| 1 |
<?php |
| 2 |
namespace epiphyt\Embed_Privacy\embed; |
| 3 |
|
| 4 |
use DOMDocument; |
| 5 |
use DOMElement; |
| 6 |
use DOMNode; |
| 7 |
use epiphyt\Embed_Privacy\data\Providers; |
| 8 |
use epiphyt\Embed_Privacy\data\Replacer; |
| 9 |
use epiphyt\Embed_Privacy\Embed_Privacy; |
| 10 |
|
| 11 |
/** |
| 12 |
* Embed replacement representation. |
| 13 |
* |
| 14 |
* @author Epiphyt |
| 15 |
* @license GPL2 |
| 16 |
* @package epiphyt\Embed_Privacy |
| 17 |
* @since 1.10.0 |
| 18 |
*/ |
| 19 |
final class Replacement { |
| 20 |
/** |
| 21 |
* @var string Original content |
| 22 |
*/ |
| 23 |
private $content = ''; |
| 24 |
|
| 25 |
/** |
| 26 |
* @var \epiphyt\Embed_privacy\embed\Provider|null Current processed provider for this replacement |
| 27 |
*/ |
| 28 |
private $provider; |
| 29 |
|
| 30 |
/** |
| 31 |
* @var \epiphyt\Embed_privacy\embed\Provider[] List of matching providers for this replacement |
| 32 |
*/ |
| 33 |
private $providers = []; |
| 34 |
|
| 35 |
/** |
| 36 |
* @var array List of replacements |
| 37 |
*/ |
| 38 |
private $replacements = []; |
| 39 |
|
| 40 |
/** |
| 41 |
* Replacement constructor |
| 42 |
* |
| 43 |
* @param string $content Original embedded content |
| 44 |
* @param string $url Embedded content URL |
| 45 |
*/ |
| 46 |
public function __construct( $content, $url = '' ) { |
| 47 |
$this->content = $content; |
| 48 |
$this->set_provider( $content, $url ); |
| 49 |
} |
| 50 |
|
| 51 |
/** |
| 52 |
* Get the content with an overlay. |
| 53 |
* |
| 54 |
* @param array $attributes Embed attributes |
| 55 |
* @param \epiphyt\Embed_Privacy\embed\Provider|null $provider Embed provider |
| 56 |
* @return string Content with embeds replaced by an overlay |
| 57 |
*/ |
| 58 |
public function get( array $attributes = [], $provider = null ) { |
| 59 |
/** |
| 60 |
* Filter the content after it has been replaced with an overlay. |
| 61 |
* |
| 62 |
* @since 1.10.0 |
| 63 |
* |
| 64 |
* @param string $content Replaced content |
| 65 |
*/ |
| 66 |
$content = (string) \apply_filters( 'embed_privacy_overlay_replaced_content', $this->content ); |
| 67 |
|
| 68 |
/** |
| 69 |
* If set to true, unknown providers are not handled via Embed Privacy. |
| 70 |
* |
| 71 |
* @since 1.5.0 |
| 72 |
* |
| 73 |
* @param bool $ignore_unknown Whether unknown providers should be ignored |
| 74 |
* @param string $content The original content |
| 75 |
*/ |
| 76 |
$ignore_unknown_providers = (bool) \apply_filters( 'embed_privacy_ignore_unknown_providers', false, $content ); |
| 77 |
|
| 78 |
// get default external content |
| 79 |
// special case for youtube-nocookie.com as it is part of YouTube provider |
| 80 |
if ( ! $ignore_unknown_providers || \str_contains( $content, 'youtube-nocookie.com' ) ) { |
| 81 |
if ( $provider instanceof Provider ) { |
| 82 |
$this->provider = $provider; |
| 83 |
$content = $this->replace( $content, $attributes ); |
| 84 |
} |
| 85 |
else { |
| 86 |
foreach ( $this->get_providers() as $provider ) { |
| 87 |
$this->provider = $provider; |
| 88 |
$content = $this->replace( $content, $attributes ); |
| 89 |
} |
| 90 |
} |
| 91 |
|
| 92 |
$this->provider = null; |
| 93 |
} |
| 94 |
|
| 95 |
return $content; |
| 96 |
} |
| 97 |
|
| 98 |
/** |
| 99 |
* Get a list of characters to replace to prevent problems with DOMDocument. |
| 100 |
* |
| 101 |
* @return array List of character replacements |
| 102 |
*/ |
| 103 |
private static function get_character_replacements() { |
| 104 |
$replacements = [ |
| 105 |
'%' => '@@epi_percentage', |
| 106 |
