|
| 1 | +#!/bin/sh |
| 2 | + |
| 3 | +### NGINX Bad Bot Blocker: setup script ################# |
| 4 | +### Copyright (C) 2017 Stuart Cardall ### |
| 5 | +### https://github.com/itoffshore ### |
| 6 | +### Licensed under the terms of the GPL2 ### |
| 7 | +########################################################## |
| 8 | + |
| 9 | +WWW=/var/www |
| 10 | +VHOST_EXT="vhost" |
| 11 | +VHOST_DIR=/etc/nginx/sites-available |
| 12 | +BOTS_DIR=/etc/nginx/bots.d |
| 13 | +CONF_DIR=/etc/nginx/conf.d |
| 14 | +MAIN_CONF=/etc/nginx/nginx.conf |
| 15 | +# setting Y / yes will whitelist only directories in $www |
| 16 | +# that look like domain.names |
| 17 | +DOT_NAMES="Y" |
| 18 | +# if you already set 'limit_conn addr' you may want to set |
| 19 | +# this to N / no. |
| 20 | +INC_DDOS="Y" |
| 21 | + |
| 22 | +####### end user configuration ########################### |
| 23 | + |
| 24 | +usage() { |
| 25 | + local script=$(basename $0) |
| 26 | + cat <<EOF |
| 27 | +$script: SETUP Nginx Bad Bot Blocker configuration in [ $MAIN_CONF ] [ $VHOST_DIR/* ] |
| 28 | +
|
| 29 | +Usage: $script [OPTIONS] |
| 30 | + [ -w ] : WWW path (default: $WWW) |
| 31 | + [ -e ] : Vhost file extension (default: .$VHOST_EXT) |
| 32 | + [ -v ] : Vhost directory (default: $VHOST_DIR) |
| 33 | + [ -b ] : Bot rules directory (default: $BOTS_DIR) |
| 34 | + [ -c ] : NGINX conf directory (default: $CONF_DIR) |
| 35 | + [ -m ] : NGINX main configuration (default: $MAIN_CONF) |
| 36 | + [ -n ] : NO whitelist of .names only (default: $DOT_NAMES) |
| 37 | + [ -d ] : NO insert of DDOS rule (default: $INC_DDOS) |
| 38 | + [ -x ] : Actually change the files (default: don't change anything) |
| 39 | + [ -h ] : this help message |
| 40 | +
|
| 41 | +Examples: |
| 42 | + $script -n (Whitelist all directory names in $WWW as domains: not just dot.name directories) |
| 43 | + $script -d (Do not insert DDOS rule: these may clash with existing 'limit_conn addr' rules) |
| 44 | + $script (Don't change anything: display results on stdout) |
| 45 | + $script -x (Change / update config files) |
| 46 | +EOF |
| 47 | + exit 0 |
| 48 | +} |
| 49 | + |
| 50 | +check_config() { |
| 51 | + local files="$*" |
| 52 | + |
| 53 | + if [ -z "$files" ]; then |
| 54 | + echo "no vhost files in: $VHOST_DIR/*.$VHOST_EXT => exiting." |
| 55 | + exit 1 |
| 56 | + fi |
| 57 | + |
| 58 | + if [ ! -f "$MAIN_CONF" ]; then |
| 59 | + echo "NGINX main configuration ('$MAIN_CONF') not found => exiting." |
| 60 | + exit 1 |
| 61 | + fi |
| 62 | +} |
| 63 | + |
| 64 | +find_vhosts() { |
| 65 | + find $VHOST_DIR -type f -name "*.$VHOST_EXT" |
| 66 | +} |
| 67 | + |
| 68 | +whitelist_ips() { |
| 69 | + local ip= conf=$BOTS_DIR/whitelist-ips.conf |
| 70 | + |
| 71 | + mkdir -p $BOTS_DIR |
| 72 | + |
| 73 | + if [ -n "$(which dig)" ]; then |
| 74 | + ip=$(dig +short myip.opendns.com @resolver1.opendns.com) |
| 75 | + if ! grep "$ip" $conf >/dev/null 2>&1; then |
| 76 | + printf "\n%-17s %-15s %-s\n" "Whitelisting ip:" "$ip" "=> $conf" |
| 77 | + if [ "$DRY_RUN" = "N" ]; then |
| 78 | + printf "%-23s %-s\n" "$ip" "0;" >> $conf |
| 79 | + fi |
| 80 | + fi |
| 81 | + else |
| 82 | + echo "WARN: dig binary missing => install bind-tools to whitelist external ip address" |
