#!/bin/sh
set -e

# NOTE: 451.36 GBs of free disk space is required to download
# each captured artefact successfully.

DB=~/Labs/modern.ie.vms.json/wayback-machine.json

set_mtime(){
	node -e '
		const [, path, mtime] = process.argv;
		const date = new Date(mtime);
		if(!+date)
			throw new Error(`Invalid date: ${mtime}`)
		require("fs").utimesSync(path, date, date);
	' "$@"
}

file_prop(){
	jq -r ".[\"$1\"] | .$2" "$DB"
}

for key in `jq -r 'keys_unsorted | .[]' "$DB"`; do
	set -- "`file_prop "$key" ts`" "$key" "${key##*/}"
	case $3 in
		*%20*) set -- "$1" "$2" "`printf %s "$3" | sed 's/%20/ /g'`";;
		'')    printf 'Invalid URL: %s\n' "$3" >&2; return 1;;
	esac
	if test -s "$3"; then
		printf >&2 'Already downloaded: %s\n' "$3"
		continue
	fi
	url="https://web.archive.org/web/$1id_/$2"
	printf >&2 'Downloading: %s\n' "$url"
	curl -# "$url" > "$3"
	stat -f %z "$3" | grep -q "`file_prop "$2" size`"
	set_mtime "$3" "`file_prop "$2" mtime`"
	chmod -w "$3"
	sha256sum "$3" >> SHA256
	sleep 6
done