<?php
/**
 * Finding the forms of the site.
 *
 * @package Dealroot
 */

namespace Dealroot\Admin;

use Dealroot\Form_Selection;
use Dealroot\Forms\Integration;
use Dealroot\Forms\Registry;

defined( 'ABSPATH' ) || exit;

/**
 * The pages to look at, and the forms found on them. The pages are opened by the browser of
 * the admin who starts the scan, as a visitor without cookies, so hosts that block requests
 * from the site to itself do not matter.
 */
class Form_Scan {

	const OPTION = 'dealroot_form_scan';

	/**
	 * Pages scanned one by one, which every later scan opens again.
	 */
	const PAGES_OPTION = 'dealroot_form_scan_pages';

	/**
	 * Most pages kept from scanning one by one.
	 */
	const MAX_EXTRA = 30;

	/**
	 * Format of the stored result. Results of an older format are scanned again on the Forms
	 * screen, because they lack details the list needs.
	 */
	const FORMAT = 2;

	/**
	 * Most pages opened by one scan.
	 */
	const MAX_PAGES = 150;

	/**
	 * Most forms kept from one scan.
	 */
	const MAX_FORMS = 200;

	/**
	 * Services whose embedded forms are recognized.
	 */
	const PROVIDERS = array( 'Calendly', 'Typeform', 'HubSpot', 'Microsoft Forms', 'Google Forms', 'Tally', 'Jotform', 'Zoho Forms', 'Cognito Forms', 'Fillout', 'Paperform', 'Cal.com', 'Acuity Scheduling', 'TidyCal', 'Airtable' );

	/**
	 * Text in a post that shows a form is placed there.
	 */
	const MARKERS = array( '[contact-form-7', 'wp:contact-form-7', '[wpforms', 'wp:wpforms', '[gravityform', 'wp:gravityforms', '[fluentform', 'wp:fluentfom', '[ninja_form', 'wp:ninja-forms', '[formidable', 'wp:formidable', '[forminator_form', 'wp:forminator', 'wp:jet-forms', '<form' );

	/**
	 * Pages to scan: the home page, the pages scanned one by one before, the latest entry and the
	 * archive of each post type for forms that are part of a template, posts and pages that hold
	 * a form, products of the shop, then the other pages. The scan adds pages it finds linked.
	 *
	 * @return string[] URLs.
	 */
	public static function targets() {
		$skip = self::shop_pages();
		$urls = array( home_url( '/' ) => true );

		foreach ( array_merge( self::extra_pages(), self::type_pages() ) as $url ) {
			$urls[ $url ] = true;
		}

		$ids = array_merge( self::posts_with_forms(), self::products() );

		foreach ( array_merge( $ids, self::pages() ) as $id ) {
			if ( in_array( (int) $id, $skip, true ) ) {
				continue;
			}

			$url = get_permalink( (int) $id );
			if ( $url ) {
				$urls[ $url ] = true;
			}

			if ( count( $urls ) >= self::MAX_PAGES ) {
				break;
			}
		}

		/**
		 * Filters the pages the form scan opens.
		 *
		 * @param string[] $urls Page URLs of this site.
		 */
		$urls = (array) apply_filters( 'dealroot_form_scan_urls', array_keys( $urls ) );

		return array_slice( array_values( array_filter( array_map( 'strval', $urls ), array( __CLASS__, 'is_own_url' ) ) ), 0, self::MAX_PAGES );
	}

	/**
	 * Two entries and the archive of every public post type, e.g. products of a shop the theme
	 * builds itself, because a form in a template shows on each page of that type. With two
	 * pages, such a form is not named after the heading of one product.
	 *
	 * @return string[] URLs.
	 */
	private static function type_pages() {
		$urls  = array();
		$types = array_diff( get_post_types( array( 'public' => true ) ), array( 'attachment', 'page' ) );

		foreach ( array_slice( array_values( $types ), 0, 20 ) as $type ) {
			$entries = get_posts(
				array(
					'post_type'    => $type,
					'post_status'  => 'publish',
					'has_password' => false,
					'numberposts'  => 2,
					'fields'       => 'ids',
				)
			);

			foreach ( $entries as $id ) {
				$urls[] = (string) get_permalink( (int) $id );
			}

			$archive = get_post_type_archive_link( $type );
			if ( $archive && 'post' !== $type ) {
				$urls[] = (string) $archive;
			}
		}

		return array_values( array_unique( array_filter( $urls ) ) );
	}

	/**
	 * Published posts of any public type whose content or Elementor data holds a form.
	 *
	 * @return int[]
	 */
	private static function posts_with_forms() {
		global $wpdb;

