Edit File by line

<?php

[0] Fix | Delete

[1] Fix | Delete

namespace Yoast\WP\SEO\Integrations\Front_End;

[2] Fix | Delete

[3] Fix | Delete

use WPSEO_Sitemaps_Router;

[4] Fix | Delete

use Yoast\WP\SEO\Conditionals\Robots_Txt_Conditional;

[5] Fix | Delete

use Yoast\WP\SEO\Helpers\Options_Helper;

[6] Fix | Delete

use Yoast\WP\SEO\Helpers\Robots_Txt_Helper;

[7] Fix | Delete

use Yoast\WP\SEO\Integrations\Integration_Interface;

[8] Fix | Delete

use Yoast\WP\SEO\Presenters\Robots_Txt_Presenter;

[9] Fix | Delete

[10] Fix | Delete

/**

[11] Fix | Delete

* Handles adding the sitemap to the `robots.txt`.

[12] Fix | Delete

[13] Fix | Delete

class Robots_Txt_Integration implements Integration_Interface {

[14] Fix | Delete

[15] Fix | Delete

/**

[16] Fix | Delete

* Holds the options helper.

[17] Fix | Delete

[18] Fix | Delete

* @var Options_Helper

[19] Fix | Delete

[20] Fix | Delete

protected $options_helper;

[21] Fix | Delete

[22] Fix | Delete

/**

[23] Fix | Delete

* Holds the robots txt helper.

[24] Fix | Delete

[25] Fix | Delete

* @var Robots_Txt_Helper

[26] Fix | Delete

[27] Fix | Delete

protected $robots_txt_helper;

[28] Fix | Delete

[29] Fix | Delete

/**

[30] Fix | Delete

* Holds the robots txt presenter.

[31] Fix | Delete

[32] Fix | Delete

* @var Robots_Txt_Presenter

[33] Fix | Delete

[34] Fix | Delete

protected $robots_txt_presenter;

[35] Fix | Delete

[36] Fix | Delete

/**

[37] Fix | Delete

* Sets the helpers.

[38] Fix | Delete

[39] Fix | Delete

* @param Options_Helper $options_helper Options helper.

[40] Fix | Delete

* @param Robots_Txt_Helper $robots_txt_helper Robots txt helper.

[41] Fix | Delete

* @param Robots_Txt_Presenter $robots_txt_presenter Robots txt presenter.

[42] Fix | Delete

[43] Fix | Delete

public function __construct( Options_Helper $options_helper, Robots_Txt_Helper $robots_txt_helper, Robots_Txt_Presenter $robots_txt_presenter ) {

[44] Fix | Delete

$this->options_helper = $options_helper;

[45] Fix | Delete

$this->robots_txt_helper = $robots_txt_helper;

[46] Fix | Delete

$this->robots_txt_presenter = $robots_txt_presenter;

[47] Fix | Delete

}

[48] Fix | Delete

[49] Fix | Delete

/**

[50] Fix | Delete

* Returns the conditionals based in which this loadable should be active.

[51] Fix | Delete

[52] Fix | Delete

* @return array

[53] Fix | Delete

[54] Fix | Delete

public static function get_conditionals() {

[55] Fix | Delete

return [ Robots_Txt_Conditional::class ];

[56] Fix | Delete

}

[57] Fix | Delete

[58] Fix | Delete

/**

[59] Fix | Delete

* Initializes the integration.

[60] Fix | Delete

[61] Fix | Delete

* This is the place to register hooks and filters.

[62] Fix | Delete

[63] Fix | Delete

* @return void

[64] Fix | Delete

[65] Fix | Delete

public function register_hooks() {

[66] Fix | Delete

\add_filter( 'robots_txt', [ $this, 'filter_robots' ], 99999 );

[67] Fix | Delete

[68] Fix | Delete

if ( $this->options_helper->get( 'deny_search_crawling' ) && ! \is_multisite() ) {

[69] Fix | Delete

\add_action( 'Yoast\WP\SEO\register_robots_rules', [ $this, 'add_disallow_search_to_robots' ], 10, 1 );

[70] Fix | Delete

}

[71] Fix | Delete

if ( $this->options_helper->get( 'deny_wp_json_crawling' ) && ! \is_multisite() ) {

