarith.c
+ * Lagarith lossless decoder
+ * @author Nathan Caldwell
+ */
+
+#include "avcodec.h"
+#include "get_bits.h"
+#include "mathops.h"
+#include "dsputil.h"
+#include "lagarithrac.h"
+
+enum LagarithFrameType {
+    FRAME_RAW           = 1,    /*!< uncompressed */
+    FRAME_U_RGB24       = 2,    /*!< unaligned RGB24 */
+    FRAME_ARITH_YUY2    = 3,    /*!< arithmetic coded YUY2 */
+    FRAME_ARITH_RGB24   = 4,    /*!< arithmetic coded RGB24 */
+    FRAME_SOLID_GRAY    = 5,    /*!< solid grayscale color frame */
+    FRAME_SOLID_COLOR   = 6,    /*!< solid non-grayscale color frame */
+    FRAME_OLD_ARITH_RGB = 7,    /*!< obsolete arithmetic coded RGB (no longer encoded by upstream since version 1.1.0) */
+    FRAME_ARITH_RGBA    = 8,    /*!< arithmetic coded RGBA */
+    FRAME_SOLID_RGBA    = 9,    /*!< solid RGBA color frame */
+    FRAME_ARITH_YV12    = 10,   /*!< arithmetic coded YV12 */
+    FRAME_REDUCED_RES   = 11,   /*!< reduced resolution YV12 frame */
+};
+
+typedef struct LagarithContext {
+    AVCodecContext *avctx;
+    AVFrame picture;
+    DSPContext dsp;
+    int zeros;                  /*!< number of consecutive zero bytes encountered */
+    int zeros_rem;              /*!< number of zero bytes remaining to output */
+} LagarithContext;
+
+/**
+ * Compute the 52bit mantissa of 1/(double)denom.
+ * This crazy format uses floats in an entropy coder and we have to match x86
+ * rounding exactly, thus ordinary floats aren't portable enough.
+ * @param denom denominator
+ * @return 52bit mantissa
+ * @see softfloat_mul
+ */
+static uint64_t softfloat_reciprocal(uint32_t denom)
+{
+    int shift = av_log2(denom - 1) + 1;
+    uint64_t ret = (1ULL << 52) / denom;
+    uint64_t err = (1ULL << 52) - ret * denom;
+    ret <<= shift;
+    err <<= shift;
+    err +=  denom / 2;
+    return ret + err / denom;
+}
+
+/**
+ * (uint32_t)(x*f), where f has the given mantissa, and exponent 0
+ * Used in combination with softfloat_reciprocal computes x/(double)denom.
+ * @param x 32bit integer factor
+ * @param mantissa mantissa of f with exponent 0
+ * @return 32bit integer value (x*f)
+ * @see softfloat_reciprocal
+ */
+static uint32_t softfloat_mul(uint32_t x, uint64_t mantissa)
+{
+    uint64_t l = x * (mantissa & 0xffffffff);
+    uint64_t h = x * (mantissa >> 32);
+    h += l >> 32;
+    l &= 0xffffffff;
+    l += 1 << av_log2(h >> 21);
+    h += l >> 32;
+    return h >> 20;
+}
+
+static uint8_t lag_calc_zero_run(int8_t x)
+{
+    return (x << 1) ^ (x >> 7);
+}
+
+static int lag_decode_prob(GetBitContext *gb, uint32_t *value)
+{
+    static const uint8_t series[] = { 1, 2, 3, 5, 8, 13, 21 };
+    int i;
+    int bit     = 0;
+    int bits    = 0;
+    int prevbit = 0;
+    unsigned val;
+
+    for (i = 0; i < 7; i++) {
+        if (prevbit && bit)
+            break;
+        prevbit = bit;
+        bit = get_bits1(gb);
+        if (bit && !prevbit)
+            bits += series[i];
+    }
+    bits--;
+    if (bits < 0 || bits > 31) {
+        *value = 0;
+        return -1;
+    } else if (bits == 0) {
+        *value = 0;
+        return 0;
+    }
+
+    val  = get_bits_long(gb, bits);
+    val |= 1 << bits;
+
+    *value = val - 1;
+
+    return 0;
+}
+
+static int lag_read_prob_header(lag_rac *rac, GetBitContext *gb)
