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

* Parses images from the given post.

[8] Fix | Delete

[9] Fix | Delete

class WPSEO_Sitemap_Image_Parser {

[10] Fix | Delete

[11] Fix | Delete

/**

[12] Fix | Delete

* Holds the home_url() value to speed up loops.

[13] Fix | Delete

[14] Fix | Delete

* @var string

[15] Fix | Delete

[16] Fix | Delete

protected $home_url = '';

[17] Fix | Delete

[18] Fix | Delete

/**

[19] Fix | Delete

* Holds site URL hostname.

[20] Fix | Delete

[21] Fix | Delete

* @var string

[22] Fix | Delete

[23] Fix | Delete

protected $host = '';

[24] Fix | Delete

[25] Fix | Delete

/**

[26] Fix | Delete

* Holds site URL protocol.

[27] Fix | Delete

[28] Fix | Delete

* @var string

[29] Fix | Delete

[30] Fix | Delete

protected $scheme = 'http';

[31] Fix | Delete

[32] Fix | Delete

/**

[33] Fix | Delete

* Cached set of attachments for multiple posts.

[34] Fix | Delete

[35] Fix | Delete

* @var array

[36] Fix | Delete

[37] Fix | Delete

protected $attachments = [];

[38] Fix | Delete

[39] Fix | Delete

/**

[40] Fix | Delete

* Holds blog charset value for use in DOM parsing.

[41] Fix | Delete

[42] Fix | Delete

* @var string

[43] Fix | Delete

[44] Fix | Delete

protected $charset = 'UTF-8';

[45] Fix | Delete

[46] Fix | Delete

/**

[47] Fix | Delete

* Set up URL properties for reuse.

[48] Fix | Delete

[49] Fix | Delete

public function __construct() {

[50] Fix | Delete

[51] Fix | Delete

$this->home_url = home_url();

[52] Fix | Delete

$parsed_home = wp_parse_url( $this->home_url );

[53] Fix | Delete

[54] Fix | Delete

if ( ! empty( $parsed_home['host'] ) ) {

[55] Fix | Delete

$this->host = str_replace( 'www.', '', $parsed_home['host'] );

[56] Fix | Delete

}

[57] Fix | Delete

[58] Fix | Delete

if ( ! empty( $parsed_home['scheme'] ) ) {

[59] Fix | Delete

$this->scheme = $parsed_home['scheme'];

[60] Fix | Delete

}

[61] Fix | Delete

[62] Fix | Delete

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

[63] Fix | Delete

}

[64] Fix | Delete

[65] Fix | Delete

/**

[66] Fix | Delete

* Get set of image data sets for the given post.

[67] Fix | Delete

[68] Fix | Delete

* @param object $post Post object to get images for.

[69] Fix | Delete

[70] Fix | Delete

* @return array

[71] Fix | Delete

[72] Fix | Delete

public function get_images( $post ) {

[73] Fix | Delete

[74] Fix | Delete

$images = [];

[75] Fix | Delete

[76] Fix | Delete

if ( ! is_object( $post ) ) {

[77] Fix | Delete

return $images;

[78] Fix | Delete

}

[79] Fix | Delete

[80] Fix | Delete

$thumbnail_id = get_post_thumbnail_id( $post->ID );

[81] Fix | Delete

[82] Fix | Delete

if ( $thumbnail_id ) {

[83] Fix | Delete

[84] Fix | Delete

$src = $this->get_absolute_url( $this->image_url( $thumbnail_id ) );

[85] Fix | Delete

$images[] = $this->get_image_item( $post, $src );

[86] Fix | Delete

}

[87] Fix | Delete

[88] Fix | Delete

/**

[89] Fix | Delete

* Filter: 'wpseo_sitemap_content_before_parse_html_images' - Filters the post content

[90] Fix | Delete

* before it is parsed for images.

[91] Fix | Delete

[92] Fix | Delete

* @param string $content The raw/unprocessed post content.

