amp-domain-checker.php

<?php
/**
 * Check domains for the status of AMP on the homepage.
 *
 * The input file needs to have one domain per line.
 *
 * USAGE:
 *
 * wp eval-file amp-domain-checker.php domains.txt
 * cat domains.txt | wp eval-file amp-domain-checker.php -
 */

use AmpProject\Dom\Document;
use AmpProject\Attribute;

global $argv;

// phpcs:disable WordPress.Security.EscapeOutput.OutputNotEscaped
// phpcs:disable WordPress.WP.GlobalVariablesOverride.Prohibited

if ( ! defined( 'WP_CLI' ) || ! isset( $argv[3] ) ) {
	printf( "Usage: wp eval-file %s FILE\n", basename( __FILE__ ) );
	exit( 1 );
}

if ( ! class_exists( 'AmpProject\Dom\Document' ) ) {
	WP_CLI::error( 'The AMP plugin must be active.' );
}

$input = $argv[3];
if ( '-' === $input ) {
	$input = 'php://stdin';
} elseif ( ! file_exists( $input ) ) {
	WP_CLI::error( "File does not exist: {$input}" );
}

$domains = explode( "\n", trim( file_get_contents( $input ) ) );

/**
 * Examine a URL for AMP status.
 *
 * @param string $url         URL.
 * @param bool   $follow_link Whether to follow the amphtml link.
 *
 * @return string[] Status.
 * @throws Exception
 */
function examine_url( $url, $follow_link = true ) {
	$response = wp_remote_get( $url );

	if ( is_wp_error( $response ) ) {
		throw new Exception( $response->get_error_code() );
	} elseif ( 200 !== wp_remote_retrieve_response_code( $response ) ) {
		throw new Exception( sprintf( 'HTTP %s', wp_remote_retrieve_response_code( $response ) ) );
	}

	$dom = new Document();
	$dom->loadHTML( wp_remote_retrieve_body( $response ) );

	$amphtml_link = $dom->xpath->query( '/html/head/link[ @rel = "amphtml" ]' )->item( 0 );

	if ( $follow_link && $amphtml_link instanceof DOMElement ) {
		$amphtml_url = $amphtml_link->getAttribute( 'href' );
		return examine_url( $amphtml_url, false );
	}

	$has_attribute = (
		$dom->documentElement->hasAttribute( Attribute::AMP )
		||
		$dom->documentElement->hasAttribute( Attribute::AMP4ADS_EMOJI )
		||
		$dom->documentElement->hasAttribute( Attribute::AMP4ADS_EMOJI_ALT )
	);

	$has_runtime = $dom->xpath->query( '/html/head/script[ @src = "https://cdn.ampproject.org/v0.js" ]' )->length > 0;

	$is_amp = $has_attribute || $has_runtime;

	$amp_generator_meta = $dom->xpath->query( '/html/head/meta[ @name = "generator" and contains( @content, "AMP Plugin" ) ]' )->item( 0 );

	if ( ! $is_amp ) {
		return [ 'NOT_AMP' ];
	} elseif ( ! $amp_generator_meta ) {
		return [ 'AMP_BUT_NOT_OURS' ];
	}

	$result = [];
	if ( ! $has_attribute ) {
		$result[] = 'DIRTY_AMP';
	} else {
		$result[] = 'AMP';
	}

	$result[] = preg_replace( '/^AMP Plugin .+?; /', '', $amp_generator_meta->getAttribute( 'content' ) );

	return $result;
}

fwrite( STDERR, sprintf( "Checking %d domains...\n", count( $domains ) ) );
foreach ( $domains as $i => $domain ) {
	echo "$i\t$domain\t";

	if ( preg_match( '#^https?://#', $domain ) ) {
		$url = $domain;
	} else {
		$url = "https://$domain/";
	}

	try {
		echo implode( "\t", examine_url( $url ) );
	} catch ( Exception $e ) {
		printf( "EXCEPTION\t%s", $e->getMessage() );
	}
	echo "\n";
}