#!/bin/sh
#C--------------------------------------------------------------------
#C- DjVuLibre-3.5
#C- Copyright (c) 2002  Leon Bottou and Yann Le Cun.
#C- Copyright (c) 2001  AT&T
#C-
#C- This software is subject to, and may be distributed under, the
#C- GNU General Public License, either Version 2 of the license,
#C- or (at your option) any later version. The license should have
#C- accompanied the software or you may obtain a copy of the license
#C- from the Free Software Foundation at http://www.fsf.org .
#C-
#C- This program is distributed in the hope that it will be useful,
#C- but WITHOUT ANY WARRANTY; without even the implied warranty of
#C- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
#C- GNU General Public License for more details.
#C-
#C- DjVuLibre-3.5 is derived from the DjVu(r) Reference Library from
#C- Lizardtech Software.  Lizardtech Software has authorized us to
#C- replace the original DjVu(r) Reference Library notice by the following
#C- text (see doc/lizard2002.djvu and doc/lizardtech2007.djvu):
#C-
#C-  ------------------------------------------------------------------
#C- | DjVu (r) Reference Library (v. 3.5)
#C- | Copyright (c) 1999-2001 LizardTech, Inc. All Rights Reserved.
#C- | The DjVu Reference Library is protected by U.S. Pat. No.
#C- | 6,058,214 and patents pending.
#C- |
#C- | This software is subject to, and may be distributed under, the
#C- | GNU General Public License, either Version 2 of the license,
#C- | or (at your option) any later version. The license should have
#C- | accompanied the software or you may obtain a copy of the license
#C- | from the Free Software Foundation at http://www.fsf.org .
#C- |
#C- | The computer code originally released by LizardTech under this
#C- | license and unmodified by other parties is deemed "the LIZARDTECH
#C- | ORIGINAL CODE."  Subject to any third party intellectual property
#C- | claims, LizardTech grants recipient a worldwide, royalty-free, 
#C- | non-exclusive license to make, use, sell, or otherwise dispose of 
#C- | the LIZARDTECH ORIGINAL CODE or of programs derived from the 
#C- | LIZARDTECH ORIGINAL CODE in compliance with the terms of the GNU 
#C- | General Public License.   This grant only confers the right to 
#C- | infringe patent claims underlying the LIZARDTECH ORIGINAL CODE to 
#C- | the extent such infringement is reasonably necessary to enable 
#C- | recipient to make, have made, practice, sell, or otherwise dispose 
#C- | of the LIZARDTECH ORIGINAL CODE (or portions thereof) and not to 
#C- | any greater extent that may be necessary to utilize further 
#C- | modifications or combinations.
#C- |
#C- | The LIZARDTECH ORIGINAL CODE is provided "AS IS" WITHOUT WARRANTY
#C- | OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED
#C- | TO ANY WARRANTY OF NON-INFRINGEMENT, OR ANY IMPLIED WARRANTY OF
#C- | MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
#C- +------------------------------------------------------------------
#C--------------------------------------------------------------------


# Step 1a -- temporary directory

tempdir=`mktemp -d --tmpdir djXXXXXX 2>/dev/null` || tempdir=''
while [ ! -d "$tempdir" -o ! -r "$tempdir" ] ; do
    tempdir="/tmp/dj"`awk 'BEGIN{srand();printf("%d",rand()*100000)}'`
    mkdir "$tempdir" || tempdir=''
done
trap 'rm -rf 2>/dev/null "$tempdir"' EXIT INT HUP QUIT


# Step 1b -- utilities

usage()
{
    cat 1>&2 <<\END
Usage:  djvudigital [options] inputfile [outputfile]
More information is displayed by typing
    djvudigital --help
END
    exit 10
}

getarg()
{
        echo "$1" | sed -e 's/^[^=]*=//'
}

getargs()
{
        echo "$1" | sed -e 's/^[^=]*=//' -e 's/,/ /g'
}

pathexpand()
{
    tmpvar="$PATH"
    tmpdir=
    while [ -n "$tmpvar" ]
    do
      IFS=':' read tmpdir tmpvar <<EOF
$tmpvar
EOF
      test -x "$tmpdir/$1" && echo "$tmpdir/$1"
    done
}

psdvifix=
checkpsdvifix()
{
    if [ -z "$psdvifix" ] 
    then
        psdvifix=psdvifix.ps
        if ( "$gsdjvu" 2>&1 -dNODISPLAY -f psdvifix.ps -c quit | \
              grep -q /undefinedfilename )
        then
            psdvifix=no
        elif ( "$gsdjvu" 2>&1 -dNODISPLAY -f psdvifix.ps -c quit | \
              grep -q WRITESYSTEMDICT )
        then
            psdvifix=no
        fi
    fi
    test "$psdvifix" != no
}