[72] Fix | Delete

\add_action( 'Yoast\WP\SEO\register_robots_rules', [ $this, 'add_disallow_wp_json_to_robots' ], 10, 1 );

[73] Fix | Delete

}

[74] Fix | Delete

if ( $this->options_helper->get( 'deny_adsbot_crawling' ) && ! \is_multisite() ) {

[75] Fix | Delete

\add_action( 'Yoast\WP\SEO\register_robots_rules', [ $this, 'add_disallow_adsbot' ], 10, 1 );

[76] Fix | Delete

}

[77] Fix | Delete

}

[78] Fix | Delete

[79] Fix | Delete

/**

[80] Fix | Delete

* Filters the robots.txt output.

[81] Fix | Delete

[82] Fix | Delete

* @param string $robots_txt The robots.txt output from WordPress.

[83] Fix | Delete

[84] Fix | Delete

* @return string Filtered robots.txt output.

[85] Fix | Delete

[86] Fix | Delete

public function filter_robots( $robots_txt ) {

[87] Fix | Delete

$robots_txt = $this->remove_default_robots( $robots_txt );

[88] Fix | Delete

$this->maybe_add_xml_sitemap();

[89] Fix | Delete

[90] Fix | Delete

/**

[91] Fix | Delete

* Filter: 'wpseo_should_add_subdirectory_multisite_xml_sitemaps' - Disabling this filter removes subdirectory sites from xml sitemaps.

[92] Fix | Delete

[93] Fix | Delete

* @since 19.8

[94] Fix | Delete

[95] Fix | Delete

* @param bool $show Whether to display multisites in the xml sitemaps.

[96] Fix | Delete

[97] Fix | Delete

if ( \apply_filters( 'wpseo_should_add_subdirectory_multisite_xml_sitemaps', true ) ) {

[98] Fix | Delete

$this->add_subdirectory_multisite_xml_sitemaps();

[99] Fix | Delete

}

[100] Fix | Delete

[101] Fix | Delete

/**

[102] Fix | Delete

* Allow registering custom robots rules to be outputted within the Yoast content block in robots.txt.

[103] Fix | Delete

[104] Fix | Delete

* @param Robots_Txt_Helper $robots_txt_helper The Robots_Txt_Helper object.

[105] Fix | Delete

[106] Fix | Delete

\do_action( 'Yoast\WP\SEO\register_robots_rules', $this->robots_txt_helper );

[107] Fix | Delete

[108] Fix | Delete

return \trim( $robots_txt . \PHP_EOL . $this->robots_txt_presenter->present() . \PHP_EOL );

[109] Fix | Delete

}

[110] Fix | Delete

[111] Fix | Delete

/**

[112] Fix | Delete

* Add a disallow rule for search to robots.txt.

[113] Fix | Delete

[114] Fix | Delete

* @param Robots_Txt_Helper $robots_txt_helper The robots txt helper.

[115] Fix | Delete

[116] Fix | Delete

* @return void

[117] Fix | Delete

[118] Fix | Delete

public function add_disallow_search_to_robots( Robots_Txt_Helper $robots_txt_helper ) {

[119] Fix | Delete

$robots_txt_helper->add_disallow( '*', '/?s=' );

[120] Fix | Delete

$robots_txt_helper->add_disallow( '*', '/page/*/?s=' );

[121] Fix | Delete

$robots_txt_helper->add_disallow( '*', '/search/' );

[122] Fix | Delete

}

[123] Fix | Delete

[124] Fix | Delete

/**

[125] Fix | Delete

* Add a disallow rule for /wp-json/ to robots.txt.

[126] Fix | Delete

[127] Fix | Delete

* @param Robots_Txt_Helper $robots_txt_helper The robots txt helper.

[128] Fix | Delete

[129] Fix | Delete

* @return void

[130] Fix | Delete

[131] Fix | Delete

public function add_disallow_wp_json_to_robots( Robots_Txt_Helper $robots_txt_helper ) {

[132] Fix | Delete

$robots_txt_helper->add_disallow( '*', '/wp-json/' );

[133] Fix | Delete

$robots_txt_helper->add_disallow( '*', '/?rest_route=' );

[134] Fix | Delete

}

[135] Fix | Delete

[136] Fix | Delete

/**

[137] Fix | Delete

* Add a disallow rule for AdsBot agents to robots.txt.

