Edit File by line

<?php

[0] Fix | Delete

/**

[1] Fix | Delete

* WPSEO plugin file.

[2] Fix | Delete

[3] Fix | Delete

* @package WPSEO\XML_Sitemaps

[4] Fix | Delete

[5] Fix | Delete

[6] Fix | Delete

/**

[7] Fix | Delete

* Renders XML output for sitemaps.

[8] Fix | Delete

[9] Fix | Delete

class WPSEO_Sitemaps_Renderer {

[10] Fix | Delete

[11] Fix | Delete

/**

[12] Fix | Delete

* XSL stylesheet for styling a sitemap for web browsers.

[13] Fix | Delete

[14] Fix | Delete

* @var string

[15] Fix | Delete

[16] Fix | Delete

protected $stylesheet = '';

[17] Fix | Delete

[18] Fix | Delete

/**

[19] Fix | Delete

* Holds the get_bloginfo( 'charset' ) value to reuse for performance.

[20] Fix | Delete

[21] Fix | Delete

* @var string

[22] Fix | Delete

[23] Fix | Delete

protected $charset = 'UTF-8';

[24] Fix | Delete

[25] Fix | Delete

/**

[26] Fix | Delete

* Holds charset of output, might be converted.

[27] Fix | Delete

[28] Fix | Delete

* @var string

[29] Fix | Delete

[30] Fix | Delete

protected $output_charset = 'UTF-8';

[31] Fix | Delete

[32] Fix | Delete

/**

[33] Fix | Delete

* If data encoding needs to be converted for output.

[34] Fix | Delete

[35] Fix | Delete

* @var bool

[36] Fix | Delete

[37] Fix | Delete

protected $needs_conversion = false;

[38] Fix | Delete

[39] Fix | Delete

/**

[40] Fix | Delete

* Set up object properties.

[41] Fix | Delete

[42] Fix | Delete

public function __construct() {

[43] Fix | Delete

$stylesheet_url = preg_replace( '/(^http[s]?:)/', '', $this->get_xsl_url() );

[44] Fix | Delete

$this->stylesheet = '<?xml-stylesheet type="text/xsl" href="' . esc_url( $stylesheet_url ) . '"?>';

[45] Fix | Delete

$this->charset = get_bloginfo( 'charset' );

[46] Fix | Delete

$this->output_charset = $this->charset;

[47] Fix | Delete

[48] Fix | Delete

if (

[49] Fix | Delete

$this->charset !== 'UTF-8'

[50] Fix | Delete

&& function_exists( 'mb_list_encodings' )

[51] Fix | Delete

&& in_array( $this->charset, mb_list_encodings(), true )

[52] Fix | Delete

) {

[53] Fix | Delete

$this->output_charset = 'UTF-8';

[54] Fix | Delete

}

[55] Fix | Delete

[56] Fix | Delete

$this->needs_conversion = $this->output_charset !== $this->charset;

[57] Fix | Delete

}

[58] Fix | Delete

[59] Fix | Delete

/**

[60] Fix | Delete

* Builds the sitemap index.

[61] Fix | Delete

[62] Fix | Delete

* @param array<string> $links Set of sitemaps index links.

[63] Fix | Delete

[64] Fix | Delete

* @return string

[65] Fix | Delete

[66] Fix | Delete

public function get_index( $links ) {

[67] Fix | Delete

[68] Fix | Delete

$xml = '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";

[69] Fix | Delete

[70] Fix | Delete

foreach ( $links as $link ) {

[71] Fix | Delete

$xml .= $this->sitemap_index_url( $link );

[72] Fix | Delete

}

[73] Fix | Delete

[74] Fix | Delete

/**

[75] Fix | Delete

* Filter to append sitemaps to the index.

[76] Fix | Delete

[77] Fix | Delete

* @param string $index String to append to sitemaps index, defaults to empty.

[78] Fix | Delete

[79] Fix | Delete

$xml .= apply_filters( 'wpseo_sitemap_index', '' );

[80] Fix | Delete

$xml .= '</sitemapindex>';

[81] Fix | Delete

[82] Fix | Delete

return $xml;

[83] Fix | Delete

}

[84] Fix | Delete

[85] Fix | Delete

/**

[86] Fix | Delete

* Builds the sitemap.

[87] Fix | Delete

[88] Fix | Delete

* @param array<string> $links Set of sitemap links.

[89] Fix | Delete

* @param string $type Sitemap type.

[90] Fix | Delete

* @param int $current_page Current sitemap page number.

