<?php
/**
 * Structural checks for a JSON-LD document.
 *
 * @package OrbitstemSEO
 */

namespace Orbitstem\SEO\Schema;

defined( 'ABSPATH' ) || exit;

/**
 * Local, offline checks: well-formed graph, resolvable @id references, and
 * the properties Google documents as required or recommended for the types
 * Orbitstem emits. Passing here does not guarantee rich results; Google's
 * Rich Results Test remains the authority.
 */
class Validator {

	/**
	 * Required properties per type.
	 *
	 * @return array<string,string[]>
	 */
	public static function required() {
		return array(
			'WebSite'           => array( 'url', 'name' ),
			'Organization'      => array( 'name', 'url' ),
			'Person'            => array( 'name' ),
			'WebPage'           => array( 'url', 'name', 'isPartOf' ),
			'CollectionPage'    => array( 'url', 'name', 'isPartOf' ),
			'SearchResultsPage' => array( 'url', 'isPartOf' ),
			'ProfilePage'       => array( 'url', 'isPartOf' ),
			'ImageObject'       => array( 'url' ),
			'Article'           => array( 'headline', 'datePublished', 'author', 'publisher' ),
			'BlogPosting'       => array( 'headline', 'datePublished', 'author', 'publisher' ),
			'NewsArticle'       => array( 'headline', 'datePublished', 'author', 'publisher' ),
			'BreadcrumbList'    => array( 'itemListElement' ),
			'FAQPage'           => array( 'mainEntity' ),
			'HowTo'             => array( 'name', 'step' ),
			'Product'           => array( 'name' ),
			'Recipe'            => array( 'name', 'recipeIngredient', 'recipeInstructions' ),
			'Event'             => array( 'name', 'startDate', 'location' ),
			'Course'            => array( 'name', 'description', 'provider' ),
			'JobPosting'        => array( 'title', 'description', 'datePosted', 'hiringOrganization', 'jobLocation' ),
			'Review'            => array( 'itemReviewed', 'reviewRating', 'author' ),
			'VideoObject'       => array( 'name', 'thumbnailUrl', 'uploadDate' ),
		);
	}

	/**
	 * Validates a decoded document.
	 *
	 * @param mixed $document Decoded JSON-LD.
	 * @return string[] Problems; empty when valid.
	 */
	public static function validate( $document ) {
		$errors = array();
		if ( ! is_array( $document ) || ! isset( $document['@context'], $document['@graph'] ) || ! is_array( $document['@graph'] ) ) {
			return array( 'Document needs @context and an @graph array.' );
		}
		if ( 'https://schema.org' !== $document['@context'] ) {
			$errors[] = '@context must be https://schema.org.';
		}

		$ids = array();
		foreach ( $document['@graph'] as $i => $node ) {
			if ( empty( $node['@type'] ) ) {
				$errors[] = "Node {$i} has no @type.";
				continue;
			}
			if ( isset( $node['@id'] ) && isset( $ids[ $node['@id'] ] ) ) {
				$errors[] = "Duplicate @id {$node['@id']}.";
			}
			self::collect_ids( $node, $ids );
		}

		$required = self::required();
		foreach ( $document['@graph'] as $node ) {
			if ( empty( $node['@type'] ) ) {
				continue;
			}
			$type = $node['@type'];
			foreach ( isset( $required[ $type ] ) ? $required[ $type ] : array() as $prop ) {
				if ( ! isset( $node[ $prop ] ) || '' === $node[ $prop ] || array() === $node[ $prop ] ) {
					$errors[] = "{$type} is missing {$prop}.";
				}
			}
			foreach ( array( 'url', 'contentUrl' ) as $prop ) {
				if ( isset( $node[ $prop ] ) && ! preg_match( '#^https?://#', $node[ $prop ] ) ) {
					$errors[] = "{$type}.{$prop} must be an absolute URL.";
				}
			}
			if ( in_array( $type, Graph::ARTICLE_TYPES, true ) ) {
				if ( isset( $node['headline'] ) && mb_strlen( $node['headline'] ) > 110 ) {
					$errors[] = "{$type}.headline is longer than 110 characters.";
				}
				if ( isset( $node['author'] ) && empty( $node['author']['name'] ) ) {
					$errors[] = "{$type}.author has no name.";
				}
				foreach ( array( 'datePublished', 'dateModified' ) as $prop ) {
					if ( isset( $node[ $prop ] ) && false === strtotime( $node[ $prop ] ) ) {
						$errors[] = "{$type}.{$prop} is not a date.";
					}
				}
			}
			if ( 'BreadcrumbList' === $type && isset( $node['itemListElement'] ) ) {
				$errors = array_merge( $errors, self::validate_breadcrumbs( $node['itemListElement'] ) );
			}
			$errors = array_merge( $errors, self::unresolved_refs( $node, $ids, $type ) );
		}

		return array_values( array_unique( $errors ) );
	}

	/**
	 * Breadcrumb positions run 1..n and every item but the last links somewhere.
	 *
	 * @param array $items ListItems.
	 * @return string[]
	 */
	private static function validate_breadcrumbs( array $items ) {
		$errors = array();
		$last   = count( $items ) - 1;
		foreach ( array_values( $items ) as $i => $item ) {
			if ( ! isset( $item['position'] ) || $item['position'] !== $i + 1 ) {
				$errors[] = 'BreadcrumbList positions must run 1, 2, 3...';
			}
			if ( empty( $item['name'] ) ) {
				$errors[] = 'BreadcrumbList item ' . ( $i + 1 ) . ' has no name.';
			}
			if ( $i !== $last && empty( $item['item'] ) ) {
				$errors[] = 'BreadcrumbList item ' . ( $i + 1 ) . ' has no URL.';
			}
		}
		return $errors;
	}

	/**
	 * Records @ids of a node and of nested definitions (objects with @type and @id).
	 *
	 * @param array $node Node.
	 * @param array $ids  Known @ids, by reference.
	 */
	private static function collect_ids( array $node, array &$ids ) {
		if ( isset( $node['@id'], $node['@type'] ) ) {
			$ids[ $node['@id'] ] = true;
		}
		foreach ( $node as $value ) {
			if ( is_array( $value ) ) {
				self::collect_ids( $value, $ids );
			}
		}
	}

	/**
	 * References like {"@id": "..."} that point at no node in the graph.
	 *
	 * @param array  $node Node.
	 * @param array  $ids  Known @ids.
	 * @param string $type Node type for messages.
	 * @return string[]
	 */
	private static function unresolved_refs( array $node, array $ids, $type ) {
		$errors = array();
		foreach ( $node as $key => $value ) {
			if ( '@id' === $key || ! is_array( $value ) ) {
				continue;
			}
			if ( 1 === count( $value ) && isset( $value['@id'] ) && ! isset( $ids[ $value['@id'] ] ) ) {
				$errors[] = "{$type}.{$key} points at missing node {$value['@id']}.";
			}
		}
		return $errors;
	}
}