[138] Fix | Delete

[139] Fix | Delete

* @param Robots_Txt_Helper $robots_txt_helper The robots txt helper.

[140] Fix | Delete

[141] Fix | Delete

* @return void

[142] Fix | Delete

[143] Fix | Delete

public function add_disallow_adsbot( Robots_Txt_Helper $robots_txt_helper ) {

[144] Fix | Delete

$robots_txt_helper->add_disallow( 'AdsBot', '/' );

[145] Fix | Delete

}

[146] Fix | Delete

[147] Fix | Delete

/**

[148] Fix | Delete

* Replaces the default WordPress robots.txt output.

[149] Fix | Delete

[150] Fix | Delete

* @param string $robots_txt Input robots.txt.

[151] Fix | Delete

[152] Fix | Delete

* @return string

[153] Fix | Delete

[154] Fix | Delete

protected function remove_default_robots( $robots_txt ) {

[155] Fix | Delete

return \preg_replace(

[156] Fix | Delete

'`User-agent: \*[\r\n]+Disallow: /wp-admin/[\r\n]+Allow: /wp-admin/admin-ajax\.php[\r\n]+`',

[157] Fix | Delete

'',

[158] Fix | Delete

$robots_txt

[159] Fix | Delete

);

[160] Fix | Delete

}

[161] Fix | Delete

[162] Fix | Delete

/**

[163] Fix | Delete

* Adds XML sitemap reference to robots.txt.

[164] Fix | Delete

[165] Fix | Delete

* @return void

[166] Fix | Delete

[167] Fix | Delete

protected function maybe_add_xml_sitemap() {

[168] Fix | Delete

// If the XML sitemap is disabled, bail.

[169] Fix | Delete

if ( ! $this->options_helper->get( 'enable_xml_sitemap', false ) ) {

[170] Fix | Delete

return;

[171] Fix | Delete

}

[172] Fix | Delete

$this->robots_txt_helper->add_sitemap( \esc_url( WPSEO_Sitemaps_Router::get_base_url( 'sitemap_index.xml' ) ) );

[173] Fix | Delete

}

[174] Fix | Delete

[175] Fix | Delete

/**

[176] Fix | Delete

* Adds subdomain multisite' XML sitemap references to robots.txt.

[177] Fix | Delete

[178] Fix | Delete

* @return void

[179] Fix | Delete

[180] Fix | Delete

protected function add_subdirectory_multisite_xml_sitemaps() {

[181] Fix | Delete

// If not on a multisite subdirectory, bail.

[182] Fix | Delete

if ( ! \is_multisite() || \is_subdomain_install() ) {

[183] Fix | Delete

return;

[184] Fix | Delete

}

[185] Fix | Delete

[186] Fix | Delete

$sitemaps_enabled = $this->get_xml_sitemaps_enabled();

[187] Fix | Delete

[188] Fix | Delete

foreach ( $sitemaps_enabled as $blog_id => $is_sitemap_enabled ) {

[189] Fix | Delete

if ( ! $is_sitemap_enabled ) {

[190] Fix | Delete

continue;

[191] Fix | Delete

}

[192] Fix | Delete

$this->robots_txt_helper->add_sitemap( \esc_url( \get_home_url( $blog_id, 'sitemap_index.xml' ) ) );

[193] Fix | Delete

}

[194] Fix | Delete

}

[195] Fix | Delete

[196] Fix | Delete

/**

[197] Fix | Delete

* Retrieves whether the XML sitemaps are enabled, keyed by blog ID.

[198] Fix | Delete

[199] Fix | Delete

* @return array

[200] Fix | Delete

[201] Fix | Delete

protected function get_xml_sitemaps_enabled() {

[202] Fix | Delete

$is_allowed = $this->is_sitemap_allowed();

[203] Fix | Delete

$blog_ids = $this->get_blog_ids();

[204] Fix | Delete

$is_enabled = [];

[205] Fix | Delete

foreach ( $blog_ids as $blog_id ) {

[206] Fix | Delete

$is_enabled[ $blog_id ] = $is_allowed && $this->is_sitemap_enabled_for( $blog_id );

[207] Fix | Delete

}

[208] Fix | Delete

[209] Fix | Delete

return $is_enabled;

[210] Fix | Delete

}

[211] Fix | Delete

[212] Fix | Delete

/**

[213] Fix | Delete

* Retrieves whether the sitemap is allowed on a sub site.