[91] Fix | Delete

[92] Fix | Delete

* @return string

[93] Fix | Delete

[94] Fix | Delete

public function get_sitemap( $links, $type, $current_page ) {

[95] Fix | Delete

[96] Fix | Delete

$urlset = '<urlset xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1" '

[97] Fix | Delete

. 'xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd '

[98] Fix | Delete

. 'http://www.google.com/schemas/sitemap-image/1.1 http://www.google.com/schemas/sitemap-image/1.1/sitemap-image.xsd" '

[99] Fix | Delete

. 'xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";

[100] Fix | Delete

[101] Fix | Delete

/**

[102] Fix | Delete

* Filters the `urlset` for all sitemaps.

[103] Fix | Delete

[104] Fix | Delete

* @param string $urlset The output for the sitemap's `urlset`.

[105] Fix | Delete

[106] Fix | Delete

$urlset = apply_filters( 'wpseo_sitemap_urlset', $urlset );

[107] Fix | Delete

[108] Fix | Delete

/**

[109] Fix | Delete

* Filters the `urlset` for a sitemap by type.

[110] Fix | Delete

[111] Fix | Delete

* @param string $urlset The output for the sitemap's `urlset`.

[112] Fix | Delete

[113] Fix | Delete

$xml = apply_filters( "wpseo_sitemap_{$type}_urlset", $urlset );

[114] Fix | Delete

[115] Fix | Delete

foreach ( $links as $url ) {

[116] Fix | Delete

$xml .= $this->sitemap_url( $url );

[117] Fix | Delete

}

[118] Fix | Delete

[119] Fix | Delete

/**

[120] Fix | Delete

* Filter to add extra URLs to the XML sitemap by type.

[121] Fix | Delete

[122] Fix | Delete

* Only runs for the first page, not on all.

[123] Fix | Delete

[124] Fix | Delete

* @param string $content String content to add, defaults to empty.

[125] Fix | Delete

[126] Fix | Delete

if ( $current_page === 1 ) {

[127] Fix | Delete

$xml .= apply_filters( "wpseo_sitemap_{$type}_content", '' );

[128] Fix | Delete

}

[129] Fix | Delete

[130] Fix | Delete

$xml .= '</urlset>';

[131] Fix | Delete

[132] Fix | Delete

return $xml;

[133] Fix | Delete

}

[134] Fix | Delete

[135] Fix | Delete

/**

[136] Fix | Delete

* Produce final XML output with debug information.

[137] Fix | Delete

[138] Fix | Delete

* @param string $sitemap Sitemap XML.

[139] Fix | Delete

[140] Fix | Delete

* @return string

[141] Fix | Delete

[142] Fix | Delete

public function get_output( $sitemap ) {

[143] Fix | Delete

[144] Fix | Delete

$output = '<?xml version="1.0" encoding="' . esc_attr( $this->output_charset ) . '"?>';

[145] Fix | Delete

[146] Fix | Delete

if ( $this->stylesheet ) {

[147] Fix | Delete

/**

[148] Fix | Delete

* Filter the stylesheet URL for the XML sitemap.

[149] Fix | Delete

[150] Fix | Delete

* @param string $stylesheet Stylesheet URL.

[151] Fix | Delete

[152] Fix | Delete

$output .= apply_filters( 'wpseo_stylesheet_url', $this->stylesheet ) . "\n";

[153] Fix | Delete

}

[154] Fix | Delete

[155] Fix | Delete

$output .= $sitemap;

[156] Fix | Delete

$output .= "\n";

[157] Fix | Delete

[158] Fix | Delete

return $output;

[159] Fix | Delete

}

[160] Fix | Delete

[161] Fix | Delete

/**

[162] Fix | Delete

* Get charset for the output.

[163] Fix | Delete

[164] Fix | Delete

* @return string

[165] Fix | Delete

[166] Fix | Delete

public function get_output_charset() {

[167] Fix | Delete

return $this->output_charset;

[168] Fix | Delete

}

[169] Fix | Delete

[170] Fix | Delete

/**

[171] Fix | Delete

* Set a custom stylesheet for this sitemap. Set to empty to just remove the default stylesheet.

[172] Fix | Delete

[173] Fix | Delete

* @param string $stylesheet Full XML-stylesheet declaration.

[174] Fix | Delete

[175] Fix | Delete

* @return void

[176] Fix | Delete

[177] Fix | Delete

public function set_stylesheet( $stylesheet ) {

[178] Fix | Delete

$this->stylesheet = $stylesheet;

[179] Fix | Delete

}

[180] Fix | Delete

[181] Fix | Delete

/**

[182] Fix | Delete

* Build the `<sitemap>` tag for a given URL.

[183] Fix | Delete

[184] Fix | Delete

* @param array<string> $url Array of parts that make up this entry.

