| 1 |
<?php |
| 2 |
/** |
| 3 |
* Resources the blocker actually matched on this site. |
| 4 |
* |
| 5 |
* The catalog is a blocking-only input: nothing on Scripts and Embeds reads it. |
| 6 |
* So a resource gated purely by a catalog pattern - a YouTube embed, a Presto |
| 7 |
* video - has no row anywhere, and an admin cannot recategorise it, exclude it, |
| 8 |
* or even see that SureCookie is acting on it. The scanner cannot fill the gap |
| 9 |
* either, because by the time it looks the blocker has already replaced the |
| 10 |
* resource with a placeholder that carries no URL. |
| 11 |
* |
| 12 |
* Record what the blocker matched while it runs, and surface those as rows. The |
| 13 |
* list is per-site and stays honest: it only ever contains resources this site |
| 14 |
* actually served. |
| 15 |
* |
| 16 |
* @package SureCookie\Inc\Modules\ScriptBlocking |
| 17 |
* @since 1.5.0 |
| 18 |
*/ |
| 19 |
|
| 20 |
namespace SureCookie\Inc\Modules\ScriptBlocking; |
| 21 |
|
| 22 |
use SureCookie\Inc\Functions\Cookie_Identity; |
| 23 |
use SureCookie\Inc\Functions\Get; |
| 24 |
use SureCookie\Inc\Modules\Services\Declared_Cookies; |
| 25 |
use SureCookie\Inc\Traits\GetInstance; |
| 26 |
|
| 27 |
if ( ! defined( 'ABSPATH' ) ) { |
| 28 |
exit; // Exit if accessed directly. |
| 29 |
} |
| 30 |
|
| 31 |
/** |
| 32 |
* Matched_Resources |
| 33 |
* |
| 34 |
* @since 1.5.0 |
| 35 |
*/ |
| 36 |
class Matched_Resources { |
| 37 |
use GetInstance; |
| 38 |
|
| 39 |
/** |
| 40 |
* Option holding the per-site matched set. |
| 41 |
*/ |
| 42 |
public const OPTION = 'surecookie_matched_resources'; |
| 43 |
|
| 44 |
/** |
| 45 |
* Upper bound on stored patterns per kind, so a site that generates unique |
| 46 |
* hostnames cannot grow this without limit. |
| 47 |
*/ |
| 48 |
private const MAX_PER_KIND = 200; |
| 49 |
|
| 50 |
/** |
| 51 |
* Patterns matched during this request, as [ kind => [ pattern => entry ] ]. |
| 52 |
* |
| 53 |
* @var array<string, array<string, array<string, string>>> |
| 54 |
*/ |
| 55 |
private array $seen = []; |
| 56 |
|
| 57 |
/** |
| 58 |
* Whether this request saw something the stored set does not have. |
| 59 |
* |
| 60 |
* @var bool |
| 61 |
*/ |
| 62 |
private bool $dirty = false; |
| 63 |
|
| 64 |
/** |
| 65 |
* Constructor. |
| 66 |
* |
| 67 |
* @since 1.5.0 |
| 68 |
*/ |
| 69 |
private function __construct() { |
| 70 |
add_filter( 'surecookie_scanned_resources', [ $this, 'merge_into_payload' ] ); |
| 71 |
add_filter( 'surecookie_scanned_cookies', [ $this, 'merge_declared_cookies' ] ); |
| 72 |
add_action( 'shutdown', [ $this, 'flush' ] ); |
| 73 |
} |
| 74 |
|
| 75 |
/** |
| 76 |
* Note that the blocker matched a resource. |
| 77 |
* |
| 78 |
* Called on every match, not only on the ones that end up parked, so a |
| 79 |
* resource an admin has already excluded still has a row to switch back. |
| 80 |
* |
| 81 |
* @since 1.5.0 |
| 82 |
* @param string $kind Resource kind ('script'|'iframe'). |
| 83 |
* @param string $subject Resource URL, or the pattern that matched. |
