<?php
/**
 * Published agent tokens and the narrow browser check.
 *
 * Tokens are the product names operators publish in their user-agent.
 * A normal phone brand that happens to contain the letters "bot" is not on this list.
 *
 * @package AgentStamp
 */

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

class AgentStamp_Agents {

	/**
	 * Agents that fetch or shop for a person. Longest tokens first so ChatGPT-User wins over a shorter neighbor.
	 *
	 * @return array<string,string> token => display name
	 */
	private static function ai_tokens() {
		return array(
			'ChatGPT-User'         => 'ChatGPT',
			'Claude-User'          => 'Claude',
			'Perplexity-User'      => 'Perplexity',
			'Google-Agent'         => 'Google Agent',
			'Meta-ExternalFetcher' => 'Meta',
			'MistralAI-User'       => 'Mistral',
			'DuckAssistBot'        => 'DuckDuckGo',
			'GrokAgent'            => 'Grok',
		);
	}

	/**
	 * AI crawlers. They index or train. They do not check out, so a checkout that names one is a claim.
	 *
	 * @return array<string,string> token => display name
	 */
	public static function ai_crawler_tokens() {
		return array(
			'OAI-SearchBot'      => 'OpenAI Search',
			'OAI-AdsBot'         => 'OpenAI Ads',
			'GPTBot'             => 'OpenAI GPTBot',
			'Claude-SearchBot'   => 'Claude Search',
			'ClaudeBot'          => 'ClaudeBot',
			'anthropic-ai'       => 'anthropic-ai',
			'PerplexityBot'      => 'PerplexityBot',
			'Google-Extended'    => 'Google-Extended',
			'GoogleOther'        => 'GoogleOther',
			'Applebot-Extended'  => 'Applebot-Extended',
			'Meta-ExternalAgent' => 'Meta crawler',
			'Amazonbot'          => 'Amazonbot',
			'Bytespider'         => 'Bytespider',
			'cohere-ai'          => 'Cohere',
			'CCBot'              => 'Common Crawl',
			'Diffbot'            => 'Diffbot',
			'YouBot'             => 'You.com',
		);
	}

	/**
	 * @param string $user_agent
	 * @return string Crawler display name, or empty.
	 */
	public static function match_ai_crawler( $user_agent ) {
		foreach ( self::ai_crawler_tokens() as $token => $name ) {
			if ( self::contains_token( $user_agent, $token ) ) {
				return $name;
			}
		}
		return '';
	}

	/**
	 * Crawlers that are not shopping agents. They must not be stamped Human.
	 *
	 * @return string[]
	 */
	public static function other_bot_tokens() {
		return array(
			'Googlebot',
			'AdsBot-Google',
			'Bingbot',
			'DuckDuckBot',
			'Baiduspider',
			'YandexBot',
			'AhrefsBot',
			'SemrushBot',
			'PetalBot',
			'facebookexternalhit',
			'Twitterbot',
			'LinkedInBot',
			'Slackbot',
			'Embedly',
			'Qwantify',
		);
	}

	/**
	 * @param string $user_agent
	 * @return string Display name, or empty when the UA is not a published agent.
	 */
	public static function match_ai_ua( $user_agent ) {
		foreach ( self::ai_tokens() as $token => $name ) {
			if ( self::contains_token( $user_agent, $token ) ) {
				return $name;
			}
		}
		$prefixed = self::match_agent_prefix( $user_agent );
		if ( '' !== $prefixed ) {
			return $prefixed;
		}
		return '';
	}

	/**
	 * Amazon's agent form: Agent/Muse. A URL path /Agent/ does not count.
	 * GrokAgent/1.0 is named from the token list, not as "1.0".
	 *
	 * @param string $user_agent
	 * @return string
	 */
	private static function match_agent_prefix( $user_agent ) {
		if ( ! is_string( $user_agent ) || '' === $user_agent ) {
			return '';
		}
		if ( ! preg_match( '/(?:^|[;(\s])Agent\/([A-Za-z][A-Za-z0-9._-]{0,40})/i', $user_agent, $match ) ) {
			return '';
		}
		return $match[1];
	}

	/**
	 * @param string $user_agent
	 * @return bool
	 */
	public static function is_other_bot( $user_agent ) {
		foreach ( self::other_bot_tokens() as $token ) {
			if ( self::contains_token( $user_agent, $token ) ) {
				return true;
			}
		}
		return false;
	}

	/**
	 * @param string $user_agent
	 * @return bool
	 */
	public static function is_human_browser( $user_agent ) {
		if ( ! is_string( $user_agent ) || strlen( $user_agent ) < 20 ) {
			return false;
		}
		if ( 0 !== strpos( $user_agent, 'Mozilla/5.0' ) ) {
			return false;
		}
		$browser = false;
		foreach ( array( 'Chrome/', 'CriOS/', 'Safari/', 'Firefox/', 'FxiOS/', 'Edg/', 'OPR/', 'SamsungBrowser/' ) as $token ) {
			if ( false !== strpos( $user_agent, $token ) ) {
				$browser = true;
				break;
			}
		}
		// In-app browsers (Instagram, Facebook, Google app, TikTok, Pinterest, Snapchat, LINE) leave out
		// "Safari/" on iOS. They are people shopping from a link.
		if ( ! $browser && false !== strpos( $user_agent, 'AppleWebKit/' ) && preg_match( '/(?:FBAN|FBAV|Instagram|GSA\/|Line\/|Pinterest|Snapchat|TikTok|musical_ly|BytedanceWebview)/', $user_agent ) ) {
			$browser = true;
		}
		if ( ! $browser ) {
			return false;
		}
		foreach ( array( 'HeadlessChrome', 'Headless', 'PhantomJS', 'Selenium', 'Puppeteer', 'Playwright' ) as $marker ) {
			if ( false !== stripos( $user_agent, $marker ) ) {
				return false;
			}
		}
		return true;
	}

	/**
	 * Display name for the host that published the signing key.
	 *
	 * @param string $host
	 * @return string
	 */
	public static function name_for_host( $host ) {
		$host = strtolower( (string) $host );
		$map  = array(
			'openai.com'     => 'ChatGPT',
			'chatgpt.com'    => 'ChatGPT',
			'oaistatic.com'  => 'ChatGPT',
			'anthropic.com'  => 'Claude',
			'claude.ai'      => 'Claude',
			'perplexity.ai'  => 'Perplexity',
			'google.com'     => 'Google',
			'googlebot.com'  => 'Google',
			'agent.bot.goog' => 'Google',
			'agentpay-key-directory.mastercard.com' => 'Mastercard Agent Pay',
			'mastercard.com' => 'Mastercard Agent Pay',
			'visa.com'       => 'Visa Intelligent Commerce',
			'amazon.com'     => 'Amazon',
			'amazonbot.amazon' => 'Amazon',
			'facebook.com'   => 'Meta',
			'meta.com'       => 'Meta',
			'apple.com'      => 'Apple',
			'bytedance.com'  => 'ByteDance',
		);
		foreach ( $map as $suffix => $name ) {
			if ( $host === $suffix || substr( $host, -strlen( '.' . $suffix ) ) === '.' . $suffix ) {
				return $name;
			}
		}
		return $host;
	}

	/**
	 * @param string $haystack
	 * @param string $token
	 * @return bool
	 */
	private static function contains_token( $haystack, $token ) {
		if ( ! is_string( $haystack ) || '' === $haystack ) {
			return false;
		}
		return false !== stripos( $haystack, $token );
	}
}