xref: /webtrees/app/Http/Middleware/BadBotBlocker.php (revision 45d54b043c9a7568180bab74fe47edd4d856511a) 
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        'Bytespider',
74        'DataForSEO',
75        'DataForSeoBot', // https://dataforseo.com/dataforseo-bot
76        'DotBot',
77        'GPTBot', // Collects training data for ChatGPT
78        'Grapeshot',
79        'Honolulu-bot', // Aggressive crawer, no info available
80        'ia_archiver',
81        'linabot', // Aggressive crawer, no info available
82        'Linguee',
83        'MegaIndex.ru',
84        'MJ12bot',
85        'netEstate NE',
86        'panscient',
87        'PetalBot',
88        'proximic',
89        'SeekportBot', // Pretends to be a search engine - but isn't
90        'SemrushBot',
91        'serpstatbot',
92        'SEOkicks',
93        'SiteKiosk',
94        'TinyTestBot',
95        'Turnitin',
96        'wp_is_mobile', // Nothing to do with wordpress
97        'XoviBot',
98        'YisouSpider',
99        'ZoominfoBot',
100    ];
101
102    /**
103     * Some search engines use reverse/forward DNS to verify the IP address.
104     *
105     * @see https://developer.amazon.com/support/amazonbot
106     * @see https://support.google.com/webmasters/answer/80553?hl=en
107     * @see https://www.bing.com/webmaster/help/which-crawlers-does-bing-use-8c184ec0
108     * @see https://www.bing.com/webmaster/help/how-to-verify-bingbot-3905dc26
109     * @see https://yandex.com/support/webmaster/robot-workings/check-yandex-robots.html
110     * @see https://www.mojeek.com/bot.html
111     * @see https://support.apple.com/en-gb/HT204683
112     */
113    private const ROBOT_REV_FWD_DNS = [
114        'Amazonbot'        => ['.crawl.amazon.com'],
115        'Applebot'         => ['.applebot.apple.com'],
116        'BingPreview'      => ['.search.msn.com'],
117        'Google'           => ['.google.com', '.googlebot.com'],
118        'Mail.RU_Bot'      => ['.mail.ru'],
119        'MicrosoftPreview' => ['.search.msn.com'],
120        'MojeekBot'        => ['.mojeek.com'],
121        'Qwantify'         => ['.search.qwant.com'],
122        'Sogou'            => ['.crawl.sogou.com'],
123        'Yahoo'            => ['.crawl.yahoo.net'],
124        'Yandex'           => ['.yandex.ru', '.yandex.net', '.yandex.com'],
125        'bingbot'          => ['.search.msn.com'],
126        'msnbot'           => ['.search.msn.com'],
127    ];
128
129    /**
130     * Some search engines only use reverse DNS to verify the IP address.
131     *
132     * @see https://help.baidu.com/question?prod_id=99&class=0&id=3001
133     * @see https://napoveda.seznam.cz/en/full-text-search/seznambot-crawler
134     * @see https://www.ionos.de/terms-gtc/faq-crawler
135     */
136    private const ROBOT_REV_ONLY_DNS = [
137        'Baiduspider' => ['.baidu.com', '.baidu.jp'],
138        'FreshBot'    => ['.seznam.cz'],
139        'IonCrawl'    => ['.1und1.org'],
140        'Neevabot'    => ['.neeva.com'],
141        'SeznamBot'   => ['.seznam.cz'],
142    ];
143
144    /**
145     * Some search engines operate from designated IP addresses.
146     *
147     * @see https://www.apple.com/go/applebot
148     * @see https://help.duckduckgo.com/duckduckgo-help-pages/results/duckduckbot
149     */
150    private const ROBOT_IPS = [
151        'AppleBot'    => [
152            '17.0.0.0/8',
153        ],
154        'Ask Jeeves'  => [
155            '65.214.45.143',
156            '65.214.45.148',
157            '66.235.124.192',
158            '66.235.124.7',
159            '66.235.124.101',
160            '66.235.124.193',
161            '66.235.124.73',
162            '66.235.124.196',
163            '66.235.124.74',
164            '63.123.238.8',
165            '202.143.148.61',
166        ],
167        'DuckDuckBot' => [
168            '23.21.227.69',
169            '50.16.241.113',
170            '50.16.241.114',
171            '50.16.241.117',
172            '50.16.247.234',
173            '52.204.97.54',
174            '52.5.190.19',
175            '54.197.234.188',
176            '54.208.100.253',
177            '54.208.102.37',
178            '107.21.1.8',
179        ],
180    ];
181
182    /**
183     * Some search engines operate from designated IP addresses.
184     *
185     * @see https://bot.seekport.com/
186     */
187    private const ROBOT_IP_FILES = [
188        'SeekportBot' => 'https://bot.seekport.com/seekportbot_ips.txt',
189    ];
190
191    /**
192     * Some search engines operate from within a designated autonomous system.