| 83 | + fi |
| 84 | +} |
| 85 | + |
| 86 | +whitelist_domains() { |
| 87 | + local domain_list= domain= domain_len= |
| 88 | + local conf=$BOTS_DIR/whitelist-domains.conf |
| 89 | + |
| 90 | + case "$DOT_NAMES" in |
| 91 | + y*|Y*) domain_list=$(find $WWW -mindepth 1 -maxdepth 1 -type d -name '*\.*' -exec basename {} \;);; |
| 92 | + *) domain_list=$(find $WWW -mindepth 1 -maxdepth 1 -type d -exec basename {} \;);; |
| 93 | + esac |
| 94 | + |
| 95 | + domain_len=$(find $WWW -mindepth 1 -maxdepth 1 -type d -exec basename {} \; \ |
| 96 | + | awk '{ print length ($0) }' | sort -nr | head -1) |
| 97 | + |
| 98 | + for domain in $domain_list; do |
| 99 | + if ! grep "$domain" $conf >/dev/null 2>&1; then |
| 100 | + printf "%-s %-$(( $domain_len +2))s %s\n" "Whitelisting domain:" "$domain" "=> $conf" |
| 101 | + if [ "$DRY_RUN" = "N" ]; then |
| 102 | + printf "%-$(( $domain_len +8))s %s\n" "\"~*$domain\"" "0;" >> $conf |
| 103 | + fi |
| 104 | + fi |
| 105 | + done |
| 106 | +} |
| 107 | + |
| 108 | +longest_str() { |
| 109 | + echo $@ | tr " " "\n" | awk '{print length ($0)}' | sort -nr | head -n1 |
| 110 | +} |
| 111 | + |
| 112 | +check_wildcard() { |
| 113 | + local file=$1 dir=$(basename $2) |
| 114 | + local check=$(grep -E "include[[:alnum:] /]+$dir/\*;" $file) |
| 115 | + echo $check |
| 116 | +} |
| 117 | + |
| 118 | +add_includes() { |
| 119 | + local ph='<<!!>>' line=$1 file=$2 conf_dir=$3 text= update= |
| 120 | + local include_list="$(echo $@ | awk '{$1=$2=$3=""}sub("^"OFS"+","")')" |
| 121 | + local col_size=$(( $(longest_str $include_list) + $(echo $conf_dir | wc -m) )) |
| 122 | + |
| 123 | + for text in $include_list; do |
| 124 | + if ! grep "$text" $file 1>/dev/null; then |
| 125 | + update='true' |
| 126 | + text="include $conf_dir/$text;" |
| 127 | + printf "%-10s %-$(( $col_size +10 ))s %s\n" "inserting:" "$text" "=> $file" |
| 128 | + if [ "$DRY_RUN" = "N" ]; then |
| 129 | + # $ph is just a placeholder so sed inserts a \t (tab) |
| 130 | + sed -i "$line i $ph \t$text $ph" $file |
| 131 | + fi |
| 132 | + fi |
| 133 | + done |
| 134 | + |
| 135 | + if [ "$DRY_RUN" = "N" ]; then |
| 136 | + if [ -n "$update" ]; then |
| 137 | + #add blank line below inserts |
| 138 | + line=$(( $line + $(echo $include_list | wc -w) )) |
| 139 | + if ! sed -n "${line}p" $file | grep ^'}' 1>/dev/null; then |
| 140 | + text="include $conf_dir/$(echo $include_list | awk '{print $1}');" |
| 141 | + sed -i "s|$text|$text\n|" $file |
| 142 | + fi |
| 143 | + |
| 144 | + #add comment above inserts |
| 145 | + text="include $conf_dir/$(echo $include_list | awk '{print $NF}');" |
| 146 | + sed -i "s|$text|# Bad Bot Blocker\n\t$text|" $file |
| 147 | + |
| 148 | + # remove placeholders |
| 149 | + sed -i "s|$ph||g" $file |
| 150 | + fi |
| 151 | + fi |
| 152 | +} |
| 153 | + |
| 154 | +find_line() { |
| 155 | + local file=$1 find_str=$2 first_last=$3 |
| 156 | + |
| 157 | + case "$first_last" in |
| 158 | + first) awk "/$find_str/{ print NR; exit }" $file;; |
| 159 | + last) awk "/$find_str/{ print NR }" $file | tail -n1;; |
| 160 | + esac |
| 161 | +} |
| 162 | + |
| 163 | +find_includes() { |
