buildhosts/buildhosts

#!/usr/bin/env bash

#
#  BuildHosts: Download & use custom hosts sources to build /etc/hosts
#
#  Written by Kevin MacMartin <prurigro@gmail.com>
#
#  Released under the MIT license
#

# Unset variables we don't expect
while read -r; do
    [[ "$REPLY" =~ ^(HOSTS_DIR|HOSTS_SYSTEM|HOSTS_CORE|HOSTS_SOURCES|HOSTS_WHITELIST|UID|HOME|PATH|SHELL|TERM|PWD|SHLVL|_)= ]] \
        || unset "${REPLY/=*}"
done < <(env)

# User variables
[[ -z "$HOSTS_DIR" ]] && HOSTS_DIR="/etc"
[[ -z "$HOSTS_SYSTEM" ]] && HOSTS_SYSTEM="$HOSTS_DIR/hosts"
[[ -z "$HOSTS_CORE" ]] && HOSTS_CORE="${HOSTS_SYSTEM}.core"
[[ -z "$HOSTS_SOURCES" ]] && HOSTS_SOURCES="${HOSTS_SYSTEM}.sources"
[[ -z "$HOSTS_WHITELIST" ]] && HOSTS_WHITELIST="${HOSTS_SYSTEM}.whitelist"

# Default list of sources (used to generate $HOSTS_SOURCES when it doesn't exist)
default_sources=(
    'https://adaway.org/hosts.txt'
    'http://winhelp2002.mvps.org/hosts.txt'
    'http://hosts-file.net/ad_servers.txt'
    'http://pgl.yoyo.org/adservers/serverlist.php?hostformat=hosts&showintro=0&mimetype=plaintext'
)

script_dependencies=('curl' 'mv' 'sed' 'sort')

# Set the script name
script_name="${0//*\/}"

# Configure colours
if [[ -t 1 ]]; then
    c_script='\e[1;35m'
    c_category='\e[1;33m'
    c_title='\e[1;34m'
    c_text='\e[1;37m'
    c_option='\e[1;31m'
    c_divider='\e[1;30m'
    c_heading='\e[1;40m'
    c_success='\e[1;42m'
    c_fail='\e[1;41m'
    c_reset='\e[0m'
else
    c_heading='|'
fi

# Help function: display help text
function buildhosts_help {
    printf "$c_text%s$c_reset%s\n" 'BuildHosts' ': Download and use custom hosts sources to build /etc/hosts'
    printf "\n$c_title%s$c_reset\n" 'USAGE'
    printf "  $c_script%s$c_reset [$c_category%s$c_reset]\n" "$script_name" 'COMMAND'
    printf "\n$c_title%s$c_reset\n" 'COMMANDS'
    printf "  $c_option%s$c_reset  $c_divider| $c_text%s$c_reset\n" 'build' 'add hosts lists to /etc/hosts'
    printf "  $c_option%s$c_reset $c_divider| $c_text%s$c_reset\n" 'revert' 'remove hosts lists from /etc/hosts'
    printf "  $c_option%s$c_reset   $c_divider| $c_text%s$c_reset\n" 'help' 'display this help'
    (( $1 )) && exit 0
}

# Error function: display error then exit unsuccessfully
function buildhosts_error {
    printf "$c_heading $c_fail%s$c_fail %s$c_reset\n" ' ! ERROR:' "$1" >&2

    [[ "$2" -eq 1 ]] && {
        printf '\n' >&2
        buildhosts_help
    }

    exit 1
}

function buildhosts_rootcheck {
    (( UID )) && buildhosts_error "$script_name must be run as root"
}

# The build function to generate or regenerate /etc/hosts using /etc/hosts.core and the list of sources
function buildhosts_build {
    # If $HOSTS_SOURCES doesn't exist, generate one using the default list of sources
    if [[ ! -f "$HOSTS_SOURCES" ]]; then
        # Fail if the user isn't root when $HOSTS_DIR or  isn't writable
        if [[ ! -w "$HOSTS_DIR" ]]; then
            buildhosts_rootcheck
        elif [[ -f "$HOSTS_SYSTEM" && ! -w "$HOSTS_SYSTEM" ]]; then
            buildhosts_rootcheck
        fi

        printf "$c_heading %s$c_reset %s\n" 'Generating Default Sources:' "$HOSTS_SOURCES"
        printf '%s\n' '# file:///etc/hosts.d/localhostlist.txt' >> "$HOSTS_SOURCES"
        printf '%s\n' '# http://domain.com/remotehostlist.txt' >> "$HOSTS_SOURCES"

        for source in "${default_sources[@]}"; do
            printf '%s\n' "$source" >> "$HOSTS_SOURCES"
        done
    fi

    # Fail if the /etc/hosts file isn't writable (and suggest using root)
    [[ -w "$HOSTS_SYSTEM" ]] \
        || buildhosts_rootcheck

    # Fail if $HOSTS_SOURCES contains no sources after trimming comments
    [[ -n $(sed 's|^\ *#.*$||' "$HOSTS_SOURCES" | tr -d "\n") ]] \
        || buildhosts_error "$HOSTS_SOURCES doesn't contain any sources"

