EncoderForm.php
2 months ago
Encoding.php
2 months ago
Filters.php
6 days ago
Validate.php
6 months ago
Filters.php
579 lines
| 1 | <?php |
| 2 | |
| 3 | namespace OnlineOptimisation\EmailEncoderBundle\Validate; |
| 4 | |
| 5 | use DOMDocument; |
| 6 | use OnlineOptimisation\EmailEncoderBundle\Traits\PluginHelper; |
| 7 | |
| 8 | class Filters |
| 9 | { |
| 10 | use PluginHelper; |
| 11 | |
| 12 | public function boot(): void |
| 13 | { |
| 14 | } |
| 15 | |
| 16 | |
| 17 | /** |
| 18 | * ###################### |
| 19 | * ### |
| 20 | * #### FILTERS |
| 21 | * ### |
| 22 | * ###################### |
| 23 | */ |
| 24 | |
| 25 | /** |
| 26 | * The main page filter function |
| 27 | * |
| 28 | * @param string $content - the content that needs to be filtered |
| 29 | * @param string $protect_using |
| 30 | * @return string - The filtered content |
| 31 | */ |
| 32 | public function filter_page( $content, $protect_using ) |
| 33 | { |
| 34 | |
| 35 | //Added in 2.0.6 |
| 36 | $content = apply_filters( 'eeb/validate/filter_page_content', $content, $protect_using ); |
| 37 | |
| 38 | $content = $this->filter_soft_dom_attributes( $content, 'char_encode' ); |
| 39 | |
| 40 | $htmlSplit = preg_split( '/(<body(([^>]*)>))/is', $content, -1, PREG_SPLIT_DELIM_CAPTURE ); |
| 41 | |
| 42 | if ( ! is_array( $htmlSplit) || count( $htmlSplit ) < 4 ) { |
| 43 | return $content; |
| 44 | } |
| 45 | |
| 46 | switch ( $protect_using ) { |
| 47 | case 'with_javascript': |
| 48 | case 'without_javascript': |
| 49 | case 'char_encode': |
| 50 | $head_encoding_method = 'char_encode'; |
| 51 | break; |
| 52 | default: |
| 53 | $head_encoding_method = 'default'; |
| 54 | break; |
| 55 | } |
| 56 | |
| 57 | //Filter head area |
| 58 | $filtered_head = $this->filter_plain_emails( $htmlSplit[0], null, $head_encoding_method ); |
| 59 | |
| 60 | //Filter body |
| 61 | //Soft attributes always need to be protected using only the char encode method since otherwise the logic breaks |
| 62 | $filtered_body = $this->filter_soft_attributes( $htmlSplit[4], 'char_encode' ); |
| 63 | $filtered_body = $this->filter_content( $filtered_body, $protect_using ); |
| 64 | |
| 65 | $filtered_content = $filtered_head . $htmlSplit[1] . $filtered_body; |
| 66 | |
| 67 | //Revalidate filtered emails that should not bbe encoded |
| 68 | $filtered_content = $this->tempEncodeAtSymbol( $filtered_content, true ); |
| 69 | |
| 70 | return $filtered_content; |
| 71 | } |
| 72 | |
| 73 | /** |
| 74 | * Filter content |
| 75 | * |
| 76 | * @param string $content |
| 77 | * @param string $protect_using |
| 78 | * @return string |
| 79 | */ |
| 80 | public function filter_content( $content, $protect_using ) |
| 81 | { |
| 82 | $filtered = $content; |
| 83 | $self = $this; |
| 84 | $encode_mailtos = (bool) $this->getSetting( 'encode_mailtos', true, 'filter_body' ); |
| 85 | $convert_plain_to_image = (bool) $this->getSetting( 'convert_plain_to_image', true, 'filter_body' ); |
| 86 | |
| 87 | //Added in 2.0.6 |
| 88 | $filtered = apply_filters( 'eeb/validate/filter_content_content', $filtered, $protect_using ); |
| 89 | |
| 90 | //Soft attributes always need to be protected using only the char encode method since otherwise the logic breaks |
| 91 | $filtered = $this->filter_soft_attributes( $filtered, 'char_encode' ); |
