Document Scanning incl OCR

#!/bin/bash

set -x
set -e

target="${1}.pdf"
num_pages=${num_pages:-1}

if [[ -z "$target" ]]; then
    echo >&2 "Usage: $0 DESTINATION"
    exit 1
fi

echo "Scanning $num_pages page(s) to $target..."

tempdir=$(mktemp -d)
scansource=$(LC_ALL=C scanimage -L \
        | grep 'HP 7650 Document scanner' \
        | sed "s,^device \`\([^']\+\)'.*,\1,")

(
    cd "$tempdir"
    scanimage -d"$scansource" -B64 --format=tiff -l0 -t0 -x215 -y297 \
        --batch-start=0 \
        --batch-count="$num_pages" \
        --batch-increment=1 \
        -b \
        --mode Color \
        --resolution 300 \
        --source ADF
)

convert="convert ${tempdir}/out0.tif"
convert "${tempdir}/out0.tif" "${tempdir}/out0.pdf"
gs="gs \
    -dBATCH \
    -q \
    -dNOPAUSE \
    -sDEVICE=pdfwrite \
    -sOutputFile=- \
    ${tempdir}/out0.pdf"

for ((i=1; i < num_pages; ++i)) do
    convert "${tempdir}/out${i}.tif" "${tempdir}/out${i}.pdf"
    convert="$convert -append ${tempdir}/out${i}.tif"
    gs="$gs ${tempdir}/out${i}.pdf"
done

$convert "${tempdir}/scanimage.pnm"

gocr -i "${tempdir}/scanimage.pnm" -o "${tempdir}/${1}.txt" -f UTF8
a2ps -Xutf-8 -i "${tempdir}/${1}.txt" -o "${tempdir}/${1}.ps"
$gs "${tempdir}/${1}.ps" > "$target"

ls -lh "$tempdir"
ls -lh "$target"
rm -r "$tempdir"