' ' => ' data-epi-spacing ', |
| 107 |
'[' => '@@epi_square_bracket_start', |
| 108 |
']' => '@@epi_square_bracket_end', |
| 109 |
'{' => '@@epi_curly_bracket_start', |
| 110 |
'}' => '@@epi_curly_bracket_end', |
| 111 |
]; |
| 112 |
|
| 113 |
/** |
| 114 |
* Filter character replacements. |
| 115 |
* |
| 116 |
* @since 1.10.0 |
| 117 |
* |
| 118 |
* @param array $replacements Current replacements |
| 119 |
*/ |
| 120 |
$replacements = (array) \apply_filters( 'embed_privacy_overlay_character_replacements', $replacements ); |
| 121 |
|
| 122 |
return $replacements; |
| 123 |
} |
| 124 |
|
| 125 |
/** |
| 126 |
* Get the current provider. |
| 127 |
* |
| 128 |
* @deprecated 1.10.4 |
| 129 |
* |
| 130 |
* @return \epiphyt\Embed_privacy\embed\Provider|null Provider object |
| 131 |
*/ |
| 132 |
public function get_provider() { |
| 133 |
\_doing_it_wrong( |
| 134 |
__METHOD__, |
| 135 |
\esc_html__( 'This method is outdated and will be removed in the future.', 'embed-privacy' ), |
| 136 |
'1.10.4' |
| 137 |
); |
| 138 |
|
| 139 |
return $this->provider; |
| 140 |
} |
| 141 |
|
| 142 |
/** |
| 143 |
* Get all providers to replace an embed of. |
| 144 |
* |
| 145 |
* @return \epiphyt\Embed_privacy\embed\Provider[] Provider object |
| 146 |
*/ |
| 147 |
public function get_providers() { |
| 148 |
return $this->providers; |
| 149 |
} |
| 150 |
|
| 151 |
/** |
| 152 |
* Replace content with an overlay and print assets. |
| 153 |
* |
| 154 |
* @param string $content Content to replace embeds in |
| 155 |
* @param array $attributes Additional attributes |
| 156 |
* @return string Replaced content |
| 157 |
*/ |
| 158 |
private function replace( $content, array $attributes ) { |
| 159 |
$new_content = $this->replace_content( $content, $attributes ); |
| 160 |
|
| 161 |
if ( $new_content !== $content ) { |
| 162 |
Embed_Privacy::get_instance()->has_embed = true; |
| 163 |
Embed_Privacy::get_instance()->frontend->print_assets(); |
| 164 |
$content = $new_content; |
| 165 |
} |
| 166 |
|
| 167 |
return $content; |
| 168 |
} |
| 169 |
|
| 170 |
/** |
| 171 |
* Replace embedded content with an overlay. |
| 172 |
* |
| 173 |
* @param string $content Content to replace embeds in |
| 174 |
* @param array $attributes Additional attributes |
| 175 |
* @return string Updated content |
| 176 |
*/ |
| 177 |
private function replace_content( $content, array $attributes ) { |
| 178 |
if ( empty( $content ) ) { |
| 179 |
return $content; |
| 180 |
} |
| 181 |
|
| 182 |
if ( ! $this->provider instanceof Provider ) { |
| 183 |
return $content; |
| 184 |
} |
| 185 |
|
| 186 |
/** |
| 187 |
* Filter whether to ignore this embed. |
| 188 |
* |
| 189 |
* @since 1.9.0 |
| 190 |
* |
| 191 |
* @param bool $ignore_embed Whether to ignore this embed |
| 192 |
* @param string $content The original content |
| 193 |
* @param string $provider_title Embed provider title |
| 194 |
* @param string $provider_name Embed provider name |
| 195 |
* @param array $attributes Additional attributes |
| 196 |
*/ |
| 197 |
$ignore_embed = (bool) \apply_filters( 'embed_privacy_ignore_embed', false, $content, $this->provider->get_title(), $this->provider->get_name(), $attributes ); |
| 198 |
|
| 199 |
if ( $ignore_embed ) { |
| 200 |
return $content; |
| 201 |
} |
| 202 |
|
| 203 |
$attributes = \wp_parse_args( $attributes, [ |
| 204 |
'additional_checks' => [], |
| 205 |
'elements' => [ 'embed', 'iframe', 'object' ], |
| 206 |
'element_attribute' => 'src', |
| 207 |
'height' => 0, |
| 208 |
'ignore_aspect_ratio' => false, |
| 209 |
'is_oembed' => false, |
