#!/bin/bash

# this is a script to transform a PDF containing a scanned book
#  into a beautiful searchable PDF :-)
# depends on convert (ImageMagick), ghostscript, pdftk, pdfjam, cuneiform and hocr2pdf (ExactImage)
# $ sudo apt-get install imagemagick ghostscript pdftk pdfjam exactimage
# get Cuneiform from their homepage

#use shared library
export LD_LIBRARY_PATH=/usr/local/lib64

echo "usage: process document.pdf orientation split left top right bottom lang author title middle"
# where orientation is one of 0,1,2,3, meaning the amount of rotation by 90°
# and split is either 0 (already single-paged) or 1 (2 book-pages per pdf-page)
# and (left top right bottom) are the ranges in mm to trim
# and lang is a language as in "cuneiform -l".
# and author,title are used for the PDF metadata
# all values relative to a resolution of 300dpi
#
# usage examples:
# ./process.sh SomeFile.pdf 0 0 0 0 2500 2000 ger SomeAuthor SomeTitle middle
# will process a PDF with one page per pdf-page, cropping to width 2500 and height 2000

pdftk "$1" burst dont_ask output pg_%04da.pdf
left=$[$4*254/10/300]
top=$[$5*254/10/300]
right=$[$6*254/10/300]
bottom=$[$7*254/10/300]
middle=$[${11}*254/10/300]
for f in pg_*a.pdf
do
    echo "pre-processing $f ..."
    convert -quiet -monochrome -normalize -density 300 "$f" "$f.bmp"
    bmpheight=`identify -format '%h' $f.bmp`
    bmpwidth=`identify -format '%w' $f.bmp`

    convert -quiet -crop $[$bmpwidth-$4-$6]x$[$bmpheight-$5-$7]+$4+$5 "$f.bmp" "$f.bmp"
    if [ "1" = "$3" ]; then #two-side
        bmpheight=`identify -format '%h' $f.bmp` #in pixels
        bmpwidth=`identify -format '%w' $f.bmp`
        height=$[($bmpheight-$4-$6)]            #in pixels
        llwidth=$[(($bmpwidth+${11})/2+$6)*254/10/300]
        rrwidth=$[(($bmpwidth+${11})/2+$4)*254/10/300]


        convert -quiet -crop $[($bmpwidth-${11})/2]x$bmpheight+0+0 $f.bmp $f.1.bmp
        ./cuneiform -l $8 -f hocr -o $f.1.hocr $f.1.bmp &>/dev/null
        if [ -e $f.1.hocr ]; then #readable
            hocr2pdf1 -i $f.1.bmp -s -o ${f%a.pdf}.1.pdf < $f.1.hocr &>/dev/null
        else #unreadable
            pdfjoin -q --rotateoversize 'false' --trim $left'mm '$bottom'mm '$llwidth'mm '$top'mm' --outfile ${f%a.pdf}.1.pdf $f
        fi

        convert -quiet -crop $[($bmpwidth-${11})/2]x$bmpheight+$[($bmpwidth+${11})/2]+0 "$f.bmp" "$f.2.bmp"
        ./cuneiform -l $8 -f hocr -o $f.2.hocr $f.2.bmp &>/dev/null
        if [ -e $f.2.hocr ]; then   #readable
            hocr2pdf1 -i "$f.2.bmp" -s -o ${f%a.pdf}.2.pdf < $f.2.hocr &>/dev/null
        else                    #unreadable
            pdfjoin -q --rotateoversize 'false' --trim $rrwidth'mm '$bottom'mm '$right'mm '$top'mm' --outfile ${f%a.pdf}.2.pdf $f
        fi
    else    #one-side
        ./cuneiform -l $8 -f hocr -o $f.hocr $f.bmp &>/dev/null
        if [ -e $f.hocr ]; then #readable
            hocr2pdf1 -i $f.bmp -s -o ${f%a.pdf}.pdf < $f.hocr &>/dev/null
        else    #unreadable
            pdfjoin -q --rotateoversize 'false' --trim $left'mm '$bottom'mm '$right'mm '$top'mm' --outfile ${f%a.pdf}.pdf $f
        fi
    fi
    rm -f pg_*.bmp $f pg_*.hocr

done

echo "[ /Title (${10})
  /Author ($9)
  /DOCINFO pdfmark" > pdfmarks

pdfjoin -q --rotateoversize 'false' --outfile $1-ocr.pdf pg_*.pdf
rm -f pg_*.pdf
rm -f doc_data.txt
rm -rf pg_*_files
gs -sDEVICE=pdfwrite -dPDFSETTINGS=/ebook -dCompatibilityLevel=1.4 -dNOPAUSE -dQUIET -dBATCH -sOutputFile=output.pdf $1-ocr.pdf pdfmarks
rm pdfmarks