unhBlock

#!/bin/sh
# based on unboundbl.sh maintained by alec armbruster (github.com/alectrocute)
# refactor by carlos hernandez (git.io/hrndz) for unbound

# list of urls containing list of domains seving ads and malware
PROGRAM="unhBlock"
UNHBLOCK_CONF="unhBlock.conf"
UNHBLOCK_CONF_DST="/etc/unbound/unbound.conf.d/$UNHBLOCK_CONF"
STATS=false
RELOAD=false
BLACKHOLE="0.0.0.0"
BLACKHOLE_RECORD_TYPE="A"

download_url() {
    url="$1"
    target_file="$2"
    printf "   Attempting to download %.185s (via curl).\n" "$url"
    if curl --output /dev/null --silent --head --fail "$url"; then
        curl -s "$url" >> "$target_file"
        cnt=$((cnt+1))
        printf " ^ Downloaded %.35s ...successfully.\n" "$url"
    else
        printf " * Error trying to download %.35s ...\n" "$url"
        failed_cnt=$((failed_cnt+1))
    fi
}


download_urls_in_list() {
    urllist="$1"
    target_file="$2"
    list_name="$(basename "$urllist")"

    printf "\n# Overview\n"
    printf " ^ %s URLs to fetchs: %s\n" "$list_name" "$urllist"

    # init counter for debugging purposes
    cnt=0
    failed_cnt=0
    printf "# Downloading URLs in %s..." "$list_name"
    while read -r url; do
        download_url "$url" "$target_file"
    done < "$urllist"
    printf "\n# Done downloading external URLs\n"
    # sort all the lists and remove any whitelist items!
    printf " ^ %d %s fetches failed." $failed_cnt "$list_name"
    printf " ^ %d %s(s) will be parsed...\n\n" $cnt "$list_name"
}


rm_cruft() {
    target_file="$1"
    printf " ^ Removing comments, empty lines, + fixing lineendings\n"
    perl -i -ne \
      '
        $line = $_;
        $line =~ s/^\s+//g;
        $line =~ s/^#.*//g;
        $line =~ s/#.*$//g;
        $line =~ s/ +$//g;
        $line =~ s/\r\n/\n/g;
        if ($line ne "\n") {
            print lc($line);
        }
      ' "${target_file}"
}


parse_domain_from_hostfile_records() {
    target_file="$1"
    printf " ^ Parse domains from hostfile like records\n"
    perl \
      -i -pe \
      '
        s/(^\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}|^::[A-z0-9_.:]+)\s+([^#\s]+)/$2/g;
      ' "${target_file}"
}


rm_duplicates() {
    target_file="$1"
    tmp_file=$(mktemp)
    printf " ^ Removing duplicates\n"
    sort "${target_file}" | uniq > "${tmp_file}"
    mv "${tmp_file}" "${target_file}"
}


rm_whitelist_from_blocklist() {
    whitelist="$1"
    blocklist="$2"
    results=$(mktemp)
    printf "\n# Removing whitelist from blocklist\n"
    grep -Fvxf "${whitelist}" "${blocklist}" > "${results}"
    mv "${results}" "${blocklist}"
}


format_as_unbound_conf() {
    target_file="$1"
    printf "\n# Formatting records as unbound local data\n"
    blackhole=$BLACKHOLE record_type=$BLACKHOLE_RECORD_TYPE \
        perl -i -ne \
          '
            $cnt++;
            if ($cnt == 1) {
                printf("server:\n")
            };
            ($domain) = split(" ", $_, -1);
            printf("local-zone: \"%s\" redirect\n", $domain);
            printf("local-data: \"%s %s %s\"\n", $domain, $ENV{record_type}, $ENV{blackhole});
          ' "${target_file}"
}


filter_domains() {
    list="$1"
    printf "\n# Filtering domains from blocklist\n"

    rm_cruft "$list"
    parse_domain_from_hostfile_records "$list"
    rm_duplicates "$list"
}


check_unbound_conf() {
    conf="$1"
    unbound-checkconf "$conf"  > /dev/null 2>&1 || \
        (printf "Unbound blocklist conf contained errors." && exit 1)
}


place_unboundBL_conf() {
    blocklist="$1"
    mv "$blocklist" "$UNHBLOCK_CONF_DST"
}


reload_unbound() {
    ! $RELOAD && return
    if unbound-control status > /dev/null 2>&1 ; then
        unbound-control reload
    fi
}


