<?php

[0] Fix | Delete

[1] Fix | Delete

namespace Smush\Core\Parser;

[2] Fix | Delete

[3] Fix | Delete

class Parser {

[4] Fix | Delete

public function get_base_url( $markup ) {

[5] Fix | Delete

$pattern = "/<base[^>]*?href\s*=\s*['\"](.*?)['\"]\s*\/\s*>/is";

[6] Fix | Delete

if ( preg_match( $pattern, $markup, $matches ) && ! empty( $matches[1] ) ) {

[7] Fix | Delete

return $matches[1];

[8] Fix | Delete

}

[9] Fix | Delete

[10] Fix | Delete

return '';

[11] Fix | Delete

}

[12] Fix | Delete

[13] Fix | Delete

public function get_inline_style_blocks( $markup ) {

[14] Fix | Delete

$pattern = '/<style\b[^>]*>(.*?)<\/style>/msi';

[15] Fix | Delete

if ( ! preg_match_all( $pattern, $markup, $matches ) ) {

[16] Fix | Delete

return array();

[17] Fix | Delete

}

[18] Fix | Delete

return $matches[1];

[19] Fix | Delete

}

[20] Fix | Delete

[21] Fix | Delete

/**

[22] Fix | Delete

* @param $markup

[23] Fix | Delete

* @param $base_url

[24] Fix | Delete

*

[25] Fix | Delete

* @return Style[]

[26] Fix | Delete

*/

[27] Fix | Delete

public function get_inline_styles( $markup, $base_url ) {

[28] Fix | Delete

$styles = array();

[29] Fix | Delete

$inline_style_blocks = $this->get_inline_style_blocks( $markup );

[30] Fix | Delete

foreach ( $inline_style_blocks as $inline_style_block ) {

[31] Fix | Delete

$image_urls = $this->get_image_urls( $inline_style_block, $base_url );

[32] Fix | Delete

$style = new Style( $inline_style_block, $image_urls );

[33] Fix | Delete

$styles[] = $style;

[34] Fix | Delete

}

[35] Fix | Delete

return $styles;

[36] Fix | Delete

}

[37] Fix | Delete

[38] Fix | Delete

/**

[39] Fix | Delete

* @param $markup string

[40] Fix | Delete

* @param $base_url

[41] Fix | Delete

*

[42] Fix | Delete

* @return Image_URL[]

[43] Fix | Delete

*/

[44] Fix | Delete

public function get_image_urls( $markup, $base_url ) {

[45] Fix | Delete

$pattern = '@(?<src>(?:https?:/|\.+)?/[^\'",\s\(\)]+\.(?<ext>jpe?g|png|gif|webp|svg))\b@is';

[46] Fix | Delete

$pattern = apply_filters( 'wp_smush_image_urls_regex', $pattern );

[47] Fix | Delete

if ( ! preg_match_all( $pattern, $markup, $matches, PREG_SET_ORDER ) ) {

[48] Fix | Delete

return array();

[49] Fix | Delete

}

[50] Fix | Delete

[51] Fix | Delete

$image_urls = array();

[52] Fix | Delete

foreach ( $matches as $match ) {

[53] Fix | Delete

if ( ! isset( $match['src'], $match['ext'] ) ) {

[54] Fix | Delete

continue;

[55] Fix | Delete

}

[56] Fix | Delete

$image_urls[] = new Image_URL(

[57] Fix | Delete

$this->remove_quote_entities( $match['src'] ),

[58] Fix | Delete

$match['ext'],

[59] Fix | Delete

$base_url

[60] Fix | Delete

);

[61] Fix | Delete

}

[62] Fix | Delete

[63] Fix | Delete

return array_values( $image_urls );

[64] Fix | Delete

}

[65] Fix | Delete

[66] Fix | Delete

public function get_tags( $markup, $tags ) {

[67] Fix | Delete

$tags_string = join( '|', $tags );

[68] Fix | Delete

[69] Fix | Delete

$matches = array();