| 92 | |
| 93 | // <option>, <textarea>, <title> can only contain text — any <span>/<script>/<img> inside |
| 94 | // them is stripped or rendered inconsistently across browsers (Firefox dropdowns show only |
| 95 | // the noscript fallback). For modes that emit HTML wrappers, pre-encode emails in those |
| 96 | // zones as entities and stash them behind placeholders so the main filter pass skips them. |
| 97 | $text_only_tokens = []; |
| 98 | $guard_text_only_zones = in_array( $protect_using, [ 'with_javascript', 'without_javascript' ], true ); |
| 99 | if ( $guard_text_only_zones ) { |
| 100 | $filtered = $this->isolate_text_only_zones( $filtered, $text_only_tokens ); |
| 101 | } |
| 102 | |
| 103 | switch ( $protect_using ) { |
| 104 | case 'char_encode': |
| 105 | $filtered = $this->filter_plain_emails( $filtered, null, 'char_encode' ); |
| 106 | break; |
| 107 | case 'strong_method': |
| 108 | $filtered = $this->filter_plain_emails( $filtered ); |
| 109 | break; |
| 110 | case 'without_javascript': |
| 111 | $filtered = $this->filter_input_fields( $filtered, $protect_using ); |
| 112 | $filtered = $this->filter_mailto_links( $filtered, 'without_javascript' ); |
| 113 | $filtered = $this->filter_custom_links( $filtered, 'without_javascript' ); |
| 114 | |
| 115 | if ( $convert_plain_to_image ) { |
| 116 | $replace_by = 'convert_image'; |
| 117 | } else { |
| 118 | $replace_by = 'use_css'; |
| 119 | } |
| 120 | |
| 121 | if ( $encode_mailtos ) { |
| 122 | if ( ! ( function_exists( 'et_fb_enabled' ) && et_fb_enabled() ) ) { |
| 123 | $filtered = $this->filter_plain_emails( $filtered, function ( $match ) use ( $self ) { |
| 124 | return $self->createProtectedMailto( $match[0], array( 'href' => 'mailto:' . $match[0] ), 'without_javascript' ); |
| 125 | }, $replace_by); |
| 126 | } else { |
| 127 | $filtered = $this->filter_plain_emails( $filtered, null, $replace_by ); |
| 128 | } |
| 129 | } else { |
| 130 | $filtered = $this->filter_plain_emails( $filtered, null, $replace_by ); |
| 131 | } |
| 132 | |
| 133 | break; |
| 134 | case 'with_javascript': |
| 135 | $filtered = $this->filter_input_fields( $filtered, $protect_using ); |
| 136 | $filtered = $this->filter_mailto_links( $filtered ); |
| 137 | $filtered = $this->filter_custom_links( $filtered ); |
| 138 | |
| 139 | if ( $convert_plain_to_image ) { |
| 140 | $replace_by = 'convert_image'; |
| 141 | } else { |
| 142 | $replace_by = 'use_javascript'; |
| 143 | } |
| 144 | |
| 145 | if ( $encode_mailtos ) { |
| 146 | if ( ! ( function_exists( 'et_fb_enabled' ) && et_fb_enabled() ) ) { |
| 147 | $filtered = $this->filter_plain_emails( $filtered, function ( $match ) use ( $self ) { |
| 148 | return $self->createProtectedMailto( $match[0], array( 'href' => 'mailto:' . $match[0] ), 'with_javascript' ); |
| 149 | }, $replace_by); |
| 150 | } else { |
| 151 | $filtered = $this->filter_plain_emails( $filtered, null, $replace_by ); |
| 152 | } |
| 153 | } else { |
| 154 | $filtered = $this->filter_plain_emails( $filtered, null, $replace_by ); |
| 155 | } |
| 156 | |
| 157 | break; |
| 158 | } |
| 159 | |
| 160 | //Revalidate filtered emails that should not be encoded |
| 161 | $filtered = $this->tempEncodeAtSymbol( $filtered, true ); |
| 162 | |
| 163 | if ( $guard_text_only_zones && ! empty( $text_only_tokens ) ) { |
