class-cssprocessor.php
1877 lines
| 1 | <?php |
| 2 | |
| 3 | namespace WordPress\DataLiberation\CSS; |
| 4 | |
| 5 | use function WordPress\Encoding\codepoint_to_utf8_bytes; |
| 6 | use function WordPress\Encoding\compat\_wp_scan_utf8; |
| 7 | use function WordPress\Encoding\utf8_ord; |
| 8 | use function WordPress\Encoding\wp_scrub_utf8; |
| 9 | |
| 10 | /** |
| 11 | * Tokenizes CSS according to the CSS Syntax Level 3 specification. |
| 12 | * |
| 13 | * This class follows the algorithm in https://www.w3.org/TR/css-syntax-3/ and |
| 14 | * exposes a pull-based API so callers can stream over large stylesheets without |
| 15 | * allocating every token up front. Each call to next_token() advances the cursor |
| 16 | * and fills in metadata (type, value, raw slice, byte offsets) that you can read |
| 17 | * through the getter methods. |
| 18 | * |
| 19 | * ## Design choices |
| 20 | * |
| 21 | * ### On-the-fly normalization |
| 22 | * |
| 23 | * The CSS Spec requires the following normalization step: |
| 24 | * |
| 25 | * > Replace any U+000D CARRIAGE RETURN (CR) code points, U+000C FORM FEED (FF) |
| 26 | * > code points, or pairs of U+000D CARRIAGE RETURN (CR) followed by U+000A LINE |
| 27 | * > FEED (LF) in input by a single U+000A LINE FEED (LF) code point. |
| 28 | * > Replace any U+0000 NULL or surrogate code points in input with U+FFFD REPLACEMENT |
| 29 | * > CHARACTER (�). |
| 30 | * |
| 31 | * This processor delays normalization as much as possible. That keeps the raw byte |
| 32 | * positions intact for accurate rewrites while still letting consumers ask for a |
| 33 | * normalized token when they need one. |
| 34 | * |
| 35 | * ### No EOF token |
| 36 | * |
| 37 | * The EOF token is a CSS parsing concept, not CSS tokenization concept. Therefore, |
| 38 | * this processor does not produce it. |
| 39 | * |
| 40 | * ### UTF-8 handling |
| 41 | * |
| 42 | * Only UTF-8 strings are supported. Invalid sequences are replaced with U+FFFD (�) |
| 43 | * using the maximal subpart approach described in |
| 44 | * https://www.unicode.org/versions/Unicode9.0.0/ch03.pdf, section 3.9 Best Practices |
| 45 | * for Using U+FFFD. |
| 46 | * |
| 47 | * ## Usage |
| 48 | * |
| 49 | * Basic iteration: |
| 50 | * |
| 51 | * ```php |
| 52 | * $css = 'width: 10px;'; |
| 53 | * $processor = CSSProcessor::create( $css ); |
| 54 | * while ( $processor->next_token() ) { |
| 55 | * echo $processor->get_normalized_token(); |
| 56 | * } |
| 57 | * // Outputs: |
| 58 | * // width: 10px; |
| 59 | * ``` |
| 60 | * |
| 61 | * Rewriting a URL while keeping the rest of the stylesheet intact: |
| 62 | * |
| 63 | * ```php |
| 64 | * $css = 'background: url(old.jpg) center / cover;'; |
| 65 | * $processor = CSSProcessor::create( $css ); |
| 66 | * while ( $processor->next_token() ) { |
| 67 | * if ( CSSProcessor::TOKEN_URL === $processor->get_token_type() ) { |
| 68 | * $processor->set_value( 'uploads/new.jpg' ); |
| 69 | * } |
| 70 | * } |
| 71 | * $result = $processor->get_updated_css(); |
| 72 | * // background: url(uploads/new.jpg) center / cover; |
| 73 | * ``` |
| 74 | * |
| 75 | * Gathering diagnostics with byte offsets: |
| 76 | * |
| 77 | * ```php |
| 78 | * $css = "color: red;\ncolor: re\nd;"; |
| 79 | * $processor = CSSProcessor::create( $css ); |
| 80 | * $bad_strings = array(); |
| 81 | * while ( $processor->next_token() ) { |
| 82 | * if ( CSSProcessor::TOKEN_BAD_STRING === $processor->get_token_type() ) { |
| 83 | * $bad_strings[] = array( |
| 84 | * 'start' => $processor->get_token_start(), |
| 85 | * 'length' => $processor->get_token_length(), |
