[FFmpeg-devel] [PATCH] CCITT Group 3 and 4 decompression

Kostya kostya.shishkov
Fri Dec 26 07:20:04 CET 2008


On Thu, Dec 25, 2008 at 07:33:35PM +0100, Michael Niedermayer wrote:
> On Thu, Dec 25, 2008 at 08:19:25AM +0200, Kostya wrote:
> > On Wed, Dec 24, 2008 at 11:00:08PM +0100, Michael Niedermayer wrote:
> > > On Wed, Dec 24, 2008 at 08:06:13AM +0200, Kostya wrote:
> > > > On Tue, Dec 23, 2008 at 08:27:31PM +0100, Michael Niedermayer wrote:
> > > > > On Tue, Dec 23, 2008 at 09:00:57AM +0200, Kostya wrote:
> > > > > > On Mon, Dec 22, 2008 at 09:06:37PM +0100, Michael Niedermayer wrote:
> > > > > > > On Mon, Dec 22, 2008 at 08:15:06PM +0200, Kostya wrote:
> > > > > > > > On Mon, Dec 22, 2008 at 01:13:53PM +0100, Michael Niedermayer wrote:
> > > > > > > > > On Mon, Dec 22, 2008 at 08:31:55AM +0200, Kostya wrote:
> > > > > > > > > > I need that for resolving issue 750 on roundup.
[...]
> > Also it does
> > not matter for decoding if there's EOL right one bit before line end -
> > there won't be sufficient data to decode that line anyway (for 2d-case
> > that single bit is read to determine line coding mode and for 1d case
> > first codeword should have length >= 4 bits).
> 
> its imporant to detect errors, for that it is important to find out if
> a sync marker is missing, this also includes the last sync marker.
> 
> i can offer to cleanup, commit and maintain CCITT fax support myself if
> you dont want to fix the issues?

Since you show interest in it, why not?
I've fixed some bugs in the code so CCITT_8.TIF from mphq decodes correctly
now. Feel free to do anything with it (including adding raw fax images
support).
 
> [...]
> > > btw, do you have some url where i can find files to test this code?
> > 
> > http://samples.mplayerhq.hu/image-samples/tiff/
> >  
> 
> thanks
> 
> 
> > > PS: merry christmas :)
> > 
> > Well, I live in a country where church calendar uses incorrect rounding
> > and was tied to the standard one only in XXth century with 13 days of difference
> > we have New Year first, then Christmas on Jan 7th and then Old New Year on
> > Jan 14th (i.e.  New Year in Old calendar). Pretty amusing, eh?
> 
> a few days off isnt that bad ...
> keep in mind that the whole calendar the world is using is a few years off,
> wiki says something between 2 and 7 years.

That happens when you set starting point to an arbitrary position.

[...]
> > @@ -103,6 +105,26 @@
> >              return -1;
> >          }
> >      }
> > +    if(s->compr == TIFF_CCITT_RLE || s->compr == TIFF_G3 || s->compr == TIFF_G4){
> > +        int i, ret = 0;
> > +        uint8_t *src2 = av_malloc(size);
> > +
> > +        for(i = 0; i < size; i++)
> > +            src2[i] = ff_reverse[src[i]];
> 
> out of mem check missing

added 
 
[checks on decoding run]
> also, what about runs of 0 ?
> if they are not allowed (except at pos 0) that should be checked for and an
> error be printed. Its easy to check for 0 before the loop and remove the 0
> code from the table ...
> Alternatively, if 0 is allowed at any place that should be supported, iam
> not sure if it is currently ...

Well, ITU T.4 does not say anything about them not occuring. That would be
ineffective but not illegal.
 
> [...]
> -- 
> Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
-------------- next part --------------
Index: libavcodec/Makefile
===================================================================
--- libavcodec/Makefile	(revision 15993)
+++ libavcodec/Makefile	(working copy)
@@ -203,7 +203,7 @@
 OBJS-$(CONFIG_THEORA_DECODER)          += vp3.o xiph.o vp3dsp.o
 OBJS-$(CONFIG_THP_DECODER)             += mjpegdec.o mjpeg.o
 OBJS-$(CONFIG_TIERTEXSEQVIDEO_DECODER) += tiertexseqv.o
-OBJS-$(CONFIG_TIFF_DECODER)            += tiff.o lzw.o
+OBJS-$(CONFIG_TIFF_DECODER)            += tiff.o lzw.o faxcompr.o
 OBJS-$(CONFIG_TIFF_ENCODER)            += tiffenc.o rle.o lzwenc.o
 OBJS-$(CONFIG_TRUEMOTION1_DECODER)     += truemotion1.o
 OBJS-$(CONFIG_TRUEMOTION2_DECODER)     += truemotion2.o
Index: libavcodec/tiff.c
===================================================================
--- libavcodec/tiff.c	(revision 16268)
+++ libavcodec/tiff.c	(working copy)
@@ -30,6 +30,7 @@
 #endif
 #include "lzw.h"
 #include "tiff.h"
+#include "faxcompr.h"
 