[93] Fix | Delete

[94] Fix | Delete

$content = apply_filters( 'wpseo_sitemap_content_before_parse_html_images', $post->post_content );

[95] Fix | Delete

[96] Fix | Delete

$unfiltered_images = $this->parse_html_images( $content );

[97] Fix | Delete

[98] Fix | Delete

foreach ( $unfiltered_images as $image ) {

[99] Fix | Delete

$images[] = $this->get_image_item( $post, $image['src'] );

[100] Fix | Delete

}

[101] Fix | Delete

[102] Fix | Delete

foreach ( $this->parse_galleries( $content, $post->ID ) as $attachment ) {

[103] Fix | Delete

$src = $this->get_absolute_url( $this->image_url( $attachment->ID ) );

[104] Fix | Delete

$images[] = $this->get_image_item( $post, $src );

[105] Fix | Delete

}

[106] Fix | Delete

[107] Fix | Delete

if ( $post->post_type === 'attachment' && wp_attachment_is_image( $post ) ) {

[108] Fix | Delete

$src = $this->get_absolute_url( $this->image_url( $post->ID ) );

[109] Fix | Delete

$images[] = $this->get_image_item( $post, $src );

[110] Fix | Delete

}

[111] Fix | Delete

[112] Fix | Delete

foreach ( $images as $key => $image ) {

[113] Fix | Delete

[114] Fix | Delete

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

[115] Fix | Delete

unset( $images[ $key ] );

[116] Fix | Delete

}

[117] Fix | Delete

}

[118] Fix | Delete

[119] Fix | Delete

/**

[120] Fix | Delete

* Filter images to be included for the post in XML sitemap.

[121] Fix | Delete

[122] Fix | Delete

* @param array $images Array of image items.

[123] Fix | Delete

* @param int $post_id ID of the post.

[124] Fix | Delete

[125] Fix | Delete

$image_list = apply_filters( 'wpseo_sitemap_urlimages', $images, $post->ID );

[126] Fix | Delete

if ( isset( $image_list ) && is_array( $image_list ) ) {

[127] Fix | Delete

$images = $image_list;

[128] Fix | Delete

}

[129] Fix | Delete

[130] Fix | Delete

return $images;

[131] Fix | Delete

}

[132] Fix | Delete

[133] Fix | Delete

/**

[134] Fix | Delete

* Get the images in the term description.

[135] Fix | Delete

[136] Fix | Delete

* @param object $term Term to get images from description for.

[137] Fix | Delete

[138] Fix | Delete

* @return array

[139] Fix | Delete

[140] Fix | Delete

public function get_term_images( $term ) {

[141] Fix | Delete

[142] Fix | Delete

$images = $this->parse_html_images( $term->description );

[143] Fix | Delete

[144] Fix | Delete

foreach ( $this->parse_galleries( $term->description ) as $attachment ) {

[145] Fix | Delete

[146] Fix | Delete

$images[] = [

[147] Fix | Delete

'src' => $this->get_absolute_url( $this->image_url( $attachment->ID ) ),

[148] Fix | Delete

];

[149] Fix | Delete

}

[150] Fix | Delete

[151] Fix | Delete

/**

[152] Fix | Delete

* Filter images to be included for the term in XML sitemap.

[153] Fix | Delete

[154] Fix | Delete

* @param array $image_list Array of image items.

[155] Fix | Delete

* @param int $term_id ID of the post.

[156] Fix | Delete

[157] Fix | Delete

$image_list = apply_filters( 'wpseo_sitemap_urlimages_term', $images, $term->term_id );

[158] Fix | Delete

if ( isset( $image_list ) && is_array( $image_list ) ) {

[159] Fix | Delete

$images = $image_list;

[160] Fix | Delete

}

[161] Fix | Delete

[162] Fix | Delete

return $images;

[163] Fix | Delete

}

[164] Fix | Delete

[165] Fix | Delete

/**

[166] Fix | Delete

* Parse `<img />` tags in content.

[167] Fix | Delete

[168] Fix | Delete

* @param string $content Content string to parse.