djvutext=
checkps2utf8()
{
    if [ -z "$djvutext" ] 
    then
        djvutext=no
        if ( "$gsdjvu" 2>&1 -dNODISPLAY -c '(ps2utf8.ps) runlibfile quit' | \
              grep -q WRITESYSTEMDICT )
        then
            djvutext="$tempdir/djvutext.ps"
            cat > $djvutext <<\EOF
(ps2utf8.ps) runlibfile currentglobal /setglobal load true setglobal 
.ps2utf8 begin /onpage { } bind def /onfont { pop pop pop } bind def
/onmark { pop pop pop pop currentx currenty currentpoint 
.djvutextmark } bind def end exec
EOF
        fi
    fi
    test "$djvutext" != no
}




# Step 2a -- locate gsdjvu executable

gsdjvu=
while read gs
do
  if [ -z "$gsdjvu" ] && ( "$gs" -h 2>&1 | grep -q djvusep )
  then
     gsdjvu="$gs"
  fi
done <<EOF
${GSDJVU}
`pathexpand gs`
`pathexpand gsdjvu`
EOF

if [ -z "$gsdjvu" ]
then
    cat 1>&2 <<\EOF
djvudigital: cannot locate suitable ghostscript executable.
+--------------------------------------------------------------------+
| DjVuDigital relies on a special ghostscript device driver, but     |
| could not find a ghostscript executable that implement this driver.|
| Please visit http://djvu.sourceforge.net/gsdjvu.html.              |
+--------------------------------------------------------------------+
EOF
    exit 10
fi

# Step 2b -- locate csepdjvu executable

csepdjvu=
while read cs
do
  if [ -z "$csepdjvu" ] && ( "$cs" -h 2>&1 | grep -q quality )
  then
      csepdjvu="$cs"
  fi
done <<EOF
${CSEPDJVU}
`pathexpand csepdjvu`
`pathexpand msepdjvu`
EOF

if [ -z "$csepdjvu" ]
then
    cat 1>&2 <<\EOF
djvudigital: cannot locate csepdjvu executable.
+--------------------------------------------------------------------+
| DjVuDigital was not able to locate the djvulibre tool "csepdjvu".   |
| Please make sure that the djvulibre tools are properly installed.  |
+--------------------------------------------------------------------+
EOF
    exit 10
fi

# Step 2c -- locate djvused executable

djvused=
while read ds
do
    if [ -z "$djvused" ] && ( "$ds" 2>&1 | grep -q -i djvulibre )
    then
        djvused="$ds"
    fi
done <<EOF
${DJVUSED}
`pathexpand djvused`
EOF

# Step 2d -- locate pdftotext executable

pdftotext=
while read pt
do
  if [ -z "$pdftotext" ] && \
     ( "$pt" -bbox foo.pdf 2>&1 | grep -q foo.pdf )
     ( "$pt" -v 2>&1 | grep -q Poppler )
  then
      pdftotext="$pt"
  fi
done <<EOF
${PDFTOTEXT}
`pathexpand pdftotext`
EOF


# Step 3 -- process arguments

gsarg0="-sDEVICE=djvusep -dNOPAUSE -dBATCH -dSAFER"
gsarg1=
gsarg2=
csepargs=
dpi="300"
gsprinted="-dPrinted"
gsepsf="-dEPSCrop"
gsverbosity=
csepverbosity='-v'
dsedverbosity=
infile=
outfile=
run=
sepfile=
popplertext=0
popplermeta=0

