hexydec\agentzero\crawlers::get()

Complexity

Conditions	11
Paths	1

Size

Total Lines	315
Code Lines	246

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

<?php
	public static function get() {
		$fn = [
			'search' => fn (string $value) : array => self::getApp($value, ['category' => 'search']),
			'ads' => fn (string $value) : array => self::getApp($value, ['category' => 'ads']),
			'validator' => fn (string $value) : array => self::getApp($value, ['category' => 'validator']),
			'feed' => fn (string $value) : array => self::getApp($value, ['category' => 'feed']),
			'crawler' => fn (string $value) : array => self::getApp($value, ['category' => 'crawler']),
			'monitor' => fn (string $value) : array => self::getApp($value, ['category' => 'monitor']),
			'scraper' => fn (string $value) : array => self::getApp($value, ['category' => 'scraper']),
			'map' => function (string $value, int $i, array $tokens) : ?array {
				if (!\str_contains($value, '://')) { // bot will be in the URL
					$parts = \explode('/', $value, 2);

					// special case
					if (\strcasecmp($parts[0], 'Spider') === 0) {
						$ua = \implode(' ', $tokens);
						if (\mb_stripos($ua, 'Screaming Frog SEO Spider') === 0) {
							$parts[0] = 'Screaming Frog SEO Spider';
						} elseif (\mb_stripos($ua, 'Sogou web spider') === 0) {
							$parts[0] = 'Sogou web spider';
						}
					}
					$category = [
						'yacybot' => 'search',
						'Googlebot' => 'search',
						'Googlebot-Mobile' => 'search',
						'Googlebot-Image' => 'search',
						'Googlebot-Video' => 'search',
						'Googlebot-News' => 'search',
						'Storebot-Google' => 'search',
						'AdsBot-Google' => 'ads',
						'AdsBot-Google-Mobile' => 'ads',
						'Bingbot' => 'search',
						'bingbot' => 'search',
						'adidxbot' => 'ads',
						'DuckDuckBot' => 'search',
						'DuckDuckGo-Favicons-Bot' => 'search',
						'coccocbot-image' => 'search',
						'coccocbot-web' => 'search',
						'Baiduspider' => 'search',
						'Applebot' => 'search',
						'YandexBot' => 'search',
						'MJ12bot' => 'search',
						'Mail.RU_Bot' => 'search',
						'HaosouSpider' => 'search',
						'360Spider' => 'search',
						'Exabot' => 'search',
						'Sogou web spider' => 'search',
						'UptimeRobot' => 'monitor',
						'PetalBot' => 'search',
						'Screaming Frog SEO Spider' => 'crawler',
						'Twitterbot' => 'feed',
						'Xbot' => 'feed',
						'Discordbot' => 'feed',
						'PRTGCloudBot' => 'monitor',
						'Bytespider' => 'search',
						'LinkedInBot' => 'feed',
						'PaperLiBot' => 'feed',
						'bitlybot' => 'feed',
						'TinEye-bot' => 'search',
						'Pinterestbot' => 'feed',
						'WebCrawler' => 'crawler',
						'webprosbot' => 'crawler',
						'GuzzleHttp' => 'scraper',
						'TelegramBot' => 'feed',
						'Ruby' => 'scraper',
						'SEMrushBot' => 'crawler',
						'Mediatoolkitbot' => 'crawler'
					];
					return self::getApp($value, [
						'category' => $category[$parts[0]] ?? null,
						'app' => $parts[0]
					]);
				}
				return null;
			}
		];
		return [
			'Yahoo! Slurp' => [
				'match' => 'exact',
				'categories' => $fn['search']
			],
			'facebookexternalhit/' => [
				'match' => 'start',
				'categories' => $fn['feed']
			],
			'Google-Site-Verification/' => [
				'match' => 'start',
				'categories' => $fn['validator']
			],
			'Google-InspectionTool/' => [
				'match' => 'start',
				'categories' => $fn['search']
			],
			'Mediapartners-Google' => [
				'match' => 'start',
				'categories' => $fn['search']
			],
			'FeedFetcher-Google' => [
				'match' => 'exact',
				'categories' => $fn['feed']
			],
			'GoogleProducer' => [
				'match' => 'exact',
				'categories' => $fn['feed']
			],
			'CFNetwork/' => [
				'match' => 'start',
				'categories' => $fn['feed']
			],
			'Siteimprove.com' => [
				'match' => 'any',
				'categories' => $fn['crawler']
			],
			'Google Page Speed Insights' => [
				'match' => 'exact',
				'categories' => $fn['validator']
			],
			'Qwantify' => [
				'match' => 'start',
				'categories' => $fn['search']
			],
			'okhttp' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'python' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'Nessus' => [