 
 typedef struct TiffContext {
@@ -41,6 +42,7 @@
     int le;
     int compr;
     int invert;
+    int fax_opts;
     int predictor;
 
     int strips, rps, sstype;
@@ -103,6 +105,30 @@
             return -1;
         }
     }
+    if(s->compr == TIFF_CCITT_RLE || s->compr == TIFF_G3 || s->compr == TIFF_G4){
+        int i, ret = 0;
+        uint8_t *src2 = av_malloc(size);
+
+        if(!src2){
+            av_log(s->avctx, AV_LOG_ERROR, "Error allocating temporary buffer\n");
+            return -1;
+        }
+        for(i = 0; i < size; i++)
+            src2[i] = ff_reverse[src[i]];
+        if(s->compr == TIFF_G3 && !(s->fax_opts & 1))
+            s->compr = TIFF_CCITT_RLE;
+        switch(s->compr){
+        case TIFF_CCITT_RLE:
+            ret = ff_ccitt_unpack_1d(s->avctx, src2, size, dst, lines, stride);
+            break;
+        case TIFF_G3:
+        case TIFF_G4:
+            ret = ff_ccitt_unpack_2d(s->avctx, src2, size, dst, lines, stride, s->compr == TIFF_G4);
+            break;
+        }
+        av_free(src2);
+        return ret;
+    }
     for(line = 0; line < lines; line++){
         if(src - ssrc > size){
             av_log(s->avctx, AV_LOG_ERROR, "Source data overread\n");
@@ -265,7 +291,12 @@
         case TIFF_RAW:
         case TIFF_PACKBITS:
         case TIFF_LZW:
+        case TIFF_CCITT_RLE:
             break;
+        case TIFF_G3:
+        case TIFF_G4:
+            s->fax_opts = 0;
+            break;
         case TIFF_DEFLATE:
         case TIFF_ADOBE_DEFLATE:
 #ifdef CONFIG_ZLIB
@@ -274,15 +305,6 @@
             av_log(s->avctx, AV_LOG_ERROR, "Deflate: ZLib not compiled in\n");
             return -1;
 #endif
-        case TIFF_G3:
-            av_log(s->avctx, AV_LOG_ERROR, "CCITT G3 compression is not supported\n");
-            return -1;
-        case TIFF_G4:
-            av_log(s->avctx, AV_LOG_ERROR, "CCITT G4 compression is not supported\n");
-            return -1;
-        case TIFF_CCITT_RLE:
-            av_log(s->avctx, AV_LOG_ERROR, "CCITT RLE compression is not supported\n");
-            return -1;
         case TIFF_JPEG:
         case TIFF_NEWJPEG:
             av_log(s->avctx, AV_LOG_ERROR, "JPEG compression is not supported\n");
@@ -373,6 +395,10 @@
             return -1;
         }
         break;
+    case TIFF_T4OPTIONS:
+    case TIFF_T6OPTIONS:
+        s->fax_opts = value;
+        break;
     }
     return 0;
 }
@@ -487,6 +513,7 @@
     avctx->coded_frame= (AVFrame*)&s->picture;
     s->picture.data[0] = NULL;
     ff_lzw_decode_open(&s->lzw);
+    ff_ccitt_unpack_init();
 
     return 0;
 }
-------------- next part --------------
/*
 * CCITT Fax Group 3 and 4 decompression
 * Copyright (c) 2008 Konstantin Shishkov
 *
 * 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
 */

/**
 * CCITT Fax Group 3 and 4 decompression
 * @file faxcompr.h
 * @author Konstantin Shishkov
 */
#ifndef AVCODEC_FAXCOMPR_H
#define AVCODEC_FAXCOMPR_H

#include "avcodec.h"

/**
 * initialize upacker code
 */
void ff_ccitt_unpack_init();

/**
 * unpack data compressed with CCITT Group 3 1-D method
 */
int ff_ccitt_unpack_1d(AVCodecContext *avctx,
                       const uint8_t *src, int srcsize,
                       uint8_t *dst, int height, int stride);

