photos2pdf/photos2pdf

#!/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"

# Number of colors images are quantized to for component detection
colors=4

# Neighbors to evaluate when detecting connected components
connected_components=8

# Area treshold for connected components detection
area_treshold=1000

# Sigmoidal contrast adjustment used when trimming
sigmoidal_contrast_trim="30,30%"

# Sigmoidal contrast adjustment visible in result images
sigmoidal_contrast="45,50%"

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

# Keep intermediate images after processing
keep_tmpfiles=true

# Paper size of all pages
paper="a4paper"

# Offset in page where image should be placed
offset="1cm 3cm"

# Factor to scale image by before placing in page
scale=0.75

# PDF backdrop template directory
templatedir=/var/lib/photos2pdf/template

# Filename of complete result PDF
result_filename="result.pdf"

# Directory of original files to process (initially unset)
dir=

##
# Functions

log() {
    level=$1
    msg=$2

    printf "%s: %s\n" "$level" "$msg" >&2
}

error() {
    log ERROR "$1"

    exit "${2:-1}"
}

warning() {
    log WARNING "$1"
}

info() {
    log INFO "$1"
}

print_help() {
    cat << 'EOF'
photos2pdf
==========

Description
-----------
Trim and brighten JPGs, produce PDF containing one photo per page.

Usage
-----
```shell
photos2pdf [OPTION ...] DIR
```

Arguments
---------
* `DIR`: Directory containing the original photos.

Options
-------
* `--simplify INT`:             Before morphology analysis, reduce
                                image dimensions by this factor
                                (default: 2).
* `--morphology STRING`:        Morphology to apply for trimming
                                (default "Open:6").
* `--morphology-shape STRING`:  Shape of morphology to apply for
                                trimming (default: "Disk").
* `--colors INT`:               Number of colors images are quantized
                                to for component detection (default: 4).
* `--connected-components INT`: Neighbors to evaluate when detecting
                                connected components (default: 8).
* `--area_treshold INT`:        Area treshold for connected components
                                detection (default: 1000).
* `--sigmoidal-contrast-trim STRING`: Sigmoidal contrast adjustment
                                used when trimming (default: "30,30%").
* `--sigmoidal-contrast STRING`: Sigmoidal contrast adjustment visible
                                in result images (default: "45,50%").
* `--posterize INT`:            Number of colors result images are
                                quantized to (default: 6).
* `--keep-tmpfiles true|false`: Keep temporary files (default: false).
* `--paper STRING`:             PDF page size (default: "a4paper").
* `--offset STRING`:            Offset in page where image should be
                                placed (default: "1cm 3cm").
* `--scale FLOAT`:              Factor to scale image by before placing
                                in page (default: 0.75).
* `--templatedir STRING`:       PDF backdrop directory (default:
                                "/var/lib/photos2pdf/template").
* `--result-filename STRING`:   Filename of complete result PDF
                                (default: "result.pdf").
EOF
}

##
# Arguments

i=1
options_done=false

while [[ $i -le $# ]] ; do
    arg=$(eval "echo \$$i")

    if ! "$options_done" ; then
        case "$arg" in
            -h|--help)
                print_help
                exit 0
            ;;
            --simplify| \
            --morphology| \
            --morphology-shape| \
            --colors| \
            --connected-components| \
            --area-treshold| \
            --sigmoidal-contrast-trim| \
            --sigmoidal-contrast| \
            --posterize| \
            --keep-tmpfiles| \
            --paper| \
            --offset| \
            --scale| \
            --templatedir| \
            --result-filename)
                var=$(echo "$arg" | sed -e 's|^\-\-||;s|\-|_|g;')
                i=$((i+1))
                arg=$(eval "echo \$$i")
                eval "$var=\"$arg\""
                i=$((i+1))
                continue
            ;;
            --)
                options_done=true
                i=$((i+1))
                continue
            ;;
            -*)
                error "Unknown option \"$arg\"; aborted."
            ;;
        esac
    fi

    case "$arg" in
        *)
            [[ -n $dir ]] && error "Only one input directory can be specified; aborted."
            dir=$arg
            i=$((i+1))
        ;;
    esac
done

[[ -z $dir ]] && error "Usage: $0 [OPTION ..] DIR"

##
# Main Program

job_id=$(printf "%s" "$origdir" | sha256sum | cut -d" " -f1)

origdir=$(readlink -f "$dir")

[[ -d $origdir ]] || error "Directory $dir not found; aborted."

