This script trims white margins out of PDF document and packs it to DjVu

pdf-trim-to-djvy.sh

#!/bin/bash
# vim: set fileencoding=utf-8 ts=2 sw=2 expandtab:
# PUBLIC DOMAIN

# This script trims white margins out of PDF document and packs it to DjVu.
# Author Sergei Astanin http://sovety.blogspot.com/
# Modified by Vsevolod K. https://github.com/precious

### Settings

DPI=150
MARGIN=10
MODE=gray # mono or gray or color
COLUMNS=1

### End of settings

prog=$0
dpidefault=$DPI
CROP=""
modedefault=$MODE
type=djvu
resizeopt=""
quality=60

function usage() {
  me=`basename "$prog"`
  monodefault=""
  graydefault=""
  colordefault=""
  if [ $modedefault == "mono" ] ; then
    monodefault="[default]"
  elif [ $modedefault == "gray" ] ; then
    graydefault="[default]"
  else
    colordefault="[default]"
  fi
  cat << END
Usage: $me [options] document.pdf

Options:
 -f <int>             the first page to process [default: 1]
 -l <int>             the last page to process
 -d <int>             resolution in DPI [default: $dpidefault]
 -q|--quality <int>   quality of images for generating pdf [default: 60]
 -c|--columns <int>   multi-column mode [default: $COLUMNS]
 -p|--crop <tXbXlXrX> crop image's top, bottom, left and/or right edges
                      example: $me --crop b40r10 file.pdf
 -m|--margin <int>    borders width [default: $MARGIN]
 -t|--type <str>      type of result document - pdf/djvu [default: djvu]
 -r|--resize <geomet> resize pages; geometry is WIDTH or xHEIGH or WIDTHxHEIGHT
                      or as presented at
          http://www.imagemagick.org/script/command-line-processing.php#geometry
 --mono               bitonal compression (black and white only) $monodefault
 --gray               DjVuPhoto compression (shades of gray images) $graydefault
 --color              DjVuPhoto compression (color images) $colordefault
 -h|--help            print this message

END
}

opts=`getopt -l "help,mono,gray,grey,color,colour,columns,crop,margin:type:resize:quality:" \
      "hf:l:d:c:p:m:t:r:q:" "$@"` && eval set -- "$opts"
while [ $# -gt 0 ] ; do
  case "$1" in
    -h|--help) usage ; exit 0 ;;
    -f) frompage=$2 ; shift 2 ;;
    -l) topage=$2 ; shift 2 ;;
    -d) DPI=$2 ; shift 2 ;;
    --mono) MODE=mono ; shift ;;
    --color|--colour) MODE=color ; shift ;;
    --gray|--grey) MODE=gray ; shift ;;
    --columns|-c) COLUMNS=$2 ; shift 2 ;;
    --crop|-p) CROP=$2 ; shift 2 ;;
    --type|-t) type=$2 ; shift 2 ;;
    --quality|-q) quality=$2 ; shift 2 ;;
    --resize|-r) resize=$2 ; shift 2 ;;
    -m|--margin) MARGIN=$2 ; shift 2 ;;
    *) shift ; break ;;
  esac
done

if [ $# -ne 1 ] ; then
  printf "Filename is required.\n"
  usage
  exit 1
fi

# absolute paths to input PDF and output DjVu files
pdf="`pwd`/$1"
filename="${pdf%.pdf}"

tmpdir=`mktemp -d /tmp/pagesXXXXX`

if [ "x" == "x${frompage}" ] ; then # first page not defined
  frompage=1
fi

if [ "x" == "x${topage}" ] ; then # last page not defined
  topage=`LANG=C pdfinfo "$pdf" | awk '/^Pages:/ {print $NF;}'`
fi

if [ $MODE == "mono" ] ; then # use bitonal compression
  monoopt="-mono"
  pmext="pbm"
  compress="cjb2"
elif [ $MODE == "gray" ] ; then  # use grayscale compression
  monoopt=""
  pmext="ppm"
  compress="c44 -crcbnone"
else # use DjVuPhoto compression (suitabable for color and grayscale images)
  monoopt=""
  pmext="ppm"
  compress="c44"
fi