+{
+    int i, j, scale_factor;
+    unsigned prob, cumulative_target;
+    unsigned cumul_prob = 0;
+    unsigned scaled_cumul_prob = 0;
+
+    rac->prob[0] = 0;
+    rac->prob[257] = UINT_MAX;
+    /* Read probabilities from bitstream */
+    for (i = 1; i < 257; i++) {
+        if (lag_decode_prob(gb, &rac->prob[i]) < 0) {
+            av_log(rac->avctx, AV_LOG_ERROR, "Invalid probability encountered.\n");
+            return -1;
+        }
+        if ((uint64_t)cumul_prob + rac->prob[i] > UINT_MAX) {
+            av_log(rac->avctx, AV_LOG_ERROR, "Integer overflow encountered in cumulative probability calculation.\n");
+            return -1;
+        }
+        cumul_prob += rac->prob[i];
+        if (!rac->prob[i]) {
+            if (lag_decode_prob(gb, &prob)) {
+                av_log(rac->avctx, AV_LOG_ERROR, "Invalid probability run encountered.\n");
+                return -1;
+            }
+            if (prob > 257 - i)
+                prob = 257 - i;
+            for (j = 0; j < prob; j++)
+                rac->prob[++i] = 0;
+        }
+    }
+
+    if (!cumul_prob) {
+        av_log(rac->avctx, AV_LOG_ERROR, "All probabilities are 0!\n");
+        return -1;
+    }
+
+    /* Scale probabilities so cumulative probability is an even power of 2. */
+    scale_factor = av_log2(cumul_prob);
+
+    if (cumul_prob & (cumul_prob - 1)) {
+        uint64_t mul = softfloat_reciprocal(cumul_prob);
+        for (i = 1; i < 257; i++) {
+            rac->prob[i] = softfloat_mul(rac->prob[i], mul);
+            scaled_cumul_prob += rac->prob[i];
+        }
+
+        scale_factor++;
+        cumulative_target = 1 << scale_factor;
+
+        if (scaled_cumul_prob > cumulative_target) {
+            av_log(rac->avctx, AV_LOG_ERROR,
+                   "Scaled probabilities are larger than target!\n");
+            return -1;
+        }
+
+        scaled_cumul_prob = cumulative_target - scaled_cumul_prob;
+
+        for (i = 1; scaled_cumul_prob; i = (i & 0x7f) + 1) {
+            if (rac->prob[i]) {
+                rac->prob[i]++;
+                scaled_cumul_prob--;
+            }
+            /* Comment from reference source:
+             * if (b & 0x80 == 0) {     // order of operations is 'wrong'; it has been left this way
+             *                          // since the compression change is negligable and fixing it
+             *                          // breaks backwards compatibilty
+             *      b =- (signed int)b;
+             *      b &= 0xFF;
+             * } else {
+             *      b++;
+             *      b &= 0x7f;
+             * }
+             */
+        }
+    }
+
+    rac->scale = scale_factor;
+
+    /* Fill probability array with cumulative probability for each symbol. */
+    for (i = 1; i < 257; i++)
+        rac->prob[i] += rac->prob[i - 1];
+
+    return 0;
+}
+
+static void add_lag_median_prediction(uint8_t *dst, uint8_t *src1,
+                                      uint8_t *diff, int w, int *left,
+                                      int *left_top)
+{
+    /* This is almost identical to add_hfyu_median_prediction in dsputil.h.
+     * However the &0xFF on the gradient predictor yealds incorrect output
+     * for lagarith.
