<?php
/**
 * Finds the URLs to scan.
 *
 * @package A11yMate
 */

namespace A11yMate\Scanner;

use A11yMate\Settings;

defined( 'ABSPATH' ) || exit;

/**
 * Homepage, published content, a sample of archives, search and 404.
 */
class Discovery {

	/**
	 * Build the URL list.
	 *
	 * @param array $config Scan config (post_types, max_pages (0 = no limit), exclude_paths, sample_archives).
	 * @return array[] Each: url, kind, post_id.
	 */
	public static function urls( array $config ) {
		$config = array_merge( Settings::all(), $config );
		// 0 means no limit: every published page is scanned.
		$max  = max( 0, (int) $config['max_pages'] );
		$list = array();
		$seen = array();
		$add  = static function ( $url, $kind, $post_id = 0 ) use ( &$list, &$seen ) {
			$key = untrailingslashit( preg_replace( '/#.*$/', '', $url ) );
			if ( isset( $seen[ $key ] ) || ! $url ) {
				return;
			}
			$seen[ $key ] = true;
			$list[]       = array(
				'url'     => $url,
				'kind'    => $kind,
				'post_id' => (int) $post_id,
			);
		};

		$add( home_url( '/' ), 'home' );
		if ( 'page' === get_option( 'show_on_front' ) && get_option( 'page_for_posts' ) ) {
			$add( get_permalink( (int) get_option( 'page_for_posts' ) ), 'archive' );
		}
		$add( add_query_arg( 's', 'accessibility', home_url( '/' ) ), 'search' );
		$add( home_url( '/a11ymate-404-check/' ), '404' );

		// Archives.
		$sample = (int) $config['sample_archives'];
		if ( $sample > 0 ) {
			foreach ( get_taxonomies( array( 'public' => true ), 'names' ) as $tax ) {
				if ( 'post_format' === $tax ) {
					continue;
				}
				$terms = get_terms(
					array(
						'taxonomy'   => $tax,
						'number'     => $sample,
						'orderby'    => 'count',
						'order'      => 'DESC',
						'hide_empty' => true,
					)
				);
				if ( is_wp_error( $terms ) ) {
					continue;
				}
				foreach ( $terms as $term ) {
					$link = get_term_link( $term );
					if ( ! is_wp_error( $link ) ) {
						$add( $link, 'archive' );
					}
				}
			}
			$authors = get_users(
				array(
					'number'              => min( 3, $sample ),
					'has_published_posts' => true,
					'fields'              => array( 'ID' ),
				)
			);
			foreach ( $authors as $u ) {
				$add( get_author_posts_url( (int) $u->ID ), 'author' );
			}
		}

		// Content, round-robin over post types so a cap never drops a whole type.
		$types = array_values( array_intersect( (array) $config['post_types'], array_keys( get_post_types( array( 'public' => true ) ) ) ) );
		$per   = array();
		foreach ( $types as $type ) {
			if ( 'attachment' === $type ) {
				continue;
			}
			$per[ $type ] = get_posts(
				array(
					'post_type'              => $type,
					'post_status'            => 'publish',
					'posts_per_page'         => $max > 0 ? $max : -1,
					'orderby'                => 'modified',
					'order'                  => 'DESC',
					'has_password'           => false,
					'fields'                 => 'ids',
					'no_found_rows'          => true,
					'update_post_term_cache' => false,
					'update_post_meta_cache' => false,
				)
			);
		}
		$i          = 0;
		$count_urls = static function () use ( &$list ) {
			return count( $list );
		};
		do {
			$any = false;
			foreach ( $per as $type => $ids ) {
				if ( isset( $ids[ $i ] ) ) {
					$any  = true;
					$link = get_permalink( $ids[ $i ] );
					if ( $link ) {
						$add( $link, $type, $ids[ $i ] );
					}
				}
			}
			++$i;
		} while ( $any && ( 0 === $max || $count_urls() < $max * 2 ) );

		/**
		 * Filter the list of URLs to scan. Add or remove entries.
		 *
		 * @param array[] $list   Each: url, kind, post_id.
		 * @param array   $config Scan config.
		 */
		$list = apply_filters( 'a11ymate_scan_urls', $list, $config );

		$list = self::exclude( $list, (string) $config['exclude_paths'] );
		$list = array_values( $list );
		return $max > 0 ? array_slice( $list, 0, $max ) : $list;
	}

	/**
	 * Apply exclusion patterns (one per line, * is a wildcard, matched against the path).
	 *
	 * @param array[] $urls    URLs.
	 * @param string  $pattern Patterns.
	 * @return array[]
	 */
	public static function exclude( array $urls, $pattern ) {
		$rules = array_filter( array_map( 'trim', preg_split( '/\R/', $pattern ) ) );
		if ( ! $rules ) {
			return $urls;
		}
		$regexes = array();
		foreach ( $rules as $rule ) {
			$regexes[] = '#' . str_replace( '\*', '.*', preg_quote( $rule, '#' ) ) . '#i';
		}
		return array_filter(
			$urls,
			static function ( $row ) use ( $regexes ) {
				$path = (string) wp_parse_url( $row['url'], PHP_URL_PATH ) . ( wp_parse_url( $row['url'], PHP_URL_QUERY ) ? '?' . wp_parse_url( $row['url'], PHP_URL_QUERY ) : '' );
				foreach ( $regexes as $re ) {
					if ( preg_match( $re, $path ) ) {
						return false;
					}
				}
				return true;
			}
		);
	}
}