#!/bin/sh
# updatehosts: fetch external hosts files and append to our own


## Settings ###################################################################
HOSTS="${1:-/etc/hosts}"
#EXCLUDE='rover.ebay.com'  # separate multiple entries with colons
#HOSTS_URL='http://someonewhocares.org/hosts/zero/hosts'  ## hosts -> 0.0.0.0
#HOSTS_URL='http://someonewhocares.org/hosts/hosts'       ## hosts -> 127.0.0.1
HOSTS_URL='http://winhelp2002.mvps.org/hosts.txt'


## Internals; here be dragons #################################################
DELIM='## END STATIC HOSTS ##'
SED_COMMENT_HOST='s/^\(127\|0\)\.0\.0\.\(1\|0\)\( \+\)%s$/#\\0\1# excluded/g'
#SED_REMOVE_LOCAL='/^#<localhost>$/,/^#<\/localhost>$/d'  # someonewhocares
SED_REMOVE_LOCAL='/^\(127\.0\.0\.1\|::1\)  localhost$/d'     # mvps
SED_SCRUB_HOSTS='/^%s$/,$d'
KEEP_EXTERNAL_LOCALHOST=
TMP_PRIMARY=
TMP_SECONDARY=
TMP_SED=
STEP=


## Functions ##################################################################

err() {
    local pattern msg

    case "$1" in
    WRITE_TO_FILE)    msg="Failed to write to file: %s"              ;;
    DOWNLOAD)         msg="Failed to download URL: %s"               ;;
    EXCLUDE)          msg="Failed to comment hosts for exclusion"    ;;
    GET_NUM_ENTRIES)  msg="Failed to parse number of entries in %s"  ;;
    TIMESTAMP)        msg="Failed to timestamp hosts file (%s)"      ;;
    VERIFY_CONSTANTS) msg="Malconfigured script; please edit %s"     ;;
    WRITE_TO_TMP)     msg="Failed to create temporary files in /tmp" ;;
    APPEND_TO_FILE)   msg="Failed to append to file (%s >> %s)"      ;;
    "")               msg="undefined!" ;;
    *)                msg="$1"         ;;
    esac

    shift 2>/dev/null
    printf -- "\n[!] ERROR: $msg\n" "$@" 1>&2
    exit 1
}


clean_up() {
    rm -f "$TMP_PRIMARY" "$TMP_SECONDARY" "$TMP_SED"
}

step() {
    STEP=$((${STEP:-0}+1))
    local pattern="$1"
    shift 2>/dev/null

    printf -- "\n[%d] $pattern " $STEP "$@"
}


scrub_old() {
    local regex
    step "Scrubbing hosts file of old additions"

    # prepare regex by inserting the delimeter into it
    regex="$(printf -- "$SED_SCRUB_HOSTS" "$DELIM")"

    #echo "sed \"$regex\" $TMP_SECONDARY" >&2
    sed_i "$regex" "$TMP_SECONDARY" || err WRITE_TO_FILE "$TMP_SECONDARY"

    # restore delim until we figure out a better sed command
    grep -q -- "$DELIM" "$TMP_SECONDARY" || echo "$DELIM" >> "$TMP_SECONDARY"
}


sed_i() {
    # sed -i is supposedly not portable so let's reinvent it
    local expression="$1"
    local _file="$2"

    sed "$expression" "$_file" > "$TMP_SED" || return 1
    mv -f "$TMP_SED" "$_file"

    return $?
}


download_hosts() {
    local num_excluded num_hosts
    step "Downloading external hosts file"

    wget -qt3 -O "$TMP_PRIMARY" "$HOSTS_URL" || err DOWNLOAD "$HOSTS_URL"

    # remove localhost section in it unless KEEP_EXTERNAL_LOCALHOST is set
    [ "$KEEP_EXTERNAL_LOCALHOST" ] || sed_i "$SED_REMOVE_LOCAL" "$TMP_PRIMARY"