		$types = array_values( array_diff( get_post_types( array( 'public' => true ) ), array( 'attachment' ) ) );
		if ( ! $types ) {
			return array();
		}

		$likes = array();
		foreach ( self::MARKERS as $marker ) {
			$likes[] = '%' . $wpdb->esc_like( $marker ) . '%';
		}

		// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
		$content = $wpdb->get_col(
			$wpdb->prepare(
				"SELECT ID FROM %i WHERE post_status = 'publish' AND post_password = '' AND post_type IN (" . implode( ',', array_fill( 0, count( $types ), '%s' ) ) . ')
				AND (' . implode( ' OR ', array_fill( 0, count( $likes ), 'post_content LIKE %s' ) ) . ')
				ORDER BY post_modified DESC LIMIT 60',
				array_merge( array( $wpdb->posts ), $types, $likes )
			)
		);

		$elementor = $wpdb->get_col(
			$wpdb->prepare(
				"SELECT p.ID FROM %i p INNER JOIN %i m ON m.post_id = p.ID
				WHERE m.meta_key = '_elementor_data' AND m.meta_value LIKE %s AND p.post_status = 'publish' AND p.post_password = ''
				AND p.post_type IN (" . implode( ',', array_fill( 0, count( $types ), '%s' ) ) . ')
				ORDER BY p.post_modified DESC LIMIT 60',
				array_merge( array( $wpdb->posts, $wpdb->postmeta, '%' . $wpdb->esc_like( '"widgetType":"form"' ) . '%' ), $types )
			)
		);
		// phpcs:enable

		return array_map( 'intval', array_merge( (array) $content, (array) $elementor ) );
	}

	/**
	 * Published pages, in menu order.
	 *
	 * @return int[]
	 */
	private static function pages() {
		return array_map(
			'intval',
			get_posts(
				array(
					'post_type'    => 'page',
					'post_status'  => 'publish',
					'has_password' => false,
					'numberposts'  => self::MAX_PAGES,
					'orderby'      => array(
						'menu_order' => 'ASC',
						'date'       => 'DESC',
					),
					'fields'       => 'ids',
				)
			)
		);
	}

	/**
	 * Pages scanned one by one.
	 *
	 * @return string[]
	 */
	public static function extra_pages() {
		return array_values( array_filter( array_map( 'strval', (array) get_option( self::PAGES_OPTION, array() ) ), array( __CLASS__, 'is_own_url' ) ) );
	}

	/**
	 * Keeps a page scanned on its own, so later scans open it again.
	 *
	 * @param string $url Page URL.
	 */
	public static function add_page( $url ) {
		$url = esc_url_raw( $url );

		if ( '' === $url || ! self::is_own_url( $url ) ) {
			return;
		}

		$pages = array_values( array_diff( self::extra_pages(), array( $url ) ) );
		array_unshift( $pages, $url );

		update_option( self::PAGES_OPTION, array_slice( $pages, 0, self::MAX_EXTRA ), false );
	}

	/**
	 * Products whose page may hold a form other products do not have: a few without a price,
	 * where shops often ask to request one, and one product of each category.
	 *
	 * @return int[]
	 */
	private static function products() {
		if ( ! post_type_exists( 'product' ) ) {
			return array();
		}

		$base = array(
			'post_type'    => 'product',
			'post_status'  => 'publish',
			'has_password' => false,
			'fields'       => 'ids',
		);

		$ids = get_posts(
			$base + array(
				'numberposts' => 3,
				'meta_query'  => array( // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_query -- Runs when the admin starts a scan.
					'relation' => 'OR',
					array(
						'key'     => '_price',
						'compare' => 'NOT EXISTS',
					),
					array(
						'key'   => '_price',
						'value' => '',
					),
				),
			)
		);

		if ( taxonomy_exists( 'product_cat' ) ) {
			$categories = get_terms(
				array(
					'taxonomy'   => 'product_cat',
					'hide_empty' => true,
					'number'     => 10,
					'fields'     => 'ids',
				)
			);

			foreach ( is_array( $categories ) ? $categories : array() as $category ) {
				$ids = array_merge(
					$ids,
					get_posts(
						$base + array(
							'numberposts' => 1,
							'tax_query'   => array( // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_tax_query -- Runs when the admin starts a scan.
								array(
									'taxonomy' => 'product_cat',
									'terms'    => (int) $category,
								),
							),
						)
					)
				);
			}
		}

		return array_values( array_unique( array_map( 'intval', $ids ) ) );
	}

	/**
	 * Cart, checkout and account pages, which only hold shop forms.
	 *
	 * @return int[]
	 */
	private static function shop_pages() {
		if ( ! function_exists( 'wc_get_page_id' ) ) {
			return array();
		}

		return array_map( 'intval', array( wc_get_page_id( 'cart' ), wc_get_page_id( 'checkout' ), wc_get_page_id( 'myaccount' ) ) );
	}

