#!/usr/bin/env bash

#############
# Variables #
#############
# Define formatting variables
fb="$(tput bold)" ; readonly fb
fr="$(tput sgr0)" ; readonly fr

# Define name regular expression
name_regex='[^a-zA-Z0-9]' ; readonly name_regex

# Create required commands array
req_cmds=(
    'wkhtmltopdf'
    'xclip'
) ; readonly req_cmds

o_loc="" ; readonly o_loc  # Set output location for PDF files
pdf_app='' ; readonly pdf_app  # Set PDF viewer application

# Create required variables array
required_vars=(
    'o_loc'
    'pdf_app'
) ; readonly required_vars


#############
# Functions #
#############
display_usage() {
    local help_msg  # Define local variable
    help_msg="Usage: '${0}'

Purpose: Save web pages to PDF files.

Requires:
    wkhtmltopdf
        html to pdf converter
    xclip
        command line interface to X selections (clipboard)

Variables to set:
    o_loc
        Output location for the PDF files.
    pdf_app
        Application to view output PDF files in."

    echo "${help_msg}"
}


# Verify string is valid URL
val_url() {
    local to_test
    to_test="${1}"
    url_regex='https?:\/\/(www\.)?[-a-zA-Z0-9@:%._\+~#=]{1,256}\.'
    url_regex+='[a-zA-Z0-9()]{1,18}\b([-a-zA-Z0-9()!@:%_\+.~#?&\/\/=]*)'
    [[ "${to_test}" =~ ${url_regex} ]] ; echo "${?}"
}


# Parse arguments
parse_args() {
    local parsed_args inv_ents

    parsed_args="$(getopt \
        -n "${0##*/}" \
        -o h \
        --long help \
        -- "${@}")"

    # Check for invalid entries
    inv_ents="${?}" ; readonly inv_ents
    if [[ "${inv_ents}" -ne 0 ]]; then
        echo ''
        display_usage
        exit 1
    fi

    # Set positional parameters using parsed argument string
    eval set -- "${parsed_args}"

    # Evaluate options
    while true; do
        case "${1}" in
            '-h' | '--help')
                display_usage
                exit 0
                ;;
            '--')
                ## End of options
                shift
                break
                ;;
        esac
    done
}


# Run initialization checks
run_init_cks() {
    local \
        err_msg \
        mis_cmds \
        req_cmd \
        unset_vars \
        required_var

    # Verify script is being run from a desktop environment
    if ! [[ "${XDG_CURRENT_DESKTOP}" ]]; then
        err_msg='\nA desktop environment must be available to '
        err_msg+='use this script.'
        echo -e "${err_msg}" 1>&2
        exit 1
    fi

    # Create missing commands array
    declare -a mis_cmds
    # Verify that required commands are installed
    for req_cmd in "${req_cmds[@]}"; do
        if ! command -v "${req_cmd}" > '/dev/null' 2>&1; then
            mis_cmds+=("${req_cmd}")
        fi
    done
    # Exit script if commands are missing
    if [[ "${mis_cmds[*]}" ]]; then
        err_msg='\nInstall the following command(s) before '
        err_msg+='running the script again:'
        echo -e "${err_msg}\n${fb}${mis_cmds[*]}${fr}" 1>&2
        exit 1
    fi

    # Create unset variables array
    declare -a unset_vars
    # Verify that required variables are set
    for required_var in "${required_vars[@]}"; do
        if ! [[ "${!required_var}" ]]; then
            unset_vars+=("${required_var}")
        fi
    done
    # Exit script if variables are not set
    if [[ "${unset_vars[*]}" ]]; then
        err_msg='\nSet the following variable(s) before '
        err_msg+='running the script again:'
        echo -e "${err_msg}\n${fb}${unset_vars[*]}${fr}" 1>&2
        exit 1
    fi
}


# Sanitize name
clean_name() {
    local cleaned
    # Replace most non-alphanumeric characters with underscores
    cleaned="${1//${name_regex}/_}"
    # Squeeze underscores and lowercase name
    cleaned="$(echo -n "${cleaned}" | tr -s '_' | tr '[:upper:]' '[:lower:]')"
    echo "${cleaned}"
}


# Define starting point for execution of the program
main() {
    parse_args "${@}"

    run_init_cks

    # Prompt for input, save to variable
    echo ''
    read -p 'Enter PDF filename and URL (q to quit): ' -r filename url
    # Check for valid selection
    until [[ "${filename}" == 'q' || "${url}" == 'q' ]]; do
        filename="$(clean_name "${filename}")"  # Sanitize filename

        while true; do
            pass_or_fail="$(val_url "${url}")"
            if [[ "${url}" == 'q' ]]; then
                exit 0
            elif [[ "${pass_or_fail}" -eq 0 ]]; then
                break
            else
                err_msg='\nInvalid URL. Try again or enter q to quit.\n'
                echo -e "${err_msg}" 1>&2

                read -p 'Enter URL (q to quit): ' -r url
            fi
        done

        # Try to generate and open PDF
        if timeout -k 30 30 wkhtmltopdf "${url}" "${o_loc}/${filename}.pdf" \
                > '/dev/null' 2>&1; then
            timeout 5 "${pdf_app}" "${o_loc}/${filename}.pdf" > \
                '/dev/null' 2>&1
        else
            # Set and display error message, place clean filename on clipboard
            err_msg="\n${fb}wkhtmltopdf timed out. Try manually "
            err_msg+="saving the page.${fr}"
            echo -e "${err_msg}" 1>&2 &&
                echo "${filename}" | xclip -selection c
        fi
        
        echo ''
        read -p 'Enter PDF filename and URL (q to quit): ' -r filename url
    done
}


###########
# Program #
###########
main "${@}"  # Start program

exit 0  # Exit script with successful exit status