Edit File by line

<?php

[0] Fix | Delete

[1] Fix | Delete

namespace Yoast\WP\SEO\Helpers\Schema;

[2] Fix | Delete

[3] Fix | Delete

/**

[4] Fix | Delete

* Class HTML_Helper.

[5] Fix | Delete

[6] Fix | Delete

class HTML_Helper {

[7] Fix | Delete

[8] Fix | Delete

/**

[9] Fix | Delete

* Sanitizes a HTML string by stripping all tags except headings, breaks, lists, links, paragraphs and formatting.

[10] Fix | Delete

[11] Fix | Delete

* @param string $html The original HTML.

[12] Fix | Delete

[13] Fix | Delete

* @return string The sanitized HTML.

[14] Fix | Delete

[15] Fix | Delete

public function sanitize( $html ) {

[16] Fix | Delete

if ( ! $this->is_non_empty_string_or_stringable( $html ) ) {

[17] Fix | Delete

if ( \is_int( $html ) || \is_float( $html ) ) {

[18] Fix | Delete

return (string) $html;

[19] Fix | Delete

}

[20] Fix | Delete

[21] Fix | Delete

return '';

[22] Fix | Delete

}

[23] Fix | Delete

[24] Fix | Delete

return \strip_tags( $html, '<h1><h2><h3><h4><h5><h6><br><ol><ul><li><a><p><b><strong><i><em>' );

[25] Fix | Delete

}

[26] Fix | Delete

[27] Fix | Delete

/**

[28] Fix | Delete

* Strips the tags in a smart way.

[29] Fix | Delete

[30] Fix | Delete

* @param string $html The original HTML.

[31] Fix | Delete

[32] Fix | Delete

* @return string The sanitized HTML.

[33] Fix | Delete

[34] Fix | Delete

public function smart_strip_tags( $html ) {

[35] Fix | Delete

if ( ! $this->is_non_empty_string_or_stringable( $html ) ) {

[36] Fix | Delete

if ( \is_int( $html ) || \is_float( $html ) ) {

[37] Fix | Delete

return (string) $html;

[38] Fix | Delete

}

[39] Fix | Delete

[40] Fix | Delete

return '';

[41] Fix | Delete

}

[42] Fix | Delete

[43] Fix | Delete

// Replace all new lines with spaces.

[44] Fix | Delete

$html = \preg_replace( '/(\r|\n)/', ' ', $html );

[45] Fix | Delete

[46] Fix | Delete

// Replace <br> tags with spaces.

[47] Fix | Delete

$html = \preg_replace( '/<br(\s*)?\/?>/i', ' ', $html );

[48] Fix | Delete

[49] Fix | Delete

// Replace closing </p> and other tags with the same tag with a space after it, so we don't end up connecting words when we remove them later.

[50] Fix | Delete

$html = \preg_replace( '/<\/(p|div|h\d)>/i', '</$1> ', $html );

[51] Fix | Delete

[52] Fix | Delete

// Replace list items with list identifiers so it still looks natural.

[53] Fix | Delete

$html = \preg_replace( '/(<li[^>]*>)/i', '$1• ', $html );

[54] Fix | Delete

[55] Fix | Delete

// Strip tags.

[56] Fix | Delete

$html = \wp_strip_all_tags( $html );

[57] Fix | Delete

[58] Fix | Delete

// Replace multiple spaces with one space.

[59] Fix | Delete

$html = \preg_replace( '!\s+!', ' ', $html );

[60] Fix | Delete

[61] Fix | Delete

return \trim( $html );

[62] Fix | Delete

}

[63] Fix | Delete

[64] Fix | Delete

/**

[65] Fix | Delete

* Verifies that the received input is either a string or stringable object.

[66] Fix | Delete

[67] Fix | Delete

* @param string $html The original HTML.

[68] Fix | Delete

[69] Fix | Delete

* @return bool

[70] Fix | Delete

[71] Fix | Delete

private function is_non_empty_string_or_stringable( $html ) {

[72] Fix | Delete

return ( \is_string( $html ) || ( \is_object( $html ) && \method_exists( $html, '__toString' ) ) ) && ! empty( $html );

[73] Fix | Delete

}

[74] Fix | Delete

}

[75] Fix | Delete

[76] Fix | Delete