| 210 |
'regex' => Replacer::extend_pattern( $this->provider->get_pattern(), $this->provider ), |
| 211 |
'strip_newlines' => ! \has_blocks( $content ), |
| 212 |
'width' => 0, |
| 213 |
] ); |
| 214 |
|
| 215 |
if ( $attributes['is_oembed'] ) { |
| 216 |
return Template::get( $this->provider, $content, $attributes ); |
| 217 |
} |
| 218 |
|
| 219 |
\libxml_use_internal_errors( true ); |
| 220 |
$dom = new DOMDocument(); |
| 221 |
$character_replacements = self::get_character_replacements(); |
| 222 |
$dom->loadHTML( |
| 223 |
'<html><meta charset="utf-8">' . \str_replace( |
| 224 |
\array_keys( $character_replacements ), |
| 225 |
\array_values( $character_replacements ), |
| 226 |
$content |
| 227 |
) . '</html>', |
| 228 |
\LIBXML_HTML_NOIMPLIED | \LIBXML_HTML_NODEFDTD |
| 229 |
); |
| 230 |
$template_dom = new DOMDocument(); |
| 231 |
// detect domain if WordPress is installed on a sub domain |
| 232 |
$host = \wp_parse_url( \home_url(), \PHP_URL_HOST ); |
| 233 |
|
| 234 |
if ( ! \filter_var( $host, \FILTER_VALIDATE_IP ) ) { |
| 235 |
$host_array = \explode( '.', \str_replace( 'www.', '', $host ) ); |
| 236 |
$tld_count = \count( $host_array ); |
| 237 |
|
| 238 |
if ( $tld_count >= 3 && \strlen( $host_array[ $tld_count - 2 ] ) === 2 ) { |
| 239 |
$host = \implode( '.', \array_splice( $host_array, $tld_count - 3, 3 ) ); |
| 240 |
} |
| 241 |
else if ( $tld_count >= 2 ) { |
| 242 |
$host = \implode( '.', \array_splice( $host_array, $tld_count - 2, $tld_count ) ); |
| 243 |
} |
| 244 |
} |
| 245 |
|
| 246 |
foreach ( $attributes['elements'] as $tag ) { |
| 247 |
$replacements = []; |
| 248 |
|
| 249 |
if ( $tag === 'object' ) { |
| 250 |
$attributes['element_attribute'] = 'data'; |
| 251 |
} |
| 252 |
|
| 253 |
/** @var \DOMElement $element */ |
| 254 |
foreach ( $dom->getElementsByTagName( $tag ) as $element ) { |
| 255 |
if ( ! Embed_Privacy::get_instance()->run_checks( $attributes['additional_checks'], $element ) ) { |
| 256 |
continue; |
| 257 |
} |
| 258 |
|
| 259 |
// ignore embeds from the same (sub-)domain |
| 260 |
if ( \preg_match( '/https?:\/\/(.*\.)?' . \preg_quote( $host, '/' ) . '/', $element->getAttribute( $attributes['element_attribute'] ) ) ) { |
| 261 |
continue; |
| 262 |
} |
| 263 |
|
| 264 |
if ( |
| 265 |
! empty( $attributes['regex'] ) |
| 266 |
&& ! \preg_match( $attributes['regex'], $element->getAttribute( $attributes['element_attribute'] ) ) |
| 267 |
) { |
| 268 |
continue; |
| 269 |
} |
| 270 |
|
| 271 |
if ( $this->provider->is_unknown() ) { |
| 272 |
$embedded_host = \wp_parse_url( $element->getAttribute( $attributes['element_attribute'] ), \PHP_URL_HOST ); |
| 273 |
|
| 274 |
// embeds with relative paths have no host |
| 275 |
// and they are local by definition, so do nothing |
| 276 |
// see https://github.com/epiphyt/embed-privacy/issues/27 |
| 277 |
if ( empty( $embedded_host ) ) { |
| 278 |
return $content; |
| 279 |
} |
| 280 |
|
| 281 |
$this->provider->set_title( $embedded_host ); |
| 282 |
$this->provider->set_name( \sanitize_title( $embedded_host ) ); |
| 283 |
|
| 284 |
// check URL for available provider |
| 285 |
foreach ( Providers::get_instance()->get_list() as $provider ) { |
| 286 |
if ( |
| 287 |
$provider->is_matching( $element->getAttribute( $attributes['element_attribute'] ) ) |
| 288 |
&& empty( $replacements ) |
| 289 |
) { |
| 290 |
continue 2; |
| 291 |
} |
| 292 |
} |
| 293 |
} |
| 294 |
|
| 295 |
/* translators: embed title */ |
| 296 |
$attributes['embed_title'] = $element->hasAttribute( 'title' ) ? $element->getAttribute( 'title' ) : ''; |