| 164 | + local file=$1 search=$2 search_first_last=$3 line= tmp=$(mktemp) |
| 165 | + local start_range=$4 start_first_last=$5 |
| 166 | + local end_range=$6 end_first_last=$7 |
| 167 | + local start=$(find_line $file $start_range $start_first_last) |
| 168 | + local end=$(find_line $file $end_range $end_first_last) |
| 169 | + |
| 170 | + sed -n ${start},${end}p ${file} > $tmp |
| 171 | + line=$(find_line $tmp $search $search_first_last) |
| 172 | + rm -f $tmp |
| 173 | + |
| 174 | + # search string not found |
| 175 | + if [ -z "$line" ]; then |
| 176 | + line=1 |
| 177 | + fi |
| 178 | + |
| 179 | + case "$search_first_last" in |
| 180 | + first) line=$(( $line + $start -1 ));; |
| 181 | + last) line=$(( $line + $start +1 ));; |
| 182 | + esac |
| 183 | + |
| 184 | + # if inserting beyond the end of the stanza |
| 185 | + if [ "$(sed -n $(( $line - 1))p $file | grep ^})" = "}" ]; then |
| 186 | + # insert blank line |
| 187 | + sed -i "$(( line - 1)) i \ " $file |
| 188 | + fi |
| 189 | + |
| 190 | + echo $line |
| 191 | +} |
| 192 | + |
| 193 | +sanitize_path() { |
| 194 | + echo $1 |tr -cd '[:alnum:] [=@=] [=.=] [=-=] [=/=] [=_=]' \ |
| 195 | + |tr -s '@.-/_' |awk '{print tolower($0)}' |
| 196 | +} |
| 197 | + |
| 198 | +sanitize_ext() { |
| 199 | + echo $1 |tr -cd '[:alnum:]' |awk '{print tolower($0)}' |
| 200 | +} |
| 201 | + |
| 202 | +check_args() { |
| 203 | + local option=$1 type=$2 arg=$3 |
| 204 | + local msg="ERROR: option '-$option' argument '$arg' requires:" |
| 205 | + |
| 206 | + case "$type" in |
| 207 | + path) if ! echo $arg | grep ^/ 1>/dev/null; then |
| 208 | + printf "$msg absolute path.\n" |
| 209 | + exit 1 |
| 210 | + fi |
| 211 | + ;; |
| 212 | + none) printf "$msg argument.\n"; exit 1;; |
| 213 | + esac |
| 214 | +} |
| 215 | + |
| 216 | +get_options() { |
| 217 | + local arg= opts= |
| 218 | + |
| 219 | + while getopts :w:e:v:b:c:m:ndxh opts "$@" |
| 220 | + do |
| 221 | + if [ -n "${OPTARG}" ]; then |
| 222 | + case "$opts" in |
| 223 | + e) arg=$(sanitize_ext ${OPTARG});; |
| 224 | + *) arg=$(sanitize_path ${OPTARG});; |
| 225 | + esac |
| 226 | + fi |
| 227 | + |
| 228 | + case "$opts" in |
| 229 | + w) WWW=$arg; check_args $opts path $arg ;; |
| 230 | + e) VHOST_EXT=$arg;; |
| 231 | + v) VHOST_DIR=$arg; check_args $opts path $arg ;; |
| 232 | + b) BOTS_DIR=$arg; check_args $opts path $arg ;; |
| 233 | + c) CONF_DIR=$arg; check_args $opts path $arg ;; |
| 234 | + m) MAIN_CONF=$arg; check_args $opts path $arg ;; |
| 235 | + n) DOT_NAMES=N ;; |
| 236 | + d) INC_DDOS=N ;; |
| 237 | + x) DRY_RUN=N ;; |
| 238 | + h) usage ;; |
| 239 | + \?) usage ;; |
| 240 | + :) check_args $OPTARG none none ;; |
| 241 | + esac |
| 242 | + done |
| 243 | +} |
| 244 | + |
| 245 | +wget_opts() { |
| 246 | + local opts= |
| 247 | + |
| 248 | + # GNU wget / Busybox 1.26.2 |
| 249 | + if wget --help 2>&1 | grep "\--spider" >/dev/null 2>&1; then |
| 250 | + opts="--spider" |
| 251 | + else # Busybox wget < 1.26.2 |
| 252 | + opts="-s" |
| 253 | + fi |
| 254 | + |
| 255 | + echo $opts |
| 256 | +} |
| 257 | + |
| 258 | +check_online() { |
| 259 | + local url=$1 options=$(wget_opts) |
| 260 | + |
| 261 | + if wget $options $url >/dev/null 2>&1; then |
| 262 | + echo "true" |