| 84 |
* @param string $service Matched service key. |
| 85 |
* @param string $category Category the pattern resolved to. |
| 86 |
* @return void |
| 87 |
*/ |
| 88 |
public function record( string $kind, string $subject, string $service, string $category ): void { |
| 89 |
$kind = $kind === 'iframe' ? 'iframe' : 'script'; |
| 90 |
|
| 91 |
// A custom rule already has a row of its own, so it is not news here. |
| 92 |
if ( strncmp( $service, 'custom-', 7 ) === 0 ) { |
| 93 |
return; |
| 94 |
} |
| 95 |
|
| 96 |
// Catalog patterns are not all hosts: `pintrk` and `firebase-settings` |
| 97 |
// ship as inline-code keywords, and giving those a scheme invents a |
| 98 |
// domain that then shows up as a resource row. |
| 99 |
if ( strpos( $subject, '.' ) === false ) { |
| 100 |
return; |
| 101 |
} |
| 102 |
|
| 103 |
$pattern = $this->host_of( $subject ); |
| 104 |
if ( $pattern === '' || $this->is_first_party( $pattern ) ) { |
| 105 |
return; |
| 106 |
} |
| 107 |
|
| 108 |
if ( isset( $this->seen[ $kind ][ $pattern ] ) ) { |
| 109 |
return; |
| 110 |
} |
| 111 |
|
| 112 |
$this->seen[ $kind ][ $pattern ] = [ |
| 113 |
'service' => $service, |
| 114 |
'category' => $category, |
| 115 |
]; |
| 116 |
|
| 117 |
$this->dirty = true; |
| 118 |
} |
| 119 |
|
| 120 |
/** |
| 121 |
* Add a row for every matched pattern the scan does not already cover. |
| 122 |
* |
| 123 |
* @since 1.5.0 |
| 124 |
* @param mixed $resources Scanned-resources payload. |
| 125 |
* @return mixed |
| 126 |
*/ |
| 127 |
public function merge_into_payload( $resources = null ) { |
| 128 |
if ( ! is_array( $resources ) ) { |
| 129 |
return $resources; |
| 130 |
} |
| 131 |
|
| 132 |
$stored = $this->stored(); |
| 133 |
|
| 134 |
foreach ( [ |
| 135 |
'script' => 'scripts', |
| 136 |
'iframe' => 'iframes', |
| 137 |
] as $kind => $bucket ) { |
| 138 |
$rows = isset( $resources[ $bucket ] ) && is_array( $resources[ $bucket ] ) ? $resources[ $bucket ] : []; |
| 139 |
$known = []; |
| 140 |
|
| 141 |
foreach ( $rows as $row ) { |
| 142 |
if ( is_array( $row ) && isset( $row['domain'] ) ) { |
| 143 |
$known[ strtolower( (string) $row['domain'] ) ] = true; |
| 144 |
} |
| 145 |
} |
| 146 |
|
| 147 |
foreach ( (array) ( $stored[ $kind ] ?? [] ) as $pattern => $entry ) { |
| 148 |
$pattern = (string) $pattern; |
| 149 |
if ( isset( $known[ strtolower( $pattern ) ] ) ) { |
| 150 |
continue; |
| 151 |
} |
| 152 |
|
| 153 |
$rows[] = [ |
| 154 |
'domain' => $pattern, |
| 155 |
'vendor' => '', |
| 156 |
'url' => '', |
| 157 |
'category' => (string) ( $entry['category'] ?? 'uncategorized' ), |
| 158 |
'source' => 'catalog', |
| 159 |
'service_slug' => (string) ( $entry['service'] ?? '' ), |
| 160 |
]; |
| 161 |
} |
| 162 |
|
| 163 |
$resources[ $bucket ] = $rows; |
| 164 |
} |
| 165 |
|
| 166 |
return $resources; |
| 167 |
} |
| 168 |
|
| 169 |
/** |
| 170 |
* Add the declared cookies of every service we matched, for services no scan |
| 171 |
* reported. |
| 172 |
* |
| 173 |
* A gated embed is replaced before any scan sees it, so its cookies were |