	/**
	 * Whether a URL points to this site.
	 *
	 * @param string $url URL.
	 * @return bool
	 */
	public static function is_own_url( $url ) {
		$host = wp_parse_url( home_url(), PHP_URL_HOST );

		return in_array( wp_parse_url( $url, PHP_URL_SCHEME ), array( 'http', 'https' ), true ) && strtolower( (string) wp_parse_url( $url, PHP_URL_HOST ) ) === strtolower( (string) $host );
	}

	/**
	 * Last result.
	 *
	 * @return array{time: int, pages: int, failed: int, forms: array}|null
	 */
	public static function last() {
		$scan = get_option( self::OPTION );

		return is_array( $scan ) && isset( $scan['forms'] ) ? $scan : null;
	}

	/**
	 * Whether the last result is of an older format.
	 *
	 * @return bool
	 */
	public static function outdated() {
		$scan = self::last();

		return $scan && self::FORMAT !== (int) ( $scan['format'] ?? 0 );
	}

	/**
	 * Checks and stores what the browser found.
	 *
	 * @param array $data  pages, failed and forms as sent by the scan script.
	 * @param bool  $merge Add the forms to those of the last scan, e.g. after scanning one more page.
	 * @return array Stored result.
	 */
	public static function store( array $data, $merge = false ) {
		$plugins = self::plugin_classes();
		$forms   = array();

		foreach ( array_slice( (array) ( $data['forms'] ?? array() ), 0, self::MAX_FORMS ) as $raw ) {
			$form = is_array( $raw ) ? self::clean_form( $raw, $plugins ) : null;

			if ( ! $form ) {
				continue;
			}

			if ( isset( $forms[ $form['key'] ] ) ) {
				$forms[ $form['key'] ]['pages'] = array_slice( array_merge( $forms[ $form['key'] ]['pages'], $form['pages'] ), 0, 20 );
				continue;
			}

			$forms[ $form['key'] ] = $form;
		}

		$pages  = max( 0, min( self::MAX_PAGES, (int) ( $data['pages'] ?? 0 ) ) );
		$failed = max( 0, min( self::MAX_PAGES, (int) ( $data['failed'] ?? 0 ) ) );
		$last   = $merge ? self::last() : null;

		if ( $last ) {
			$known = array();
			foreach ( $last['forms'] as $form ) {
				$known[ $form['key'] ] = $form;
			}

			foreach ( $forms as $key => $form ) {
				if ( isset( $known[ $key ] ) ) {
					$known[ $key ]['pages'] = self::clean_pages( array_merge( $known[ $key ]['pages'], $form['pages'] ) );

					if ( 'plugin' !== $form['kind'] && $known[ $key ]['name'] !== $form['name'] ) {
						$known[ $key ]['name'] = self::shared_name( (string) $known[ $key ]['name'], (string) $form['name'] );
					}
				} else {
					$known[ $key ] = $form;
				}
			}

			$forms   = array_slice( $known, 0, self::MAX_FORMS );
			$pages  += (int) $last['pages'];
			$failed += (int) $last['failed'];
		}

		$scan = array(
			'format' => $last && self::FORMAT !== (int) ( $last['format'] ?? 0 ) ? 0 : self::FORMAT,
			'time'   => time(),
			'pages'  => $pages,
			'failed' => $failed,
			'forms'  => array_values( $forms ),
		);

		update_option( self::OPTION, $scan, false );

		return $scan;
	}

	/**
	 * One found form, or null when it is not valid.
	 *
	 * @param array $raw     Form from the browser.
	 * @param array $plugins Plugin key => integration class.
	 * @return array|null
	 */
	private static function clean_form( array $raw, array $plugins ) {
		$kind = (string) ( $raw['kind'] ?? '' );
		$name = mb_substr( sanitize_text_field( (string) ( $raw['name'] ?? '' ) ), 0, 191 );
		$form = array(
			'kind'     => $kind,
			'key'      => '',
			'plugin'   => '',
			'id'       => '',
			'selector' => '',
			'provider' => '',
			'via'      => '',
			'tag'      => 'newsletter' === ( $raw['tag'] ?? '' ) ? 'newsletter' : '',
			'name'     => $name,
			'pages'    => self::clean_pages( (array) ( $raw['pages'] ?? array() ) ),
		);

