#!/bin/bash

NC='\033[0m'
RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[0;33m'

msg_info() {
    echo -e "${GREEN}$1${NC}"
}

msg_warn() {
    echo -e "${YELLOW}$1${NC}"
}

msg_error() {
    echo -e "${RED}$1${NC}"
}

print_help() {
    echo
    echo "scan2pdf - turn document pictures into searchable PDFs"
    echo
    echo "    scan2pdf uses unpaper to clean the input image and"
    echo "    tesseract to extract the text and bundle everything in a"
    echo "    single PDF. imagemagick is used to prepare the input for"
    echo "    unpaper, and sxiv to selectively repeat the cleaning"
    echo "    process, when needed. Results may vary, depending on the"
    echo "    quality of the input image and the options used with"
    echo "    unpaper."
    echo
    echo "USAGE:"
    echo "    unpaper [OPTION ...] FILE [...]"
    echo
    echo "OPTIONs are:"
    echo "    -o | --output <name>"
    echo "        provide a name for the output file (without extension)"
    echo "    -h | --help"
    echo "        print this help"
    echo
}

batch_convert() {
    echo
    echo "Batch converting scans to PNM."
    for IMG in "${INPUT[@]}"
    do
        IMG_BASE=`basename $IMG`
        IMG_BASE_PNM="${IMG_BASE%.*}.pnm"
        echo "Converting $IMG -> $TEMP/$IMG_BASE_PNM"
        convert "$IMG" "$TEMP/$IMG_BASE_PNM"
        TO_PROCESS+=( "$TEMP/$IMG_BASE_PNM" )
    done
}

batch_unpaper() {
    echo
    echo "Batch unpaper scans."
    # Prompt user to get additional unpaper options. It is possible to
    # access the man page as well.
    UNPAPER_MORE_OPTIONS=()
    echo "Using the following options: ${UNPAPER_OPTIONS[@]}"
    echo "Specify additional options (quit|man|<option value>):"
    QUIT=0
    while [ "$QUIT" -eq 0 ]; do
        echo -n "> "
        read OPTION
        case "$OPTION" in
            q|quit)
                QUIT=1
                ;;
            m|man)
                man unpaper
                ;;
            *)
                UNPAPER_MORE_OPTIONS=( $OPTION )
                ;;
        esac
        echo "Using the following options: ${UNPAPER_OPTIONS[@]} ${UNPAPER_MORE_OPTIONS[@]}"
    done
    for IMG_PNM in "$@"
    do
        IMG_BASE_PNM=`basename "$IMG_PNM"`
        UNPAPER_IMG_BASE_PNM="unpaper_$IMG_BASE_PNM"
        # Possibly remove temp files from previous iterations
        rm -f "$TEMP/$UNPAPER_IMG_BASE_PNM"
        unpaper "${UNPAPER_OPTIONS[@]}" "${UNPAPER_MORE_OPTIONS[@]}" "$TEMP/$IMG_BASE_PNM" "$TEMP/$UNPAPER_IMG_BASE_PNM"
    done
    TO_PROCESS=()
    REVIEW=`\ls "$TEMP"/unpaper_* | sxiv -iopqt | sed "s|^$TEMP/unpaper_|$TEMP/|"`
    for FILE in $REVIEW
    do
        TO_PROCESS+=( "$FILE" )
    done
}

TEMP=`mktemp -d`
UNPAPER_OPTIONS=(--grayfilter-size 10,10 --grayfilter-step 5,5 --post-border 100,100,100,100 --border-align top --border-margin 100,100 --post-size a4)

INPUT=()
TO_PROCESS=()
OUTPUT=out

while [[ $# -gt 0 ]]; do
    case $1 in
        -h|--help)
            print_help
            exit 0
            ;;
        -o|--output)
            if [[ $# -ge 2 ]]; then
                OUTPUT=$2
                shift ; shift
            else
                msg_error "No value found for flag $1!"
                print_help
                exit 2
            fi
            ;;
        *)
            INPUT+=( "$1")
            shift
            ;;
    esac
done

# Convert all images to PNM
batch_convert "$@"

# Batch unpaper scans until user is ok with the result
while [[ ${TO_PROCESS[@]} ]]; do
    batch_unpaper "${TO_PROCESS[@]}"
done

echo
echo "Run tesseract on processed scans."
pushd "$TEMP" >/dev/null
\ls unpaper_* | tesseract - output --dpi 300 pdf
popd >/dev/null

echo
echo "Removing temporary files."
mv "$TEMP"/output.pdf ./
rm -rf "$TEMP"