for n
do
  case $n in 
      --*) 
          ;;
      -*)  
          n="-$n" 
          ;;
  esac
  case $n in
      --help)
          man djvudigital
          exit 10
          ;;
      --check)
          echo 1>&2 "Using: $gsdjvu"
          echo 1>&2 "  and: $csepdjvu"
          test -n "$djvused" && echo 1>&2 "  and: $djvused"
          test -n "$pdftotext" && echo 1>&2 "  and: $pdftotext"
          exit 0
          ;;
      --dpi=[0-9]*)
          dpi="`getarg $n`"
          ;;
      --verbose|--v)
          gsverbosity=''
          csepverbosity='-vv'
          dsedverbosity='-v'
          ;;
      --dryrun)
          run="echo +"
          ;;
      --sepfile)
          sepfile=yes
          ;;
      --quiet|--q)
          gsverbosity='-q'
          csepverbosity=''
          ;;
      --psrotate=0)
          gsarg2="-c << /Orientation 0 >> setpagedevice"
          ;;
      --psrotate=90)
          gsarg2="-c << /Orientation 3 >> setpagedevice"
          ;;
      --psrotate=180)
          gsarg2="-c << /Orientation 2 >> setpagedevice"
          ;;
      --psrotate=270)
          gsarg2="-c << /Orientation 1 >> setpagedevice"
          ;;
      --epsf=no)
          gsepsf=
          ;;
      --epsf=ignore)
          gsepsf="-dNOEPS"
          ;;
      --epsf=fit)
          gsepsf="-dEPSFitPage"
          ;;
      --epsf=crop)
          gsepsf="-dEPSCrop"
          ;;
      --words)
          gsarg0="$gsarg0 -dProvideUnicode -dExtractText"
          if checkpsdvifix; then
            gsarg1="-f $psdvifix"
          elif checkps2utf8; then
            # Modern gs do not like DELAYBIND
            gsarg0="$gsarg0 -dDELAYBIND -dWRITESYSTEMDICT"
            gsarg1="-f $djvutext"
          fi
          ;;
      --lines)
          gsarg0="$gsarg0 -dProvideUnicode -dExtractText"
          if checkpsdvifix; then
            gsarg1="-f $psdvifix"
          elif checkps2utf8; then
            # Modern gs do not like DELAYBIND
            gsarg0="$gsarg0 -dDELAYBIND -dWRITESYSTEMDICT"
            gsarg1="-f $djvutext"
          fi
          csepargs="$csepargs -t"
          ;;
      --poppler=*)
          for arg in `getargs $n` ; do
            case "$arg" in 
              text) popplertext=1 ;;
              meta) popplermeta=1 ;;
                 *) echo 1>&2 "djvudigital: unrecognized option --poppler=$arg"
                    usage ;; 
            esac
          done
          ;;
      --pdf=screen)
          gsprinted="-dPrinted=false"
          ;;
      --pdf=printed)
          gsprinted="-dPrinted=true"
          ;;
      --exact-color)
          gsarg0="-dUseCIEColor $gsarg0"
          ;;
      --threshold=*)
          gsarg0=" -dThreshold=`getarg $n` $gsarg0"
          ;;
      --bg-subsample=*)
          gsarg0=" -dBgSubsample=`getarg $n` $gsarg0"
          ;;
      --bg-slices=*)
          csepargs=" -q `getarg $n` $csepargs"
          ;;
      --fg-colors=*)
          gsarg0=" -dFgColors=`getarg $n` $gsarg0"
          ;;
      --fg-image-colors=*)
          gsarg0=" -dFgImgColors=`getarg $n` $gsarg0"
          ;;
      --gsarg=*)
          gsarg0=" `getargs $n` $gsarg0"
          ;;
      --cseparg=*)
          csepargs="$csepargs `getargs $n`"
          ;;
      -*)
          echo 1>&2 "djvudigital: unrecognized option $n"
          usage
          ;;
      *)
          if [ -z "$infile" ] ; then
              infile=$n
          elif [ -z "$outfile" ] ; then
              outfile=$n
          else
              usage
          fi
          ;;
  esac
done


# Step 4 -- check input filename

if [ -z "$infile" ] ; then
    usage
elif [ ! -r "$infile" -a -z "$run" ]; then
    echo "djvudigital: cannot open $infile for reading" 1>&2
    exit 10
fi

if [ -z "$outfile" ]
then
    outfile="$infile"
    for ext in gz GZ ps PS eps EPS pdf PDF
    do
       case "$outfile" in
         *.$ext)
            outfile=`basename "$outfile" .$ext`
            ;;
       esac
    done
    if [ "$sepfile" = "yes" ] 
    then
      outfile="$outfile.sep"
    else
      outfile="$outfile.djvu"
    fi
fi

