| 1 |
<?php |
| 2 |
namespace ABlocksCookieConsent; |
| 3 |
|
| 4 |
if ( ! defined( 'ABSPATH' ) ) { |
| 5 |
exit; |
| 6 |
} |
| 7 |
|
| 8 |
use ABlocks\Performance\ScriptGate; |
| 9 |
|
| 10 |
/** |
| 11 |
* Layer two: everything that never went through `wp_enqueue_script`. |
| 12 |
* |
| 13 |
* This is the layer that decides whether the feature works. GTM, GA4 and the |
| 14 |
* Meta Pixel are, on the overwhelming majority of sites, pasted into a header |
| 15 |
* hook or injected by another plugin — none of that passes through |
| 16 |
* `script_loader_tag`, so layer one never sees it. |
| 17 |
* |
| 18 |
* Buffering the whole page is the highest-risk thing this addon does, and the |
| 19 |
* mitigations are deliberate rather than incidental: |
| 20 |
* |
| 21 |
* - it only runs when the addon is on and the mode is opt-in; |
| 22 |
* - it only rewrites a script a rule matched with confidence, and leaves |
| 23 |
* everything else exactly as it found it; |
| 24 |
* - anything third-party it could not classify is reported rather than guessed |
| 25 |
* at, so an incomplete configuration is visible instead of silent; |
| 26 |
* - `dry_run` runs the whole pass and rewrites nothing, which is how a site |
| 27 |
* owner finds out what would break before it breaks. |
| 28 |
*/ |
| 29 |
class Buffer { |
| 30 |
|
| 31 |
/** |
| 32 |
* Script types that actually execute. A `application/ld+json` block or an |
| 33 |
* `text/html` template is not JavaScript and must be left alone — gating a |
| 34 |
* template would silently change the page. |
| 35 |
* |
| 36 |
* @var array |
| 37 |
*/ |
| 38 |
private static $executable_types = [ |
| 39 |
'', |
| 40 |
'text/javascript', |
| 41 |
'application/javascript', |
| 42 |
'text/ecmascript', |
| 43 |
'application/ecmascript', |
| 44 |
'module', |
| 45 |
]; |
| 46 |
|
| 47 |
public static function init() { |
| 48 |
if ( ! Helper::is_gating_active() || ! Helper::get( 'buffer_gating', true ) ) { |
| 49 |
return; |
| 50 |
} |
| 51 |
|
| 52 |
$self = new self(); |
| 53 |
// Started as early as a theme's output can be, so this buffer is the |
| 54 |
// outermost one and its callback therefore runs on the finished |
| 55 |
// document — after every inner buffer has flushed into it. |
| 56 |
add_action( 'template_redirect', [ $self, 'start' ], -9999 ); |
| 57 |
} |
| 58 |
|
| 59 |
public function start() { |
| 60 |
if ( is_feed() || is_embed() || is_robots() ) { |
| 61 |
return; |
| 62 |
} |
| 63 |
ob_start( [ $this, 'filter' ] ); |
| 64 |
} |
| 65 |
|
| 66 |
/** |
| 67 |
* @param string $html The rendered document. |
| 68 |
* @return string |
| 69 |
*/ |
| 70 |
public function filter( $html ) { |
| 71 |
if ( ! is_string( $html ) || '' === $html ) { |
| 72 |
return $html; |
| 73 |
} |
| 74 |
// Cheap bail-outs first: this callback runs on every page. |
| 75 |
if ( false === stripos( $html, '<script' ) ) { |
| 76 |
return $html; |
| 77 |
} |
| 78 |
if ( ! $this->is_html_response() ) { |
| 79 |
return $html; |
| 80 |
} |
| 81 |
|
| 82 |
$dry_run = (bool) Helper::get( 'dry_run', false ); |
| 83 |
|
| 84 |
$filtered = preg_replace_callback( |
| 85 |
'#<script\b([^>]*)>(.*?)</script\s*>#is', |
| 86 |
function ( $match ) use ( $dry_run ) { |
| 87 |
return $this->process( $match, $dry_run ); |
| 88 |
}, |
| 89 |
$html |
| 90 |
); |
| 91 |
|
| 92 |
// A backtrack limit or a catastrophic pattern returns null. Serving the |
| 93 |
// original page ungated is bad; serving an empty page is worse. |
| 94 |
return null === $filtered ? $html : $filtered; |
| 95 |
} |
| 96 |
|
| 97 |
/** |
| 98 |