[214] Fix | Delete

[215] Fix | Delete

* @return bool

[216] Fix | Delete

[217] Fix | Delete

protected function is_sitemap_allowed() {

[218] Fix | Delete

$options = \get_network_option( null, 'wpseo_ms' );

[219] Fix | Delete

if ( ! $options || ! isset( $options['allow_enable_xml_sitemap'] ) ) {

[220] Fix | Delete

// Default is enabled.

[221] Fix | Delete

return true;

[222] Fix | Delete

}

[223] Fix | Delete

[224] Fix | Delete

return (bool) $options['allow_enable_xml_sitemap'];

[225] Fix | Delete

}

[226] Fix | Delete

[227] Fix | Delete

/**

[228] Fix | Delete

* Retrieves whether the sitemap is enabled on a site.

[229] Fix | Delete

[230] Fix | Delete

* @param int $blog_id The blog ID.

[231] Fix | Delete

[232] Fix | Delete

* @return bool

[233] Fix | Delete

[234] Fix | Delete

protected function is_sitemap_enabled_for( $blog_id ) {

[235] Fix | Delete

if ( ! $this->is_yoast_active_on( $blog_id ) ) {

[236] Fix | Delete

return false;

[237] Fix | Delete

}

[238] Fix | Delete

[239] Fix | Delete

$options = \get_blog_option( $blog_id, 'wpseo' );

[240] Fix | Delete

if ( ! $options || ! isset( $options['enable_xml_sitemap'] ) ) {

[241] Fix | Delete

// Default is enabled.

[242] Fix | Delete

return true;

[243] Fix | Delete

}

[244] Fix | Delete

[245] Fix | Delete

return (bool) $options['enable_xml_sitemap'];

[246] Fix | Delete

}

[247] Fix | Delete

[248] Fix | Delete

/**

[249] Fix | Delete

* Determines whether Yoast SEO is active.

[250] Fix | Delete

[251] Fix | Delete

* @param int $blog_id The blog ID.

[252] Fix | Delete

[253] Fix | Delete

* @return bool

[254] Fix | Delete

[255] Fix | Delete

protected function is_yoast_active_on( $blog_id ) {

[256] Fix | Delete

return \in_array( 'wordpress-seo/wp-seo.php', (array) \get_blog_option( $blog_id, 'active_plugins', [] ), true ) || $this->is_yoast_active_for_network();

[257] Fix | Delete

}

[258] Fix | Delete

[259] Fix | Delete

/**

[260] Fix | Delete

* Determines whether Yoast SEO is active for the entire network.

[261] Fix | Delete

[262] Fix | Delete

* @return bool

[263] Fix | Delete

[264] Fix | Delete

protected function is_yoast_active_for_network() {

[265] Fix | Delete

$plugins = \get_network_option( null, 'active_sitewide_plugins' );

[266] Fix | Delete

if ( isset( $plugins['wordpress-seo/wp-seo.php'] ) ) {

[267] Fix | Delete

return true;

[268] Fix | Delete

}

[269] Fix | Delete

[270] Fix | Delete

return false;

[271] Fix | Delete

}

[272] Fix | Delete

[273] Fix | Delete

/**

[274] Fix | Delete

* Retrieves the blog IDs of public, "active" sites on the network.

[275] Fix | Delete

[276] Fix | Delete

* @return array

[277] Fix | Delete

[278] Fix | Delete

protected function get_blog_ids() {

[279] Fix | Delete

$criteria = [

[280] Fix | Delete

'archived' => 0,

[281] Fix | Delete

'deleted' => 0,

[282] Fix | Delete

'public' => 1,

[283] Fix | Delete

'spam' => 0,

[284] Fix | Delete

'fields' => 'ids',

[285] Fix | Delete

'network_id' => \get_current_network_id(),

[286] Fix | Delete

];

[287] Fix | Delete

[288] Fix | Delete

return \get_sites( $criteria );

[289] Fix | Delete

}

[290] Fix | Delete

}

[291] Fix | Delete

[292] Fix | Delete