#!/bin/sh
# soulcoud: save an url on archive.org and send a copy by email to read later
# the email address is read from environment SOULCOUD_MAIL
# Author: prx <prx@si3t.ch>
# licence: MIT
# Require: curl, w3m or reader
# one may replace mail command with msmtp or else
# improve with wkhtml2pdf or other snapshot tool

usage()
{
	printf "usage:\n\t"
	printf "%s (-h) <url> <title>\n" "$0"
	printf "\t-h : print help\n"
	printf "\nexample:\n"
	printf "\t%s 'http://who-is-clarkkent.net' Superman id?\n" "$0"
	printf "\nThe page is sent to the mail address set in environment variable SOULCOUD_MAIL\n"
	exit
}

archive_url()
{
	# archive the url and output the page link on archive.org
	curl -s "https://web.archive.org/$1" | 
		sed -n 's/.*href="\([^"]*\).*/\1/p'
}

dump_page()
{
	# use reader or lynx to echo page content
	#reader "$1"
	w3m -F -dump -O utf-8 -T text/html "$1"
}

if [ -z "${SOULCOUD_MAIL}" ]; then
	printf "You must set the environment variable SOULCOUD_MAIL\n"
	printf "Example: in ~/.profile, enter\n"
	printf "\texport SOULCOUD_MAIL=bruce.wayne.corp\n"
	exit
fi

while getopts 'h' c
do
    case $c in
    	h) usage ;;
    esac
done
shift $((OPTIND - 1))

test $# -lt 1 && usage

url="$(printf "$*" | cut -d' ' -f1)"
title="$(printf "$*" | cut -d' ' -f2-)"
test -z "${title}" && title="${url}"

(
	# save the page with lynx
	dump_page "${url}"
	printf "\n---\n"
	printf "url: ${url}\n"
	# then save on archive.org
	case "${url}" in
		http* )
		archived="$(archive_url ${url})"
		if [ $? -eq 0 ]; then
			printf "Saved on: https://web.archive.org/%s\n" "${archived}"
		fi
		;;
	esac
) | mail -s "${title}" "${SOULCOUD_MAIL}"