[185] Fix | Delete

[186] Fix | Delete

* @return string

[187] Fix | Delete

[188] Fix | Delete

protected function sitemap_index_url( $url ) {

[189] Fix | Delete

[190] Fix | Delete

$date = null;

[191] Fix | Delete

[192] Fix | Delete

if ( ! empty( $url['lastmod'] ) ) {

[193] Fix | Delete

$date = YoastSEO()->helpers->date->format( $url['lastmod'] );

[194] Fix | Delete

}

[195] Fix | Delete

[196] Fix | Delete

$url['loc'] = htmlspecialchars( $url['loc'], ENT_COMPAT, $this->output_charset, false );

[197] Fix | Delete

[198] Fix | Delete

$output = "\t<sitemap>\n";

[199] Fix | Delete

$output .= "\t\t<loc>" . $url['loc'] . "</loc>\n";

[200] Fix | Delete

$output .= empty( $date ) ? '' : "\t\t<lastmod>" . htmlspecialchars( $date, ENT_COMPAT, $this->output_charset, false ) . "</lastmod>\n";

[201] Fix | Delete

$output .= "\t</sitemap>\n";

[202] Fix | Delete

[203] Fix | Delete

return $output;

[204] Fix | Delete

}

[205] Fix | Delete

[206] Fix | Delete

/**

[207] Fix | Delete

* Build the `<url>` tag for a given URL.

[208] Fix | Delete

[209] Fix | Delete

* Public access for backwards compatibility reasons.

[210] Fix | Delete

[211] Fix | Delete

* @param array<string> $url Array of parts that make up this entry.

[212] Fix | Delete

[213] Fix | Delete

* @return string

[214] Fix | Delete

[215] Fix | Delete

public function sitemap_url( $url ) {

[216] Fix | Delete

[217] Fix | Delete

$date = null;

[218] Fix | Delete

[219] Fix | Delete

if ( ! empty( $url['mod'] ) ) {

[220] Fix | Delete

// Create a DateTime object date in the correct timezone.

[221] Fix | Delete

$date = YoastSEO()->helpers->date->format( $url['mod'] );

[222] Fix | Delete

}

[223] Fix | Delete

[224] Fix | Delete

$output = "\t<url>\n";

[225] Fix | Delete

$output .= "\t\t<loc>" . $this->encode_and_escape( $url['loc'] ) . "</loc>\n";

[226] Fix | Delete

$output .= empty( $date ) ? '' : "\t\t<lastmod>" . htmlspecialchars( $date, ENT_COMPAT, $this->output_charset, false ) . "</lastmod>\n";

[227] Fix | Delete

[228] Fix | Delete

if ( empty( $url['images'] ) ) {

[229] Fix | Delete

$url['images'] = [];

[230] Fix | Delete

}

[231] Fix | Delete

[232] Fix | Delete

foreach ( $url['images'] as $img ) {

[233] Fix | Delete

[234] Fix | Delete

if ( empty( $img['src'] ) ) {

[235] Fix | Delete

continue;

[236] Fix | Delete

}

[237] Fix | Delete

[238] Fix | Delete

$output .= "\t\t<image:image>\n";

[239] Fix | Delete

$output .= "\t\t\t<image:loc>" . $this->encode_and_escape( $img['src'] ) . "</image:loc>\n";

[240] Fix | Delete

$output .= "\t\t</image:image>\n";

[241] Fix | Delete

}

[242] Fix | Delete

unset( $img );

[243] Fix | Delete

[244] Fix | Delete

$output .= "\t</url>\n";

[245] Fix | Delete

[246] Fix | Delete

/**

[247] Fix | Delete

* Filters the output for the sitemap URL tag.

[248] Fix | Delete

[249] Fix | Delete

* @param string $output The output for the sitemap url tag.

[250] Fix | Delete

* @param array $url The sitemap URL array on which the output is based.

[251] Fix | Delete

[252] Fix | Delete

return apply_filters( 'wpseo_sitemap_url', $output, $url );

[253] Fix | Delete

}

[254] Fix | Delete

[255] Fix | Delete

/**

[256] Fix | Delete

* Ensure the URL is encoded per RFC3986 and correctly escaped for use in an XML sitemap.

[257] Fix | Delete

[258] Fix | Delete

* This method works around a two quirks in esc_url():

[259] Fix | Delete

* 1. `esc_url()` leaves schema-relative URLs alone, while according to the sitemap specs,

[260] Fix | Delete

* the URL must always begin with a protocol.

[261] Fix | Delete

* 2. `esc_url()` escapes ampersands as `&` instead of the more common `&`.

[262] Fix | Delete

* According to the specs, `&` should be used, and even though this shouldn't

[263] Fix | Delete

* really make a difference in practice, to quote Jono: "I'd be nervous about &

[264] Fix | Delete

* given how many weird and wonderful things eat sitemaps", so better safe than sorry.