193     *
194     * @see https://developers.facebook.com/docs/sharing/webmasters/crawler
195     * @see https://www.facebook.com/peering/
196     */
197    private const ROBOT_ASNS = [
198        'facebook' => ['AS32934', 'AS63293'],
199        'twitter'  => ['AS13414'],
200    ];
201
202    /**
203     * @param ServerRequestInterface  $request
204     * @param RequestHandlerInterface $handler
205     *
206     * @return ResponseInterface
207     */
208    public function process(ServerRequestInterface $request, RequestHandlerInterface $handler): ResponseInterface
209    {
210        $ua      = Validator::serverParams($request)->string('HTTP_USER_AGENT', '');
211        $ip      = Validator::attributes($request)->string('client-ip');
212        $address = IPFactory::parseAddressString($ip);
213        assert($address instanceof AddressInterface);
214
215        foreach (self::BAD_ROBOTS as $robot) {
216            if (str_contains($ua, $robot)) {
217                return $this->response();
218            }
219        }
220
221        foreach (self::ROBOT_REV_FWD_DNS as $robot => $valid_domains) {
222            if (str_contains($ua, $robot) && !$this->checkRobotDNS($ip, $valid_domains, false)) {
223                return $this->response();
224            }
225        }
226
227        foreach (self::ROBOT_REV_ONLY_DNS as $robot => $valid_domains) {
228            if (str_contains($ua, $robot) && !$this->checkRobotDNS($ip, $valid_domains, true)) {
229                return $this->response();
230            }
231        }
232
233        foreach (self::ROBOT_IPS as $robot => $valid_ip_ranges) {
234            if (str_contains($ua, $robot)) {
235                foreach ($valid_ip_ranges as $ip_range) {
236                    $range = IPFactory::parseRangeString($ip_range);
237
238                    if ($range instanceof RangeInterface && $range->contains($address)) {
239                        continue 2;
240                    }
241                }
242
243                return $this->response();
244            }
245        }
246
247        foreach (self::ROBOT_IP_FILES as $robot => $url) {
248            if (str_contains($ua, $robot)) {
249                $valid_ip_ranges = $this->fetchIpRangesForUrl($robot, $url);
250
251                foreach ($valid_ip_ranges as $ip_range) {
252                    $range = IPFactory::parseRangeString($ip_range);
253
254                    if ($range instanceof RangeInterface && $range->contains($address)) {
255                        continue 2;
256                    }
257                }
258
259                return $this->response();
260            }
261        }
262
263        foreach (self::ROBOT_ASNS as $robot => $asns) {
264            foreach ($asns as $asn) {
265                if (str_contains($ua, $robot)) {
266                    foreach ($this->fetchIpRangesForAsn($asn) as $range) {
267                        if ($range->contains($address)) {
268                            continue 2;
269                        }
270                    }
271
272                    return $this->response();
273                }
274            }
275        }
276
277        // Allow sites to block access from entire networks.
278        $block_asn = Validator::attributes($request)->string('block_asn', '');
279        preg_match_all('/(AS\d+)/', $block_asn, $matches);
280
281        foreach ($matches[1] as $asn) {
282            foreach ($this->fetchIpRangesForAsn($asn) as $range) {
283                if ($range->contains($address)) {
284                    return $this->response();
285                }
286            }
287        }
288
289        return $handler->handle($request);
290    }
291
292    /**
293     * Check that an IP address belongs to a robot operator using a forward/reverse DNS lookup.
294     *
295     * @param string        $ip
296     * @param array<string> $valid_domains
297     * @param bool          $reverse_only
298     *
299     * @return bool
300     */
301    private function checkRobotDNS(string $ip, array $valid_domains, bool $reverse_only): bool
302    {
303        $host = gethostbyaddr($ip);
304
305        if ($host === false) {
306            return false;
307        }
308
309        foreach ($valid_domains as $domain) {
310            if (str_ends_with($host, $domain)) {
311                return $reverse_only || $ip === gethostbyname($host);
312            }
313        }
314
315        return false;
316    }
317
318    /**
319     * Perform a whois search for an ASN.
320     *
321     * @param string $asn The autonomous system number to query
322     *
323     * @return array<RangeInterface>
324     */
325    private function fetchIpRangesForAsn(string $asn): array
326    {
327        return Registry::cache()->file()->remember('whois-asn-' . $asn, static function () use ($asn): array {
328            $mapper = static fn (AsnRouteInfo $route_info): ?RangeInterface => IPFactory::parseRangeString($route_info->route ?: $route_info->route6);
329
330            try {
331                $loader = new CurlLoader(self::WHOIS_TIMEOUT);
332                $whois  = new Whois($loader);
333                $info   = $whois->loadAsnInfo($asn);
334                $routes = $info->routes;
335                $ranges = array_map($mapper, $routes);
336
337                return array_filter($ranges);
338            } catch (Throwable) {
339                return [];
340            }
341        }, random_int(self::WHOIS_TTL_MIN, self::WHOIS_TTL_MAX));
342    }
343
344    /**
345     * Fetch a list of IP addresses from a remote file.
346     *
347     * @param string $ua
348     * @param string $url
349     *
350     * @return array<string>
351     */
352    private function fetchIpRangesForUrl(string $ua, string $url): array
353    {
354        return Registry::cache()->file()->remember('url-ip-list-' . $ua, static function () use ($url): array {
355            try {
356                $client   = new Client();
357                $response = $client->get($url, ['timeout' => 5]);
358                $contents = $response->getBody()->getContents();
359
360                preg_match_all(self::REGEX_IPV4, $contents, $matches);
361
362                return $matches[0];
363            } catch (GuzzleException) {
364                return [];
365            }
366        }, random_int(self::WHOIS_TTL_MIN, self::WHOIS_TTL_MAX));
367    }
368
369    /**
370     * @return ResponseInterface
371     */
372    private function response(): ResponseInterface
373    {
374        return response('Not acceptable', StatusCodeInterface::STATUS_NOT_ACCEPTABLE);
375    }
376}
377