# Step 5 -- execute command

if [ "$csepverbosity" != "" -o "$gsverbosity" != "-q" ] 
then
    test -z "$run" && echo "DJVUDIGITAL --- DjVuLibre-3.5"
fi

if [ "$sepfile" = "yes" ] 
then
  backend="$outfile"
else
  backend="| "'"'"$csepdjvu"'"'" -d $dpi"
  backend="$backend $csepverbosity $csepargs - "'"'"$outfile"'"'
fi


case "$infile" in
    *.gz|*.GZ)
        if test -z "$run" ; then
          gzip -d -c "$infile" | \
          "$gsdjvu" "-r$dpi" $gsverbosity $gsprinted $gsepsf \
            "-sOutputFile=$backend" $gsarg0 $gsarg1 $gsarg2 -_ -c quit
        else
          $run gzip -d -c '"'"$infile"'"' '|' "" \
             "$gsdjvu" "-r$dpi" $gsverbosity $gsprinted $gsepsf "" \
             "-sOutputFile=""'""$backend""'" "" \
             $gsarg0 $gsarg1 $gsarg2 -_ -c quit
        fi
        ;;
    *)
        if test -z "$run" ; then
          "$gsdjvu" "-r$dpi" $gsverbosity $gsprinted $gsepsf \
            "-sOutputFile=$backend" $gsarg0 $gsarg1 $gsarg2 \
            -f "$infile" -c quit
        else
          $run "$gsdjvu" "-r$dpi" $gsverbosity $gsprinted $gsepsf "" \
             "-sOutputFile=""'""$backend""'" "" \
             $gsarg0 $gsarg1 $gsarg2 -f '"'"$infile"'"' -c quit
        fi
        ;;
esac


# Step 6 -- Postprocess djvu file with metadata/text found by poppler

if ( file "$infile" | grep -q -i pdf ) \
    && test "$popplertext$popplermeta" != "00"
then

    # check for djvused
    if [ -z "$djvused" ]
    then
        cat 1>&2 <<\EOF
djvudigital: cannot locate djvused executable.
+--------------------------------------------------------------------+
| DjVuDigital was not able to locate the djvulibre tool "djvused".   |
| This tool is needed to use the --poppler=(text,data) options.      |
| Please make sure that the djvulibre tools are properly installed.  |
+--------------------------------------------------------------------+
EOF
        exit 10
    fi

    # check for pdftotext
    if [ -z "$pdftotext" ] 
    then
        cat 1>&2 <<\EOF
djvudigital: cannot locate pdftotext executable.
+--------------------------------------------------------------------+
| DjVuDigital was not able to locate the poppler tool "pdftotext".   |
| This tool is needed to use the --poppler=(text,data) options.      |
| Please make sure that the poppler tools are properly installed.    |
+--------------------------------------------------------------------+
EOF
        exit 10
    fi

    # xml parser for awk :-)
    xml2dsed="$tempdir/xml2dsed.awk"
    cat > "$xml2dsed" <<\EOF

# initializations

function _esc_init() {
    _ctrl = ""
    for (i=1; i<32; i++) { _ctrl = _ctrl sprintf("%c",i) }
    _esc = "[\"\\\\&" _ctrl "]"
    _ctrl = "[" _ctrl "]"
}

function _ord_init() {
    for (i=0; i<256; i++) {
	_ord[sprintf("%c",i)]=i
    }
}

function _amp_init() {
    _amp["&amp;"] = "&"
    _amp["&lt;"] = "<"
    _amp["&gt;"] = ">"
    _amp["&apos;"] = "'"
    _amp["&quot;"] = "\\\""
}

BEGIN {
    _esc_init()
    _ord_init()
    _amp_init()
    delete meta
    pheight=0
    pwidth=0
    location=""
    context=""
    content=""
    pageno=0
    dpi = 300 # use awk -f xml2dsed.awk dpi=xxx to override
    dometa=1  # use awk -f xml2dsed.awk dometa=0 to override
    dotext=1  # use awk -f xml2dsed.awk dotext=0 to override
    RS=">"    # xml parsing wants record delimiter set to ">"
}

# return character code

function ord(str) {
    return _ord[substr(str,1,1)]
}

# print properly escaped c string

function pstr(str,tmp) {
    printf("\"")
    while (str) {
	tmp = match(str,_esc) # char classes do not always work
	if (tmp == 0) {
	    printf("%s",str)
	    str = ""
	} else if (tmp > 1) {
	    printf("%s",substr(str,1,tmp-1))
	    str = substr(str,tmp)
	} else {
	    tmp = match(str,"^&[a-z]*;")
	    if (tmp != 1) { tmp = "" }
	    if (tmp) { tmp = _amp[substr(str,RSTART,RLENGTH)] }
	    if (tmp) {
		printf("%s",tmp)
		str = substr(str,RLENGTH+1)
	    } else {
		printf("\\%03o", ord(str))
		str = substr(str,2)
	    }
	}
    }
    printf("\"")
}