				'match' => 'start',
				'categories' => $fn['monitor']
			],
			'Chrome-Lighthouse' => [
				'match' => 'start',
				'categories' => $fn['validator']
			],
			'PingdomTMS/' => [
				'match' => 'start',
				'categories' => $fn['monitor']
			],
			'Pingdom.com' => [
				'match' => 'start',
				'categories' => function (string $value) : array {
					$version = \explode('_', \trim($value, '_'));
					return [
						'type' => 'robot',
						'category' => 'monitor',
						'app' => 'Pingdom.com Bot',
						'appversion' => \end($version)
					];
				}
			],
			'proximic' => [
				'match' => 'exact',
				'categories' => $fn['ads']
			],
			'WordPress' => [
				'match' => 'start',
				'categories' => $fn['monitor']
			],
			'PRTG Network Monitor' => [
				'match' => 'exact',
				'categories' => $fn['monitor']
			],
			'Site24x7' => [
				'match' => 'exact',
				'categories' => $fn['monitor']
			],
			'adbeat.com' => [
				'match' => 'start',
				'categories' => fn (string $value) : array => [
					'type' => 'robot',
					'category' => 'ads',
					'app' => 'Adbeat',
					'url' => $value
				]
			],
			'MicrosoftPreview/' => [
				'match' => 'start',
				'categories' => $fn['feed']
			],
			'Let\'s Encrypt validation server' => [
				'match' => 'exact',
				'categories' => $fn['validator']
			],
			'Expanse' => [
				'match' => 'start',
				'categories' => $fn['crawler']
			],
			'WhatsApp/' => [
				'match' => 'start',
				'categories' => $fn['feed']
			],
			'Apache-HttpClient/' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'eCairn-Grabber/' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'SEOkicks' => [
				'match' => 'exact',
				'categories' => $fn['crawler']
			],
			'PostmanRuntime/' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'axios/' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'Rogerbot/' => [
				'match' => 'start',
				'categories' => $fn['crawler']
			],
			'Go-http-client/' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'DashLinkPreviews/' => [
				'match' => 'start',
				'categories' => $fn['feed']
			],
			'Microsoft Office' => [
				'match' => 'start',
				'categories' => function (string $value, int $i, array $tokens) : array {
					$data = [
						'type' => 'robot',
						'category' => 'feed'
					];
					if (\str_contains($value, '/')) {
						foreach (\array_slice($tokens, $i + 1) AS $item) {
							if (\str_starts_with($item, 'Microsoft ')) {
								$parts = \explode(' ', $item);
								$data['app'] = $parts[0].' '.$parts[1];
								if (isset($parts[2])) {
									$data['appversion'] = $parts[2];
								}
								break;
							}
						}
						if (!isset($data['app'])) {
							$parts = \explode('/', $value, 2);
							$data['app'] = $parts[0];
							if (!isset($data['appversion'])) {
								$data['appversion'] = $parts[1];
							}
						}
					} else {
						$parts = \explode(' ', $value);
						$data['app'] = $parts[0].' '.$parts[1].' '.$parts[2];
						$data['appversion'] = $parts[3] ?? null;
					}
					return $data;
				}
			],
			'PycURL/' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'lua-resty-http/' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'Snapchat/' => [
				'match' => 'start',
				'categories' => $fn['feed']
			],
			'HTTPClient/' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'WhatsApp/' => [
				'match' => 'any',
				'categories' => $fn['feed']
			],
			'Hootsuite-Authoring/' => [
				'match' => 'start',
				'categories' => $fn['feed']
			],
			'ApacheBench/' => [
				'match' => 'start',
				'categories' => $fn['validator']
			],
			'Asana/' => [
				'match' => 'start',
				'categories' => $fn['feed']
			],
			'Java/' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'curl/' => [
				'match' => 'start',
				'categories' => $fn['scraper']
			],
			'feed' => [
				'match' => 'any',
				'categories' => $fn['feed']
			],
			'spider' => [
				'match' => 'any',
				'categories' => $fn['map']
			],
			'crawler' => [
				'match' => 'any',
				'categories' => $fn['map']
			],
			'bot' => [
				'match' => 'any',
				'categories' => $fn['map']
			],
}