[169] Fix | Delete

[170] Fix | Delete

* @return array

[171] Fix | Delete

[172] Fix | Delete

private function parse_html_images( $content ) {

[173] Fix | Delete

[174] Fix | Delete

$images = [];

[175] Fix | Delete

[176] Fix | Delete

if ( ! class_exists( 'DOMDocument' ) ) {

[177] Fix | Delete

return $images;

[178] Fix | Delete

}

[179] Fix | Delete

[180] Fix | Delete

if ( empty( $content ) ) {

[181] Fix | Delete

return $images;

[182] Fix | Delete

}

[183] Fix | Delete

[184] Fix | Delete

// Prevent DOMDocument from bubbling warnings about invalid HTML.

[185] Fix | Delete

libxml_use_internal_errors( true );

[186] Fix | Delete

[187] Fix | Delete

$post_dom = new DOMDocument();

[188] Fix | Delete

$post_dom->loadHTML( '<?xml encoding="' . $this->charset . '">' . $content );

[189] Fix | Delete

[190] Fix | Delete

// Clear the errors, so they don't get kept in memory.

[191] Fix | Delete

libxml_clear_errors();

[192] Fix | Delete

[193] Fix | Delete

/**

[194] Fix | Delete

* Image attribute.

[195] Fix | Delete

[196] Fix | Delete

* @var DOMElement $img

[197] Fix | Delete

[198] Fix | Delete

foreach ( $post_dom->getElementsByTagName( 'img' ) as $img ) {

[199] Fix | Delete

[200] Fix | Delete

$src = $img->getAttribute( 'src' );

[201] Fix | Delete

[202] Fix | Delete

if ( empty( $src ) ) {

[203] Fix | Delete

continue;

[204] Fix | Delete

}

[205] Fix | Delete

[206] Fix | Delete

$class = $img->getAttribute( 'class' );

[207] Fix | Delete

[208] Fix | Delete

if ( // This detects WP-inserted images, which we need to upsize. R.

[209] Fix | Delete

! empty( $class )

[210] Fix | Delete

&& ( strpos( $class, 'size-full' ) === false )

[211] Fix | Delete

&& preg_match( '|wp-image-(?P<id>\d+)|', $class, $matches )

[212] Fix | Delete

&& get_post_status( $matches['id'] )

[213] Fix | Delete

) {

[214] Fix | Delete

$query_params = wp_parse_url( $src, PHP_URL_QUERY );

[215] Fix | Delete

$src = $this->image_url( $matches['id'] );

[216] Fix | Delete

[217] Fix | Delete

if ( $query_params ) {

[218] Fix | Delete

$src .= '?' . $query_params;

[219] Fix | Delete

}

[220] Fix | Delete

}

[221] Fix | Delete

[222] Fix | Delete

$src = $this->get_absolute_url( $src );

[223] Fix | Delete

[224] Fix | Delete

if ( strpos( $src, $this->host ) === false ) {

[225] Fix | Delete

continue;

[226] Fix | Delete

}

[227] Fix | Delete

[228] Fix | Delete

if ( $src !== esc_url( $src, null, 'attribute' ) ) {

[229] Fix | Delete

continue;

[230] Fix | Delete

}

[231] Fix | Delete

[232] Fix | Delete

$images[] = [

[233] Fix | Delete

'src' => $src,

[234] Fix | Delete

];

[235] Fix | Delete

}

[236] Fix | Delete

[237] Fix | Delete

return $images;

[238] Fix | Delete

}

[239] Fix | Delete

[240] Fix | Delete

/**

[241] Fix | Delete

* Parse gallery shortcodes in a given content.

[242] Fix | Delete

[243] Fix | Delete

* @param string $content Content string.

[244] Fix | Delete

* @param int $post_id Optional. ID of post being parsed.

[245] Fix | Delete

[246] Fix | Delete

* @return array Set of attachment objects.

[247] Fix | Delete

[248] Fix | Delete

protected function parse_galleries( $content, $post_id = 0 ) {

[249] Fix | Delete

[250] Fix | Delete

$attachments = [];

[251] Fix | Delete

$galleries = $this->get_content_galleries( $content );

[252] Fix | Delete

[253] Fix | Delete

foreach ( $galleries as $gallery ) {

[254] Fix | Delete

[255] Fix | Delete

$id = $post_id;

[256] Fix | Delete

[257] Fix | Delete

if ( ! empty( $gallery['id'] ) ) {

[258] Fix | Delete

$id = intval( $gallery['id'] );

[259] Fix | Delete

}

[260] Fix | Delete

[261] Fix | Delete

// Forked from core gallery_shortcode() to have exact same logic. R.

