scanner-doc

#!/bin/bash
# Utility for making quick scans for documents, all pages combined into a
# single OCR'ed PDF. Also check out tiff2pdf and pdfsandwich

# TODO djvu
set -euo pipefail

DPI=150

TMP=$(mktemp -d "${TMPDIR:-/tmp}/$(basename "$0").XXXXX")
function finalize {
    rm -rf "${TMP}"
}
trap finalize EXIT

ARGS=(
    -colorspace gray
    #-sigmoidal-contrast 20,50%
    #-posterize 8
)

SCANS=()
while true; do
    echo "Starting scan..." >&2
    SCAN="$(mktemp --tmpdir="$TMP" --dry-run "XXXXX.pdf")"
    scanimage \
        --progress \
        --resolution ${DPI} \
        -l 2 -t 2 -x 213 -y 295 \
        --format tiff \
    | convert TIFF:- \
        -strip \
        "${ARGS[@]}" \
        -background white \
        -page a4 \
        PDF:- \
    | ocrmypdf - "$SCAN"

    if [ $? -eq 0 ]; then
        echo "Scan finished." >&2
        SCANS+=("${SCAN}")
    else
        echo "Scan failed." >&2
    fi

    read -r -p "Add another page? [Y/n]" confirm
    case $confirm in
        [nN]*) break ;;
        *) continue ;;
    esac
done

# Combine pages
pdftk ${SCANS[@]} cat output "${1:-untitled}"