<?php
/**
 * Reads the site's own published content and lists every link that points to
 * another website, marked dofollow, nofollow, sponsored or UGC.
 *
 * Runs entirely inside WordPress. Nothing is sent anywhere.
 *
 * @package AnyBacklink
 */

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

/**
 * Outbound link scanner.
 */
class AnyBacklink_Scanner {

	const CACHE_KEY = 'anybacklink_scan';
	/* Most recent posts read per scan, so a big site never times out. */
	const MAX_POSTS = 500;
	/* Most links kept for the report. Counts still cover every link. */
	const MAX_LINKS = 2000;

	/**
	 * The last scan, or null if there is none yet.
	 *
	 * @return array|null
	 */
	public static function cached() {
		$scan = get_transient( self::CACHE_KEY );
		return is_array( $scan ) ? $scan : null;
	}

	/**
	 * Read the content and save the result for a day.
	 *
	 * @return array
	 */
	public static function run() {
		$home  = anybacklink_home_host();
		$types = array_values( array_diff( get_post_types( array( 'public' => true ) ), array( 'attachment' ) ) );

		$query = new WP_Query(
			array(
				'post_type'              => $types,
				'post_status'            => 'publish',
				'posts_per_page'         => self::MAX_POSTS,
				'orderby'                => 'modified',
				'order'                  => 'DESC',
				'no_found_rows'          => true,
				'update_post_meta_cache' => false,
				'update_post_term_cache' => false,
			)
		);

		$counts  = array(
			'follow'    => 0,
			'nofollow'  => 0,
			'sponsored' => 0,
			'ugc'       => 0,
		);
		$links   = array();
		$sites   = array();
		$follows = array();

		foreach ( $query->posts as $post ) {
			foreach ( self::links_in( $post->post_content, $home ) as $link ) {
				++$counts[ $link['type'] ];
				$sites[ $link['host'] ] = true;
				if ( 'follow' === $link['type'] ) {
					$follows[ $link['host'] ] = true;
				}
				if ( count( $links ) < self::MAX_LINKS ) {
					$link['post'] = $post->ID;
					$links[]      = $link;
				}
			}
		}

		$scan = array(
			'scanned_at'   => time(),
			'posts'        => count( $query->posts ),
			'counts'       => $counts,
			'total'        => array_sum( $counts ),
			'sites'        => count( $sites ),
			'follow_sites' => count( $follows ),
			'links'        => $links,
		);
		set_transient( self::CACHE_KEY, $scan, DAY_IN_SECONDS );
		return $scan;
	}

	/**
	 * Every link in one piece of content that leaves this site.
	 *
	 * @param string $html Post content.
	 * @param string $home This site's domain.
	 * @return array[] Each with url, host and type.
	 */
	public static function links_in( $html, $home ) {
		$out = array();
		if ( '' === trim( (string) $html ) || false === stripos( $html, '<a' ) ) {
			return $out;
		}
		$tags = new WP_HTML_Tag_Processor( $html );
		while ( $tags->next_tag( 'a' ) ) {
			$href = $tags->get_attribute( 'href' );
			if ( ! is_string( $href ) || ! preg_match( '#^(https?:)?//#i', trim( $href ) ) ) {
				continue; // Relative, mailto:, tel:, # and empty links stay on the site.
			}
			$href = trim( $href );
			$host = strtolower( (string) wp_parse_url( 0 === strpos( $href, '//' ) ? 'https:' . $href : $href, PHP_URL_HOST ) );
			$host = 0 === strpos( $host, 'www.' ) ? substr( $host, 4 ) : $host;
			if ( '' === $host || $host === $home || self::ends_with( $host, '.' . $home ) ) {
				continue;
			}
			$rel   = $tags->get_attribute( 'rel' );
			$out[] = array(
				'url'  => $href,
				'host' => $host,
				'type' => self::type_of( is_string( $rel ) ? $rel : '' ),
			);
		}
		return $out;
	}

	/**
	 * Sponsored and UGC win over nofollow, as Google reads them that way.
	 *
	 * @param string $rel The rel attribute.
	 * @return string follow, nofollow, sponsored or ugc.
	 */
	public static function type_of( $rel ) {
		$tokens = preg_split( '/\s+/', strtolower( trim( $rel ) ) );
		if ( in_array( 'sponsored', $tokens, true ) ) {
			return 'sponsored';
		}
		if ( in_array( 'ugc', $tokens, true ) ) {
			return 'ugc';
		}
		if ( in_array( 'nofollow', $tokens, true ) ) {
			return 'nofollow';
		}
		return 'follow';
	}

	/**
	 * PHP 7.4 safe ends-with.
	 *
	 * @param string $haystack Text.
	 * @param string $needle   Ending.
	 * @return bool
	 */
	private static function ends_with( $haystack, $needle ) {
		$len = strlen( $needle );
		return $len > 0 && substr( $haystack, -$len ) === $needle;
	}
}