#!/bin/sh

# Version:    1.4.1
# Author:     Héctor Molinero Fernández <hector@molinero.xyz>
# Repository: https://github.com/zant95/hblock
# License:    MIT, https://opensource.org/licenses/MIT

set -eu
export LC_ALL=C

# Default options
output='/etc/hosts'
redirection='0.0.0.0'
backup=false
backupDir=''
lenient=false
ignoreDownloadError=false
color=auto
quiet=false

# shellcheck disable=SC2039
hostname=${HOSTNAME-$(uname -n)}

# Default header block
header=$(cat <<-EOF
	127.0.0.1       localhost $hostname
	255.255.255.255 broadcasthost
	::1             localhost ip6-localhost ip6-loopback
	fe00::0         ip6-localnet
	ff00::0         ip6-mcastprefix
	ff02::1         ip6-allnodes
	ff02::2         ip6-allrouters
	ff02::3         ip6-allhosts
EOF
)

# Default custom block
custom=''

# Default blocklist block
blocklist=''

# Default sources
sources=$(cat <<-'EOF'
	https://raw.githubusercontent.com/zant95/hmirror/master/data/adaway.org/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/adblock-nocoin-list/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/adguard-simplified/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/disconnect.me-ad/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/disconnect.me-malvertising/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/disconnect.me-malware/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/disconnect.me-tracking/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/eth-phishing-detect/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/fademind-add.2o7net/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/fademind-add.dead/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/fademind-add.risk/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/fademind-add.spam/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/kadhosts/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/malwaredomainlist.com/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/malwaredomains.com-immortaldomains/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/malwaredomains.com-justdomains/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/matomo.org-spammers/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/mitchellkrogza-badd-boyz-hosts/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/pgl.yoyo.org/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/ransomwaretracker.abuse.ch/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/someonewhocares.org/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/spam404.com/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/stevenblack/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/winhelp2002.mvps.org/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/zerodot1-coinblockerlists-browser/list.txt
	https://raw.githubusercontent.com/zant95/hmirror/master/data/zeustracker.abuse.ch/list.txt
EOF
)

# Default whitelist (POSIX basic regex)
# Examples:
#   \.com$              -> all domains that ends with '.com'.
#   ^example            -> all domains that starts with 'example'.
#   ^sub\.example\.org$ -> domain 'sub.example.org'.
whitelist=$(cat <<-'EOF'
EOF
)

# Default blacklist
blacklist=$(cat <<-'EOF'
EOF
)

# Methods
printStdout() {
	if [ "$quiet" != true ]; then
		# shellcheck disable=SC2059
		printf -- "$@"
	fi
}

printStderr() {
	# shellcheck disable=SC2059
	>&2 printf -- "$@"
}

logInfo() {
	printStdout '   - %s\n' "$@"
}

logAction() {
	if [ "$color" = true ]; then
		printStdout '\033[1;33m + \033[1;32m%s \033[0m\n' "$@"
	else
		printStdout ' + %s \n' "$@"
	fi
}

logError() {
	if [ "$color" = true ]; then
		printStderr '\033[1;33m + \033[1;31m%s \033[0m\n' "$@"
	else
		printStderr ' + %s \n' "$@"
	fi
}

getBoolVal() {
	[ "$1" = true ] && s='yes' || s='no'
	printf -- '%s' "$s"
}

checkBinary() {
	command -v -- "$@" >/dev/null 2>&1
}

fetchUrl() {
	userAgent='Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Firefox/52.0'
	if checkBinary curl; then
		curl -fsSL -A "$userAgent" -- "$@"
	else
		wget -qO- -U "$userAgent" -- "$@"
	fi
}

writeFile() {
	if [ -d "$2" ]; then
		logError "Cannot write '$2': is a directory"
		exit 1
	elif ([ -e "$2" ] && [ -w "$2" ]) || touch -- "$2" >/dev/null 2>&1; then
		printf -- '%s\n' "$1" | tee -- "$2" >/dev/null
	elif checkBinary sudo; then
		printf -- '%s\n' "$1" | sudo tee -- "$2" >/dev/null
	else
		logError "Cannot write '$2': permission denied"
		exit 1
	fi
}