| 86 | * 'value' => $processor->get_unnormalized_token(), |
| 87 | * ); |
| 88 | * } |
| 89 | * } |
| 90 | * ``` |
| 91 | * |
| 92 | * @see https://www.w3.org/TR/css-syntax-3/#tokenization |
| 93 | */ |
| 94 | class CSSProcessor { |
| 95 | /** |
| 96 | * Token type constants matching the CSS Syntax Level 3 specification. |
| 97 | * |
| 98 | * @see https://www.w3.org/TR/css-syntax-3/#tokenization |
| 99 | */ |
| 100 | public const TOKEN_WHITESPACE = 'whitespace-token'; |
| 101 | public const TOKEN_COMMENT = 'comment'; |
| 102 | public const TOKEN_STRING = 'string-token'; |
| 103 | |
| 104 | /** |
| 105 | * BAD-STRING tokens occur when a string contains an unescaped newline. |
| 106 | * |
| 107 | * Valid strings: "hello", 'world', "line1\Aline2" (escaped newline) |
| 108 | * Invalid (produces bad-string): "hello |
| 109 | * world" (literal newline breaks the string) |
| 110 | * |
| 111 | * The processor stops at the newline and produces a bad-string token for error recovery. |
| 112 | * |
| 113 | * @see https://www.w3.org/TR/css-syntax-3/#typedef-bad-string-token |
| 114 | */ |
| 115 | public const TOKEN_BAD_STRING = 'bad-string-token'; |
| 116 | public const TOKEN_HASH = 'hash-token'; |
| 117 | public const TOKEN_DELIM = 'delim-token'; |
| 118 | public const TOKEN_NUMBER = 'number-token'; |
| 119 | public const TOKEN_PERCENTAGE = 'percentage-token'; |
| 120 | public const TOKEN_DIMENSION = 'dimension-token'; |
| 121 | public const TOKEN_AT_KEYWORD = 'at-keyword-token'; |
| 122 | public const TOKEN_COLON = 'colon-token'; |
| 123 | public const TOKEN_SEMICOLON = 'semicolon-token'; |
| 124 | public const TOKEN_COMMA = 'comma-token'; |
| 125 | public const TOKEN_LEFT_PAREN = '(-token'; |
| 126 | public const TOKEN_RIGHT_PAREN = ')-token'; |
| 127 | public const TOKEN_LEFT_BRACKET = '[-token'; |
| 128 | public const TOKEN_RIGHT_BRACKET = ']-token'; |
| 129 | public const TOKEN_LEFT_BRACE = '{-token'; |
| 130 | public const TOKEN_RIGHT_BRACE = '}-token'; |
| 131 | public const TOKEN_FUNCTION = 'function-token'; |
| 132 | |
| 133 | /** |
| 134 | * URL tokens represent unquoted URLs in url() notation. |
| 135 | * |
| 136 | * Valid: url(image.jpg), url(https://example.com) |
| 137 | * Quoted URLs are parsed as url( + string-token + ), not url-token. |
| 138 | * |
| 139 | * @see https://www.w3.org/TR/css-syntax-3/#typedef-url-token |
| 140 | */ |
| 141 | public const TOKEN_URL = 'url-token'; |
| 142 | |
| 143 | /** |
| 144 | * BAD-URL tokens occur when a URL contains invalid characters. |
| 145 | * |
| 146 | * Invalid characters: quotes ("), apostrophes ('), parentheses (() |
| 147 | * Example invalid: url(image(.jpg) or url(image".jpg) |
| 148 | * |
| 149 | * When detected, the processor consumes everything up to ) or EOF. |
| 150 | * This prevents the bad URL from breaking subsequent tokens. |
| 151 | * |
| 152 | * @see https://www.w3.org/TR/css-syntax-3/#typedef-bad-url-token |
| 153 | */ |
| 154 | public const TOKEN_BAD_URL = 'bad-url-token'; |
| 155 | |
| 156 | /** |
| 157 | * Identifier tokens, such as `color`, `margin-top`, `red`, |
| 158 | * `inherit`, `--my-var`, `\escaped`, `über` (Unicode), etc. |
| 159 | * |
| 160 | * They can contain: letters, digits, hyphens, underscores, non-ASCII, escapes |
| 161 | * and cannot start with a digit (unless preceded by a hyphen). |
| 162 | * |
| 163 | * @see https://www.w3.org/TR/css-syntax-3/#typedef-ident-token |
| 164 | */ |
| 165 | public const TOKEN_IDENT = 'ident-token'; |
| 166 | |
| 167 | /** |
| 168 | * CDC (Comment Delimiter Close) token: --> |
| 169 | * |
| 170 | * Legacy token from when CSS was embedded in HTML <style> tags |