[70] Fix | Delete

if ( preg_match_all( '/<(' . $tags_string . ').*?\/\1>/s', $markup, $matches, PREG_PATTERN_ORDER ) ) {

[71] Fix | Delete

return $matches[0];

[72] Fix | Delete

}

[73] Fix | Delete

[74] Fix | Delete

return array();

[75] Fix | Delete

}

[76] Fix | Delete

[77] Fix | Delete

public function get_block_by_tag( $markup, $tag ) {

[78] Fix | Delete

$pattern = "/<$tag\b([^>]*>.*)<\/$tag>/is";

[79] Fix | Delete

if ( ! preg_match( $pattern, $markup, $matches ) ) {

[80] Fix | Delete

return $markup;

[81] Fix | Delete

}

[82] Fix | Delete

[83] Fix | Delete

return $matches[1];

[84] Fix | Delete

}

[85] Fix | Delete

[86] Fix | Delete

public function get_composite_elements( $markup, $base_url ) {

[87] Fix | Delete

$composite_elements = array();

[88] Fix | Delete

$tag_names = array( 'picture' );

[89] Fix | Delete

foreach ( $tag_names as $tag_name ) {

[90] Fix | Delete

$html_elements = $this->get_tags( $markup, array( $tag_name ) );

[91] Fix | Delete

foreach ( $html_elements as $html_element ) {

[92] Fix | Delete

$elements = $this->get_elements_with_image_attributes( $html_element, $base_url );

[93] Fix | Delete

if ( ! empty( $elements ) ) {

[94] Fix | Delete

$composite_elements[] = new Composite_Element( $html_element, $tag_name, $elements );

[95] Fix | Delete

}

[96] Fix | Delete

}

[97] Fix | Delete

}

[98] Fix | Delete

return $composite_elements;

[99] Fix | Delete

}

[100] Fix | Delete

[101] Fix | Delete

/**

[102] Fix | Delete

* @param $markup

[103] Fix | Delete

* @param $base_url

[104] Fix | Delete

*

[105] Fix | Delete

* @return Element[]

[106] Fix | Delete

*/

[107] Fix | Delete

public function get_elements_with_image_attributes( $markup, $base_url ) {

[108] Fix | Delete

$pattern = '@(?<element><(?:(?<img>img)\b[^>]+|(?<tag>[a-zA-Z]+)\b[^>]+\.(?:jpe?g|png|gif|webp|svg)[^>]+)>)@is';

[109] Fix | Delete

$pattern = apply_filters( 'wp_smush_images_from_content_regex', $pattern );

[110] Fix | Delete

if ( ! preg_match_all( $pattern, $markup, $matches, PREG_SET_ORDER ) ) {

[111] Fix | Delete

return array();

[112] Fix | Delete

}

[113] Fix | Delete

[114] Fix | Delete

$elements = array();

[115] Fix | Delete

foreach ( $matches as $item ) {

[116] Fix | Delete

$element = $item['element'];

[117] Fix | Delete

$tag_name = ! empty( $item['img'] ) ? $item['img'] : $item['tag'];

[118] Fix | Delete

$attributes = $this->get_element_attributes( $element, $base_url );

[119] Fix | Delete

$background = $this->get_element_background_image( $element, $base_url );

[120] Fix | Delete

$css_properties = $background ? array( $background ) : array();

[121] Fix | Delete

[122] Fix | Delete

// TODO: Support CSS variable with images.