+     */
+    int i;
+    uint8_t l, lt;
+
+    l  = *left;
+    lt = *left_top;
+
+    for (i = 0; i < w; i++) {
+        l = mid_pred(l, src1[i], l + src1[i] - lt) + diff[i];
+        lt = src1[i];
+        dst[i] = l;
+    }
+
+    *left     = l;
+    *left_top = lt;
+}
+
+static void lag_pred_line(LagarithContext *l, uint8_t *buf,
+                          int width, int stride, int line)
+{
+    int L, TL;
+
+    if (!line) {
+        /* Left prediction only for first line */
+        L = l->dsp.add_hfyu_left_prediction(buf + 1, buf + 1,
+                                            width - 1, buf[0]);
+        return;
+    } else if (line == 1) {
+        /* Second line, left predict first pixel, the rest of the line is median predicted */
+        /* FIXME: In the case of RGB this pixel is top predicted */
+        TL = buf[-stride];
+    } else {
+        /* Top left is 2 rows back, last pixel */
+        TL = buf[width - (2 * stride) - 1];
+    }
+    /* Left pixel is actually prev_row[width] */
+    L = buf[width - stride - 1];
+
+    add_lag_median_prediction(buf, buf - stride, buf,
+                              width, &L, &TL);
+}
+
+static int lag_decode_line(LagarithContext *l, lag_rac *rac,
+                           uint8_t *dst, int width, int stride,
+                           int esc_count)
+{
+    int i = 0;
+    int ret = 0;
+
+    if (!esc_count)
+        esc_count = -1;
+
+    /* Output any zeros remaining from the previous run */
+handle_zeros:
+    if (l->zeros_rem) {
+        int count = FFMIN(l->zeros_rem, width - i);
+        memset(dst + i, 0, count);
+        i += count;
+        l->zeros_rem -= count;
+    }
+
+    while (i < width) {
+        dst[i] = lag_get_rac(rac);
+        ret++;
+
+        if (dst[i])
+            l->zeros = 0;
+        else
+            l->zeros++;
+
+        i++;
+        if (l->zeros == esc_count) {
+            int index = lag_get_rac(rac);
+            ret++;
+
+            l->zeros = 0;
+
+            l->zeros_rem = lag_calc_zero_run(index);
+            goto handle_zeros;
+        }
+    }
+    return ret;
+}
+
+static int lag_decode_zero_run_line(LagarithContext *l, uint8_t *dst,
+                                    const uint8_t *src, int width,
+                                    int esc_count)
+{
+    int i = 0;
+    int count;
+    uint8_t zero_run = 0;
+    const uint8_t *start = src;
+    uint8_t mask1 = -(esc_count < 2);
+    uint8_t mask2 = -(esc_count < 3);
+    uint8_t *end = dst + (width - 2);
+
+output_zeros:
+    if (l->zeros_rem) {
+        count = FFMIN(l->zeros_rem, width - i);
+        memset(dst, 0, count);
+        l->zeros_rem -= count;
+        dst += count;
+    }
+
+    while (dst < end) {
+        i = 0;
+        while (!zero_run && dst + i < end) {
+            i++;
+            zero_run =
+                !(src[i] | (src[i + 1] & mask1) | (src[i + 2] & mask2));
+        }
+        if (zero_run) {
+            zero_run = 0;
+            i += esc_count;
+            memcpy(dst, src, i);
+            dst += i;
+            l->zeros_rem = lag_calc_zero_run(src[i]);
+
+            src += i + 1;
+            goto output_zeros;
+        } else {
+            memcpy(dst, src, i);
+            src += i;
+        }
+    }
+    return start - src;
+}
+
+
+
+static int lag_decode_arith_plane(LagarithContext *l, uint8_t *dst,
+                                  int width, int height, int stride,
+                                  const uint8_t *src, int src_size)
+{
+    int i = 0;
+    int read = 0;
+    uint32_t length;
+    uint32_t offset = 1;
+    int esc_count = src[0];
+    GetBitContext gb;
+    lag_rac rac;
+
+    rac.avctx = l->avctx;
+    l->zeros = 0;
+
+    if (esc_count < 4) {
+        length = width * height;
+        if (esc_count && AV_RL32(src + 1) < length) {
+            length = AV_RL32(src + 1);
+            offset += 4;
+        }
+
+        init_get_bits(&gb, src + offset, src_size * 8);
+
+        if (lag_read_prob_header(&rac, &gb) < 0)
+            return -1;
+
+        lag_rac_init(&rac, &gb, length - stride);
+
+        for (i = 0; i < height; i++)
+            read += lag_decode_line(l, &rac, dst + (i * stride), width,
+                                    stride, esc_count);
+
+        if (read > length)
+            av_log(l->avctx, AV_LOG_WARNING,
+                   "Output more bytes than length (%d of %d)\n", read,
+                   length);
+    } else if (esc_count < 8) {
+        esc_count -= 4;
+        if (esc_count > 0) {
+            /* Zero run coding only, no range coding. */
+            for (i = 0; i < height; i++)
+                src += lag_decode_zero_run_line(l, dst + (i * stride), src,
+                                                width, esc_count);
+        } else {
+            /* Plane is stored uncompressed */
+            for (i = 0; i < height; i++) {
+                memcpy(dst + (i * stride), src, width);
+                src += width;
+            }
+        }
+    } else if (esc_count == 0xff) {
+        /* Plane is a solid run of 0 bytes */
+        for (i = 0; i < height; i++)
+            memset(dst + i * stride, 0, width);
+    } else {
+        av_log(l->avctx, AV_LOG_ERROR,
+               "Invalid zero run escape code! (%#x)\n", esc_count);
+        return -1;
+    }
+
+    for (i = 0; i < height; i++) {
+        lag_pred_line(l, dst, width, stride, i);
+        dst += stride;
+    }
+
+    return 0;
+}
+
+/**
+ * Decode a frame.