| 171 | * and needed to be hidden from old browsers using HTML comments: |
| 172 | * |
| 173 | * <style> |
| 174 | * <!-- |
| 175 | * body { color: red; } |
| 176 | * --> |
| 177 | * </style> |
| 178 | * |
| 179 | * Modern CSS no longer needs these, but they're preserved for compatibility. |
| 180 | * In stylesheets, they're typically treated like whitespace. |
| 181 | * |
| 182 | * @see https://www.w3.org/TR/css-syntax-3/#typedef-CDC-token |
| 183 | */ |
| 184 | public const TOKEN_CDC = 'CDC-token'; |
| 185 | |
| 186 | /** |
| 187 | * CDO (Comment Delimiter Open) token: <!-- |
| 188 | * |
| 189 | * Legacy token from when CSS was embedded in HTML <style> tags. |
| 190 | * See TOKEN_CDC for full explanation of HTML comment compatibility. |
| 191 | * |
| 192 | * @see https://www.w3.org/TR/css-syntax-3/#typedef-CDO-token |
| 193 | */ |
| 194 | public const TOKEN_CDO = 'CDO-token'; |
| 195 | |
| 196 | /** |
| 197 | * @var string |
| 198 | */ |
| 199 | private $css; |
| 200 | |
| 201 | /** |
| 202 | * @var int |
| 203 | */ |
| 204 | private $length = 0; |
| 205 | |
| 206 | /** |
| 207 | * @var int |
| 208 | */ |
| 209 | private $at = 0; |
| 210 | |
| 211 | /** |
| 212 | * The type of the current token. One of the self::TOKEN_* constants. |
| 213 | * |
| 214 | * @var string|null |
| 215 | */ |
| 216 | private $token_type = null; |
| 217 | |
| 218 | /** |
| 219 | * The byte offset at which the current token starts. |
| 220 | * |
| 221 | * Example: |
| 222 | * |
| 223 | * background-image: url(https://example.com/image.jpg); |
| 224 | * ^ token_starts_at |
| 225 | * |
| 226 | * @var int|null |
| 227 | */ |
| 228 | private $token_starts_at = null; |
| 229 | |
| 230 | /** |
| 231 | * The byte length of the current token. |
| 232 | * |
| 233 | * Example: |
| 234 | * |
| 235 | * background-image: url(https://example.com/image.jpg); |
| 236 | * ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ |
| 237 | * token_length |
| 238 | * |
| 239 | * @var int|null |
| 240 | */ |
| 241 | private $token_length = null; |
| 242 | |
| 243 | /** |
| 244 | * The byte offset at which the value of the current token starts. |
| 245 | * |
| 246 | * It is used for STRING and URL tokens. For example: |
| 247 | * |
| 248 | * background-image: url(https://example.com/image.jpg); |
| 249 | * ^ token_value_starts_at |
| 250 | * |
| 251 | * @var int|null |
| 252 | */ |
| 253 | private $token_value_starts_at = null; |
| 254 | |
| 255 | /** |
| 256 | * The byte offset at which the value of the current token starts. |
| 257 | * |
| 258 | * It is relevant for STRING and URL tokens. For example: |
| 259 | * |
| 260 | * background-image: url(https://example.com/image.jpg); |
| 261 | * ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ |
| 262 | * token_value_length |
| 263 | * |
| 264 | * @var int|null |
| 265 | */ |
| 266 | private $token_value_length = null; |
| 267 | |
| 268 | /** |
| 269 | * The string value of the current token. |
| 270 | * |
| 271 | * For numbers, this is a float. |
| 272 | * For identifiers/functions/strings/URLs with escapes, this is a decoded string. |
| 273 | * Otherwise, it's null and the value is computed from token indices. |
| 274 | * |
| 275 | * @var string|float|null |
| 276 | */ |
| 277 | private $token_value = null; |
| 278 | |
| 279 | /** |
| 280 | * The unit of the current token, e.g. "px", "em", "deg", etc. |
| 281 | * |
| 282 | * @var string|null |
| 283 | */ |
| 284 | private $token_unit = null; |
| 285 | |
| 286 | /** |
| 287 | * Lexical replacements to apply to input CSS document. |
| 288 | * |