| 174 |
* missing from All Cookies and from the public cookie policy even though the |
| 175 |
* site demonstrably loads that service once a visitor consents. |
| 176 |
* |
| 177 |
* @since 1.5.0 |
| 178 |
* @param mixed $cookies Cookies grouped by category id. |
| 179 |
* @return mixed |
| 180 |
*/ |
| 181 |
public function merge_declared_cookies( $cookies = null ) { |
| 182 |
if ( ! is_array( $cookies ) ) { |
| 183 |
return $cookies; |
| 184 |
} |
| 185 |
|
| 186 |
$services = $this->matched_services(); |
| 187 |
if ( empty( $services ) ) { |
| 188 |
return $cookies; |
| 189 |
} |
| 190 |
|
| 191 |
$seen = []; |
| 192 |
foreach ( $cookies as $rows ) { |
| 193 |
foreach ( (array) $rows as $row ) { |
| 194 |
if ( is_array( $row ) && isset( $row['name'] ) ) { |
| 195 |
$seen[ Cookie_Identity::key_for( $row ) ] = true; |
| 196 |
} |
| 197 |
} |
| 198 |
} |
| 199 |
|
| 200 |
foreach ( Declared_Cookies::get_instance()->build_for_services( $services ) as $category => $rows ) { |
| 201 |
foreach ( (array) $rows as $row ) { |
| 202 |
if ( ! is_array( $row ) || empty( $row['name'] ) ) { |
| 203 |
continue; |
| 204 |
} |
| 205 |
|
| 206 |
// Cookie_Identity strips the leading dot, so a declared |
| 207 |
// `.youtube.com` row dedupes against an observed `youtube.com` |
| 208 |
// one instead of both reaching the public cookie policy. |
| 209 |
$key = Cookie_Identity::key_for( $row ); |
| 210 |
if ( isset( $seen[ $key ] ) ) { |
| 211 |
continue; |
| 212 |
} |
| 213 |
|
| 214 |
$seen[ $key ] = true; |
| 215 |
$cookies[ $category ][] = $row; |
| 216 |
} |
| 217 |
} |
| 218 |
|
| 219 |
return $cookies; |
| 220 |
} |
| 221 |
|
| 222 |
/** |
| 223 |
* Persist anything new this request saw. |
| 224 |
* |
| 225 |
* Writes only when a pattern is genuinely new, so a settled site stops |
| 226 |
* touching the option entirely after the first few page views. |
| 227 |
* |
| 228 |
* @since 1.5.0 |
| 229 |
* @return void |
| 230 |
*/ |
| 231 |
public function flush(): void { |
| 232 |
if ( ! $this->dirty ) { |
| 233 |
return; |
| 234 |
} |
| 235 |
|
| 236 |
$this->dirty = false; |
| 237 |
$stored = $this->stored(); |
| 238 |
$changed = false; |
| 239 |
|
| 240 |
foreach ( $this->seen as $kind => $patterns ) { |
| 241 |
foreach ( $patterns as $pattern => $entry ) { |
| 242 |
if ( isset( $stored[ $kind ][ $pattern ] ) ) { |
| 243 |
continue; |
| 244 |
} |
| 245 |
if ( count( $stored[ $kind ] ?? [] ) >= self::MAX_PER_KIND ) { |
| 246 |
break; |
| 247 |
} |
| 248 |
|
| 249 |
$stored[ $kind ][ $pattern ] = $entry; |
| 250 |
$changed = true; |
| 251 |
} |
| 252 |
} |
| 253 |
|
| 254 |
if ( $changed ) { |
| 255 |
update_option( self::OPTION, $stored, false ); |
| 256 |
} |
| 257 |
} |
| 258 |
|
| 259 |
/** |
| 260 |
* Distinct catalog service keys recorded on this site. |
| 261 |
* |
| 262 |
* @since 1.5.0 |
| 263 |
* @return array<int, string> |
| 264 |
*/ |
| 265 |
private function matched_services(): array { |
| 266 |
$services = []; |
| 267 |
|
| 268 |
foreach ( $this->stored() as $patterns ) { |
