photos2pdf/02-convert.sh

#!/bin/bash

##
# Configuration

# Before morphology analysis, reduce image dimensions by this factor
simplify=2

# Morphology to apply to reduce content to main area
morphology="Open:6"
morphology_shape="Disk"

# Color simplification to apply (50x50% means black/white)
random_treshold="50x50%"

# Treshold for connected components detection
area_treshold=500

# Parameters for sigmoidal contrast adjustment for trimming
sigmoidal_contrast_trim="35,30%"

# Parameters for sigmoidal contrast adjustment for result images
sigmoidal_contrast="45,50%"

# Number of colors result images are quantized to
posterize=6

# Keep intermediate images after processing (note that they will be removed in
# any case at the next run of this script)
keep_tmpfiles=true

##
# Main Program

dir="$(dirname "$(readlink -f "$0")")"
tmpdir="$dir/tmp"
indir="$dir/in"
comdir="$dir/com"
outdir="$dir/out"
resize_percent=$((100/simplify))

if ! cd "$dir" ; then
    echo "ERROR: Could not change to directory \"$dir\"; aborted." >&2
    exit 1
elif ! mkdir -p "$comdir" "$outdir" "$tmpdir" ; then
    echo "ERROR: Could not create working directories; aborted." >&2
    exit 1
elif ! rm -f -- "$comdir/"*.* ; then
    echo "ERROR: Could not delete existing component files; aborted." >&2
    exit 1
elif ! rm -f -- "$tmpdir/"*.* ; then
    echo "ERROR: Could not delete existing tempfiles; aborted." >&2
    exit 1
elif ! rm -f -- "$outdir/"*.* ; then
    echo "ERROR: Could not delete existing results; aborted." >&2
    exit 1
else
    rv=0

    for jpg in "$indir/"*.jpg ; do
        base=$(basename "$jpg" .jpg)
        tmp1="$tmpdir/$base.tmp1.png"
        tmp2="$tmpdir/$base.tmp2.png"
        com="$tmpdir/$base.com.png"
        trimmed="$tmpdir/$base.trimmed.png"
        inf="$tmpdir/$base.txt"
        out="$outdir/$base.png"

        if ! convert \
            "$jpg" \
            -colorspace RGB \
            -sigmoidal-contrast "$sigmoidal_contrast_trim" \
            -resize "$resize_percent%" \
            -channel All -random-threshold "$random_treshold" \
            -morphology "$morphology" "$morphology_shape" \
            "$tmp1"
        then
            echo "ERROR: Could not create test image \"$tmp1\"; skipped." >&2
            rv=1
            continue
        fi

        draw=()

        while read -r primitive ; do
            draw+=("-fill" "white" "-draw" "$primitive")
        done < <(
            convert \
                "$tmp1" \
                -define connected-components:verbose=true \
                -define connected-components:exclude-header=true \
                -define connected-components:sort-order=increasing \
                -define connected-components:area-threshold="$area_treshold" \
                -virtual-pixel None \
                -connected-components 8 \
                -auto-level \
                "$com" | \
            perl -e '
                while(<>){
                    chomp;
                    next unless /(\d+)x(\d+)\+(\d+)\+(\d+)/;
                    printf
                        STDERR
                        "DEBUG: component: $_\n";
                    printf
                        "rectangle %d,%d,%d,%d\n",
                        $1, $2, $1+$3, $2+$4;
                }

            ' | \
            sort | \
            uniq
        )

        printf "DEBUG: connected components in \"%s\":" "$tmp1"
        for d in "${draw[@]}" ; do
            printf " \"%s\"" "$d"
        done
        printf "\n"

        if ! convert \
            "$tmp1" \
            -fill black -colorize 100 \
            "${draw[@]}" \
            "$tmp2"
        then
            echo "ERROR: Could not convert \"$tmp1\" to \"$tmp2\"; skipped." >&2
            rv=1
            continue
        elif ! convert \
            "$tmp2" \
            -trim info: \
            > "$inf"
        then
            echo "ERROR: Could not determine trim info from \"$tmp2\"; skipped." >&2
            rv=1
            continue
        fi

        if "$keep_tmpfiles" ; then
            if ! convert \
                "$tmp2" \
                -trim \
                "$trimmed"
            then
                echo "ERROR: Could not test trim \"$tmp2\"; skipped." >&2
                rv=1
                continue
            fi
        fi

        # upscale crop area from smaller temporary picture sizes
        crop=$(simplify=$simplify perl -MEnv -e '
            while(<>) {
                chomp();
                s/(\d+)/$1*${simplify}/eg;
                /(\d+)x(\d+) \d+x\d+\+(\d+)\+(\d+)/ && do {
                    print $1."x".$2."+".$3."+".$4;
                }
            }
        ' < "$inf")

        if ! "$keep_tmpfiles" ; then
            if ! rm -f -- "$tmp1" "$tmp2" "$trimmed" "$inf" ; then
                echo "WARNING: Could not remove temporary files in \"$dir/tmp\"." >&2
                rv=1
            fi
        fi

        echo "Processing $base: crop=$crop ..."

        if ! convert \
            "$jpg" \
            -crop "$crop" \
            -sigmoidal-contrast "$sigmoidal_contrast" \
            +dither -posterize "$posterize" \
            "$out"
        then
            echo "ERROR: Could not process \"$jpg\"; skipped." >&2
            rv=1
            continue
        fi
    done

    exit "$rv"
fi