if [[ -n $TMP ]] ; then
    tmp_base=$TMP/photos2pdf/$job_id
elif [[ -n $XDG_RUNTIME_DIR ]] ; then
    tmp_base=$XDG_RUNTIME_DIR/photos2pdf/$job_id
else
    tmp_base=/tmp/photos2pdf/$job_id
fi

indir="$tmp_base/in"
tmpdir="$tmp_base/tmp"
outdir="$tmp_base/out"
pdfdir="$tmp_base/pdf"

if ! test -d "$origdir" ; then
    error "Directory \"$origdir\" not found; aborted."
elif ! mkdir -p "$indir" "$outdir" "$pdfdir" "$tmpdir" ; then
    error "Could not create working directories; aborted."
elif ! rm -f -- "$indir/"*.* ; then
    error "Could not delete existing input files; aborted."
elif ! rm -f -- "$tmpdir/"*.* ; then
    error "Could not delete existing tempfiles; aborted."
elif ! rm -f -- "$outdir/"*.* ; then
    error "Could not delete existing results; aborted."
elif ! rm -f -- "$pdfdir/"*.* ; then
    error "Could not delete existing PDFs; aborted."
fi

resize_percent=$((100/simplify))

i=1
rv=0

while read -r jpg ; do
    renamed=$(printf "%02d" "$i").jpg

    if cp "$jpg" "$indir/$renamed" ; then
        info "Copied \"$jpg\" to \"$renamed\"."
    else
        warning "Could not copy \"$jpg\" to \"$renamed\"."
        rv=1
    fi

    i=$((i+1))
done < <(ls --reverse --sort time -- "$origdir/"*.jpg)

if [[ $rv -ne 0 ]] ; then
    error "One or more errors during input file naming; aborted." "$rv"
fi

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 "$resize_percent%" \
        -colors "$colors" \
        -morphology "$morphology" "$morphology_shape" \
        "$tmp1"
    then
        warning "Could not create test image \"$tmp1\"; skipped."
        rv=1
        continue
    fi

    draw=()
    v=255

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

        ' | \
        sort | \
        uniq
    )

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

    if "$keep_tmpfiles" ; then
        if ! convert \
            "$tmp2" \
            -trim \
            "$trimmed"
        then
            warning "Could not test trim \"$tmp2\"; skipped."
            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")

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

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

if [[ $rv -ne 0 ]] ; then
    error "One or more errors during image conversion; aborted." "$rv"
fi

for png in "out/"*.png ; do
    info "Processing \"$png\" ..."

    read -r width height < <(
        identify -verbose "$png" | \
        perl -ne '/Geometry: (\d+)x(\d+)/ && print "$1 $2"'
    )

    if [[ -z $width ]] || [[ -z $height ]] ; then
        warning "Could not identify width or height of \"$png\"; skipped."
        rv=1
        continue
    elif [[ $width -gt $height ]] ; then
        backdrop="$templatedir/a4paper/landscape.pdf"
        jam_landscape=--landscape
    else
        backdrop="$templatedir/a4paper/portrait.pdf"
        jam_landscape=--no-landscape
    fi

    base=$(basename "$png" .png)
    pdf="$tmpdir/$base.pdf"
    jammed="$tmpdir/$base-pdfjam.pdf"
    backdropped="$tmpdir/$base.backdropped.pdf"

    if ! convert "$png" "$pdf" ; then
        warning "Could not convert \"$png\" to PDF; skipped."
        rv=1
        continue
    elif !  pdfjam \
        --quiet \
        --outfile "$tmpdir" \
        --paper "$paper" \
        "$jam_landscape" \
        --scale "$scale" \
        --offset "$offset" \
        "$pdf"
    then
        warning "Could not align \"$pdf\"; skipped."
        rv=1
        continue
    elif ! pdftk \
        "$jammed" \
        stamp "$backdrop" \
        output "$backdropped"
    then
        warning "Could not watermark \"$pdf\"; skipped."
        rv=1
        continue
    fi
done

input=()

while read -r pdf ; do
    input+=("$pdf")
done < <(ls -- "$tmpdir/"*.backdropped.pdf)

if ! pdftk "${input[@]}" cat output "$pdfdir/$result_filename" ; then
    warning "Could not concatenate \"$result_filename\"."
    rv=1
elif ! "$keep_tmpfiles" ; then
    if ! rm -f -- "$tmp_base" ; then
        warning "Could not remove temporary files."
        rv=1
    fi
fi

if [[ $rv -ne 0 ]] ; then
    error "One or more errors; aborted." "$rv"
fi