<?php
/**
 * Page Analyzer — classifies any WordPress post/page into one of 18 named
 * page types using a priority-ordered signal stack.
 *
 * Signal priority (first match wins):
 *   1. User override  — _schema_ai_page_type post meta
 *   2. WordPress core — post type + static front page option
 *   3. URL slug       — regex patterns on the permalink path
 *   4. Content signals — Q&A ratio, service card count, NAP+map detection
 *   5. Fallback        — generic_page
 *
 * Works in both front-end and admin (save_post) contexts.
 *
 * @package Schema_AI
 * @since   1.1.4
 */

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

class Schema_AI_Page_Analyzer {

	/**
	 * Post meta key for manual page-type override.
	 */
	const OVERRIDE_META_KEY = '_schema_ai_page_type';

	/**
	 * Valid page type slugs.
	 *
	 * @var string[]
	 */
	public static $page_types = array(
		'homepage',
		'about_page',
		'contact_page',
		'location_page',
		'service_hub',
		'service_page',
		'blog_post',
		'blog_archive',
		'product_page',
		'dedicated_faq_page',
		'legal_page',
		// Added in v1.1.5 for Phase 6 site-type-aware schema.
		'event_page',
		'course_page',
		'recipe_page',
		'podcast_episode',
		'software_product_page',
		'portfolio_item',
		'generic_page',
	);

	/**
	 * Classify a post/page into one of the 12 page types.
	 *
	 * Falls back gracefully through the signal stack, stopping at the first
	 * confident match. Returns 'generic_page' when nothing else matches.
	 *
	 * @param int $post_id WordPress post ID.
	 * @return string One of the page type slugs in self::$page_types.
	 */
	public function classify( $post_id ) {
		$post = get_post( $post_id );
		if ( ! $post ) {
			return 'generic_page';
		}

		// Level 1 — explicit user override.
		$override = $this->check_user_override( $post_id );
		if ( $override ) {
			return $override;
		}

		// Level 2 — WordPress core / post type checks.
		$core = $this->check_core( $post_id, $post );
		if ( $core ) {
			return $core;
		}

		// Level 3 — URL slug patterns.
		$url = $this->check_url_patterns( $post_id );
		if ( $url ) {
			return $url;
		}

		// Level 4 — content signals.
		$content = $this->check_content_signals( $post );
		if ( $content ) {
			return $content;
		}

		// Level 5 — fallback.
		return 'generic_page';
	}

	// -------------------------------------------------------------------------
	// Level 1: User override
	// -------------------------------------------------------------------------

	/**
	 * Check for a manually-set page type override in post meta.
	 *
	 * @param int $post_id Post ID.
	 * @return string|null Valid page type or null.
	 */
	private function check_user_override( $post_id ) {
		$value = get_post_meta( $post_id, self::OVERRIDE_META_KEY, true );

		if ( $value && in_array( $value, self::$page_types, true ) ) {
			return $value;
		}

		return null;
	}

	// -------------------------------------------------------------------------
	// Level 2: WordPress core / post type signals
	// -------------------------------------------------------------------------

	/**
	 * Classify using WordPress core logic and post type.
	 *
	 * Works in both front-end (using is_* conditionals) and admin
	 * (using post data and options directly).
	 *
	 * @param int     $post_id Post ID.
	 * @param WP_Post $post    Post object.
	 * @return string|null Page type or null.
	 */
	private function check_core( $post_id, $post ) {
		// Front-end context — use WP conditionals where available.
		if ( ! is_admin() ) {
			if ( is_front_page() ) {
				return 'homepage';
			}
			if ( is_singular( 'post' ) ) {
				return 'blog_post';
			}
			if ( is_home() || is_archive() ) {
				return 'blog_archive';
			}
			if ( is_singular( 'product' ) ) {
				return 'product_page';
			}
		}

		// Admin / generation context — derive from post data.
		if ( 'post' === $post->post_type ) {
			return 'blog_post';
		}

		if ( 'product' === $post->post_type ) {
			return 'product_page';
		}

		// Event CPTs (The Events Calendar, EventON, WP Events Manager).
		if ( in_array( $post->post_type, array( 'tribe_events', 'em_event', 'mec-events', 'mp-event' ), true ) ) {
			return 'event_page';
		}

		// Course CPTs (LearnDash, LifterLMS, Tutor LMS, LearnPress).
		if ( in_array( $post->post_type, array( 'sfwd-courses', 'lp_course', 'tutor_course', 'course' ), true ) ) {
			return 'course_page';
		}