if [ ! -z "$CROP" ] ; then # process crop options and write them into appropriate variables
  for param in t b l r ; do 
    export $param=`echo $CROP|grep -Po "(?<=$param)\d+"`
  done
  
  if [ "$l" -o "$t" ] ; then cropopt="$cropopt -crop +${l:-0}+${t:-0}" ; fi
  if [ "$r" -o "$b" ] ; then cropopt="$cropopt -crop -${r:-0}-${b:-0}" ; fi
  if [ -z "$cropopt" ] ; then echo "invalid crop arguments" ; usage ; exit 1 ; fi
  
fi

if [ ! -z "$resize" ] ; then # set appropriate resize option for mogrify
  resizeopt="-resize $resize"
fi

# rasterize all pages: pdftoppm -r $DPI
# cut margins: mogrify ... -trim +repage
# add narrow margin: mogrify ... -bordercolor white -border $MARGIN
# compress with color-enabled djvu: c44
# display progress with page numbers (skipped pages indicated with "∙")
( cd $tmpdir &&
  (
    # for every page
    i=$frompage
    while [ $i -le $topage ] ; do
      echo -n "$i" && \
      pdftoppm -r $DPI $monoopt -f $i -l $i "$pdf" page && \
      fname=`ls -1t page-*.$pmext | head -1` && \
      if [ ! -z "$cropopt" ] ; then mogrify -quiet $cropopt +repage "$fname" ; fi && \
      mogrify -quiet -trim +repage $resizeopt "$fname" && \
      ( # skip empty pages
        if [ `identify -format "%k" "$fname"` -gt 1 ] ; then
           ####### mogrify -quiet -resize 800 "$fname"
           if [ $COLUMNS -ne 1 ] ; then  # cut page in COLUMNS x n pieces
             w=$(($(identify -format "%w" "$fname")/COLUMNS))  # crop width
             h=$(((2*MARGIN+w)*4/3-2*MARGIN))  # crop height, aspect 4:3
             pageh=$(identify -format "%h" "$fname")
             n=$((pageh/h + 1))  # number of vertical slices
             overlap=$(((h*n-pageh)/(n-1)))
             dh=$((h-overlap))
             state=("/" "-" "\\" "|")
             for c in `seq 0 $((COLUMNS-1))` ; do  # for every column
               for j in `seq 0 $((n-1))` ; do # for every partial image
                 pagename=`printf "%06d_%d_%02d.$pmext" $i $c $j`
                 djvuname=`printf "%06d_%d_%02d.djvu" $i $c $j`
                 convert "$fname" -crop ${w}x${h}+$((c*w))+$((j*dh)) \
                     -bordercolor white -border $MARGIN "$pagename"
                 $compress -dpi $DPI "$pagename" "$djvuname"
                 rm "$pagename"
                 printf "\b%s" ${state[$(((c*n+j)%4))]}
               done
             done
             printf " "
           else # keep page as is
             if [ "$MARGIN" != "0" ] ; then
               mogrify -bordercolor white -border $MARGIN "$fname"
             fi
             case "$type" in
               pdf) convert -quality $quality "$fname" `printf "%06d.jpg" $i` ;;
               djvu) $compress -dpi $DPI "$fname" `printf "%06d.djvu" $i` ;;
             esac
             printf " "
           fi
        else
           echo -n " skiped [empty page]"
        fi
      ) && rm page-*.$pmext && i=$((i+1))
    done
  ) && (
    echo -e "\nmerging..."
    case "$type" in
      djvu) djvm -c "$filename.djvu" *.djvu ;;
      pdf) convert *.jpg "${filename}_.pdf" ;;
    esac    
  )
) && rm -rf "$tmpdir" || (
#)  || (
  printf "Failure\nTemporary directory left: %s\n" $tmpdir
  exit 2
)