[123] Fix | Delete

if ( empty( $attributes ) && empty( $css_properties ) ) {

[124] Fix | Delete

continue;

[125] Fix | Delete

}

[126] Fix | Delete

[127] Fix | Delete

$elements[] = new Element( $element, $tag_name, $attributes, $css_properties );

[128] Fix | Delete

}

[129] Fix | Delete

[130] Fix | Delete

return $elements;

[131] Fix | Delete

}

[132] Fix | Delete

[133] Fix | Delete

/**

[134] Fix | Delete

* @param $element

[135] Fix | Delete

* @param $base_url

[136] Fix | Delete

*

[137] Fix | Delete

* @return Element_Attribute[]

[138] Fix | Delete

*/

[139] Fix | Delete

public function get_element_attributes( $element, $base_url ) {

[140] Fix | Delete

$image_attributes = array();

[141] Fix | Delete

[142] Fix | Delete

$pattern = '#\b(?<name>(?:data-(?:[a-z0-9_-]+-)?)?(?:[a-z0-9_-]+))\s*=\s*(["\'])(?<value>[^\'"]+)\2#is';

[143] Fix | Delete

$pattern = apply_filters( 'wp_smush_image_attributes_regex', $pattern );

[144] Fix | Delete

[145] Fix | Delete

if ( ! preg_match_all( $pattern, $element, $matches, PREG_SET_ORDER ) ) {

[146] Fix | Delete

return $image_attributes;

[147] Fix | Delete

}

[148] Fix | Delete

[149] Fix | Delete

foreach ( $matches as $attr_data ) {

[150] Fix | Delete

if ( ! isset( $attr_data['name'], $attr_data['value'] ) ) {

[151] Fix | Delete

continue;

[152] Fix | Delete

}

[153] Fix | Delete

$attr_name = $attr_data['name'];

[154] Fix | Delete

$attr_value = trim( $attr_data['value'] );

[155] Fix | Delete

if ( ! $this->is_safe( $attr_name ) || ! $this->is_safe( $attr_value ) ) {

[156] Fix | Delete

continue;

[157] Fix | Delete

}

[158] Fix | Delete

$image_urls = $this->get_image_urls( $attr_value, $base_url );

[159] Fix | Delete

[160] Fix | Delete

$image_attributes[ $attr_name ] = new Element_Attribute(

[161] Fix | Delete

$attr_name,

[162] Fix | Delete

$attr_value,

[163] Fix | Delete

$attr_data[0],

[164] Fix | Delete

$image_urls

[165] Fix | Delete

);

[166] Fix | Delete

}

[167] Fix | Delete

[168] Fix | Delete

return $image_attributes;

[169] Fix | Delete

}

[170] Fix | Delete

[171] Fix | Delete

public function get_element_background_image( $element, $base_url ) {

[172] Fix | Delete

if ( ! strpos( $element, 'background' ) ) {

[173] Fix | Delete

return null;

[174] Fix | Delete

}

[175] Fix | Delete

[176] Fix | Delete

$style = $this->get_element_attribute_value( $element, 'style' );

[177] Fix | Delete

if ( empty( $style ) ) {

[178] Fix | Delete

return null;

[179] Fix | Delete

}

[180] Fix | Delete

[181] Fix | Delete

/**

[182] Fix | Delete

* Background image regex supports:

[183] Fix | Delete

*

[184] Fix | Delete

* 1. background or background-image.

[185] Fix | Delete

* 2. Multiple background images.

[186] Fix | Delete

*

[187] Fix | Delete

* background: url(img_flwr.gif) right bottom no-repeat, url(paper.gif) left top repeat;

[188] Fix | Delete

* background-image: url(&quot;image1.png&quot;), url(https://sample.com/image2.png?lossy=2&amp;strip=1&amp;webp=1), linear-gradient(to right, rgba(30, 75, 115, 1), rgba(255, 255, 255, 0));

[189] Fix | Delete

*/

[190] Fix | Delete

[191] Fix | Delete

// Regex rule to get all inline style after background(-image) property.