| 297 |
$attributes['embed_url'] = $element->getAttribute( $attributes['element_attribute'] ); |
| 298 |
$attributes['height'] = $element->hasAttribute( 'height' ) ? $element->getAttribute( 'height' ) : 0; |
| 299 |
$attributes['width'] = $element->hasAttribute( 'width' ) ? $element->getAttribute( 'width' ) : 0; |
| 300 |
|
| 301 |
// get overlay template as DOM element |
| 302 |
$template_dom->loadHTML( |
| 303 |
'<html><meta charset="utf-8">' . \str_replace( '%', '%_epi_', Template::get( $this->provider, $dom->saveHTML( $element ), $attributes ) ) . '</html>', |
| 304 |
\LIBXML_HTML_NOIMPLIED | \LIBXML_HTML_NODEFDTD |
| 305 |
); |
| 306 |
$overlay = null; |
| 307 |
|
| 308 |
/** @var \DOMElement $div */ |
| 309 |
foreach ( $template_dom->getElementsByTagName( 'div' ) as $div ) { |
| 310 |
if ( \stripos( $div->getAttribute( 'class' ), 'embed-privacy-container' ) !== false ) { |
| 311 |
$overlay = $div; |
| 312 |
break; |
| 313 |
} |
| 314 |
} |
| 315 |
|
| 316 |
// store the elements to replace (see regressive loop down below) |
| 317 |
if ( $overlay instanceof DOMNode || $overlay instanceof DOMElement ) { |
| 318 |
$replacements[] = [ |
| 319 |
'element' => $element, |
| 320 |
'replace' => $dom->importNode( $overlay, true ), |
| 321 |
]; |
| 322 |
} |
| 323 |
|
| 324 |
// reset embed provider name |
| 325 |
if ( $this->provider->is_unknown() ) { |
| 326 |
$this->provider->set_name( '' ); |
| 327 |
$this->provider->set_title( '' ); |
| 328 |
} |
| 329 |
} |
| 330 |
|
| 331 |
if ( ! empty( $replacements ) ) { |
| 332 |
$this->replacements = \array_merge( $this->replacements, $replacements ); |
| 333 |
Embed_Privacy::get_instance()->has_embed = true; |
| 334 |
$elements = $dom->getElementsByTagName( $tag ); |
| 335 |
$i = $elements->length - 1; |
| 336 |
|
| 337 |
// use regressive loop for replaceChild() |
| 338 |
// see: https://www.php.net/manual/en/domnode.replacechild.php#50500 |
| 339 |
while ( $i > -1 ) { |
| 340 |
$element = $elements->item( $i ); |
| 341 |
|
| 342 |
foreach ( $replacements as $replacement ) { |
| 343 |
if ( $replacement['element'] === $element ) { |
| 344 |
$element->parentNode->replaceChild( $replacement['replace'], $replacement['element'] ); // phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase |
| 345 |
} |
| 346 |
} |
| 347 |
|
| 348 |
--$i; |
| 349 |
} |
| 350 |
|
| 351 |
$content = $dom->saveHTML( $dom->documentElement ); // phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase |
| 352 |
} |
| 353 |
} |
| 354 |
|
| 355 |
\libxml_use_internal_errors( false ); |
| 356 |
|
| 357 |
$i = -1; |
| 358 |
|
| 359 |
// embeds for other elements need to be handled manually |
| 360 |
if ( |
| 361 |
empty( $this->replacements ) |
| 362 |
&& ! empty( $attributes['regex'] ) |
| 363 |
&& ! $this->provider->is_unknown() |
| 364 |
&& ! $this->provider->is_disabled() |
| 365 |
&& \preg_match_all( $attributes['regex'], $content, $matches ) >= 1 |
| 366 |
) { |
| 367 |
foreach ( $matches[0] as $matched_content ) { |
| 368 |
++$i; |
| 369 |
|
| 370 |
if ( \str_contains( $matched_content, 'embed-privacy-' ) ) { |
| 371 |
continue; |
| 372 |
} |
| 373 |
|
| 374 |
if ( empty( $matches['original_pattern'][ $i ] ) ) { |
| 375 |
continue; |
| 376 |
} |
| 377 |
|
| 378 |
// the original pattern must not be inside a href attribute |
| 379 |
if ( \str_contains( $matched_content, 'href="' . $matches['original_pattern'][ $i ] ) ) { |
| 380 |
continue; |
| 381 |
} |
| 382 |
|
| 383 |
// if the content contains an embed wrapper class, that means that the |