+ * @param avctx codec context
+ * @param data output AVFrame
+ * @param data_size size of output data or 0 if no picture is returned
+ * @param avpkt input packet
+ * @return number of consumed bytes on success or negative if decode fails
+ */
+static int lag_decode_frame(AVCodecContext *avctx,
+                            void *data, int *data_size, AVPacket *avpkt)
+{
+    const uint8_t *buf = avpkt->data;
+    int buf_size = avpkt->size;
+    LagarithContext *l = avctx->priv_data;
+    AVFrame *const p = &l->picture;
+    uint8_t frametype = 0;
+    uint32_t offset_gu = 0, offset_bv = 0, offset_ry = 9;
+
+    AVFrame *picture = data;
+
+    if (p->data[0])
+        avctx->release_buffer(avctx, p);
+
+    p->reference = 0;
+    p->key_frame = 1;
+
+    frametype = buf[0];
+
+    offset_gu = AV_RL32(buf + 1);
+    offset_bv = AV_RL32(buf + 5);
+
+    switch (frametype) {
+    case FRAME_ARITH_YV12:
+        avctx->pix_fmt = PIX_FMT_YUV420P;
+
+        if (avctx->get_buffer(avctx, p) < 0) {
+            av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
+            return -1;
+        }
+
+        lag_decode_arith_plane(l, p->data[0], avctx->width, avctx->height,
+                               p->linesize[0], buf + offset_ry,
+                               buf_size);
+        lag_decode_arith_plane(l, p->data[2], avctx->width / 2,
+                               avctx->height / 2, p->linesize[2],
+                               buf + offset_gu, buf_size);
+        lag_decode_arith_plane(l, p->data[1], avctx->width / 2,
+                               avctx->height / 2, p->linesize[1],
+                               buf + offset_bv, buf_size);
+        break;
+    default:
+        av_log(avctx, AV_LOG_ERROR,
+               "Unsupported Lagarith frame type: %#x\n", frametype);
+        return -1;
+    }
+
+    *picture = *p;
+    *data_size = sizeof(AVFrame);
+
+    return buf_size;
+}
+
+static av_cold int lag_decode_init(AVCodecContext *avctx)
+{
+    LagarithContext *l = avctx->priv_data;
+    l->avctx = avctx;
+
+    dsputil_init(&l->dsp, avctx);
+
+    return 0;
+}
+
+static av_cold int lag_decode_end(AVCodecContext *avctx)
+{
+    LagarithContext *l = avctx->priv_data;
+
+    if (l->picture.data[0])
+        avctx->release_buffer(avctx, &l->picture);
+
+    return 0;
+}
+
+AVCodec lagarith_decoder = {
+    "lagarith",
+    CODEC_TYPE_VIDEO,
+    CODEC_ID_LAGARITH,
+    sizeof(LagarithContext),
+    lag_decode_init,
+    NULL,
+    lag_decode_end,
+    lag_decode_frame,
+    CODEC_CAP_DR1,
+    .long_name = NULL_IF_CONFIG_SMALL("Lagarith lossless"),
+};
diff --git a/libavcodec/lagarithrac.c b/libavcodec/lagarithrac.c
new file mode 100644
index 0000000000..0cbc3b84df
--- /dev/null
+++ b/libavcodec/lagarithrac.c
@@ -0,0 +1,58 @@
+/*
+ * Lagarith range decoder
+ * Copyright (c) 2009 Nathan Caldwell <saintdev (at) gmail.com>
+ * Copyright (c) 2009 David Conrad
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * 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
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/**
+ * @file libavcodec/lagarithrac.c
+ * Lagarith range decoder
+ * @author Nathan Caldwell
+ * @author David Conrad
+ */
+
+#include "get_bits.h"
+#include "lagarithrac.h"
+
+void lag_rac_init(lag_rac *l, GetBitContext *gb, int length)
+{
+    int i, j;
+
+    /* According to reference decoder "1st byte is garbage",
+     * however, it gets skipped by the call to align_get_bits()
+     */
+    align_get_bits(gb);
+    l->bytestream_start =
+    l->bytestream       = gb->buffer + get_bits_count(gb) / 8;
+    l->bytestream_end   = l->bytestream_start + length;
+
+    l->range        = 0x80;
+    l->low          = *l->bytestream >> 1;
+    l->hash_shift   = FFMAX(l->scale - 8, 0);
+