[262] Fix | Delete

if ( ! empty( $gallery['ids'] ) ) {

[263] Fix | Delete

$gallery['include'] = $gallery['ids'];

[264] Fix | Delete

}

[265] Fix | Delete

[266] Fix | Delete

$gallery_attachments = $this->get_gallery_attachments( $id, $gallery );

[267] Fix | Delete

[268] Fix | Delete

$attachments = array_merge( $attachments, $gallery_attachments );

[269] Fix | Delete

}

[270] Fix | Delete

[271] Fix | Delete

return array_unique( $attachments, SORT_REGULAR );

[272] Fix | Delete

}

[273] Fix | Delete

[274] Fix | Delete

/**

[275] Fix | Delete

* Retrieves galleries from the passed content.

[276] Fix | Delete

[277] Fix | Delete

* Forked from core to skip executing shortcodes for performance.

[278] Fix | Delete

[279] Fix | Delete

* @param string $content Content to parse for shortcodes.

[280] Fix | Delete

[281] Fix | Delete

* @return array A list of arrays, each containing gallery data.

[282] Fix | Delete

[283] Fix | Delete

protected function get_content_galleries( $content ) {

[284] Fix | Delete

[285] Fix | Delete

$galleries = [];

[286] Fix | Delete

[287] Fix | Delete

if ( ! preg_match_all( '/' . get_shortcode_regex( [ 'gallery' ] ) . '/s', $content, $matches, PREG_SET_ORDER ) ) {

[288] Fix | Delete

return $galleries;

[289] Fix | Delete

}

[290] Fix | Delete

[291] Fix | Delete

foreach ( $matches as $shortcode ) {

[292] Fix | Delete

[293] Fix | Delete

$attributes = shortcode_parse_atts( $shortcode[3] );

[294] Fix | Delete

[295] Fix | Delete

if ( $attributes === '' ) { // Valid shortcode without any attributes. R.

[296] Fix | Delete

$attributes = [];

[297] Fix | Delete

}

[298] Fix | Delete

[299] Fix | Delete

$galleries[] = $attributes;

[300] Fix | Delete

}

[301] Fix | Delete

[302] Fix | Delete

return $galleries;

[303] Fix | Delete

}

[304] Fix | Delete

[305] Fix | Delete

/**

[306] Fix | Delete

* Get image item array with filters applied.

[307] Fix | Delete

[308] Fix | Delete

* @param WP_Post $post Post object for the context.

[309] Fix | Delete

* @param string $src Image URL.

[310] Fix | Delete

[311] Fix | Delete

* @return array

[312] Fix | Delete

[313] Fix | Delete

protected function get_image_item( $post, $src ) {

[314] Fix | Delete

[315] Fix | Delete

$image = [];

[316] Fix | Delete

[317] Fix | Delete

/**

[318] Fix | Delete

* Filter image URL to be included in XML sitemap for the post.

[319] Fix | Delete

[320] Fix | Delete

* @param string $src Image URL.

[321] Fix | Delete

* @param object $post Post object.

[322] Fix | Delete

[323] Fix | Delete

$image['src'] = apply_filters( 'wpseo_xml_sitemap_img_src', $src, $post );

[324] Fix | Delete

[325] Fix | Delete

/**

[326] Fix | Delete

* Filter image data to be included in XML sitemap for the post.

[327] Fix | Delete

[328] Fix | Delete

* @param array $image {

[329] Fix | Delete

* Array of image data.

[330] Fix | Delete

[331] Fix | Delete

* @type string $src Image URL.

[332] Fix | Delete

* }

[333] Fix | Delete

[334] Fix | Delete

* @param object $post Post object.

[335] Fix | Delete

[336] Fix | Delete

return apply_filters( 'wpseo_xml_sitemap_img', $image, $post );

[337] Fix | Delete

}

[338] Fix | Delete

[339] Fix | Delete

/**

[340] Fix | Delete

* Get attached image URL with filters applied. Adapted from core for speed.