| 164 | $filtered = strtr( $filtered, $text_only_tokens ); |
| 165 | } |
| 166 | |
| 167 | return $filtered; |
| 168 | } |
| 169 | |
| 170 | /** |
| 171 | * Pre-encode emails inside text-only HTML zones (<option>, <textarea>, <title>) using |
| 172 | * entity encoding and replace each zone with an opaque placeholder. The main filter pass |
| 173 | * leaves the placeholders alone; the caller restores them at the end. |
| 174 | * |
| 175 | * @param string $content |
| 176 | * @param array<string, string> $tokens Populated with placeholder => replacement pairs. |
| 177 | * @return string Content with text-only zones replaced by placeholder tokens. |
| 178 | */ |
| 179 | private function isolate_text_only_zones( string $content, array &$tokens ): string |
| 180 | { |
| 181 | $result = preg_replace_callback( |
| 182 | '/(<(option|textarea|title)\b[^>]*>)(.*?)(<\/\2\s*>)/is', |
| 183 | function ( array $match ) use ( &$tokens ) { |
| 184 | $inner_encoded = $this->filter_plain_emails( $match[3], null, 'char_encode', false ); |
| 185 | $token = "\x00EEB_TEXT_ONLY_" . count( $tokens ) . "\x00"; |
| 186 | $tokens[ $token ] = $match[1] . $inner_encoded . $match[4]; |
| 187 | |
| 188 | return $token; |
| 189 | }, |
| 190 | $content |
| 191 | ); |
| 192 | |
| 193 | return $result ?? $content; |
| 194 | } |
| 195 | |
| 196 | /** |
| 197 | * Emails will be replaced by '*protected email*' |
| 198 | * @param string $content |
| 199 | * @param string|callable $replace_by Optional |
| 200 | * @param string $protection_method Optional |
| 201 | * @param mixed $show_encoded_check Optional |
| 202 | * @return string |
| 203 | */ |
| 204 | public function filter_plain_emails( $content, $replace_by = null, $protection_method = 'default', $show_encoded_check = 'default' ) |
| 205 | { |
| 206 | |
| 207 | if ( $show_encoded_check === 'default' ) { |
| 208 | $show_encoded_check = (bool) $this->getSetting( 'show_encoded_check', true ); |
| 209 | } |
| 210 | |
| 211 | if ( $replace_by === null ) { |
| 212 | $replace_by = (string) $this->getSetting( 'protection_text', true ); |
| 213 | } |
| 214 | |
| 215 | $self = $this; |
| 216 | |
| 217 | return preg_replace_callback( $this->settings()->get_email_regex(), function ( $matches ) use ( $replace_by, $protection_method, $show_encoded_check, $self ) { |
| 218 | // workaround to skip responsive image names containing @ |
| 219 | $extention = strtolower( $matches[4] ); |
| 220 | $excludedList = array( |
| 221 | '.jpg', |
| 222 | '.jpeg', |
| 223 | '.png', |
| 224 | '.gif', |
| 225 | '.svg', |
| 226 | '.webp', |
| 227 | '.bmp', |
| 228 | '.tiff', |
| 229 | '.avif', |
| 230 | ); |
| 231 | |
| 232 | //Added in 2.1.1 |
| 233 | $excludedList = apply_filters( 'eeb/validate/excluded_image_urls', $excludedList ); |
| 234 | |
| 235 | if ( in_array( $extention, $excludedList ) ) { |
| 236 | return $matches[0]; |
| 237 | } |
| 238 | |
| 239 | if ( is_callable( $replace_by ) ) { |
| 240 | return call_user_func( $replace_by, $matches, $protection_method ); |
| 241 | } |
| 242 | |
| 243 | if ( $protection_method === 'char_encode' ) { |
| 244 | $protected_return = antispambot( $matches[0] ); |
| 245 | } elseif ( $protection_method === 'convert_image' ) { |
| 246 | |
| 247 | $image_link = $self->generateEmailImageUrl( $matches[0] ); |
