xref: /webtrees/app/Http/Middleware/BadBotBlocker.php (revision cf891017dd40a0de9e99bfa64db680a95043750a) 
1<?php
2
3/**
4 * webtrees: online genealogy
5 * Copyright (C) 2023 webtrees development team
6 * This program is free software: you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by
8 * the Free Software Foundation, either version 3 of the License, or
9 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 * You should have received a copy of the GNU General Public License
15 * along with this program. If not, see <https://www.gnu.org/licenses/>.
16 */
17
18declare(strict_types=1);
19
20namespace Fisharebest\Webtrees\Http\Middleware;
21
22use Fig\Http\Message\StatusCodeInterface;
23use Fisharebest\Webtrees\Registry;
24use Fisharebest\Webtrees\Validator;
25use GuzzleHttp\Client;
26use GuzzleHttp\Exception\GuzzleException;
27use Iodev\Whois\Loaders\CurlLoader;
28use Iodev\Whois\Modules\Asn\AsnRouteInfo;
29use Iodev\Whois\Whois;
30use IPLib\Address\AddressInterface;
31use IPLib\Factory as IPFactory;
32use IPLib\Range\RangeInterface;
33use Psr\Http\Message\ResponseInterface;
34use Psr\Http\Message\ServerRequestInterface;
35use Psr\Http\Server\MiddlewareInterface;
36use Psr\Http\Server\RequestHandlerInterface;
37use Throwable;
38
39use function array_filter;
40use function array_map;
41use function assert;
42use function gethostbyaddr;
43use function gethostbyname;
44use function preg_match_all;
45use function random_int;
46use function response;
47use function str_contains;
48use function str_ends_with;
49
50/**
51 * Middleware to block bad robots before they waste our valuable CPU cycles.
52 */
53class BadBotBlocker implements MiddlewareInterface
54{
55    private const REGEX_OCTET = '(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)';
56    private const REGEX_IPV4  = '/\\b' . self::REGEX_OCTET . '(?:\\.' . self::REGEX_OCTET . '){3}\\b/';
57
58    // Cache whois requests.  Try to avoid all caches expiring at the same time.
59    private const WHOIS_TTL_MIN = 28 * 86400;
60    private const WHOIS_TTL_MAX = 35 * 86400;
61    private const WHOIS_TIMEOUT = 5;
62
63    // Bad robots - SEO optimisers, advertisers, etc.  This list is shared with robots.txt.
64    public const BAD_ROBOTS = [
65        'admantx',
66        'Adsbot',
67        'AhrefsBot',
68        'Amazonbot', // Until it understands crawl-delay and noindex / nofollow
69        'AspiegelBot',
70        'Awario', // Brand management
71        'Barkrowler',
72        'BLEXBot',
73        'DataForSEO',
74        'DataForSeoBot', // https://dataforseo.com/dataforseo-bot
75        'DotBot',
76        'Grapeshot',
77        'Honolulu-bot', // Aggressive crawer, no info available
78        'ia_archiver',
79        'linabot', // Aggressive crawer, no info available
80        'Linguee',
81        'MegaIndex.ru',
82        'MJ12bot',
83        'netEstate NE',
84        'panscient',
85        'PetalBot',
86        'proximic',
87        'SeekportBot', // Pretends to be a search engine - but isn't
88        'SemrushBot',
89        'serpstatbot',
90        'SEOkicks',
91        'SiteKiosk',
92        'Turnitin',
93        'wp_is_mobile', // Nothing to do with wordpress
94        'XoviBot',
95        'ZoominfoBot',
96    ];
97
98    /**
99     * Some search engines use reverse/forward DNS to verify the IP address.
100     *
101     * @see https://developer.amazon.com/support/amazonbot
102     * @see https://support.google.com/webmasters/answer/80553?hl=en
103     * @see https://www.bing.com/webmaster/help/which-crawlers-does-bing-use-8c184ec0
104     * @see https://www.bing.com/webmaster/help/how-to-verify-bingbot-3905dc26
105     * @see https://yandex.com/support/webmaster/robot-workings/check-yandex-robots.html
106     * @see https://www.mojeek.com/bot.html
107     * @see https://support.apple.com/en-gb/HT204683
108     */
109    private const ROBOT_REV_FWD_DNS = [
110        'Amazonbot'        => ['.crawl.amazon.com'],
111        'Applebot'         => ['.applebot.apple.com'],
112        'BingPreview'      => ['.search.msn.com'],
113        'Google'           => ['.google.com', '.googlebot.com'],
114        'Mail.RU_Bot'      => ['.mail.ru'],
115        'MicrosoftPreview' => ['.search.msn.com'],
116        'MojeekBot'        => ['.mojeek.com'],
117        'Qwantify'         => ['.search.qwant.com'],
118        'Sogou'            => ['.crawl.sogou.com'],
119        'Yahoo'            => ['.crawl.yahoo.net'],
120        'Yandex'           => ['.yandex.ru', '.yandex.net', '.yandex.com'],
121        'bingbot'          => ['.search.msn.com'],
122        'msnbot'           => ['.search.msn.com'],
123    ];
124
125    /**
126     * Some search engines only use reverse DNS to verify the IP address.