* Decide what to do with one script element. |
| 99 |
* |
| 100 |
* @param array $match Regex match: 0 whole, 1 attributes, 2 contents. |
| 101 |
* @param bool $dry_run Report instead of rewrite. |
| 102 |
* @return string |
| 103 |
*/ |
| 104 |
private function process( $match, $dry_run ) { |
| 105 |
$whole = $match[0]; |
| 106 |
$attrs = $match[1]; |
| 107 |
$code = $match[2]; |
| 108 |
$open = '<script ' . ltrim( $attrs ) . '>'; |
| 109 |
|
| 110 |
if ( ScriptGate::is_gated( $open ) ) { |
| 111 |
return $whole; |
| 112 |
} |
| 113 |
// The consent scripts must never gate themselves. |
| 114 |
if ( false !== stripos( $attrs, 'data-ablocks-consent' ) ) { |
| 115 |
return $whole; |
| 116 |
} |
| 117 |
if ( ! $this->is_executable( $attrs ) ) { |
| 118 |
return $whole; |
| 119 |
} |
| 120 |
|
| 121 |
$src = $this->attribute( $attrs, 'src' ); |
| 122 |
|
| 123 |
if ( $src ) { |
| 124 |
$category = Gating::match_src( $src ); |
| 125 |
$subject = $src; |
| 126 |
$source = 'external'; |
| 127 |
} else { |
| 128 |
$category = Gating::match_inline( $code ); |
| 129 |
$subject = $code; |
| 130 |
$source = 'inline'; |
| 131 |
} |
| 132 |
|
| 133 |
if ( '' === $category ) { |
| 134 |
// Not classified. Report a third-party script so the gap is |
| 135 |
// visible; leave a first-party or unrecognised inline script alone |
| 136 |
// and say nothing, because reporting every one is noise. |
| 137 |
if ( $src && Gating::is_third_party( $src ) ) { |
| 138 |
Report::add( $src, '', $source ); |
| 139 |
} |
| 140 |
return $whole; |
| 141 |
} |
| 142 |
|
| 143 |
if ( $dry_run ) { |
| 144 |
Report::add( $subject, $category, $source ); |
| 145 |
return $whole; |
| 146 |
} |
| 147 |
|
| 148 |
return ScriptGate::rewrite_tag( $open, 'text/plain', [ 'data-ablocks-consent' => $category ] ) |
| 149 |
. $code |
| 150 |
. '</script>'; |
| 151 |
} |
| 152 |
|
| 153 |
/** |
| 154 |
* Whether this script would run if left alone. |
| 155 |
* |
| 156 |
* @param string $attrs Raw attribute string. |
| 157 |
* @return bool |
| 158 |
*/ |
| 159 |
private function is_executable( $attrs ) { |
| 160 |
$type = strtolower( trim( (string) $this->attribute( $attrs, 'type' ) ) ); |
| 161 |
return in_array( $type, self::$executable_types, true ); |
| 162 |
} |
| 163 |
|
| 164 |
/** |
| 165 |
* Pull one attribute's value out of a raw attribute string. |
| 166 |
* |
| 167 |
* @param string $attrs Raw attribute string. |
| 168 |
* @param string $name Attribute name. |
| 169 |
* @return string |
| 170 |
*/ |
| 171 |
private function attribute( $attrs, $name ) { |
| 172 |
if ( preg_match( '/\s' . preg_quote( $name, '/' ) . '\s*=\s*("([^"]*)"|\'([^\']*)\'|([^\s>]+))/i', ' ' . $attrs, $found ) ) { |
| 173 |
foreach ( [ 2, 3, 4 ] as $group ) { |
| 174 |
if ( isset( $found[ $group ] ) && '' !== $found[ $group ] ) { |
| 175 |
return $found[ $group ]; |
| 176 |
} |
| 177 |
} |
| 178 |
} |
| 179 |
return ''; |
| 180 |
} |
| 181 |
|
| 182 |
/** |
| 183 |
* Only rewrite documents that are actually HTML. A plugin that hijacks the |
| 184 |
* request to return XML or a file download still passes through this |
| 185 |
* buffer, and must come out the other side untouched. |
| 186 |
* |
| 187 |
* @return bool |
| 188 |
*/ |
| 189 |
private function is_html_response() { |
| 190 |
foreach ( headers_list() as $header ) { |
| 191 |
if ( 0 === stripos( $header, 'content-type:' ) ) { |
| 192 |
return (bool) stripos( $header, 'text/html' ); |
| 193 |
} |
| 194 |
} |
| 195 |
// No explicit header means PHP's default, which is text/html. |
| 196 |
return true; |
| 197 |
} |
| 198 |
} |
| 199 |
|
