wkhtmltopdf

makepdf

#!/bin/bash
#
# Create a pdf from HTML or URL


cleanup(){
    # check whether $TEMPDIR exists and is a subdir of /tmp
    if [[ -d $TEMPDIR && $(dirname "$TEMPDIR") = /tmp ]]; then
        rm -rf "$TEMPDIR"
    fi
}

error(){
    echo "ERROR: $@" >&2
}


getbinaries(){
    local binaries bpath bversion flag tpath tversion

    readarray -t binaries < <(which -a wkhtmltopdf)

    if (( OPTION_NEW == 1 )) || (( OPTION_OLD == 1 )); then
        declare -Al binmap
        for bpath in "${binaries[@]}"; do
            bversion=$(getversion "$bpath")
            binmap["$bversion"]=$bpath
        done

        (( OPTION_NEW == 1 )) && flag=r

        tversion=$(printf '%s\n' "${!binmap[@]}" | sort -V${flag} | head -n1)
        tpath=${binmap["$tversion"]}

        binaries=("$tpath")
    fi

    printf '%s\n' "${binaries[@]}"
}


getistring(){
    # get major version number
    local mversion=$(getversion "$1" | cut -d. -f2)

    if (( mversion >= 12 )); then
        echo '--load-error-handling ignore'
    else
        echo '--ignore-load-errors'
    fi
}


getoutput(){
    local bn destination fn output_determined output_supplied url

    destination=$DESTINATION
    url=$1
    output_supplied=$2

    if [[ -n $output_supplied ]]; then

        # check whether supplied output is a destination directory
        if [[ -d $output_supplied ]]; then
            destination=$output_supplied
        # check whether supplied output is a complete path
        elif [[ -d $(dirname "$output_supplied") && $(dirname "$output_supplied") != . ]]; then
            output_determined=$output_supplied
        # check whether supplied output is a filename
        else
            output_determined=$destination/$output_supplied
        fi

    fi

    if [[ -z $output_determined ]]; then
        bn=$(basename "$url")
        if [[ $bn = "$url" ]]; then
            fn=$RANDOM
        else
            fn=$(sed 's/^\(.*\)\..*$/\1/' <<<"$bn")
        fi
        output_determined=${destination}/${fn}.pdf
    fi

    echo "$output_determined"
}


getversion(){
    "$1" --version | awk '/wkhtmltopdf/ {print $2; exit}'
}


makepdf(){
    local binaries binary istring output tempfile url

    url=$1

    # determine actual destination path
    output=$(getoutput "$url" "$2")

    # ensure destination path does not already exist
    if [[ -e $output ]]; then
        error "'$output' already exists"
        return 1
    fi

    # ensure temp dir is removed upon exit
    trap cleanup EXIT

    # create temp dir
    TEMPDIR=$(mktemp --directory --tmpdir makepdf-XXXXXX)

    # generate temp file path (do not actually create the file yet)
    tempfile=$(mktemp --dry-run --tmpdir="$TEMPDIR" XXXXXX.pdf)

    readarray -t binaries < <(getbinaries)

    for binary in "${binaries[@]}"; do

        if (( OPTION_IGNORE == 1 )); then
            istring=$(getistring "$binary")
        fi

        # execute wkhtmltopdf with options
        "$binary" $istring "$url" "$tempfile"
        STATUS=$?

        if [[ -f $tempfile ]]; then
            (( OPTION_OPEN == 1 )) && ! opener "$tempfile" && break

            mkdir -p "$(dirname "$output")" && mv "$tempfile" "$output" && realpath "$output" && break
        else
            error "$binary failed to process '$url'"
            return 1
        fi

    done
}


opener(){
    local reply

    "$PDFTOOL" "$1"

    read -n1 -p "Keep file? " reply
    echo

    [[ $reply = [Yy] ]]
}


usage(){
    cat <<-EOF
	Usage: $PROGRAM [OPTION]... URL [OUTPUT]

	Leverage wkhtmltopdf to create a PDF from URL or HTML.

	  -i, --ignore           ignore load errors
	      --new              use newest version of wkhtmltopdf
	      --old              use oldest version of wkhtmltopdf
	  -o, --open             open pdf with $(basename "$PDFTOOL") after creation

	$program accepts the following wkhtmltopdf options:
	  -l  --lowquality
	  -n  --disable-javascript

	default binary: ${DEFAULT:-UNAVAILABLE}
	default destination: ${DESTINATION}

	EOF
    exit 0
}


version(){
    cat <<-EOF
	$PROGRAM $VERSION

	Copyright (c) 2015 Six ([email protected])
	EOF
    exit 0
}


PROGRAM=${0##*/}
VERSION='0.0.3'
DEFAULT=$(which wkhtmltopdf)
DESTINATION=$HOME/Downloads/wkhtmltopdf
PDFTOOL='evince'

if (( $# == 1 )); then
    if [[ $1 =~ ^(-h|--help|help)$ ]]; then
        usage
    elif [[ $1 =~ ^(-v|--version|version)$ ]]; then
        version
    fi
fi

shopts='i,l,n,o'
lopts='disable-javascript,ignore,lowquality,new,old,open'
params=$(getopt -o "$shopts" -l "$lopts" -n "$program" -- "$@") || exit 1

eval set -- "$params"

# parse options
for arg in "$@"; do
    if (( last == 1 )); then
        wkargs+=("$arg")
    else
        case $arg in
            -i|--ignore)
                OPTION_IGNORE=1
                ;;
            --new)
                OPTION_NEW=1
                ;;
            --old)
                OPTION_OLD=1
                ;;
            -o|--open)
                OPTION_OPEN=1
                ;;
            --)
                last=1
                ;;
        esac
    fi
done

# check whether arguments were supplied
if (( ${#wkargs[@]} == 0 )); then
    echo "$PROGRAM: missing operand" >&2
    echo "Try '$PROGRAM --help' for more information." >&2
    exit 1
# ensure --new and --old are not combined
elif (( OPTION_NEW == 1 )) && (( OPTION_OLD == 1 )); then
    error "'--new' and '--old' may not be used concurrently"
    exit 1
# ensure wkhtmltopdf is installed
elif [[ -z $DEFAULT ]]; then
    error "wkhtmltopdf is not installed"
    exit 1
fi

makepdf "${wkargs[@]}"

exit "${STATUS:-1}"