127     *
128     * @see https://help.baidu.com/question?prod_id=99&class=0&id=3001
129     * @see https://napoveda.seznam.cz/en/full-text-search/seznambot-crawler
130     * @see https://www.ionos.de/terms-gtc/faq-crawler
131     */
132    private const ROBOT_REV_ONLY_DNS = [
133        'Baiduspider' => ['.baidu.com', '.baidu.jp'],
134        'FreshBot'    => ['.seznam.cz'],
135        'IonCrawl'    => ['.1und1.org'],
136        'Neevabot'    => ['.neeva.com'],
137        'SeznamBot'   => ['.seznam.cz'],
138    ];
139
140    /**
141     * Some search engines operate from designated IP addresses.
142     *
143     * @see https://www.apple.com/go/applebot
144     * @see https://help.duckduckgo.com/duckduckgo-help-pages/results/duckduckbot
145     */
146    private const ROBOT_IPS = [
147        'AppleBot'    => [
148            '17.0.0.0/8',
149        ],
150        'Ask Jeeves'  => [
151            '65.214.45.143',
152            '65.214.45.148',
153            '66.235.124.192',
154            '66.235.124.7',
155            '66.235.124.101',
156            '66.235.124.193',
157            '66.235.124.73',
158            '66.235.124.196',
159            '66.235.124.74',
160            '63.123.238.8',
161            '202.143.148.61',
162        ],
163        'DuckDuckBot' => [
164            '23.21.227.69',
165            '50.16.241.113',
166            '50.16.241.114',
167            '50.16.241.117',
168            '50.16.247.234',
169            '52.204.97.54',
170            '52.5.190.19',
171            '54.197.234.188',
172            '54.208.100.253',
173            '54.208.102.37',
174            '107.21.1.8',
175        ],
176    ];
177
178    /**
179     * Some search engines operate from designated IP addresses.
180     *
181     * @see https://bot.seekport.com/
182     */
183    private const ROBOT_IP_FILES = [
184        'SeekportBot' => 'https://bot.seekport.com/seekportbot_ips.txt',
185    ];
186
187    /**
188     * Some search engines operate from within a designated autonomous system.
189     *
190     * @see https://developers.facebook.com/docs/sharing/webmasters/crawler
191     * @see https://www.facebook.com/peering/
192     */
193    private const ROBOT_ASNS = [
194        'facebook' => ['AS32934', 'AS63293'],
195        'twitter'  => ['AS13414'],
196    ];
197
198    /**
199     * @param ServerRequestInterface  $request
200     * @param RequestHandlerInterface $handler
201     *
202     * @return ResponseInterface
203     */
204    public function process(ServerRequestInterface $request, RequestHandlerInterface $handler): ResponseInterface
205    {
206        $ua      = Validator::serverParams($request)->string('HTTP_USER_AGENT', '');
207        $ip      = Validator::attributes($request)->string('client-ip');
208        $address = IPFactory::parseAddressString($ip);
209        assert($address instanceof AddressInterface);
210
211        foreach (self::BAD_ROBOTS as $robot) {
212            if (str_contains($ua, $robot)) {
213                return $this->response();
214            }
215        }
216
217        foreach (self::ROBOT_REV_FWD_DNS as $robot => $valid_domains) {
218            if (str_contains($ua, $robot) && !$this->checkRobotDNS($ip, $valid_domains, false)) {
219                return $this->response();
220            }
221        }
222
223        foreach (self::ROBOT_REV_ONLY_DNS as $robot => $valid_domains) {
224            if (str_contains($ua, $robot) && !$this->checkRobotDNS($ip, $valid_domains, true)) {
225                return $this->response();
226            }
227        }
228
229        foreach (self::ROBOT_IPS as $robot => $valid_ip_ranges) {
230            if (str_contains($ua, $robot)) {
231                foreach ($valid_ip_ranges as $ip_range) {
232                    $range = IPFactory::parseRangeString($ip_range);
233
234                    if ($range instanceof RangeInterface && $range->contains($address)) {
235                        continue 2;
236                    }
237                }
238
239                return $this->response();
240            }
241        }
242
243        foreach (self::ROBOT_IP_FILES as $robot => $url) {
244            if (str_contains($ua, $robot)) {
245                $valid_ip_ranges = $this->fetchIpRangesForUrl($robot, $url);
246
247                foreach ($valid_ip_ranges as $ip_range) {
248                    $range = IPFactory::parseRangeString($ip_range);
249
250                    if ($range instanceof RangeInterface && $range->contains($address)) {
251                        continue 2;
252                    }
253                }
254
255                return $this->response();
256            }
257        }
258
259        foreach (self::ROBOT_ASNS as $robot => $asns) {
260            foreach ($asns as $asn) {
261                if (str_contains($ua, $robot)) {
262                    foreach ($this->fetchIpRangesForAsn($asn) as $range) {
263                        if ($range->contains($address)) {
264                            continue 2;
265                        }
266                    }
267
268                    return $this->response();
269                }
270            }
271        }
272
273        // Allow sites to block access from entire networks.