+    for (i = j = 0; i < 256; i++) {
+        unsigned r = i << l->hash_shift;
+        while (l->prob[j + 1] <= r)
+            j++;
+        l->range_hash[i] = j;
+    }
+
+    /* Add conversion factor to hash_shift so we don't have to in lag_get_rac. */
+    l->hash_shift += 23;
+}
diff --git a/libavcodec/lagarithrac.h b/libavcodec/lagarithrac.h
new file mode 100644
index 0000000000..d985b60353
--- /dev/null
+++ b/libavcodec/lagarithrac.h
@@ -0,0 +1,116 @@
+/*
+ * Lagarith range decoder
+ * Copyright (c) 2009 Nathan Caldwell <saintdev (at) gmail.com>
+ * Copyright (c) 2009 David Conrad
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * 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
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/**
+ * @file libavcodec/lagarithrac.h
+ * Lagarith range decoder
+ * @author Nathan Caldwell
+ * @author David Conrad
+ */
+
+#ifndef AVCODEC_LAGARITHRAC_H
+#define AVCODEC_LAGARITHRAC_H
+
+#include <stdint.h>
+#include "libavutil/common.h"
+#include "libavutil/intreadwrite.h"
+#include "avcodec.h"
+#include "get_bits.h"
+
+typedef struct lag_rac {
+    AVCodecContext *avctx;
+    unsigned low;
+    unsigned range;
+    unsigned scale;             /*!< Number of bits of precision in range. */
+    unsigned hash_shift;        /*!< Number of bits to shift to calculate hash for radix search. */
+
+    const uint8_t *bytestream_start;  /*!< Start of input bytestream. */
+    const uint8_t *bytestream;        /*!< Current position in input bytestream. */
+    const uint8_t *bytestream_end;    /*!< End position of input bytestream. */
+
+    uint32_t prob[258];         /*!< Table of cumulative probability for each symbol. */
+    uint8_t  range_hash[256];   /*!< Hash table mapping upper byte to approximate symbol. */
+} lag_rac;
+
+void lag_rac_init(lag_rac *l, GetBitContext *gb, int length);
+
+/* TODO: Optimize */
+static inline void lag_rac_refill(lag_rac *l)
+{
+    while (l->range <= 0x800000) {
+        l->low   <<= 8;
+        l->range <<= 8;
+        l->low |= 0xff & (AV_RB16(l->bytestream) >> 1);
+        if (l->bytestream < l->bytestream_end)
+            l->bytestream++;
+    }
+}
+
+/**
+ * Decode a single byte from the compressed plane described by *l.
+ * @param l pointer to lag_rac for the current plane
+ * @return next byte of decoded data
+ */
+static inline uint8_t lag_get_rac(lag_rac *l)
+{
+    unsigned range_scaled, low_scaled, div;
+    int val;
+    uint8_t shift;
+
+    lag_rac_refill(l);
+
+    range_scaled = l->range >> l->scale;
+
+    if (l->low < range_scaled * l->prob[255]) {
+        /* val = 0 is frequent enough to deserve a shortcut */
+        if (l->low < range_scaled * l->prob[1]) {
+            val = 0;
+        } else {
+            /* FIXME __builtin_clz is ~20% faster here, but not allowed in generic code. */
+            shift = 30 - av_log2(range_scaled);
+            div = ((range_scaled << shift) + (1 << 23) - 1) >> 23;
+            /* low>>24 ensures that any cases too big for exact FASTDIV are
+             * under- rather than over-estimated
+             */
+            low_scaled = FASTDIV(l->low - (l->low >> 24), div);
+            shift -= l->hash_shift;
+            shift &= 31;
+            low_scaled = (low_scaled << shift) | (low_scaled >> (32 - shift));
+            /* low_scaled is now a lower bound of low/range_scaled */
+            val = l->range_hash[(uint8_t) low_scaled];
+            while (l->low >= range_scaled * l->prob[val + 1])
+                val++;
+        }
+
+        l->range = range_scaled * (l->prob[val + 1] - l->prob[val]);
+    } else {
+        val = 255;
+        l->range -= range_scaled * l->prob[255];
+    }
+
+    l->low -= range_scaled * l->prob[val];
+
+    return val;
+}
+
+
+#endif /* AVCODEC_LAGARITHRAC_H */