| 289 | * Tracks modifications to be applied to the CSS, such as changing URL values. |
| 290 | * Each entry is an associative array with 'start', 'length', and 'text' keys. |
| 291 | * |
| 292 | * @var array[] |
| 293 | */ |
| 294 | private $lexical_updates = array(); |
| 295 | |
| 296 | /** |
| 297 | * Constructor for the CSS processor. |
| 298 | * |
| 299 | * Do not instantiate directly. Use CSSProcessor::create() instead. |
| 300 | * |
| 301 | * @param string $css CSS source to tokenize. |
| 302 | */ |
| 303 | private function __construct( string $css ) { |
| 304 | $this->css = $css; |
| 305 | $this->length = strlen( $css ); |
| 306 | } |
| 307 | |
| 308 | /** |
| 309 | * Creates a CSS processor for the given CSS string. |
| 310 | * |
| 311 | * Use this method to create a CSS processor instance. |
| 312 | * |
| 313 | * ## Current Support |
| 314 | * |
| 315 | * - The only supported document encoding is `UTF-8`, which is the default value. |
| 316 | * |
| 317 | * @param string $css CSS source to tokenize. |
| 318 | * @param string $encoding Text encoding of the document; must be default of 'UTF-8'. |
| 319 | * @return static|null The created processor if successful, otherwise null. |
| 320 | */ |
| 321 | public static function create( string $css, string $encoding = 'UTF-8' ) { |
| 322 | if ( 'UTF-8' !== $encoding ) { |
| 323 | return null; |
| 324 | } |
| 325 | |
| 326 | return new static( $css ); |
| 327 | } |
| 328 | |
| 329 | /** |
| 330 | * Moves to the next token in the CSS stream. |
| 331 | * |
| 332 | * Implements the main tokenization loop, consuming the next token from the input stream. |
| 333 | * |
| 334 | * @see https://www.w3.org/TR/css-syntax-3/#consume-token |
| 335 | * |
| 336 | * @return bool Whether a token was found. |
| 337 | */ |
| 338 | public function next_token(): bool { |
| 339 | $this->after_token(); |
| 340 | |
| 341 | // Bale out once we reach the end. |
| 342 | if ( $this->at >= $this->length ) { |
| 343 | return false; |
| 344 | } |
| 345 | |
| 346 | /* |
| 347 | * CSS comments. They are not preserved as tokens in the specification, but we |
| 348 | * still track them. |
| 349 | * |
| 350 | * @see https://www.w3.org/TR/css-syntax-3/#consume-comment |
| 351 | */ |
| 352 | if ( |
| 353 | $this->at + 1 < $this->length && |
| 354 | '/' === $this->css[ $this->at ] && |
| 355 | '*' === $this->css[ $this->at + 1 ] |
| 356 | ) { |
| 357 | $this->token_type = self::TOKEN_COMMENT; |
| 358 | $this->token_starts_at = $this->at; |
| 359 | $this->token_value_starts_at = $this->at; |
| 360 | |
| 361 | $end = strpos( $this->css, '*/', $this->at + 2 ); |
| 362 | $this->at = false !== $end ? $end + 2 : $this->length; |
| 363 | $this->token_length = $this->at - $this->token_starts_at; |
| 364 | $this->token_value_length = $this->token_length - 4; |
| 365 | return true; |
| 366 | } |
| 367 | |
| 368 | /* |
| 369 | * Whitespace tokens. |
| 370 | * |
| 371 | * We consider U+000A LINE FEED, U+0009 CHARACTER TABULATION, and U+0020 SPACE bytes covered by the spec. |
| 372 | * In addition, we also capture U+000D CARRIAGE RETURN and U+000C FORM FEED that are normally converted to |
| 373 | * U+000A LINE FEED during the preprocessing phase. |
| 374 | * |
| 375 | * @see https://www.w3.org/TR/css-syntax-3/#newline |
| 376 | * @see https://www.w3.org/TR/css-syntax-3/#whitespace |
| 377 | */ |
| 378 | $whitespace_length = strspn( $this->css, "\t\n\f\r ", $this->at ); |
| 379 | if ( $whitespace_length > 0 ) { |
| 380 | $this->token_type = self::TOKEN_WHITESPACE; |
| 381 | $this->token_length = $whitespace_length; |
| 382 | $this->token_starts_at = $this->at; |