showHelp() {
	if [ $# -eq 0 ]; then
		printStdout '%s\n' "$(cat <<-'EOF'
			Usage: hblock [options...]
			 -O, --output FILE            Hosts file location (default: /etc/hosts)
			 -R, --redirection IP         Destination IP for all entries in the blocklist
			                              (default: 0.0.0.0)
			 -H, --header HEADER          Content to be included at the beginning of the
			                              hosts file. You can use the output of any other
			                              command (e.g. "$(cat header.txt)")
			 -S, --sources URLS           Sources to be used to generate the blocklist
			                              (whitespace separated URLs)
			 -W, --whitelist ENTRIES      Entries to be removed from the blocklist
			                              (whitespace separated POSIX BREs)
			 -B, --blacklist ENTRIES      Entries to be added to the blocklist
			                              (whitespace separated domain names)
			 -b, --backup [DIRECTORY]     Make a time-stamped backup in DIRECTORY
			                              (default: output file directory)
			 -l, --lenient                Match any IP address from sources, although it
			                              will be replaced by the destination IP
			                              (default: 0.0.0.0, 127.0.0.1 or none)
			 -i, --ignore-download-error  Do not abort if a download error occurs
			 -c, --color auto|true|false  Colorize the output (default: auto)
			 -q, --quiet                  Suppress non-error messages
			 -v, --version                Show version number and quit
			 -h, --help                   Show this help and quit
		EOF
		)"
		exit 0
	else
		[ -n "$1" ] && printStderr '%s\n' "Illegal option $1"
		printStderr '%s\n' "Try 'hblock --help' for more information"
		exit 1
	fi
}

showVersion() {
	printStdout '%s\n' '1.4.1'
	exit 0
}

main() {
	# Transform long options to short ones
	for opt in "$@"; do
		shift
		case "$opt" in
			'--output')                set -- "$@" '-O' ;;
			'--redirection')           set -- "$@" '-R' ;;
			'--header')                set -- "$@" '-H' ;;
			'--sources')               set -- "$@" '-S' ;;
			'--whitelist')             set -- "$@" '-W' ;;
			'--blacklist')             set -- "$@" '-B' ;;
			'--backup')                set -- "$@" '-b' ;;
			'--lenient')               set -- "$@" '-l' ;;
			'--ignore-download-error') set -- "$@" '-i' ;;
			'--color')                 set -- "$@" '-c' ;;
			'--quiet')                 set -- "$@" '-q' ;;
			'--version')               set -- "$@" '-v' ;;
			'--help')                  set -- "$@" '-h' ;;
			*)                         set -- "$@" "$opt"
		esac
	done
	# Set omitted arguments to empty strings
	for opt in "$@"; do
		shift
		case "$opt" in
			-*b)
				if a="$*"; [ -z "$a" ] || [ "${a#\-}x" != "${a}x" ]
					then set -- "$@" "$opt" ''
					else set -- "$@" "$opt"
				fi
				;;
			*) set -- "$@" "$opt"
		esac
	done
	# Read options
	OPTIND=1
	while getopts ':O:R:H:S:W:B:b:lic:qvh-:' opt; do
		case "$opt" in
			'O') output="$OPTARG" ;;
			'R') redirection="$OPTARG" ;;
			'H') header="$OPTARG" ;;
			'S') sources="$OPTARG" ;;
			'W') whitelist="$OPTARG" ;;
			'B') blacklist="$OPTARG" ;;
			'b') backup=true; backupDir="$OPTARG" ;;
			'l') lenient=true ;;
			'i') ignoreDownloadError=true ;;
			'c') color="$OPTARG" ;;
			'q') quiet=true ;;
			'v') showVersion ;;
			'h') showHelp ;;
			'-') showHelp "--$OPTARG" ;;
			*)   showHelp "-$OPTARG"
		esac
	done

	# Check color support
	if [ "$color" = auto ]; then
		[ -t 1 ] && color=true || color=false
	fi

	logAction 'Configuration:'

	logInfo "Hosts file: $output"
	logInfo "Backup: $(getBoolVal "$backup")"
	logInfo "Destination IP: $redirection"
	logInfo "Lenient: $(getBoolVal "$lenient")"
	logInfo "Ignore download error: $(getBoolVal "$ignoreDownloadError")"

	logAction 'Downloading lists...'

	if ! checkBinary curl && ! checkBinary wget; then
		logError 'Either wget or curl are required for this script'
		exit 1
	fi
	for url in $sources; do
		logInfo "$url"
		content=$(fetchUrl "$url") || true

		if [ -z "$content" ] && [ "$ignoreDownloadError" != true ]; then
			logError 'Download failed'
			exit 1
		fi

		blocklist=$(printf -- '%s\n%s' "$blocklist" "$content")
		unset content
	done

	logAction 'Parsing lists...'