		switch ( $kind ) {
			case 'plugin':
				$plugin = (string) ( $raw['plugin'] ?? '' );
				$key    = Form_Selection::key( $plugin, (string) ( $raw['id'] ?? '' ) );

				if ( ! isset( $plugins[ $plugin ] ) || '' === $key ) {
					return null;
				}

				$form['plugin'] = $plugin;
				$form['id']     = (string) $raw['id'];
				$form['key']    = $key;
				$title          = ctype_digit( $form['id'] ) ? $plugins[ $plugin ]::form_title( (int) $form['id'] ) : '';
				$form['name']   = '' !== $title ? mb_substr( $title, 0, 191 ) : $name;
				break;

			case 'custom':
				$form['selector'] = Form_Selection::sanitize_selector( (string) ( $raw['selector'] ?? '' ) );

				if ( '' === $form['selector'] ) {
					return null;
				}

				$form['key'] = 'custom:' . $form['selector'];
				break;

			case 'coded':
				// Why it is always saved: "attribute" for data-dealroot in the theme, otherwise the
				// selector from the settings that matched.
				$via              = (string) ( $raw['via'] ?? '' );
				$form['via']      = 'attribute' === $via ? $via : Form_Selection::sanitize_selector( $via );
				$form['selector'] = Form_Selection::sanitize_selector( (string) ( $raw['selector'] ?? '' ) );
				$form['key']      = 'coded:' . md5( '' !== $form['selector'] ? $form['selector'] : $name );
				break;

			case 'auto':
				$plugin = (string) ( $raw['plugin'] ?? '' );

				if ( ! isset( $plugins[ $plugin ] ) ) {
					return null;
				}

				$form['plugin'] = $plugin;
				$form['key']    = 'auto:' . $plugin;
				break;

			case 'embed':
				$provider = (string) ( $raw['provider'] ?? '' );

				if ( ! in_array( $provider, self::PROVIDERS, true ) ) {
					return null;
				}

				$form['provider'] = $provider;
				$form['name']     = $provider;
				$form['plugin']   = (string) array_search( $provider, Registry::EMBEDS, true );
				// Embeds whose submissions reach the page can be switched off like a form plugin.
				$form['key'] = '' !== $form['plugin'] ? $form['plugin'] . ':all' : 'embed:' . sanitize_key( $provider );
				break;

			default:
				return null;
		}

		return $form;
	}

	/**
	 * What the names of one form on several pages share, e.g. "Bestellanfrage" for
	 * "Bestellanfrage für Ring" and "Bestellanfrage für Armreif". Same as in the scan script.
	 *
	 * @param string $first  Name.
	 * @param string $second Name.
	 * @return string
	 */
	private static function shared_name( $first, $second ) {
		$a     = explode( ' ', $first );
		$b     = explode( ' ', $second );
		$words = array();

		foreach ( $a as $i => $word ) {
			if ( ! isset( $b[ $i ] ) || $b[ $i ] !== $word ) {
				break;
			}
			$words[] = $word;
		}

		$shared = preg_replace( '/\s+(for|für|zu|zum|zur|de|du|des|pour|per|para|por|voor|van|dla|na|o|для|на|о|про)$/iu', '', implode( ' ', $words ) );
		$shared = preg_replace( '/[\s:–—-]+$/u', '', (string) $shared );

		return mb_strlen( (string) $shared ) >= 3 ? (string) $shared : $first;
	}

	/**
	 * Pages a form was found on.
	 *
	 * @param array $pages URL and title of each page.
	 * @return array<int, array{url: string, title: string}>
	 */
	private static function clean_pages( array $pages ) {
		$clean = array();

		foreach ( array_slice( $pages, 0, 20 ) as $page ) {
			$url = esc_url_raw( (string) ( is_array( $page ) ? ( $page['url'] ?? '' ) : '' ) );

			if ( '' === $url || ! self::is_own_url( $url ) || isset( $clean[ $url ] ) ) {
				continue;
			}

			$clean[ $url ] = array(
				'url'   => $url,
				'title' => mb_substr( sanitize_text_field( (string) ( $page['title'] ?? '' ) ), 0, 191 ),
			);
		}

		return array_values( $clean );
	}

	/**
	 * Integrations whose forms the scan can recognize, by plugin key.
	 *
	 * @return array<string, string>
	 */
	public static function plugin_classes() {
		$classes = array();

		foreach ( Registry::integrations() as $class ) {
			if ( is_a( $class, Integration::class, true ) && 'woocommerce' !== $class::key() ) {
				$classes[ $class::key() ] = $class;
			}
		}

		return $classes;
	}
}