| 248 | if ( ! empty( $image_link ) ) { |
| 249 | $protected_return = '<img src="' . $image_link . '" />'; |
| 250 | } else { |
| 251 | $protected_return = antispambot( $matches[0] ); |
| 252 | } |
| 253 | |
| 254 | } elseif ( $protection_method === 'use_javascript' ) { |
| 255 | $protection_text = (string) $this->getSetting( 'protection_text', true ); |
| 256 | $protected_return = $this->dynamicJsEmailEncoding( $matches[0], $protection_text ); |
| 257 | } elseif ( $protection_method === 'use_css' ) { |
| 258 | $protection_text = (string) $this->getSetting( 'protection_text', true ); |
| 259 | // $protected_return = $this->validate()->encoding->encode_email_css( $matches[0], $protection_text ); |
| 260 | $protected_return = $this->validate()->encoding->encode_email_css( $matches[0] ); |
| 261 | } elseif ( $protection_method === 'no_encoding' ) { |
| 262 | $protected_return = $matches[0]; |
| 263 | } else { |
| 264 | $protected_return = $replace_by; |
| 265 | } |
| 266 | |
| 267 | // mark link as successfully encoded (for admin users) |
| 268 | if ( current_user_can( $this->getAdminCap( 'frontend-display-security-check' ) ) && $show_encoded_check ) { |
| 269 | $protected_return .= $this->getEncodedEmailIcon(); |
| 270 | } |
| 271 | |
| 272 | return $protected_return; |
| 273 | }, $content ) ?? ''; |
| 274 | } |
| 275 | |
| 276 | /** |
| 277 | * Filter passed input fields |
| 278 | * |
| 279 | * @param string $content |
| 280 | * @return string |
| 281 | */ |
| 282 | public function filter_input_fields( string $content, string $encoding_method = 'default' ) |
| 283 | { |
| 284 | $strong_encoding = (bool) $this->getSetting( 'input_strong_protection', true, 'filter_body' ); |
| 285 | |
| 286 | $callback_encode_input_fields = function ( $match ) use ( $encoding_method, $strong_encoding ) { |
| 287 | $input = $match[0]; |
| 288 | $email = $match[2]; |
| 289 | |
| 290 | //Only allow strong encoding if javascript is supported |
| 291 | if ( $encoding_method === 'without_javascript' ) { |
| 292 | $strong_encoding = false; |
| 293 | } |
| 294 | |
| 295 | return $this->validate()->encoding->encode_input_field( $input, $email, $strong_encoding ); |
| 296 | }; |
| 297 | |
| 298 | $regexpInputField = '/<input([^>]*)value=["\'][\s+]*' . $this->settings()->get_email_regex( true ) . '[\s+]*["\']([^>]*)>/is'; |
| 299 | |
| 300 | return preg_replace_callback( $regexpInputField, $callback_encode_input_fields, $content ) ?? ''; |
| 301 | } |
| 302 | |
| 303 | /** |
| 304 | * @param string $content |
| 305 | * @param string $protection_method |
| 306 | * @return string |
| 307 | */ |
| 308 | public function filter_mailto_links( string $content, ?string $protection_method = null ) |
| 309 | { |
| 310 | $self = $this; |
| 311 | |
| 312 | $callbackEncodeMailtoLinks = function ( $match ) use ( $self, $protection_method ) { |
| 313 | $attrs = $this->helper()->parse_html_attributes( $match[1] ); |
| 314 | return $self->createProtectedMailto( $match[4], $attrs, $protection_method ); |
| 315 | }; |
| 316 | |
| 317 | $regexpMailtoLink = '/<a[\s+]*(([^>]*)href=["\']mailto\:([^>]*)["\' ])>(.*?)<\/a[\s+]*>/is'; |
| 318 | |
| 319 | return preg_replace_callback( $regexpMailtoLink, $callbackEncodeMailtoLinks, $content ) ?? ''; |
| 320 | } |
| 321 | |
| 322 | /** |
| 323 | * @param string $content |