/**
 * unpack data compressed with CCITT Group 3 2-D or Group 4 method
 */
int ff_ccitt_unpack_2d(AVCodecContext *avctx,
                       const uint8_t *src, int srcsize,
                       uint8_t *dst, int height, int stride, int g4);

#endif /* AVCODEC_FAXCOMPR_H */
-------------- next part --------------
/*
 * CCITT Fax Group 3 and 4 decompression
 * Copyright (c) 2008 Konstantin Shishkov
 *
 * 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
 */

/**
 * CCITT Fax Group 3 and 4 decompression
 * @file faxcompr.c
 * @author Konstantin Shishkov
 */
#include "avcodec.h"
#include "bitstream.h"
#include "faxcompr.h"

#define CCITT_SYMS 104

static const uint16_t ccitt_syms[CCITT_SYMS] = {
    0,    1,    2,    3,    4,    5,    6,    7,    8,    9,   10,   11,   12,
   13,   14,   15,   16,   17,   18,   19,   20,   21,   22,   23,   24,   25,
   26,   27,   28,   29,   30,   31,   32,   33,   34,   35,   36,   37,   38,
   39,   40,   41,   42,   43,   44,   45,   46,   47,   48,   49,   50,   51,
   52,   53,   54,   55,   56,   57,   58,   59,   60,   61,   62,   63,   64,
  128,  192,  256,  320,  384,  448,  512,  576,  640,  704,  768,  832,  896,
  960, 1024, 1088, 1152, 1216, 1280, 1344, 1408, 1472, 1536, 1600, 1664, 1728,
 1792, 1856, 1920, 1984, 2048, 2112, 2176, 2240, 2304, 2368, 2432, 2496, 2560
};

static const uint8_t ccitt_codes_bits[2][CCITT_SYMS] =
{
  {
    0x35, 0x07, 0x07, 0x08, 0x0B, 0x0C, 0x0E, 0x0F, 0x13, 0x14, 0x07, 0x08, 0x08,
    0x03, 0x34, 0x35, 0x2A, 0x2B, 0x27, 0x0C, 0x08, 0x17, 0x03, 0x04, 0x28, 0x2B,
    0x13, 0x24, 0x18, 0x02, 0x03, 0x1A, 0x1B, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
    0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x04, 0x05, 0x0A, 0x0B, 0x52, 0x53, 0x54,
    0x55, 0x24, 0x25, 0x58, 0x59, 0x5A, 0x5B, 0x4A, 0x4B, 0x32, 0x33, 0x34, 0x1B,
    0x12, 0x17, 0x37, 0x36, 0x37, 0x64, 0x65, 0x68, 0x67, 0xCC, 0xCD, 0xD2, 0xD3,
    0xD4, 0xD5, 0xD6, 0xD7, 0xD8, 0xD9, 0xDA, 0xDB, 0x98, 0x99, 0x9A, 0x18, 0x9B,
    0x08, 0x0C, 0x0D, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x1C, 0x1D, 0x1E, 0x1F
  },
  {
    0x37, 0x02, 0x03, 0x02, 0x03, 0x03, 0x02, 0x03, 0x05, 0x04, 0x04, 0x05, 0x07,
    0x04, 0x07, 0x18, 0x17, 0x18, 0x08, 0x67, 0x68, 0x6C, 0x37, 0x28, 0x17, 0x18,
    0xCA, 0xCB, 0xCC, 0xCD, 0x68, 0x69, 0x6A, 0x6B, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6,
    0xD7, 0x6C, 0x6D, 0xDA, 0xDB, 0x54, 0x55, 0x56, 0x57, 0x64, 0x65, 0x52, 0x53,
    0x24, 0x37, 0x38, 0x27, 0x28, 0x58, 0x59, 0x2B, 0x2C, 0x5A, 0x66, 0x67, 0x0F,
    0xC8, 0xC9, 0x5B, 0x33, 0x34, 0x35, 0x6C, 0x6D, 0x4A, 0x4B, 0x4C, 0x4D, 0x72,
    0x73, 0x74, 0x75, 0x76, 0x77, 0x52, 0x53, 0x54, 0x55, 0x5A, 0x5B, 0x64, 0x65,
    0x08, 0x0C, 0x0D, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x1C, 0x1D, 0x1E, 0x1F
  }
};