    # comment our exclusions and catch returned count
    num_excluded=$(exclude "$TMP_PRIMARY") || err EXCLUDE

    # count number of uncommented lines in our temp file
    num_hosts=$(get_num_entries "$TMP_PRIMARY") || err GET_NUM_ENTRIES

    printf -- '\n... %d host definitions loaded%s' \
        $num_hosts "${num_excluded:+" ($num_excluded commented out")}"
}


append() {
    step "Adding loaded hosts"

    cat "$TMP_PRIMARY" >> "$TMP_SECONDARY" || \
    err APPEND_TO_FILE "$TMP_PRIMARY" "$TMP_SECONDARY"
}


get_num_entries() {
    local _file
    _file="${1:-"$TMP_PRIMARY"}"
    cat "$_file" | grep -v '^[ \t]*#\|^$' | wc -l
    return $?
}


timestamp() {
    cat >> "$TMP_SECONDARY" << _EOF

# [${0##*/}] external hosts file as below added on $(date)

_EOF
    [ $? -eq 0 ] || err TIMESTAMP "$TMP_SECONDARY"
}


exclude() {
    [ "$EXCLUDE" ] || return 0
    local host regex before after delta
    local _file="$1"
    local commented=0
    local IFS=':'

    for host in $EXCLUDE; do
        unset regex before after delta
        regex="$(printf -- "$SED_COMMENT_HOST" "$host")"

        before=$(get_num_entries "$_file")
        sed_i "$regex" "$_file" || return 1
        after=$(get_num_entries "$_file")

        delta=$((before-$after))
        commented=$((commented+$delta))
    done

    [ $commented -gt 0 ] && echo $commented
    return 0
}


sanity_check() {
    # constants set?
    ([ "$HOSTS" ]     && [ "$SED_COMMENT_HOST" ]  && \
     [ "$DELIM" ]     && [ "$SED_REMOVE_LOCAL" ]  && \
     [ "$HOSTS_URL" ] && [ "$SED_SCRUB_HOSTS"  ] ) || err VERIFY_CONSTANTS "$0"

    # does the hosts file exist to begin with. if not, have our downloaded list
    # retain its basic localhost entries (127.0.0.1 etc)
    [ ! -e "$HOSTS" ] && KEEP_EXTERNAL_LOCALHOST=1

    # delimeter present in hosts? else append
    grep -q -- "$DELIM" "$TMP_SECONDARY" || echo "$DELIM" >> "$TMP_SECONDARY"

    return 0
}


create_temp() {
    step "Creating working copy of %s along with temporary files" "$HOSTS"

    TMP_PRIMARY="$(  mktemp /tmp/hosts.primary.XXXXXX)" || err WRITE_TO_TMP
    TMP_SECONDARY="$(mktemp /tmp/hosts.secondary.XXXXXX)"
    TMP_SED="$(      mktemp /tmp/hosts.sed.XXXXXX)"
    [ -e "$HOSTS" ] && cp "$HOSTS" "$TMP_SECONDARY"

    return $?
}


commit() {
    step "Committing all changes to %s (may need root)" "$HOSTS"

    cp -f "$TMP_SECONDARY" "$HOSTS" 2>/dev/null && return 0

    # if we're still here we failed, so try sudo and then su -c
    printf -- '\n... failed; retrying with sudo\n'
    sudo cp -f "$TMP_SECONDARY" "$HOSTS" 2>/dev/null && return 0

    printf '\n... failed; retrying with su -c\n'
    su -c "cp -f "$TMP_SECONDARY" "$HOSTS"" 2>/dev/null && return 0

    err WRITE_TO_FILE "$HOSTS"
}


finish_with_flair() {
    step "All done, no errors\n"
    return 0
}


main() {
    trap 'clean_up' EXIT

    create_temp
    sanity_check

    download_hosts
    scrub_old
    timestamp
    append
    commit

    finish_with_flair
    return $?
}


## Execution start ############################################################

main
exit $?