| 324 | * @param string $protection_method |
| 325 | * @return string |
| 326 | */ |
| 327 | public function filter_custom_links( string $content, ?string $protection_method = null ) |
| 328 | { |
| 329 | $self = $this; |
| 330 | $custom_href_attr = (string) $this->getSetting( 'custom_href_attr', true ); |
| 331 | |
| 332 | if ( ! empty( $custom_href_attr ) ) { |
| 333 | $custom_attr_list = explode( ',', $custom_href_attr ); |
| 334 | foreach ( $custom_attr_list as $s_attr ) { |
| 335 | $attr_name = trim( $s_attr ); |
| 336 | |
| 337 | $callbackEncodeCustomLinks = function ( $match ) use ( $self, $protection_method ) { |
| 338 | $attrs = shortcode_parse_atts( $match[1] ); |
| 339 | return $self->createProtectedHrefAtt( $match[4], $attrs, $protection_method ); |
| 340 | }; |
| 341 | |
| 342 | $regexpMailtoLink = '/<a[\s+]*(([^>]*)href=["\']' . addslashes( $attr_name ) . '\:([^>]*)["\' ])>(.*?)<\/a[\s+]*>/is'; |
| 343 | |
| 344 | $content = preg_replace_callback( $regexpMailtoLink, $callbackEncodeCustomLinks, $content ) ?? ''; |
| 345 | } |
| 346 | } |
| 347 | |
| 348 | return $content; |
| 349 | } |
| 350 | |
| 351 | /** |
| 352 | * Emails will be replaced by '*protected email*' |
| 353 | * |
| 354 | * @param string $content |
| 355 | * @param string $protection_type |
| 356 | * @return string |
| 357 | */ |
| 358 | public function filter_rss( string $content, ?string $protection_type ) |
| 359 | { |
| 360 | |
| 361 | if ( $protection_type === 'strong_method' ) { |
| 362 | $filtered = $this->filter_plain_emails( $content ); |
| 363 | } else { |
| 364 | $filtered = $this->filter_plain_emails( $content, null, 'char_encode' ); |
| 365 | } |
| 366 | |
| 367 | return $filtered; |
| 368 | } |
| 369 | |
| 370 | /** |
| 371 | * Filter plain emails using soft attributes |
| 372 | * |
| 373 | * @param string $content - the content that should be soft filtered |
| 374 | * @param string $protection_method - The method (E.g. char_encode) |
| 375 | * @return string |
| 376 | */ |
| 377 | public function filter_soft_attributes( string $content, string $protection_method ) |
| 378 | { |
| 379 | $soft_attributes = (array) $this->settings()->get_soft_attribute_regex(); |
| 380 | |
| 381 | foreach ( $soft_attributes as $ident => $regex ) { |
| 382 | |
| 383 | // $attributes = array(); |
| 384 | preg_match_all( $regex, $content, $attributes ); |
| 385 | |
| 386 | // if ( is_array( $attributes ) && isset( $attributes[0] ) ) { |
| 387 | foreach ( $attributes[0] as $single ) { |
| 388 | |
| 389 | if ( empty( $single ) ) { |
| 390 | continue; |
| 391 | } |
| 392 | |
| 393 | $content = str_replace( $single, $this->filter_plain_emails( $single, null, $protection_method, false ), $content ); |
| 394 | } |
| 395 | // } |
| 396 | |
| 397 | } |
| 398 | |
| 399 | return $content; |
| 400 | } |
| 401 | |
| 402 | /** |
| 403 | * Filter plain emails using soft dom attributes |
| 404 | * |
| 405 | * @param string $content - the content that should be soft filtered |
| 406 | * @param string $protection_method - The method (E.g. char_encode) |
| 407 | * @return string |
| 408 | */ |
| 409 | public function filter_soft_dom_attributes( $content, $protection_method ) |
| 410 | { |
| 411 | $content = (string) $content; |
| 412 | |
| 413 | $no_script_tags = (bool) $this->getSetting( 'no_script_tags', true, 'filter_body' ); |