static const uint8_t ccitt_codes_lens[2][CCITT_SYMS] =
{
  {
     8,  6,  4,  4,  4,  4,  4,  4,  5,  5,  5,  5,  6,  6,  6,  6,  6,  6,  7,  7,
     7,  7,  7,  7,  7,  7,  7,  7,  7,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,
     8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,  8,
     8,  8,  8,  8,  5,  5,  6,  7,  8,  8,  8,  8,  8,  8,  9,  9,  9,  9,  9,  9,
     9,  9,  9,  9,  9,  9,  9,  9,  9,  6,  9, 11, 11, 11, 12, 12, 12, 12, 12, 12,
    12, 12, 12, 12
  },
  {
    10,  3,  2,  2,  3,  4,  4,  5,  6,  6,  7,  7,  7,  8,  8,  9, 10, 10, 10, 11,
    11, 11, 11, 11, 11, 11, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12,
    12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12,
    12, 12, 12, 12, 10, 12, 12, 12, 12, 12, 12, 13, 13, 13, 13, 13, 13, 13, 13, 13,
    13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 11, 11, 11, 12, 12, 12, 12, 12, 12,
    12, 12, 12, 12
  }
};

static const uint8_t ccitt_group3_2d_bits[11] = {
    1, 1, 2, 2, 2, 1, 3, 3, 3, 1, 1
};

static const uint8_t ccitt_group3_2d_lens[11] = {
    4, 3, 7, 6, 3, 1, 3, 6, 7, 7, 9
};

static VLC ccitt_vlc[2], ccitt_group3_2d_vlc;

av_cold void ff_ccitt_unpack_init()
{
    static VLC_TYPE code_table1[528][2];
    static VLC_TYPE code_table2[648][2];
    int i;
    static int initialized = 0;

    if(initialized)
        return;
    ccitt_vlc[0].table = code_table1;
    ccitt_vlc[0].table_allocated = 528;
    ccitt_vlc[1].table = code_table2;
    ccitt_vlc[1].table_allocated = 648;
    for(i = 0; i < 2; i++){
        init_vlc_sparse(&ccitt_vlc[i], 9, CCITT_SYMS,
                        ccitt_codes_lens[i], 1, 1,
                        ccitt_codes_bits[i], 1, 1, 
                        ccitt_syms, 2, 2,
                        INIT_VLC_USE_NEW_STATIC);
    }
    INIT_VLC_STATIC(&ccitt_group3_2d_vlc, 9, 11,
                    ccitt_group3_2d_lens, 1, 1,
                    ccitt_group3_2d_bits, 1, 1, 512);
    initialized = 1;
}


static int decode_group3_1d_line(AVCodecContext *avctx, GetBitContext *gb,
                                 int pix_left, int *runs)
{
    int mode = 0, run = 0;
    unsigned int t;
    for(;;){
        t = get_vlc2(gb, ccitt_vlc[mode].table, 9, 2);
        run += t;
        if(t < 64){
            pix_left -= run;
            *runs++ = run;
            if(pix_left <= 0){
                if(!pix_left)
                    break;
                runs[-1] = 0;
                av_log(avctx, AV_LOG_ERROR, "Run went out of bounds\n");
                return -1;
            }
            run = 0;
            mode = !mode;
        }else if((int)t == -1){
            av_log(avctx, AV_LOG_ERROR, "Incorrect code\n");
            return -1;
        }
    }
    *runs++ = 0;
    return 0;
}

static int decode_group3_2d_line(AVCodecContext *avctx, GetBitContext *gb,
                                 int pix_left, int *runs, const int *ref)
{
    int mode = 0, offs = 0, run = 0, saved_run = 0, t;
    int run_off = *ref++;
    int *run_start = runs;