	if [ -n "$blocklist" ]; then
		logInfo 'Remove carriage return'
		blocklist=$(printf -- '%s' "$blocklist" | tr -d '\r')

		logInfo 'Transform to lowercase'
		blocklist=$(printf -- '%s' "$blocklist" | tr '[:upper:]' '[:lower:]')

		logInfo 'Remove comments'
		blocklist=$(printf -- '%s' "$blocklist" | sed 's/#.*//')

		logInfo 'Trim spaces'
		blocklist=$(printf -- '%s' "$blocklist" | sed \
			-e 's/^[[:blank:]]*//' \
			-e 's/[[:blank:]]*$//'
		)

		logInfo 'Match hosts lines'
		if [ "$lenient" = true ]; then
			# This regex would be ideal, but it is not POSIX-compliant.
			# ipOctetRegex='\(25[0-5]\|2[0-4][0-9]\|[01]\{0,1\}[0-9][0-9]\{0,1\}\)'
			# ipRegex="\\($ipOctetRegex\\.$ipOctetRegex\\.$ipOctetRegex\\.$ipOctetRegex\\)"
			ipRegex='\([0-9]\{1,3\}\.\)\{3\}[0-9]\{1,3\}'
		else
			# Same as above.
			# ipRegex='\(\(0\.0\.0\.0\)\|\(127\.0\.0\.1\)\)'
			ipRegex='\(0\.0\.0\.0\)\{0,1\}\(127\.0\.0\.1\)\{0,1\}'
		fi
		domainRegex='\([0-9a-z_-]\{1,63\}\.\)\{1,\}[a-z][0-9a-z_-]\{1,62\}'
		blocklist=$(printf -- '%s' "$blocklist" | sed -n "/^\\(${ipRegex}[[:blank:]]\\{1,\\}\\)\\{0,1\\}$domainRegex$/p")

		logInfo 'Remove reserved TLDs'
		blocklist=$(printf -- '%s' "$blocklist" | sed \
			-e '/\.example$/d' \
			-e '/\.invalid$/d' \
			-e '/\.local$/d' \
			-e '/\.localdomain$/d' \
			-e '/\.localhost$/d' \
			-e '/\.test$/d'
		)

		logInfo 'Remove destination IPs'
		blocklist=$(printf -- '%s' "$blocklist" | sed 's/^.\{1,\}[[:blank:]]\{1,\}//')

		if [ -n "$whitelist" ]; then
			logInfo 'Apply whitelist'
			for domain in $whitelist; do
				blocklist=$(printf -- '%s' "$blocklist" | sed "/$domain/d")
			done
		fi
	fi

	if [ -n "$blacklist" ]; then
		logInfo 'Apply blacklist'
		for domain in $blacklist; do
			blocklist=$(printf -- '%s\n%s' "$blocklist" "$domain")
		done
	fi

	# This domain is used to check if hBlock is disabled
	blocklist=$(printf -- '%s\n%s' "$blocklist" 'hblock-check.molinero.xyz')

	logInfo 'Remove empty lines'
	blocklist=$(printf -- '%s' "$blocklist" | sed '/^$/d')

	logInfo 'Sort entries'
	blocklist=$(printf -- '%s' "$blocklist" | sort | uniq)

	if [ -n "$redirection" ]; then
		logInfo 'Add new destination IP'
		# Escape string literal for use as the replacement string in sed
		escapedRedirection=$(printf -- '%s' "$redirection" | sed 's/[&/\]/\\&/g')
		blocklist=$(printf -- '%s' "$blocklist" | sed "s/^/$escapedRedirection /")
	fi

	if [ -f "$output" ]; then
		content=$(cat -- "$output")

		# Get custom section
		logAction 'Reading custom section...'
		custom=$(
			printf -- '%s' "$content" | sed '/^#.*<custom>/,/^#.*<\/custom>/!d;//d' &&
			# "user-defined" is deprecated, use "custom" instead
			printf -- '%s' "$content" | sed '/^#.*<user-defined>/,/^#.*<\/user-defined>/!d;//d'
		)

		# Backup procedure
		if [ "$backup" = true ]; then
			logAction 'Backing up hosts file...'
			[ -z "$backupDir" ] && backupDir=$(dirname -- "$output")
			writeFile "$content" "$backupDir/$(basename -- "$output").$(date +%s).bak"
		fi

		unset content
	fi

	generationDate=$(date -u)
	blocklistCount=$([ -n "$blocklist" ] && printf -- '%s\n' "$blocklist" | wc -l | tr -d '[:blank:]' || printf '0')

	logAction 'Generating hosts file...'

	# Hosts file
	hosts=$(cat <<-EOF
		# Author:          Héctor Molinero Fernández <hector@molinero.xyz>
		# Repository:      https://github.com/zant95/hblock
		# Last updated:    $generationDate
		# Blocked domains: $blocklistCount
		$(
			[ -n "$header" ]    && printf -- '\n%s\n%s\n%s\n' '# <header>'    "$header"    '# </header>'
			[ -n "$custom" ]    && printf -- '\n%s\n%s\n%s\n' '# <custom>'    "$custom"    '# </custom>'
			[ -n "$blocklist" ] && printf -- '\n%s\n%s\n%s\n' '# <blocklist>' "$blocklist" '# </blocklist>'
		)
	EOF
	)

	writeFile "$hosts" "$output"
	logAction "$blocklistCount blocked domains!"
}

main "$@"