| 383 | $this->at += $whitespace_length; |
| 384 | return true; |
| 385 | } |
| 386 | |
| 387 | /* |
| 388 | * String tokens with either " or ' as delimiters. |
| 389 | * |
| 390 | * @see https://www.w3.org/TR/css-syntax-3/#consume-string-token |
| 391 | */ |
| 392 | if ( '"' === $this->css[ $this->at ] || "'" === $this->css[ $this->at ] ) { |
| 393 | return $this->consume_string(); |
| 394 | } |
| 395 | |
| 396 | $char = $this->css[ $this->at ]; |
| 397 | $this->token_starts_at = $this->at; |
| 398 | |
| 399 | /* |
| 400 | * U+0023 NUMBER SIGN (#) |
| 401 | * |
| 402 | * A hash token is created when # is followed by an ident code point or valid escape. |
| 403 | * This is commonly used for hex colors (#fff) or ID selectors (#header). |
| 404 | * |
| 405 | * @see https://www.w3.org/TR/css-syntax-3/#consume-token |
| 406 | */ |
| 407 | if ( '#' === $char ) { |
| 408 | if ( $this->at + 1 < $this->length ) { |
| 409 | if ( |
| 410 | $this->consume_ident_codepoint( $this->at + 1 ) > 0 || |
| 411 | // The next two input code points are a valid escape. |
| 412 | $this->is_valid_escape( $this->at + 1 ) |
| 413 | ) { |
| 414 | // Create a <hash-token>. |
| 415 | ++$this->at; |
| 416 | |
| 417 | // We skip this check as we don't track the type flag: |
| 418 | // > If the next 3 input code points would start an ident sequence, |
| 419 | // > set the <hash-token>'s type flag to "id". |
| 420 | |
| 421 | // Consume an ident sequence, and set the <hash-token>'s value to the returned string. |
| 422 | $this->consume_ident_sequence(); |
| 423 | $this->token_type = self::TOKEN_HASH; |
| 424 | $this->token_length = $this->at - $this->token_starts_at; |
| 425 | return true; |
| 426 | } |
| 427 | } |
| 428 | // Otherwise, return a <delim-token> with its value set to the current input code point. |
| 429 | ++$this->at; |
| 430 | $this->token_type = self::TOKEN_DELIM; |
| 431 | $this->token_length = 1; |
| 432 | return true; |
| 433 | } |
| 434 | |
| 435 | /* |
| 436 | * Simple single-byte tokens |
| 437 | * |
| 438 | * These characters form their own tokens when encountered. |
| 439 | * Note: ( tokens here are not function tokens - those are handled |
| 440 | * in consume_ident_like() when ( follows an identifier. |
| 441 | * |
| 442 | * @see https://www.w3.org/TR/css-syntax-3/#tokenization |
| 443 | */ |
| 444 | $simple = array( |
| 445 | '(' => self::TOKEN_LEFT_PAREN, |
| 446 | ')' => self::TOKEN_RIGHT_PAREN, |
| 447 | ',' => self::TOKEN_COMMA, |
| 448 | ':' => self::TOKEN_COLON, |
| 449 | ';' => self::TOKEN_SEMICOLON, |
| 450 | '[' => self::TOKEN_LEFT_BRACKET, |
| 451 | ']' => self::TOKEN_RIGHT_BRACKET, |
| 452 | '{' => self::TOKEN_LEFT_BRACE, |
| 453 | '}' => self::TOKEN_RIGHT_BRACE, |
| 454 | ); |
| 455 | if ( isset( $simple[ $char ] ) ) { |
| 456 | ++$this->at; |
| 457 | $this->token_type = $simple[ $char ]; |
| 458 | $this->token_length = 1; |
| 459 | return true; |
| 460 | } |
| 461 | |
| 462 | /* |
| 463 | * U+0040 COMMERCIAL AT (@) |
| 464 | * |
| 465 | * An at-keyword is @ followed by an identifier, used for at-rules like |
| 466 | * @media, @import, @keyframes, etc. |
| 467 | * |
| 468 | * @see https://www.w3.org/TR/css-syntax-3/#consume-token |
| 469 | */ |
| 470 | if ( '@' === $char ) { |
| 471 | ++$this->at; |
| 472 | // If the next 3 input code points after the @ would start an ident sequence, |
| 473 | // consume an ident sequence, create an <at-keyword-token> with its value set to the returned value, |
| 474 | // and return it. |
| 475 | if ( $this->check_if_3_code_points_start_an_ident_sequence( $this->at ) ) { |