    while(pix_left > 0){
        int cmode = get_vlc2(gb, ccitt_group3_2d_vlc.table, 9, 1);
        if(cmode == -1){
            av_log(avctx, AV_LOG_ERROR, "Incorrect mode VLC\n");
            return -1;
        }
        //sync line pointers
        if(runs != run_start)while(run_off <= offs && run_off < avctx->width){
            run_off += *ref++;
            run_off += *ref++;
        }
        if(!cmode){//pass mode
            run_off += *ref++;
            run = run_off - offs;
            run_off += *ref++;
            pix_left -= run;
            if(pix_left < 0){
                av_log(avctx, AV_LOG_ERROR, "Run went out of bounds\n");
                return -1;
            }
            offs += run;
            saved_run += run;
        }else if(cmode == 1){//horizontal mode
            int k;
            for(k = 0; k < 2; k++){
                run = 0;
                for(;;){
                    t = get_vlc2(gb, ccitt_vlc[mode].table, 9, 2);
                    if(t == -1){
                        av_log(avctx, AV_LOG_ERROR, "Incorrect code\n");
                        return -1;
                    }
                    run += t;
                    if(t < 64)
                        break;
                }
                *runs++ = run + saved_run;
                saved_run = 0;
                if(pix_left < run){
                    av_log(avctx, AV_LOG_ERROR, "Run went out of bounds\n");
                    return -1;
                }
                pix_left -= run;
                offs += run;
                mode = !mode;
            }
        }else if(cmode == 9 || cmode == 10){
            av_log(avctx, AV_LOG_ERROR, "Special modes are not supported (yet)\n");
            return -1;
        }else{//vertical mode
            run = run_off - offs + (cmode - 5);
            if(cmode >= 5)
                run_off += *ref++;
            else
                run_off -= *--ref;
            pix_left -= run;
            if(pix_left < 0){
                av_log(avctx, AV_LOG_ERROR, "Run went out of bounds\n");
                return -1;
            }
            offs += run;
            *runs++ = run + saved_run;
            saved_run = 0;
            mode = !mode;
        }
    }
    *runs++ = saved_run;
    *runs++ = 0;
    return 0;
}

static void put_line(uint8_t *dst, int size, int width, const int *runs)
{
    PutBitContext pb;
    int run, mode = 0, pix_left = width, run_idx = 0;

    init_put_bits(&pb, dst, size*8);
    while(pix_left > 0){
        run = runs[run_idx++];
        if(!run){
            mode = !mode;
            continue;
        }
        pix_left -= run;
        for(; run > 16; run -= 16)
            put_sbits(&pb, 16, -mode);
        put_sbits(&pb, run, -mode);
        mode = !mode;
    }
}

static int find_group3_syncmarker(GetBitContext *gb, int srcsize)
{
    int state = get_bits(gb, 12);
    int rem = srcsize - get_bits_count(gb);
    while((state & 0xFFF) != 1){
        state = (state << 1) | get_bits1(gb);
        if(--rem <= 0)
            return -1;
    }
    return 0;
}

int ff_ccitt_unpack_1d(AVCodecContext *avctx,
                       const uint8_t *src, int srcsize,
                       uint8_t *dst, int height, int stride)
{
    int j;
    GetBitContext gb;
    int *runs, *last;

    runs = av_malloc(avctx->width * sizeof(runs[0]));
    last = av_malloc(avctx->width * sizeof(last[0]));
    init_get_bits(&gb, src, srcsize*8);
    for(j = 0; j < height; j++){
        if(find_group3_syncmarker(&gb, srcsize*8) < 0)
            break;
        if(decode_group3_1d_line(avctx, &gb, avctx->width, runs) < 0){
            put_line(dst, stride, avctx->width, last);
        }else{
            put_line(dst, stride, avctx->width, runs);
            FFSWAP(int*, last, runs);
        }
        dst += stride;
    }
    av_free(runs);
    av_free(last);
    return 0;
}

int ff_ccitt_unpack_2d(AVCodecContext *avctx,
                       const uint8_t *src, int srcsize,
                       uint8_t *dst, int height, int stride, int g4)
{
    int j;
    GetBitContext gb;
    int *runs, *ref;
    int ret;

    runs = av_malloc((avctx->width + 2) * sizeof(runs[0]));
    ref  = av_malloc((avctx->width + 2) * sizeof(ref[0]));
    ref[0] = avctx->width;
    ref[1] = 0;
    ref[2] = 0;
    init_get_bits(&gb, src, srcsize*8);
    for(j = 0; j < height; j++){
        if(g4){
            ret = decode_group3_2d_line(avctx, &gb, avctx->width, runs, ref);
            if(ret < 0){
                av_free(runs);
                av_free(ref);
                return -1;
            }
        }else{
            if(find_group3_syncmarker(&gb, srcsize*8) < 0)
                break;
            if(get_bits1(&gb))
                ret = decode_group3_1d_line(avctx, &gb, avctx->width, runs);
            else
                ret = decode_group3_2d_line(avctx, &gb, avctx->width, runs, ref);
        }
        if(ret < 0){
            put_line(dst, stride, avctx->width, ref);
        }else{
            put_line(dst, stride, avctx->width, runs);
            FFSWAP(int*, runs, ref);
        }
        dst += stride;
    }
    av_free(runs);
    av_free(ref);
    return 0;
}



More information about the ffmpeg-devel mailing list