| 414 | $no_attribute_validation = (bool) $this->getSetting( 'no_attribute_validation', true, 'filter_body' ); |
| 415 | |
| 416 | if ( $content !== '' ) { |
| 417 | |
| 418 | if ( class_exists( 'DOMDocument' ) ) { |
| 419 | $dom = new DOMDocument(); |
| 420 | @$dom->loadHTML($content); |
| 421 | |
| 422 | //Filter html attributes |
| 423 | if ( ! $no_attribute_validation ) { |
| 424 | $allNodes = $dom->getElementsByTagName('*'); |
| 425 | foreach ( $allNodes as $snote ) { |
| 426 | if ( $snote->hasAttributes() ) { |
| 427 | foreach ( $snote->attributes as $attr ) { |
| 428 | if ( $attr->nodeName == 'href' || $attr->nodeName == 'src' ) { |
| 429 | continue; |
| 430 | } |
| 431 | |
| 432 | if ( $attr->nodeValue !== null && strpos( $attr->nodeValue, '@' ) !== false ) { |
| 433 | // $single_tags = array(); |
| 434 | preg_match_all( '/' . $attr->nodeName . '=["\']([^"]*)["\']/i', $content, $single_tags ); |
| 435 | |
| 436 | // if ( is_array( $single_tags ) && isset( $single_tags[0] ) ) { |
| 437 | foreach ( $single_tags[0] as $single ) { |
| 438 | |
| 439 | if ( empty( $single ) ) { |
| 440 | continue; |
| 441 | } |
| 442 | |
| 443 | $content = str_replace( $single, $this->filter_plain_emails( $single, null, $protection_method, false ), $content ); |
| 444 | } |
| 445 | // } |
| 446 | |
| 447 | } |
| 448 | } |
| 449 | } |
| 450 | } |
| 451 | } |
| 452 | |
| 453 | //Keep for now |
| 454 | //Soft-encode scripts |
| 455 | // $script = $dom->getElementsByTagName('script'); |
| 456 | // if ( ! empty( $script ) ) { |
| 457 | // $scripts_encoded = true; |
| 458 | |
| 459 | // if ( ! $no_script_tags ) { |
| 460 | // foreach( $script as $item ) { |
| 461 | // $content = str_replace( $item->nodeValue, $this->filter_plain_emails( $item->nodeValue, null, $protection_method, false ), $content ); |
| 462 | // } |
| 463 | // } else { |
| 464 | // foreach( $script as $item ) { |
| 465 | // $content = str_replace( $item->nodeValue, $this->temp_encode_at_symbol( $item->nodeValue ), $content ); |
| 466 | // } |
| 467 | // } |
| 468 | // } |
| 469 | |
| 470 | } |
| 471 | |
| 472 | //Validate script tags for better encoding |
| 473 | $pattern = '/<script\b[^>]*>(.*?)<\/script>/is'; |
| 474 | |
| 475 | // On very large pages — or hosts with a low pcre.backtrack_limit — this |
| 476 | // lazy regex can bail out with PREG_BACKTRACK_LIMIT_ERROR and return no |
| 477 | // matches. That silently disables @-masking inside <script> blocks, so the |
| 478 | // encoder later injects a raw </script> into inline JSON data blobs (e.g. a |
| 479 | // page builder's FAQ/search payload), prematurely closing the host script |
| 480 | // and dumping its contents as visible text below the page. Fall back to a |
| 481 | // linear, backtrack-free scan so the guard keeps working on those pages. |
| 482 | if ( preg_match_all( $pattern, $content, $matches ) === false || preg_last_error() !== PREG_NO_ERROR ) { |
| 483 | $matches = $this->extract_script_blocks( $content ); |
| 484 | } |
| 485 | |
| 486 | if ( ! empty( $matches[1] ) ) { |
| 487 | if ( ! $no_script_tags ) { |
| 488 | foreach ( $matches[1] as $key => $item ) { |
| 489 | |
| 490 | //Don't do anything if something doesn't add up |
| 491 | if ( ! isset( $matches[0][ $key ] ) ) { |
| 492 | continue; |