| 384 |
// embed is broken |
| 385 |
if ( |
| 386 |
$this->provider->is_system() |
| 387 |
&& \str_contains( $matched_content, 'class="wp-block-embed__wrapper' ) |
| 388 |
) { |
| 389 |
return $content; |
| 390 |
} |
| 391 |
|
| 392 |
$content = \str_replace( |
| 393 |
$matched_content, |
| 394 |
Template::get( |
| 395 |
$this->provider, |
| 396 |
$matched_content, |
| 397 |
$attributes |
| 398 |
), |
| 399 |
$content |
| 400 |
); |
| 401 |
} |
| 402 |
} |
| 403 |
|
| 404 |
// decode to make sure there is nothing left encoded if replacements have been made |
| 405 |
// otherwise, content is untouched by DOMDocument, and we don't need a decoding |
| 406 |
// only required for WPBakery Page Builder |
| 407 |
if ( ! empty( $this->replacements ) && \str_contains( 'vc_row', $content ) ) { |
| 408 |
$content = \rawurldecode( $content ); |
| 409 |
} |
| 410 |
|
| 411 |
return \str_replace( |
| 412 |
\array_merge( |
| 413 |
[ |
| 414 |
'<html><meta charset="utf-8">', |
| 415 |
'</html>', |
| 416 |
'%20data-epi-spacing%20', |
| 417 |
'%_epi_20data-epi-spacing%_epi_20', // % has been replaced with %_epi_ after replacing spaces |
| 418 |
], |
| 419 |
\array_values( $character_replacements ) |
| 420 |
), |
| 421 |
\array_merge( |
| 422 |
[ |
| 423 |
'', |
| 424 |
'', |
| 425 |
' ', |
| 426 |
' ', |
| 427 |
], |
| 428 |
\array_keys( $character_replacements ) |
| 429 |
), |
| 430 |
$content |
| 431 |
); |
| 432 |
} |
| 433 |
|
| 434 |
/** |
| 435 |
* Set the provider for this overlay. |
| 436 |
* |
| 437 |
* @param string $content Content to get the provider from |
| 438 |
* @param string $url URL to the embedded content |
| 439 |
*/ |
| 440 |
private function set_provider( $content, $url = '' ) { |
| 441 |
$current_provider = null; |
| 442 |
$providers = Providers::get_instance()->get_list(); |
| 443 |
|
| 444 |
foreach ( $providers as $provider ) { |
| 445 |
if ( |
| 446 |
! $provider->is_matching( |
| 447 |
$content, |
| 448 |
Replacer::extend_pattern( $provider->get_pattern(), $provider ) |
| 449 |
) |
| 450 |
&& ( empty( $url ) || ! $provider->is_matching( $url ) ) |
| 451 |
) { |
| 452 |
continue; |
| 453 |
} |
| 454 |
|
| 455 |
$current_provider = $provider; |
| 456 |
|
| 457 |
/** |
| 458 |
* Filter the overlay provider. |
| 459 |
* |
| 460 |
* @since 1.10.0 |
| 461 |
* |
| 462 |
* @param \epiphyt\Embed_Privacy\embed\Provider $provider Current provider |
| 463 |
* @param string $content Content to get the provider from |
| 464 |
* @param string $url URL to the embedded content |
| 465 |
*/ |
| 466 |
$this->providers[] = \apply_filters( 'embed_privacy_overlay_provider', $current_provider, $content, $url ); |
| 467 |
} |
| 468 |
|
| 469 |
// support unknown oEmbed provider |
| 470 |
// see https://github.com/epiphyt/embed-privacy/issues/89 |
| 471 |
if ( $current_provider === null && ! empty( $url ) ) { |
| 472 |
$parsed_url = \wp_parse_url( $url ); |
| 473 |
$provider = isset( $parsed_url['host'] ) ? $parsed_url['host'] : ''; |
| 474 |
$current_provider = new Provider(); |
| 475 |
$current_provider->set_name( $provider ); |
| 476 |
$current_provider->set_title( $provider ); |
| 477 |
} |
| 478 |
|
| 479 |
// unknown embeds |
| 480 |
if ( $current_provider === null ) { |
| 481 |
$current_provider = new Provider(); |
| 482 |
} |
| 483 |
|
| 484 |
/** |
| 485 |
* This filter is documented in inc/embed/class-replacement.php. |
| 486 |
*/ |
| 487 |
$this->providers[] = \apply_filters( 'embed_privacy_overlay_provider', $current_provider, $content, $url ); |
| 488 |
} |
| 489 |
} |
| 490 |
|