    # If $HOSTS_CORE doesn't exist and $HOSTS_SYSTEM does, move $HOSTS_SYSTEM to $HOSTS_CORE
    [[ -f "$HOSTS_SYSTEM" && ! -f "$HOSTS_CORE" ]] && {
        printf "$c_heading %s $c_reset %s\n" 'Moving:' "$HOSTS_SYSTEM to $HOSTS_CORE"
        install -Dm644 "$HOSTS_SYSTEM" "$HOSTS_CORE"
    }

    # Generate the hosts list using the URLs in the $HOSTS_SOURCES
    unset temp_hosts

    while read -r source; do
        [[ -n "$source" ]] && {
            printf "$c_heading %s $c_reset $source\n" 'Downloading:'

            # Download the the current source into $source_data and fail if the result is empty
            source_data=$(curl -C - -s "$source" | tr -d "\r")

            [[ -n "$source_data" ]] \
                || buildhosts_error "Could not download list @ $source"

            # If this isn't the first source, add a newline at the top
            [[ -n "$temp_hosts" ]] \
                && source_data=$(printf '\n%s\n' "$source_data")

            # Strip comments, then add the source to the end of $temp_hosts
            temp_hosts="$temp_hosts$(sed 's|\ *#.*$||' <<< "$source_data")"
        }
    done < <(sed 's|^\ *#.*$||' "$HOSTS_SOURCES")

    # Sort and remove duplicates, change 0.0.0.0 to 127.0.0.1, and remove non-127.0.0.1 redirections
    temp_hosts="$(sort -u < <(sed 's|\t| |;s|^\ *[0-9]*\.[0-9]*\.[0-9]*\.[0-9]*|127\.0\.0\.1|;s|^127\.0\.0\.1\ \ *localhost\ *$||;s|^\ *[^1].*$||' <<< "$temp_hosts"))"

    # Add the system hosts file to the hosts list and warn if $HOSTS_CORE is missing
    printf "$c_heading %s $c_reset %s\n" 'WRITING:' "$HOSTS_CORE and $(wc -l <<< "$temp_hosts") host source entries to $HOSTS_SYSTEM"

    if [[ -f "$HOSTS_CORE" ]]; then
        temp_hosts="$(<"$HOSTS_CORE")$(printf '\n\n%s\n' "# Generated Host List ($script_name)")$temp_hosts"
    else
        printf "$c_heading $c_fail %s$c_reset %s" 'WARNING:' "core hosts file $HOSTS_CORE does not exist"
    fi

    # Exclude domains in the whitelist
    [[ -f "$HOSTS_WHITELIST" ]] && {
        while read -r; do
            temp_hosts="$(sed "/^[0-9][0-9\.]*[0-9] $REPLY/d" <<< "$temp_hosts")"
        done <"$HOSTS_WHITELIST"
    }

    # Write $temp_hosts to $HOSTS_SYSTEM if it's not empty
    [[ -n "$temp_hosts" ]] \
        && printf '%s\n' "$temp_hosts" > "$HOSTS_SYSTEM"

    if [[ -n $(<"$HOSTS_SYSTEM") ]]; then
        printf "$c_heading $c_success %s $c_reset\n" 'DONE!'
    else
        printf "$c_heading $c_fail %s $c_reset %s\n" 'FAILED...' '(reverting hosts file)'
        install -Dm644 "$HOSTS_CORE" "$HOSTS_SYSTEM"
        buildhosts_error "$HOSTS_SYSTEM could not be created"
        exit 1
    fi
}

function buildhosts_revert {
    # Fail if $HOSTS_CORE doesn't exist
    [[ -f "$HOSTS_CORE" ]] \
        || buildhosts_error "$HOSTS_CORE does not exist, cannot revert"

    # Fail if $HOSTS_CORE is empty when all comments are removed
    [[ -n $(sed 's|^\ *#.*$||' "$HOSTS_CORE" | tr -d '\n') ]] \
        || buildhosts_error "$HOSTS_CORE contains no information, cannot revert"

    # Move $HOSTS_CORE to $HOSTS_SYSTEM
    printf "$c_heading %s$c_reset %s\n" 'MOVING:' "$HOSTS_CORE to $HOSTS_SYSTEM"

    if mv "$HOSTS_CORE" "$HOSTS_SYSTEM"; then
        printf "$c_heading $c_success %s$c_reset\n" 'DONE!'
    else
        buildhosts_error "$HOSTS_CORE could not be moved to $HOSTS_SYSTEM"
    fi
}

# Check for missing dependencies and fail if any exist
declare -a missing_dependencies=()

for dep in "${script_dependencies[@]}"; do
    type -P "$dep" >/dev/null \
        || missing_dependencies=( ${missing_dependencies[@]} "$dep" )
done

[[ -n "${missing_dependencies[*]}" ]] && {
    buildhosts_error "Missing dependencies: $(
        for (( x=0; x < ${#missing_dependencies[@]}; x++ )); do
            printf '%s' "${missing_dependencies[$x]}"
            (( (( x + 1 )) < ${#missing_dependencies[@]} )) && printf '%s' ', '
        done
    )"
}

# Fail if run with no commands
[[ -z "$1" ]] && buildhosts_error 'A valid command must be provided' 1

# Run the command entered by the user
case "$1" in
    build)
        buildhosts_build
        ;;
    revert)
        buildhosts_revert
        ;;
    help|--help|-h)
        buildhosts_help 1
        ;;
    *)
        buildhosts_error 'Invalid command' 1
        ;;
esac