[192] Fix | Delete

$pattern = '#(?<!-)\b(?<property>background(?:-image)?):(?<value>[^;:]*?url\s*\([^>]+\)[^=:;]*);{0,1}#is';

[193] Fix | Delete

$pattern = apply_filters( 'wp_smush_background_images_regex', $pattern );

[194] Fix | Delete

[195] Fix | Delete

if ( ! preg_match_all( $pattern, $style, $matches, PREG_SET_ORDER ) ) {

[196] Fix | Delete

return null;

[197] Fix | Delete

}

[198] Fix | Delete

[199] Fix | Delete

$bg_property = null;

[200] Fix | Delete

foreach ( $matches as $bg_image_data ) {

[201] Fix | Delete

if ( ! isset( $bg_image_data['property'], $bg_image_data['value'] ) ) {

[202] Fix | Delete

continue;

[203] Fix | Delete

}

[204] Fix | Delete

$bg_image_property = $bg_image_data['property'];

[205] Fix | Delete

$bg_image_value = $bg_image_data['value'];

[206] Fix | Delete

if ( ! $this->is_safe( $bg_image_property ) || ! $this->is_safe( $bg_image_value ) ) {

[207] Fix | Delete

continue;

[208] Fix | Delete

}

[209] Fix | Delete

$image_urls = $this->get_image_urls( $bg_image_value, $base_url );

[210] Fix | Delete

if ( empty( $image_urls ) ) {

[211] Fix | Delete

continue;

[212] Fix | Delete

}

[213] Fix | Delete

[214] Fix | Delete

// background-image:url(&quot;image1.png&quot;); background-repeat: no-repeat.

[215] Fix | Delete

if ( substr_count( $bg_image_value, ';' ) ) {

[216] Fix | Delete

$bg_image_value = $this->extract_background_image( $bg_image_value );

[217] Fix | Delete

}

[218] Fix | Delete

[219] Fix | Delete

// An element only has one background, so let try to get the latest one.

[220] Fix | Delete

$bg_property = new Element_CSS_Property(

[221] Fix | Delete

$bg_image_data[0],

[222] Fix | Delete

$bg_image_property,

[223] Fix | Delete

$bg_image_value,

[224] Fix | Delete

$image_urls

[225] Fix | Delete

);

[226] Fix | Delete

}

[227] Fix | Delete

[228] Fix | Delete

return $bg_property;

[229] Fix | Delete

}

[230] Fix | Delete

[231] Fix | Delete

/**

[232] Fix | Delete

* Extract background image value from the inline style.

[233] Fix | Delete

*

[234] Fix | Delete

* Input: linear-gradient(to right, rgba(30, 75, 115, 1), rgba(255, 255, 255, 0)), url(&quot;image1.png&quot;) right bottom no-repeat,

[235] Fix | Delete

* url(https://sample.com/image2.png?lossy=2&amp;strip=1&amp;webp=1) left top repeat; background-color: #fff;

[236] Fix | Delete

* Output: linear-gradient(to right, rgba(30, 75, 115, 1), rgba(255, 255, 255, 0)), url(&quot;image1.png&quot;) right bottom no-repeat,

[237] Fix | Delete

* url(https://sample.com/image2.png?lossy=2&amp;strip=1&amp;webp=1) left top repeat;

[238] Fix | Delete

*/

[239] Fix | Delete

public function extract_background_image( $bg_image_value ) {

[240] Fix | Delete

$urls = array();

[241] Fix | Delete

$bg_image_value_without_url = preg_replace_callback(

[242] Fix | Delete

'#url\s*\(([^\)]+)\)#is',

[243] Fix | Delete

function ( $matches ) use ( &$urls ) {

[244] Fix | Delete

if ( ! empty( $matches[1] ) ) {

[245] Fix | Delete

$url = $matches[1];

[246] Fix | Delete

$urls[] = $url;

[247] Fix | Delete

$count = count( $urls );

[248] Fix | Delete

$matches[0] = str_replace( $url, "[URL{$count}]", $matches[0] );

[249] Fix | Delete

}

[250] Fix | Delete

return $matches[0];

[251] Fix | Delete

},

[252] Fix | Delete

$bg_image_value

[253] Fix | Delete

);

