<?php
namespace Yipresser\AdminOptimizer\Modules;

// Exit if accessed directly.
if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

/**
 * Static, filterable registry of known AI/search bot user agents grouped
 * by category (search / agent / training).
 */
final class AI_Bot_Registry {

	const CATEGORY_SEARCH   = 'search';
	const CATEGORY_AGENT    = 'agent';
	const CATEGORY_TRAINING = 'training';

	/**
	 * Maps each category to its Content-Signal field name.
	 * Order here is the order fields are emitted in the generated
	 * Content-Signal line.
	 */
	const CONTENT_SIGNAL_FIELD_MAP = [
		self::CATEGORY_SEARCH   => 'search',
		self::CATEGORY_TRAINING => 'ai-train',
		self::CATEGORY_AGENT    => 'ai-input',
	];

	/** Bumped whenever the bundled bot list below changes. Display-only. */
	const REGISTRY_VERSION = '2026.09.1';

	/**
	 * Get the full bot list, filterable by third parties.
	 *
	 * @return array<int, array{user_agent: string, category: string, operator: string, verified: bool}>
	 */
	public static function get_bots(): array {
		return apply_filters( 'adminoptim_ai_bot_registry', self::bundled_bots() );
	}

	/**
	 * Bundled, hand-maintained bot list.
	 *
	 * @return array<int, array{user_agent: string, category: string, operator: string, verified: bool}>
	 */
	private static function bundled_bots(): array {
		return [
			// -- Search --
			[
				'user_agent' => 'Googlebot',
				'category'   => self::CATEGORY_SEARCH,
				'operator'   => 'Google',
				'verified'   => true,
			],
			[
				'user_agent' => 'Bingbot',
				'category'   => self::CATEGORY_SEARCH,
				'operator'   => 'Microsoft',
				'verified'   => true,
			],
			[
				'user_agent' => 'DuckDuckBot',
				'category'   => self::CATEGORY_SEARCH,
				'operator'   => 'DuckDuckGo',
				'verified'   => true,
			],
			[
				'user_agent' => 'Applebot',
				'category'   => self::CATEGORY_SEARCH,
				'operator'   => 'Apple',
				'verified'   => true,
			],
			[
				'user_agent' => 'OAI-SearchBot',
				'category'   => self::CATEGORY_SEARCH,
				'operator'   => 'OpenAI',
				'verified'   => true,
			],

			// -- Agent --
			[
				'user_agent' => 'ChatGPT-User',
				'category'   => self::CATEGORY_AGENT,
				'operator'   => 'OpenAI',
				'verified'   => true,
			],
			[
				'user_agent' => 'PerplexityBot',
				'category'   => self::CATEGORY_AGENT,
				'operator'   => 'Perplexity',
				'verified'   => true,
			],
			[
				'user_agent' => 'Claude-User',
				'category'   => self::CATEGORY_AGENT,
				'operator'   => 'Anthropic',
				'verified'   => true,
			],

			// -- Training --
			[
				'user_agent' => 'GPTBot',
				'category'   => self::CATEGORY_TRAINING,
				'operator'   => 'OpenAI',
				'verified'   => true,
			],
			[
				'user_agent' => 'Google-Extended',
				'category'   => self::CATEGORY_TRAINING,
				'operator'   => 'Google',
				'verified'   => true,
			],
			[
				'user_agent' => 'CCBot',
				'category'   => self::CATEGORY_TRAINING,
				'operator'   => 'Common Crawl',
				'verified'   => false,
			],
			[
				'user_agent' => 'Bytespider',
				'category'   => self::CATEGORY_TRAINING,
				'operator'   => 'ByteDance',
				'verified'   => false,
			],
			[
				'user_agent' => 'anthropic-ai',
				'category'   => self::CATEGORY_TRAINING,
				'operator'   => 'Anthropic',
				'verified'   => true,
			],
			[
				'user_agent' => 'cohere-ai',
				'category'   => self::CATEGORY_TRAINING,
				'operator'   => 'Cohere',
				'verified'   => true,
			],
			[
				'user_agent' => 'Diffbot',
				'category'   => self::CATEGORY_TRAINING,
				'operator'   => 'Diffbot',
				'verified'   => false,
			],
			[
				'user_agent' => 'omgili',
				'category'   => self::CATEGORY_TRAINING,
				'operator'   => 'Webz.io',
				'verified'   => false,
			],
			[
				'user_agent' => 'omgilibot',
				'category'   => self::CATEGORY_TRAINING,
				'operator'   => 'Webz.io',
				'verified'   => false,
			],
		];
	}

	/**
	 * Get bots belonging to a single category.
	 *
	 * @param string $category One of the CATEGORY_* constants.
	 * @return array<int, array{user_agent: string, category: string, operator: string, verified: bool}>
	 */
	public static function get_by_category( string $category ): array {
		return array_values(
			array_filter(
				self::get_bots(),
				fn( $bot ) => $bot['category'] === $category
			)
		);
	}
}