| 263 | + fi |
| 264 | +} |
| 265 | + |
| 266 | +main() { |
| 267 | + local include_url= file= line= file_list= |
| 268 | + local CONF_FILES= VHOST_INCLUDES= |
| 269 | + local REPO=https://raw.githubusercontent.com/mitchellkrogza/nginx-ultimate-bad-bot-blocker/master |
| 270 | + |
| 271 | + # require root |
| 272 | + if [ "$(id -u)" != "0" ]; then |
| 273 | + echo "This script must be run as root" 1>&2 |
| 274 | + exit 1 |
| 275 | + fi |
| 276 | + |
| 277 | + # parse command line |
| 278 | + get_options $@ |
| 279 | + include_url=$REPO/include_filelist.txt |
| 280 | + |
| 281 | + # check repo is online & source includes |
| 282 | + printf "Checking url: $include_url\n" |
| 283 | + if [ -n "$(check_online $include_url)" ]; then |
| 284 | + local tmp=$(mktemp) |
| 285 | + wget -q $include_url -O $tmp |
| 286 | + # use period not source in POSIX shell |
| 287 | + . $tmp 2>/dev/null |
| 288 | + rm -f $tmp |
| 289 | + else |
| 290 | + printf "Repo down or missing: $include_url\n" |
| 291 | + exit 1 |
| 292 | + fi |
| 293 | + |
| 294 | + # double check we have some files sourced |
| 295 | + if [ -z "$CONF_FILES" ] || [ -z "$VHOST_INCLUDES" ]; then |
| 296 | + printf "Error sourcing variables from: $include_url\n" |
| 297 | + exit 1 |
| 298 | + fi |
| 299 | + |
| 300 | + # configure ddos include |
| 301 | + case "$INC_DDOS" in |
| 302 | + n*|N*) VHOST_INCLUDES=$(echo $VHOST_INCLUDES | sed 's|ddos.conf||');; |
| 303 | + esac |
| 304 | + |
| 305 | + # gather vhosts |
| 306 | + file_list=$(find_vhosts) |
| 307 | + check_config $file_list |
| 308 | + |
| 309 | + # by default do not change any files |
| 310 | + if [ -z "$DRY_RUN" ]; then |
| 311 | + printf "\n** Dry Run ** | not updating files | -x or --exec to change files\n\n" |
| 312 | + else |
| 313 | + printf "\n" |
| 314 | + fi |
| 315 | + |
| 316 | + # update vhosts |
| 317 | + for file in $file_list; do |
| 318 | + line=$(find_includes $file include last server_ last location first ) |
| 319 | + if [ -n "$(check_wildcard $file $BOTS_DIR)" ]; then |
| 320 | + local col_size=$(( $(longest_str $file_list) + $(echo $BOTS_DIR | wc -m) )) |
| 321 | + printf "%-10s %-$(( $col_size -19 ))s %s\n" "WARN:" "$BOTS_DIR/* detected" "=> $file" |
| 322 | + else # don't use wildcards in vhost files |
| 323 | + add_includes $line $file $BOTS_DIR $VHOST_INCLUDES |
| 324 | + fi |
| 325 | + done |
| 326 | + |
| 327 | + # update main config |
| 328 | + line=$(find_includes $MAIN_CONF include last http first '\}' last ) |
| 329 | + if [ -n "$(check_wildcard $MAIN_CONF $CONF_DIR)" ]; then |
| 330 | + local col_size=$(( $(longest_str $file_list) + $(echo $CONF_DIR | wc -m) )) |
| 331 | + printf "%-10s %-$(( $col_size -19 ))s %s\n" "INFO:" "$CONF_DIR/* detected" "=> $MAIN_CONF" |
| 332 | + else # wildcard conf.d ok in nginx.conf |
| 333 | + add_includes $line $MAIN_CONF $CONF_DIR $CONF_FILES |
| 334 | + fi |
| 335 | + |
| 336 | + whitelist_ips |
| 337 | + |
| 338 | + if [ -d $WWW ]; then |
| 339 | + whitelist_domains |
| 340 | + else |
| 341 | + printf "\nWeb directory not found ('$WWW'): not whitelisting domains.\n" |
| 342 | + fi |
| 343 | +} |
| 344 | + |
| 345 | +## START ## |
| 346 | +main $@ |
| 347 | +exit $? |
0 commit comments