make_ocr.sh

make_ocr.sh

#!/bin/bash

basefolder="/home/ashley/Development/personal/vasulka-archive-archive/ocr"

for i in $(find * -iname '*.pdf');
do
        if [ `dirname $i` != "." ]
        then
                dirpath="${i%/*}"
                dir_arr=(`echo $dirpath | tr "/" "\n"`)
                path=""
                for x in "${dir_arr[@]}"
                do
                        if [ -z "$path" ]
                        then
                                path=$x
                                mkdir -p $basefolder$path
                        else
                                path=$path"/"$x
                                mkdir -p $basefolder$path
                        fi
                done
                ext="."${i##*.}
                output=${i/$ext/".txt"}
                if [ ! -f $basefolder$output ] || [ $i -nt $basefolder$output ]
                then
                        echo $i
                        pdftotext -enc ASCII7 $i $basefolder$output
                fi
        else
                ext="."${i##*.}
                output=${i/$ext/".txt"}
                if [ ! -f $basefolder$output ] || [ $i -nt $basefolder$output ]
                then
                        echo $i
                        pdftotext -enc ASCII7 $i $basefolder$output
                fi
        fi
done