| 269 |
foreach ( $patterns as $entry ) { |
| 270 |
$service = (string) ( $entry['service'] ?? '' ); |
| 271 |
// Scan-detected rows carry their own cookies already. |
| 272 |
if ( $service !== '' && strncmp( $service, 'scan_', 5 ) !== 0 ) { |
| 273 |
$services[ $service ] = true; |
| 274 |
} |
| 275 |
} |
| 276 |
} |
| 277 |
|
| 278 |
return array_keys( $services ); |
| 279 |
} |
| 280 |
|
| 281 |
/** |
| 282 |
* Whether a host belongs to this site. |
| 283 |
* |
| 284 |
* Presto's self-hosted and audio providers resolve to a local file, and a |
| 285 |
* first-party host recorded as a resource can be always-loaded by an admin, |
| 286 |
* which would exempt every same-host resource on the site. |
| 287 |
* |
| 288 |
* @since 1.5.0 |
| 289 |
* @param string $host Lowercased host. |
| 290 |
* @return bool |
| 291 |
*/ |
| 292 |
private function is_first_party( string $host ): bool { |
| 293 |
$site = Blocker::without_www( strtolower( (string) wp_parse_url( home_url(), PHP_URL_HOST ) ) ); |
| 294 |
$host = Blocker::without_www( $host ); |
| 295 |
|
| 296 |
if ( $site === '' ) { |
| 297 |
return false; |
| 298 |
} |
| 299 |
|
| 300 |
return $host === $site || substr( $host, - ( strlen( $site ) + 1 ) ) === '.' . $site; |
| 301 |
} |
| 302 |
|
| 303 |
/** |
| 304 |
* Host of a resource URL, or of a bare pattern. |
| 305 |
* |
| 306 |
* Recording the host rather than the catalog pattern keeps the row honest: |
| 307 |
* it names what this site actually contacted, which is also the value the |
| 308 |
* admin UI writes an exclusion or override against. |
| 309 |
* |
| 310 |
* @since 1.5.0 |
| 311 |
* @param string $subject Resource URL, or a bare blocking pattern. |
| 312 |
* @return string Lowercased host, or '' when there is none to record. |
| 313 |
*/ |
| 314 |
private function host_of( string $subject ): string { |
| 315 |
$subject = trim( $subject ); |
| 316 |
|
| 317 |
// A data: URI names no host, and a root-relative path is first-party. |
| 318 |
if ( $subject === '' || stripos( $subject, 'data:' ) === 0 || strpos( $subject, '/' ) === 0 ) { |
| 319 |
return strpos( $subject, '//' ) === 0 ? $this->host_of( 'https:' . $subject ) : ''; |
| 320 |
} |
| 321 |
|
| 322 |
if ( strpos( $subject, '://' ) === false ) { |
| 323 |
$subject = 'https://' . $subject; |
| 324 |
} |
| 325 |
|
| 326 |
$host = wp_parse_url( $subject, PHP_URL_HOST ); |
| 327 |
|
| 328 |
return is_string( $host ) ? strtolower( $host ) : ''; |
| 329 |
} |
| 330 |
|
| 331 |
/** |
| 332 |
* The stored set, normalised to both kinds. |
| 333 |
* |
| 334 |
* @since 1.5.0 |
| 335 |
* @return array<string, array<string, array<string, string>>> |
| 336 |
*/ |
| 337 |
private function stored(): array { |
| 338 |
$stored = Get::option( self::OPTION, [], 'array' ); |
| 339 |
|
| 340 |
return [ |
| 341 |
'script' => is_array( $stored['script'] ?? null ) ? $stored['script'] : [], |
| 342 |
'iframe' => is_array( $stored['iframe'] ?? null ) ? $stored['iframe'] : [], |
| 343 |
]; |
| 344 |
} |
| 345 |
} |
| 346 |
|