# sax-like callbacks

function charData(str) {
    if (context == "title") {
	meta["Title"] = str
    } else if (context == "word") {
	gsub(_ctrl," ",str)      # kill control characters
	gsub("\302\240"," ",str) # nbsp in utf-8
	gsub(/  */," ",str)      # simplify spaces
	gsub(/^ /,"",str)        # simplify spaces
	gsub(/ $/,"",str)        # simplify spaces
	if (match(str,/[^ ]/)) { content = str } else { content = "" }
    }
}

function startElement(tag,attrs) {
    if (tag=="head" && ! context && dometa) {
	context = "head"
    } else if (tag=="title" && context=="head") {
	context = "title"
    } else if (tag=="meta" && attrs["name"] && attrs["content"] ) {
	meta[attrs["name"]] = attrs["content"]
    } else if (tag=="body" && ! context && dotext) {
	context = "body"
    } else if (tag=="page" && context == "body") {
	pageno = pageno + 1
	pwidth = attrs["width"] * dpi / 72
	pheight = attrs["height"] * dpi / 72
	context = "page"
	printf("select %d\n", pageno)
	printf("set-txt\n")
	printf("(page %d %d %d %d\n", 0, 0, pwidth, pheight)
    } else if (tag=="word" && context == "page") {
	if (attrs["xMin"] && attrs["xMax"] && attrs["yMin"] && attrs["yMax"]) {
	    context = "word"
	    location = sprintf("%d %d %d %d",
			       attrs["xMin"] * dpi / 72,
			       pheight - attrs["yMin"] * dpi / 72,
			       attrs["xMax"] * dpi / 72,
			       pheight - attrs["yMax"] * dpi / 72 )
	}
    }
}

function endElement(tag) {
    if (tag == "title" && context == "title") {
	context = "head"
    } else if (tag == "head" && context == "head" && length(meta)>0) {
	context = ""
	printf("create-shared-ant\n")
	printf("set-meta\n")
	for (i in meta) {
	    printf("%s\t",i)
	    pstr(meta[i])
	    printf("\n")
	}
	printf(".\n")
    } else if (tag == "page" && context == "page") {
	context = "body"
	printf(")\n.\n")
    } else if (tag == "word" && context == "word" && content) {
	context = "page"
	printf("  (word %s ", location)
	pstr(content)
	printf(")\n")
	content = ""
    }
}

# xml parser (for pdftotext!)
{
    str = $0
    match(str,/^[ \n\r\t\f]*/)
    if (RSTART == 1 && RLENGTH > 0) {
	str = substr(str,RLENGTH+1)
    }
    if (! match(str,/</)) {
	charData(str)
    } else {
	if (RSTART > 1) {
	    arg = substr(str,1,RSTART-1)
	    str = substr(str,RSTART)
	    charData(arg)
	}
	match(str, "^</?[a-zA-Z][a-zA-Z0-9_]*")
	if (RSTART == 1) {
	    tag = substr(str,2,RLENGTH-1)
	    arg = substr(str,RLENGTH+1)
	    # parse attrs
	    delete attrs
	    if (match(arg,"/$")) { arg = substr(arg,1,RSTART-1) }
	    while (arg && match(arg,/[a-zA-Z][a-zA-Z0-9_]*=/)) {
		attr = substr(arg,RSTART,RLENGTH-1)
		arg = substr(arg,RSTART+RLENGTH)
		if (match(arg,/^[ \n\r\f\t]*"/) && match(arg,/"[^"]*"/)) {
		    attrs[attr] = substr(arg,RSTART+1,RLENGTH-2)
		    arg = substr(arg,RSTART+RLENGTH)
		} else if (match(arg,/^[^ \n\r\g\t]*/)) {
		    attrs[attr] = substr(arg,RSTART,RLENGTH)
		    arg = substr(arg,RSTART+RLENGTH)
		}
	    }
	    # callbacks
	    if (match(tag,"^/")) {
		endElement(substr(tag,2))
	    } else {
		startElement(tag,attrs)
		if (match(str,"/$")) { endElement(tag) }
	    }
	} 
    }	    
}
EOF

    # prepare djvused script
    dsedscript="$tempdir/dsed"
    if test -z "$run" 
    then
        "$pdftotext" -bbox "$infile" "-" \
            | awk -f $xml2dsed dpi=$dpi dometa=$popplermeta dotext=$popplertext \
            > "$dsedscript"
    else
        $run "$pdftotext" -bbox "$infile" "-" \
            \| awk -f $xml2dsed dpi=$dpi dometa=$popplermeta dotext=$popplertext \
            \> "$dsedscript"
    fi
    
    # execute dsed script
    $run "$djvused" $dsedverbosity -f "$dsedscript" -s "$outfile"
fi