# create unbound conf
create_and_install_unbound_conf() {
    printf "[Starting unHblock update]"
    blocklist="$(mktemp)"
    download_urls_in_list "$blacklist_sources" "$blocklist"
    filter_domains "$blocklist"

    if [ -n "$whitelist_sources"  ]; then
        whitelist="$(mktemp)"
        download_urls_in_list "$whitelist_sources" "$whitelist"
        filter_domains "$whitelist"

        rm_whitelist_from_blocklist "$whitelist" "$blocklist"
    fi

    format_as_unbound_conf "$blocklist"
    check_unbound_conf "$blocklist"
    place_unboundBL_conf "$blocklist"

    printf "+ %s update complete!\n\n" "$PROGRAM"
}


# to be expanded in the future, stats() or -stats displays
# the amount of domains on the included blocklist.
show_stats() {
    ! $STATS && return
    domains_total=$(grep -Ec "^local-data" "$UNHBLOCK_CONF_DST")
    printf "\n# Blocked stats\n"
    printf " ^ %s domains are being blocked.\n\n" "$domains_total"
    exit 0
}


# displays usage settings for manual usage, if desired
show_usage() {
    printf "\nUsage: %s [-s] -u [-w whitelist] blaclist\n\n" "$PROGRAM"
    printf " -h, --help            Display usage instructions\n"
    printf " -b, --blackhole       Blackhole destination (0.0.0.0).\n"
    printf " -r, --record          Blackhole record type (A).\n"
    printf " -d, --dest            System path to place/source Unbound foramtted blocklist(s).\n"
    printf " -s, --stats           Display basic statistics of blocklist(s).\n"
    printf " -w, --whitelist file  Whitelist sources to omit form blocklist.\n\n"
}


check_arguments() {
    $STATS && return
    if [ -z "$blacklist_sources" ] || [ ! -f "$blacklist_sources" ]; then
        printf "Missing blocklist file.\n"
        exit 1
    fi

    if [ -n "$whitelist_sources" ]; then

        [ -f "$whitelist_sources" ] || \
            (printf "File containing whitelist sources not found.\n" \
            && exit 1)
    fi
}


parse_arguments() {
    # shell script functionality
    while getopts ":w:b:t:d:hrs-:" opt; do
      case $opt in
        w)
          whitelist_sources="$OPTARG"
          ;;
        b)
          BLACKHOLE="$OPTARG"
          ;;
        t)
          BLACKHOLE_RECORD_TYPE="$OPTARG"
          ;;
        d)
          UNHBLOCK_CONF_DST="$OPTARG"
          ;;
        r)
          RELOAD=true
          ;;
        s)
          STATS=true
          ;;
        h)
          show_usage >&2
          exit
          ;;
        - )
          case $OPTARG in
            whitelist)
              whitelist_sources="$OPTARG"
              ;;
            blackhole)
              BLACKHOLE="$OPTARG"
              ;;
            type)
              BLACKHOLE_RECORD_TYPE="$OPTARG"
              ;;
            dest)
              UNHBLOCK_CONF_DST="$OPTARG"
              ;;
            reload)
              RELOAD=true
              ;;
            stats)
              STATS=true
              ;;
            help)
              show_usage >&2
              exit
              ;;
            *)
              echo "Invalid option: --$OPTARG" >&2
              exit 2
              ;;
          esac
          ;;

        :)
          echo 'Missing argument!' >&2
          exit 1
           ;;

        \?)
          echo "Invalid option: -$OPTARG" >&2
          exit 2
          ;;
      esac
    done
    shift $((OPTIND-1))
    blacklist_sources="$1"
    check_arguments
}


check_sys_prereqs() {
    command -v unbound-checkconf > /dev/null || \
        (echo "Missing unbound-checkconf" && exit 1)
}


main() {
    check_sys_prereqs
    parse_arguments "$@"
    show_stats
    create_and_install_unbound_conf
    reload_unbound
}


# Support sourcing
if [ "$(basename -- "$0")" = "$PROGRAM" ]; then
    main "$@"
fi