| 493 | } |
| 494 | |
| 495 | $org_script = $matches[0][ $key ]; |
| 496 | |
| 497 | //Only encode emails when a CDATA is given to not cause any break within the scripts |
| 498 | if ( strpos( $item, '<![CDATA' ) !== false ) { |
| 499 | $validated_script = str_replace( $item, $this->filter_plain_emails( $item, null, $protection_method, false ), $org_script ); |
| 500 | } else { |
| 501 | $validated_script = str_replace( $item, $this->tempEncodeAtSymbol( $item ), $org_script ); |
| 502 | } |
| 503 | |
| 504 | $content = str_replace( $org_script, $validated_script, $content ); |
| 505 | } |
| 506 | } else { |
| 507 | foreach ( $matches[1] as $key => $item ) { |
| 508 | |
| 509 | //Don't do anything if something doesn't add up |
| 510 | if ( ! isset( $matches[0][ $key ] ) ) { |
| 511 | continue; |
| 512 | } |
| 513 | |
| 514 | $org_script = $matches[0][ $key ]; |
| 515 | $validated_script = str_replace( $item, $this->tempEncodeAtSymbol( $item ), $org_script ); |
| 516 | |
| 517 | $content = str_replace( $org_script, $validated_script, $content ); |
| 518 | } |
| 519 | } |
| 520 | } |
| 521 | |
| 522 | } |
| 523 | |
| 524 | return $content; |
| 525 | } |
| 526 | |
| 527 | /** |
| 528 | * Linear, backtrack-free extraction of <script>…</script> blocks. |
| 529 | * |
| 530 | * Mirrors the capture shape of preg_match_all( '/<script\b[^>]*>(.*?)<\/script>/is' ) |
| 531 | * — index 0 holds the full tags, index 1 the bodies — but without the lazy |
| 532 | * quantifier, so it cannot hit pcre.backtrack_limit on very large pages. Used as a |
| 533 | * fallback when that regex errors out (see filter_soft_dom_attributes). |
| 534 | * |
| 535 | * @param string $content |
| 536 | * @return array{0: array<int, string>, 1: array<int, string>} |
| 537 | */ |
| 538 | private function extract_script_blocks( string $content ) |
| 539 | { |
| 540 | $full = array(); |
| 541 | $bodies = array(); |
| 542 | $length = strlen( $content ); |
| 543 | $offset = 0; |
| 544 | |
| 545 | while ( ( $open = stripos( $content, '<script', $offset ) ) !== false ) { |
| 546 | |
| 547 | // Enforce the regex's \b: the char after "script" must be a non-word char, |
| 548 | // so tags like <scripting> are not treated as <script>. |
| 549 | $after = $open + 7; // strlen( '<script' ) |
| 550 | $boundary = ( $after < $length ) ? $content[ $after ] : ''; |
| 551 | if ( $boundary !== '' && ( ctype_alnum( $boundary ) || $boundary === '_' ) ) { |
| 552 | $offset = $after; |
| 553 | continue; |
| 554 | } |
| 555 | |
| 556 | // End of the opening tag ( first '>' — matches the regex's [^>]*> ). |
| 557 | $tag_end = strpos( $content, '>', $open ); |
| 558 | if ( $tag_end === false ) { |
| 559 | break; |
| 560 | } |
| 561 | $body_start = $tag_end + 1; |
| 562 | |
| 563 | // First literal </script> ( matches the lazy .*? up to <\/script> ). |
| 564 | $close = stripos( $content, '</script>', $body_start ); |
| 565 | if ( $close === false ) { |
| 566 | break; |
| 567 | } |
| 568 | $full_end = $close + 9; // strlen( '</script>' ) |
| 569 | |
| 570 | $full[] = substr( $content, $open, $full_end - $open ); |
| 571 | $bodies[] = substr( $content, $body_start, $close - $body_start ); |
| 572 | |
| 573 | $offset = $full_end; |
| 574 | } |
| 575 | |
| 576 | return array( $full, $bodies ); |
| 577 | } |
| 578 | } |
| 579 |