[254] Fix | Delete

[255] Fix | Delete

if ( ! empty( $urls ) && preg_match( '/[^;]+/is', $bg_image_value_without_url, $matches ) ) {

[256] Fix | Delete

$bg_image_value = $matches[0];

[257] Fix | Delete

foreach ( $urls as $index => $url ) {

[258] Fix | Delete

$index ++;

[259] Fix | Delete

$bg_image_value = str_replace( "[URL{$index}]", $url, $bg_image_value );

[260] Fix | Delete

}

[261] Fix | Delete

}

[262] Fix | Delete

[263] Fix | Delete

return $bg_image_value;

[264] Fix | Delete

}

[265] Fix | Delete

[266] Fix | Delete

private function remove_quote_entities( $image ) {

[267] Fix | Delete

// Quote entities.

[268] Fix | Delete

$quotes = apply_filters( 'wp_smush_background_image_quotes', array( '&quot;', '&#034;', '&#039;', '&apos;' ) );

[269] Fix | Delete

[270] Fix | Delete

$image = trim( $image );

[271] Fix | Delete

if ( empty( $image ) || strlen( $image ) < 6 ) {

[272] Fix | Delete

return $image;

[273] Fix | Delete

}

[274] Fix | Delete

[275] Fix | Delete

// Remove the starting quotes.

[276] Fix | Delete

if ( in_array( substr( $image, 0, 6 ), $quotes, true ) ) {

[277] Fix | Delete

$image = substr( $image, 6 );

[278] Fix | Delete

}

[279] Fix | Delete

[280] Fix | Delete

// Remove the ending quotes.

[281] Fix | Delete

if ( in_array( substr( $image, - 6 ), $quotes, true ) ) {

[282] Fix | Delete

$image = substr( $image, 0, - 6 );

[283] Fix | Delete

}

[284] Fix | Delete

[285] Fix | Delete

return $image;

[286] Fix | Delete

}

[287] Fix | Delete

[288] Fix | Delete

public function add_element_attribute( $element, $name, $value = null ) {

[289] Fix | Delete

$closing = false === strpos( $element, '/>' ) ? '>' : ' />';

[290] Fix | Delete

$quotes = false === strpos( $element, '"' ) ? '\'' : '"';

[291] Fix | Delete

[292] Fix | Delete

if ( ! is_null( $value ) ) {

[293] Fix | Delete

$element = rtrim( $element, $closing ) . " {$name}={$quotes}{$value}{$quotes}{$closing}";

[294] Fix | Delete

} else {

[295] Fix | Delete

$element = rtrim( $element, $closing ) . " {$name}{$closing}";

[296] Fix | Delete

}

[297] Fix | Delete

[298] Fix | Delete

return $element;

[299] Fix | Delete

}

[300] Fix | Delete

[301] Fix | Delete

public function remove_element_attribute( $element, $attribute ) {

[302] Fix | Delete

return preg_replace( '/\s' . $attribute . '=[\'"](.*?)[\'"]/i', '', $element );

[303] Fix | Delete

}

[304] Fix | Delete

[305] Fix | Delete

/**

[306] Fix | Delete

* @param $markup

[307] Fix | Delete

* @param $base_url

[308] Fix | Delete

*

[309] Fix | Delete

* @return Element[]

[310] Fix | Delete

*/

[311] Fix | Delete

public function get_iframe_elements( $markup, $base_url ) {

[312] Fix | Delete

if ( strpos( $markup, '<iframe' ) === false ) {

[313] Fix | Delete

return array();

[314] Fix | Delete

}

[315] Fix | Delete

[316] Fix | Delete

// Iframe tag has srcdocs attribute which might contains HTML code.