		// Podcast CPTs (Seriously Simple Podcasting, PowerPress, Castos).
		if ( in_array( $post->post_type, array( 'podcast', 'castos_podcast', 'powerpress' ), true ) ) {
			return 'podcast_episode';
		}

		// Portfolio CPTs (Jetpack, custom portfolio plugins).
		if ( in_array( $post->post_type, array( 'jetpack-portfolio', 'portfolio', 'work', 'project' ), true ) ) {
			return 'portfolio_item';
		}

		// Is this post set as the static front page?
		if ( 'page' === $post->post_type
			&& 'page' === get_option( 'show_on_front' )
			&& (int) get_option( 'page_on_front' ) === $post_id
		) {
			return 'homepage';
		}

		// Is this the blog posts index page?
		if ( 'page' === $post->post_type
			&& 'page' === get_option( 'show_on_front' )
			&& (int) get_option( 'page_for_posts' ) === $post_id
		) {
			return 'blog_archive';
		}

		return null;
	}

	// -------------------------------------------------------------------------
	// Level 3: URL slug patterns
	// -------------------------------------------------------------------------

	/**
	 * Map URL slug patterns to page types via regex.
	 *
	 * Operates on the URL path portion of the permalink so it is
	 * domain-independent.
	 *
	 * @param int $post_id Post ID.
	 * @return string|null Page type or null.
	 */
	private function check_url_patterns( $post_id ) {
		$permalink = get_permalink( $post_id );

		if ( ! $permalink ) {
			return null;
		}

		$path = wp_parse_url( $permalink, PHP_URL_PATH );

		if ( ! $path ) {
			return null;
		}

		$path = strtolower( rtrim( $path, '/' ) );

		// Order matters — more specific patterns first.
		$patterns = array(
			// Multi-location: /locations/city-name/ or /location/city-name/
			'#/locations?/[^/]+/?$#'              => 'location_page',
			// Service detail: /services/service-name/
			'#/services?/[^/]+/?$#'               => 'service_page',
			// Service hub: /services/ or /our-services/
			'#/(our-?)?services?/?$#'              => 'service_hub',
			// About
			'#/(about|about-us|who-we-are)/?$#'   => 'about_page',
			// Contact
			'#/(contact|contact-us|get-in-touch)/?$#' => 'contact_page',
			// FAQ
			'#/(faq|faqs|frequently-asked-questions)/?$#' => 'dedicated_faq_page',
			// Legal
			'#/(privacy|privacy-policy|terms|terms-of-service|terms-and-conditions|cookies|cookie-policy|disclaimer|legal)/?$#' => 'legal_page',
			// Events
			'#/events?/[^/]+/?$#'                 => 'event_page',
			// Courses
			'#/courses?/[^/]+/?$#'                => 'course_page',
			'#/lessons?/[^/]+/?$#'                => 'course_page',
			// Podcast episodes
			'#/episodes?/[^/]+/?$#'               => 'podcast_episode',
			'#/podcast/[^/]+/?$#'                 => 'podcast_episode',
			// Portfolio items
			'#/(portfolio|work|projects?)/[^/]+/?$#' => 'portfolio_item',
			// Software product/features/pricing (sub-pages only, not the root)
			'#/(features?|pricing)/?$#'           => 'software_product_page',
		);

		foreach ( $patterns as $regex => $type ) {
			if ( preg_match( $regex, $path ) ) {
				return $type;
			}
		}

		return null;
	}

	// -------------------------------------------------------------------------
	// Level 4: Content signals
	// -------------------------------------------------------------------------

	/**
	 * Infer page type from content structure.
	 *
	 * Checks:
	 * - Q&A heading ratio ≥ 60% → dedicated_faq_page
	 * - ≥ 3 linked service blocks → service_hub
	 * - NAP block + contact form + map embed → contact_page
	 * - Single H1 + service-flavoured content → service_page
	 *
	 * @param WP_Post $post Post object.
	 * @return string|null Page type or null.
	 */
	private function check_content_signals( $post ) {
		$content = $post->post_content;

		if ( empty( $content ) ) {
			return null;
		}

		// FAQ ratio: count H2/H3 ending in "?" vs total H2/H3.
		preg_match_all( '/<h[23][^>]*>(.*?)<\/h[23]>/is', $content, $headings );
		$total_h23    = count( $headings[1] );
		$question_h23 = 0;

		if ( $total_h23 > 0 ) {
			foreach ( $headings[1] as $text ) {
				if ( substr( rtrim( wp_strip_all_tags( $text ) ), -1 ) === '?' ) {
					$question_h23++;
				}
			}

			if ( ( $question_h23 / $total_h23 ) >= 0.6 ) {
				return 'dedicated_faq_page';
			}
		}