[265] Fix | Delete

[266] Fix | Delete

* @link https://www.sitemaps.org/protocol.html#xmlTagDefinitions

[267] Fix | Delete

* @link https://www.sitemaps.org/protocol.html#escaping

[268] Fix | Delete

* @link https://developer.wordpress.org/reference/functions/esc_url/

[269] Fix | Delete

[270] Fix | Delete

* @param string $url URL to encode and escape.

[271] Fix | Delete

[272] Fix | Delete

* @return string

[273] Fix | Delete

[274] Fix | Delete

protected function encode_and_escape( $url ) {

[275] Fix | Delete

$url = $this->encode_url_rfc3986( $url );

[276] Fix | Delete

$url = esc_url( $url );

[277] Fix | Delete

$url = str_replace( '&', '&', $url );

[278] Fix | Delete

$url = str_replace( ''', ''', $url );

[279] Fix | Delete

[280] Fix | Delete

if ( strpos( $url, '//' ) === 0 ) {

[281] Fix | Delete

// Schema-relative URL for which esc_url() does not add a scheme.

[282] Fix | Delete

$url = 'http:' . $url;

[283] Fix | Delete

}

[284] Fix | Delete

[285] Fix | Delete

return $url;

[286] Fix | Delete

}

[287] Fix | Delete

[288] Fix | Delete

/**

[289] Fix | Delete

* Apply some best effort conversion to comply with RFC3986.

[290] Fix | Delete

[291] Fix | Delete

* @param string $url URL to encode.

[292] Fix | Delete

[293] Fix | Delete

* @return string

[294] Fix | Delete

[295] Fix | Delete

protected function encode_url_rfc3986( $url ) {

[296] Fix | Delete

[297] Fix | Delete

if ( filter_var( $url, FILTER_VALIDATE_URL ) ) {

[298] Fix | Delete

return $url;

[299] Fix | Delete

}

[300] Fix | Delete

[301] Fix | Delete

$path = wp_parse_url( $url, PHP_URL_PATH );

[302] Fix | Delete

[303] Fix | Delete

if ( ! empty( $path ) && $path !== '/' ) {

[304] Fix | Delete

$encoded_path = explode( '/', $path );

[305] Fix | Delete

[306] Fix | Delete

// First decode the path, to prevent double encoding.

[307] Fix | Delete

$encoded_path = array_map( 'rawurldecode', $encoded_path );

[308] Fix | Delete

[309] Fix | Delete

$encoded_path = array_map( 'rawurlencode', $encoded_path );

[310] Fix | Delete

$encoded_path = implode( '/', $encoded_path );

[311] Fix | Delete

[312] Fix | Delete

$url = str_replace( $path, $encoded_path, $url );

[313] Fix | Delete

}

[314] Fix | Delete

[315] Fix | Delete

$query = wp_parse_url( $url, PHP_URL_QUERY );

[316] Fix | Delete

[317] Fix | Delete

if ( ! empty( $query ) ) {

[318] Fix | Delete

[319] Fix | Delete

parse_str( $query, $parsed_query );

[320] Fix | Delete

[321] Fix | Delete

$parsed_query = http_build_query( $parsed_query, '', '&', PHP_QUERY_RFC3986 );

[322] Fix | Delete

[323] Fix | Delete

$url = str_replace( $query, $parsed_query, $url );

[324] Fix | Delete

}

[325] Fix | Delete

[326] Fix | Delete

return $url;

[327] Fix | Delete

}

[328] Fix | Delete

[329] Fix | Delete

/**

[330] Fix | Delete

* Retrieves the XSL URL that should be used in the current environment

[331] Fix | Delete

[332] Fix | Delete

* When home_url and site_url are not the same, the home_url should be used.

[333] Fix | Delete

* This is because the XSL needs to be served from the same domain, protocol and port

[334] Fix | Delete

* as the XML file that is loading it.

[335] Fix | Delete

[336] Fix | Delete

* @return string The XSL URL that needs to be used.

[337] Fix | Delete

[338] Fix | Delete

protected function get_xsl_url() {

[339] Fix | Delete

if ( home_url() !== site_url() ) {

[340] Fix | Delete

return home_url( 'main-sitemap.xsl' );

[341] Fix | Delete

}

[342] Fix | Delete

[343] Fix | Delete

[344] Fix | Delete

* Fallback to circumvent a cross-domain security problem when the XLS file is

[345] Fix | Delete

* loaded from a different (sub)domain.

[346] Fix | Delete

[347] Fix | Delete

if ( strpos( plugins_url(), home_url() ) !== 0 ) {

[348] Fix | Delete

return home_url( 'main-sitemap.xsl' );

[349] Fix | Delete

}

[350] Fix | Delete

[351] Fix | Delete

return plugin_dir_url( WPSEO_FILE ) . 'css/main-sitemap.xsl';

[352] Fix | Delete

}

[353] Fix | Delete

}

[354] Fix | Delete

[355] Fix | Delete