274        $block_asn = Validator::attributes($request)->string('block_asn', '');
275        preg_match_all('/(AS\d+)/', $block_asn, $matches);
276
277        foreach ($matches[1] as $asn) {
278            foreach ($this->fetchIpRangesForAsn($asn) as $range) {
279                if ($range->contains($address)) {
280                    return $this->response();
281                }
282            }
283        }
284
285        return $handler->handle($request);
286    }
287
288    /**
289     * Check that an IP address belongs to a robot operator using a forward/reverse DNS lookup.
290     *
291     * @param string        $ip
292     * @param array<string> $valid_domains
293     * @param bool          $reverse_only
294     *
295     * @return bool
296     */
297    private function checkRobotDNS(string $ip, array $valid_domains, bool $reverse_only): bool
298    {
299        $host = gethostbyaddr($ip);
300
301        if ($host === false) {
302            return false;
303        }
304
305        foreach ($valid_domains as $domain) {
306            if (str_ends_with($host, $domain)) {
307                return $reverse_only || $ip === gethostbyname($host);
308            }
309        }
310
311        return false;
312    }
313
314    /**
315     * Perform a whois search for an ASN.
316     *
317     * @param string $asn The autonomous system number to query
318     *
319     * @return array<RangeInterface>
320     */
321    private function fetchIpRangesForAsn(string $asn): array
322    {
323        return Registry::cache()->file()->remember('whois-asn-' . $asn, static function () use ($asn): array {
324            $mapper = static fn (AsnRouteInfo $route_info): ?RangeInterface => IPFactory::parseRangeString($route_info->route ?: $route_info->route6);
325
326            try {
327                $loader = new CurlLoader(self::WHOIS_TIMEOUT);
328                $whois  = new Whois($loader);
329                $info   = $whois->loadAsnInfo($asn);
330                $routes = $info->routes;
331                $ranges = array_map($mapper, $routes);
332
333                return array_filter($ranges);
334            } catch (Throwable) {
335                return [];
336            }
337        }, random_int(self::WHOIS_TTL_MIN, self::WHOIS_TTL_MAX));
338    }
339
340    /**
341     * Fetch a list of IP addresses from a remote file.
342     *
343     * @param string $ua
344     * @param string $url
345     *
346     * @return array<string>
347     */
348    private function fetchIpRangesForUrl(string $ua, string $url): array
349    {
350        return Registry::cache()->file()->remember('url-ip-list-' . $ua, static function () use ($url): array {
351            try {
352                $client   = new Client();
353                $response = $client->get($url, ['timeout' => 5]);
354                $contents = $response->getBody()->getContents();
355
356                preg_match_all(self::REGEX_IPV4, $contents, $matches);
357
358                return $matches[0];
359            } catch (GuzzleException) {
360                return [];
361            }
362        }, random_int(self::WHOIS_TTL_MIN, self::WHOIS_TTL_MAX));
363    }
364
365    /**
366     * @return ResponseInterface
367     */
368    private function response(): ResponseInterface
369    {
370        return response('Not acceptable', StatusCodeInterface::STATUS_NOT_ACCEPTABLE);
371    }
372}
373