		// Service hub: ≥3 internal links that look like service pages.
		preg_match_all( '/<a\s[^>]*href=["\']([^"\']+)["\'][^>]*>(.*?)<\/a>/is', $content, $links );
		$service_links = 0;
		$site_url      = home_url();

		if ( ! empty( $links[1] ) ) {
			foreach ( $links[1] as $href ) {
				if ( strpos( $href, $site_url ) === 0 || strpos( $href, '/' ) === 0 ) {
					$link_path = strtolower( wp_parse_url( $href, PHP_URL_PATH ) ?? '' );
					// Any internal sub-page link under a services-type path.
					if ( preg_match( '#/services?/#', $link_path ) ) {
						$service_links++;
					}
				}
			}
		}

		if ( $service_links >= 3 ) {
			return 'service_hub';
		}

		// Contact page: has a contact-form shortcode/block AND phone/address indicators.
		$has_form    = ( false !== stripos( $content, 'contact-form' )
						|| false !== stripos( $content, 'wpforms' )
						|| false !== stripos( $content, 'gravity_form' )
						|| false !== stripos( $content, 'cf7' ) );
		$has_map     = ( false !== stripos( $content, 'google.com/maps' )
						|| false !== stripos( $content, 'maps.google' ) );
		$has_address = (bool) preg_match( '/\d{1,5}\s+\w[\w\s,\.]+(?:st|street|ave|avenue|blvd|boulevard|rd|road|dr|drive|ln|lane|way|court|ct|pl|place)/i', wp_strip_all_tags( $content ) );

		if ( $has_form && ( $has_map || $has_address ) ) {
			return 'contact_page';
		}

		// Service page: single H1 + service/pricing/CTA language.
		$service_keywords = array( 'service', 'contractor', 'repair', 'install', 'replace', 'estimate', 'quote', 'hire' );
		$text             = strtolower( wp_strip_all_tags( $content ) );
		$keyword_matches  = 0;

		foreach ( $service_keywords as $kw ) {
			if ( false !== strpos( $text, $kw ) ) {
				$keyword_matches++;
			}
		}

		if ( $keyword_matches >= 3 ) {
			return 'service_page';
		}

		// Recipe: ingredients + cooking instructions signals.
		$recipe_keywords = array( 'ingredients', 'instructions', 'prep time', 'cook time', 'servings', 'recipe', 'tablespoon', 'teaspoon', 'preheat' );
		$recipe_matches  = 0;
		foreach ( $recipe_keywords as $kw ) {
			if ( false !== strpos( $text, $kw ) ) {
				$recipe_matches++;
			}
		}
		if ( $recipe_matches >= 4 ) {
			return 'recipe_page';
		}

		// Podcast episode: audio embed + episode language.
		$has_audio    = ( false !== stripos( $content, 'powerpress' )
						|| false !== stripos( $content, 'simplecast' )
						|| false !== stripos( $content, 'buzzsprout' )
						|| false !== stripos( $content, 'podbean' )
						|| preg_match( '/<audio[^>]+>/i', $content ) );
		$podcast_keywords = array( 'episode', 'listen', 'transcript', 'show notes', 'subscribe' );
		$podcast_matches  = 0;
		foreach ( $podcast_keywords as $kw ) {
			if ( false !== strpos( $text, $kw ) ) {
				$podcast_matches++;
			}
		}
		if ( $has_audio && $podcast_matches >= 2 ) {
			return 'podcast_episode';
		}

		// Course page: curriculum/enrollment language.
		$course_keywords = array( 'enroll', 'curriculum', 'lesson', 'module', 'course', 'certificate', 'syllabus', 'instructor' );
		$course_matches  = 0;
		foreach ( $course_keywords as $kw ) {
			if ( false !== strpos( $text, $kw ) ) {
				$course_matches++;
			}
		}
		if ( $course_matches >= 4 ) {
			return 'course_page';
		}

		return null;
	}

	// -------------------------------------------------------------------------
	// Utility
	// -------------------------------------------------------------------------

	/**
	 * Save a manual page-type override to post meta.
	 *
	 * Called from the meta box save.
	 *
	 * @param int    $post_id  Post ID.
	 * @param string $type     One of the page type slugs, or empty to clear.
	 * @return void
	 */
	public function save_override( $post_id, $type ) {
		if ( empty( $type ) || ! in_array( $type, self::$page_types, true ) ) {
			delete_post_meta( $post_id, self::OVERRIDE_META_KEY );
		} else {
			update_post_meta( $post_id, self::OVERRIDE_META_KEY, $type );
		}
	}
}