[341] Fix | Delete

[342] Fix | Delete

* @param int $post_id ID of the post.

[343] Fix | Delete

[344] Fix | Delete

* @return string

[345] Fix | Delete

[346] Fix | Delete

private function image_url( $post_id ) {

[347] Fix | Delete

[348] Fix | Delete

static $uploads;

[349] Fix | Delete

[350] Fix | Delete

if ( empty( $uploads ) ) {

[351] Fix | Delete

$uploads = wp_upload_dir();

[352] Fix | Delete

}

[353] Fix | Delete

[354] Fix | Delete

if ( $uploads['error'] !== false ) {

[355] Fix | Delete

return '';

[356] Fix | Delete

}

[357] Fix | Delete

[358] Fix | Delete

$file = get_post_meta( $post_id, '_wp_attached_file', true );

[359] Fix | Delete

[360] Fix | Delete

if ( empty( $file ) ) {

[361] Fix | Delete

return '';

[362] Fix | Delete

}

[363] Fix | Delete

[364] Fix | Delete

// Check that the upload base exists in the file location.

[365] Fix | Delete

if ( strpos( $file, $uploads['basedir'] ) === 0 ) {

[366] Fix | Delete

$src = str_replace( $uploads['basedir'], $uploads['baseurl'], $file );

[367] Fix | Delete

}

[368] Fix | Delete

elseif ( strpos( $file, 'wp-content/uploads' ) !== false ) {

[369] Fix | Delete

$src = $uploads['baseurl'] . substr( $file, ( strpos( $file, 'wp-content/uploads' ) + 18 ) );

[370] Fix | Delete

}

[371] Fix | Delete

else {

[372] Fix | Delete

// It's a newly uploaded file, therefore $file is relative to the baseurl.

[373] Fix | Delete

$src = $uploads['baseurl'] . '/' . $file;

[374] Fix | Delete

}

[375] Fix | Delete

[376] Fix | Delete

return apply_filters( 'wp_get_attachment_url', $src, $post_id );

[377] Fix | Delete

}

[378] Fix | Delete

[379] Fix | Delete

/**

[380] Fix | Delete

* Make absolute URL for domain or protocol-relative one.

[381] Fix | Delete

[382] Fix | Delete

* @param string $src URL to process.

[383] Fix | Delete

[384] Fix | Delete

* @return string

[385] Fix | Delete

[386] Fix | Delete

protected function get_absolute_url( $src ) {

[387] Fix | Delete

[388] Fix | Delete

if ( empty( $src ) || ! is_string( $src ) ) {

[389] Fix | Delete

return $src;

[390] Fix | Delete

}

[391] Fix | Delete

[392] Fix | Delete

if ( YoastSEO()->helpers->url->is_relative( $src ) === true ) {

[393] Fix | Delete

[394] Fix | Delete

if ( $src[0] !== '/' ) {

[395] Fix | Delete

return $src;

[396] Fix | Delete

}

[397] Fix | Delete

[398] Fix | Delete

// The URL is relative, we'll have to make it absolute.

[399] Fix | Delete

return $this->home_url . $src;

[400] Fix | Delete

}

[401] Fix | Delete

[402] Fix | Delete

if ( strpos( $src, 'http' ) !== 0 ) {

[403] Fix | Delete

// Protocol relative URL, we add the scheme as the standard requires a protocol.

[404] Fix | Delete

return $this->scheme . ':' . $src;

[405] Fix | Delete

}

[406] Fix | Delete

[407] Fix | Delete

return $src;

[408] Fix | Delete

}

[409] Fix | Delete

[410] Fix | Delete

/**

[411] Fix | Delete

* Returns the attachments for a gallery.

[412] Fix | Delete

[413] Fix | Delete

* @param int $id The post ID.

[414] Fix | Delete

* @param array $gallery The gallery config.

[415] Fix | Delete

[416] Fix | Delete

* @return array The selected attachments.

[417] Fix | Delete

[418] Fix | Delete

protected function get_gallery_attachments( $id, $gallery ) {

[419] Fix | Delete

[420] Fix | Delete

// When there are attachments to include.

