<?php
/**
 * Which pages a scan looks at.
 *
 * @package WorkMoreCookieCrumb
 */

namespace WorkMore\CookieCrumb\Scan;

/**
 * A sample, not a crawl: the front page, the newest entry of every kind of
 * content the site publishes, and the pages where shops and sites tend to
 * load something extra — plus the addresses the owner adds. Each is opened
 * once in the owner's own browser, so the list stays short.
 */
final class Pages {

	/**
	 * Pages one scan opens at most.
	 */
	public const MAX = 12;

	/**
	 * The option that keeps the addresses the owner added.
	 */
	public const MORE_OPTION = 'wmcc_scan_pages';

	/**
	 * The pages to scan: the sample, then the owner's own.
	 *
	 * @return array<int, string> Addresses on this site, each once.
	 */
	public static function all(): array {
		return array_slice( self::onSite( array_merge( self::sample(), self::more() ) ), 0, self::MAX );
	}

	/**
	 * The sample of the site's pages.
	 *
	 * @return array<int, string>
	 */
	public static function sample(): array {
		$pages = array( home_url( '/' ) );

		if ( 'page' === get_option( 'show_on_front' ) && (int) get_option( 'page_for_posts' ) > 0 ) {
			$pages[] = (string) get_permalink( (int) get_option( 'page_for_posts' ) );
		}

		// The newest entry of each kind of content: a post, a page, a product.
		foreach ( get_post_types( array( 'public' => true ) ) as $type ) {
			if ( 'attachment' === $type || ! is_post_type_viewable( $type ) ) {
				continue;
			}

			$newest = get_posts(
				array(
					'post_type'        => $type,
					'post_status'      => 'publish',
					'has_password'     => false,
					'numberposts'      => 1,
					'fields'           => 'ids',
					'suppress_filters' => false,
				)
			);

			if ( array() !== $newest ) {
				$pages[] = (string) get_permalink( (int) $newest[0] );
			}
		}

		// A shop's own pages: carts and checkouts load what other pages do not.
		if ( function_exists( 'wc_get_page_permalink' ) ) {
			foreach ( array( 'shop', 'cart', 'checkout', 'myaccount' ) as $page ) {
				$pages[] = (string) wc_get_page_permalink( $page );
			}
		}

		/**
		 * Filters the sample of pages a scan opens.
		 *
		 * @param array<int, string> $pages Addresses on this site.
		 */
		$pages = apply_filters( 'wmcc_scan_pages', $pages );

		return self::onSite( is_array( $pages ) ? $pages : array() );
	}

	/**
	 * The addresses the owner added.
	 *
	 * @return array<int, string>
	 */
	public static function more(): array {
		$more = get_option( self::MORE_OPTION );

		return self::onSite( is_array( $more ) ? $more : array() );
	}

	/**
	 * Keep the addresses the owner added (those on this site, at most MAX).
	 *
	 * @param array<int, mixed> $addresses What was typed.
	 * @return array<int, string> What was kept.
	 */
	public static function saveMore( array $addresses ): array {
		$more = array_slice( self::onSite( $addresses ), 0, self::MAX );

		if ( array() === $more ) {
			delete_option( self::MORE_OPTION );
		} else {
			update_option( self::MORE_OPTION, $more, false );
		}

		return $more;
	}

	/**
	 * Addresses of this site only, each once, without fragments. A path
	 * ("/shop/") is taken as a page of this site.
	 *
	 * @param array<int, mixed> $addresses Addresses.
	 * @return array<int, string>
	 */
	private static function onSite( array $addresses ): array {
		$host  = strtolower( (string) wp_parse_url( home_url(), PHP_URL_HOST ) );
		$pages = array();

		foreach ( $addresses as $address ) {
			if ( ! is_string( $address ) ) {
				continue;
			}

			$address = trim( $address );

			if ( '' !== $address && '/' === $address[0] && '/' !== substr( $address, 1, 1 ) ) {
				$address = home_url( $address );
			}

			$address = esc_url_raw( (string) strtok( $address, '#' ), array( 'http', 'https' ) );

			if ( '' === $address || strtolower( (string) wp_parse_url( $address, PHP_URL_HOST ) ) !== $host ) {
				continue;
			}

			// The marks of a scan itself are not part of an address to scan.
			$address = remove_query_arg( array( Request::QUERY, Request::TOKEN, Request::INDEX, '_wpnonce' ), $address );

			$pages[ $address ] = $address;
		}//end foreach

		return array_values( $pages );
	}
}