[317] Fix | Delete

$pattern = '#<iframe\b[^>]*\s(?<attr>src\s*=\s*(\'|")(?<src>[^\'"]+)\2)[^>]*>#is';

[318] Fix | Delete

$pattern = apply_filters( 'wp_smush_iframes_regex', $pattern );

[319] Fix | Delete

$iframes = array();

[320] Fix | Delete

[321] Fix | Delete

if ( ! preg_match_all( $pattern, $markup, $matches, PREG_SET_ORDER ) ) {

[322] Fix | Delete

return $iframes;

[323] Fix | Delete

}

[324] Fix | Delete

[325] Fix | Delete

foreach ( $matches as $iframe_data ) {

[326] Fix | Delete

if ( empty( $iframe_data['attr'] ) || empty( $iframe_data['src'] ) ) {

[327] Fix | Delete

continue;

[328] Fix | Delete

}

[329] Fix | Delete

[330] Fix | Delete

$attributes = $this->get_element_attributes( $iframe_data[0], $base_url );

[331] Fix | Delete

$iframe_element = new Element( $iframe_data[0], 'iframe', $attributes, array() );

[332] Fix | Delete

[333] Fix | Delete

$iframes[] = $iframe_element;

[334] Fix | Delete

}

[335] Fix | Delete

[336] Fix | Delete

return $iframes;

[337] Fix | Delete

}

[338] Fix | Delete

[339] Fix | Delete

public function get_element_attribute_value( $element_markup, $attribute_name ) {

[340] Fix | Delete

if ( strpos( $element_markup, $attribute_name ) === false ) {

[341] Fix | Delete

return '';

[342] Fix | Delete

}

[343] Fix | Delete

[344] Fix | Delete

$pattern = '#' . $attribute_name . '\s*=\s*([\'|"])(?<value>(?:(?!\1).)*)\1#is';

[345] Fix | Delete

if ( ! preg_match_all( $pattern, $element_markup, $matches, PREG_SET_ORDER ) ) {

[346] Fix | Delete

return '';

[347] Fix | Delete

}

[348] Fix | Delete

[349] Fix | Delete

return empty( $matches[0]['value'] )

[350] Fix | Delete

? ''

[351] Fix | Delete

: $matches[0]['value'];

[352] Fix | Delete

}

[353] Fix | Delete

[354] Fix | Delete

private function is_safe( $str ) {

[355] Fix | Delete

$str = trim( $str );

[356] Fix | Delete

return $this->sanitize_value( $str ) === $str;

[357] Fix | Delete

}

[358] Fix | Delete

[359] Fix | Delete

/**

[360] Fix | Delete

* This is almost the same as {@see _sanitize_text_fields()} but it doesn't remove percent encoded values because they are valid.

[361] Fix | Delete

*/

[362] Fix | Delete

private function sanitize_value( $str ) {

[363] Fix | Delete

if ( is_object( $str ) || is_array( $str ) ) {

[364] Fix | Delete

return '';

[365] Fix | Delete

}

[366] Fix | Delete

[367] Fix | Delete

$str = (string) $str;

[368] Fix | Delete

[369] Fix | Delete

$filtered = wp_check_invalid_utf8( $str );

[370] Fix | Delete

[371] Fix | Delete

if ( str_contains( $filtered, '<' ) ) {

[372] Fix | Delete

$filtered = wp_pre_kses_less_than( $filtered );

[373] Fix | Delete

// This will strip extra whitespace for us.

[374] Fix | Delete

$filtered = wp_strip_all_tags( $filtered );

[375] Fix | Delete

[376] Fix | Delete

/*

[377] Fix | Delete

* Use HTML entities in a special case to make sure that

[378] Fix | Delete

* later newline stripping stages cannot lead to a functional tag.

[379] Fix | Delete

*/

[380] Fix | Delete

$filtered = str_replace( "<\n", "&lt;\n", $filtered );

[381] Fix | Delete

}

[382] Fix | Delete

/**

[383] Fix | Delete

* Skip removal of percent encoded values {@see _sanitize_text_fields}

[384] Fix | Delete

*/

[385] Fix | Delete

return trim( $filtered );

[386] Fix | Delete

}

[387] Fix | Delete

}

[388] Fix | Delete

[389] Fix | Delete