[421] Fix | Delete

if ( ! empty( $gallery['include'] ) ) {

[422] Fix | Delete

return $this->get_gallery_attachments_for_included( $gallery['include'] );

[423] Fix | Delete

}

[424] Fix | Delete

[425] Fix | Delete

// When $id is empty, just return empty array.

[426] Fix | Delete

if ( empty( $id ) ) {

[427] Fix | Delete

return [];

[428] Fix | Delete

}

[429] Fix | Delete

[430] Fix | Delete

return $this->get_gallery_attachments_for_parent( $id, $gallery );

[431] Fix | Delete

}

[432] Fix | Delete

[433] Fix | Delete

/**

[434] Fix | Delete

* Returns the attachments for the given ID.

[435] Fix | Delete

[436] Fix | Delete

* @param int $id The post ID.

[437] Fix | Delete

* @param array $gallery The gallery config.

[438] Fix | Delete

[439] Fix | Delete

* @return array The selected attachments.

[440] Fix | Delete

[441] Fix | Delete

protected function get_gallery_attachments_for_parent( $id, $gallery ) {

[442] Fix | Delete

$query = [

[443] Fix | Delete

'posts_per_page' => -1,

[444] Fix | Delete

'post_parent' => $id,

[445] Fix | Delete

];

[446] Fix | Delete

[447] Fix | Delete

// When there are posts that should be excluded from result set.

[448] Fix | Delete

if ( ! empty( $gallery['exclude'] ) ) {

[449] Fix | Delete

$query['post__not_in'] = wp_parse_id_list( $gallery['exclude'] );

[450] Fix | Delete

}

[451] Fix | Delete

[452] Fix | Delete

return $this->get_attachments( $query );

[453] Fix | Delete

}

[454] Fix | Delete

[455] Fix | Delete

/**

[456] Fix | Delete

* Returns an array with attachments for the post IDs that will be included.

[457] Fix | Delete

[458] Fix | Delete

* @param array $included_ids Array with IDs to include.

[459] Fix | Delete

[460] Fix | Delete

* @return array The found attachments.

[461] Fix | Delete

[462] Fix | Delete

protected function get_gallery_attachments_for_included( $included_ids ) {

[463] Fix | Delete

$ids_to_include = wp_parse_id_list( $included_ids );

[464] Fix | Delete

$attachments = $this->get_attachments(

[465] Fix | Delete

[

[466] Fix | Delete

'posts_per_page' => count( $ids_to_include ),

[467] Fix | Delete

'post__in' => $ids_to_include,

[468] Fix | Delete

]

[469] Fix | Delete

);

[470] Fix | Delete

[471] Fix | Delete

$gallery_attachments = [];

[472] Fix | Delete

foreach ( $attachments as $val ) {

[473] Fix | Delete

$gallery_attachments[ $val->ID ] = $val;

[474] Fix | Delete

}

[475] Fix | Delete

[476] Fix | Delete

return $gallery_attachments;

[477] Fix | Delete

}

[478] Fix | Delete

[479] Fix | Delete

/**

[480] Fix | Delete

* Returns the attachments.

[481] Fix | Delete

[482] Fix | Delete

* @param array $args Array with query args.

[483] Fix | Delete

[484] Fix | Delete

* @return array The found attachments.

[485] Fix | Delete

[486] Fix | Delete

protected function get_attachments( $args ) {

[487] Fix | Delete

$default_args = [

[488] Fix | Delete

'post_status' => 'inherit',

[489] Fix | Delete

'post_type' => 'attachment',

[490] Fix | Delete

'post_mime_type' => 'image',

[491] Fix | Delete

[492] Fix | Delete

// Defaults taken from function get_posts.

[493] Fix | Delete

'orderby' => 'date',

[494] Fix | Delete

'order' => 'DESC',

[495] Fix | Delete

'meta_key' => '',

[496] Fix | Delete

'meta_value' => '',

[497] Fix | Delete

'suppress_filters' => true,

[498] Fix | Delete

'ignore_sticky_posts' => true,

[499] Fix | Delete