| 476 | $this->consume_ident_sequence(); |
| 477 | $this->token_type = self::TOKEN_AT_KEYWORD; |
| 478 | $this->token_length = $this->at - $this->token_starts_at; |
| 479 | return true; |
| 480 | } else { |
| 481 | // Otherwise, return a <delim-token> with its value set to the current input code point. |
| 482 | $this->token_type = self::TOKEN_DELIM; |
| 483 | $this->token_length = 1; |
| 484 | return true; |
| 485 | } |
| 486 | } |
| 487 | |
| 488 | /* |
| 489 | * Numbers start with digits, the plus sign, minus sign, and decimal point. |
| 490 | * |
| 491 | * @see https://www.w3.org/TR/css-syntax-3/#starts-with-a-number |
| 492 | */ |
| 493 | if ( $this->would_next_3_code_points_start_a_number() ) { |
| 494 | return $this->consume_numeric(); |
| 495 | } |
| 496 | |
| 497 | /* |
| 498 | * U+002D HYPHEN-MINUS (-) |
| 499 | */ |
| 500 | if ( '-' === $char ) { |
| 501 | // This case is covered above: |
| 502 | // > If the input stream starts with a number. |
| 503 | |
| 504 | /* |
| 505 | * If followed by another hyphen and >, this is a CDC token (-->) |
| 506 | * |
| 507 | * Comment Delimiter Close - legacy HTML comment syntax in CSS. |
| 508 | * |
| 509 | * @see https://www.w3.org/TR/css-syntax-3/#CDC-token-diagram |
| 510 | */ |
| 511 | if ( |
| 512 | $this->at + 2 < $this->length && |
| 513 | '-' === $this->css[ $this->at + 1 ] && |
| 514 | '>' === $this->css[ $this->at + 2 ] |
| 515 | ) { |
| 516 | // Consume them and return a <CDC-token>. |
| 517 | $this->at += 3; |
| 518 | $this->token_type = self::TOKEN_CDC; |
| 519 | $this->token_length = 3; |
| 520 | return true; |
| 521 | } |
| 522 | |
| 523 | // Otherwise, if the input stream starts with an ident sequence, |
| 524 | // reconsume the current input code point, consume an ident-like |
| 525 | // token, and return it. |
| 526 | if ( $this->check_if_3_code_points_start_an_ident_sequence( $this->at ) ) { |
| 527 | return $this->consume_ident_like(); |
| 528 | } |
| 529 | |
| 530 | // Otherwise, return a <delim-token> with its value set to the current input code point. |
| 531 | ++$this->at; |
| 532 | $this->token_type = self::TOKEN_DELIM; |
| 533 | $this->token_length = 1; |
| 534 | return true; |
| 535 | } |
| 536 | |
| 537 | /* |
| 538 | * U+003C LESS-THAN SIGN (<) |
| 539 | * If followed by !--, this is a CDO token (<!--) |
| 540 | * |
| 541 | * Comment Delimiter Open - legacy HTML comment syntax in CSS. |
| 542 | * |
| 543 | * @see https://www.w3.org/TR/css-syntax-3/#CDO-token-diagram |
| 544 | */ |
| 545 | if ( '<' === $char && $this->at + 3 < $this->length && |
| 546 | '!' === $this->css[ $this->at + 1 ] && |
| 547 | '-' === $this->css[ $this->at + 2 ] && |
| 548 | '-' === $this->css[ $this->at + 3 ] ) { |
| 549 | // Consume them and return a <CDO-token>. |
| 550 | $this->at += 4; |
| 551 | $this->token_type = self::TOKEN_CDO; |
| 552 | $this->token_length = 4; |
| 553 | return true; |
| 554 | } |
| 555 | |
| 556 | /* |
| 557 | * Ident-start code point |
| 558 | * |
| 559 | * If the input stream starts with an ident sequence, reconsume the current |
| 560 | * input code point, consume an ident-like token, and return it. |
| 561 | * |
| 562 | * Could be an identifier, function, or url() token. |
| 563 | * |
| 564 | * @see https://www.w3.org/TR/css-syntax-3/#consume-ident-like-token |
| 565 | */ |
| 566 | if ( $this->check_if_3_code_points_start_an_ident_sequence( $this->at ) ) { |
| 567 | return $this->consume_ident_like(); |
| 568 | } |
| 569 | |
| 570 | /* |
| 571 | * Delim token (delimiter) |
| 572 | * |
| 573 | * Any code point that doesn't match above rules becomes a delim token. |
| 574 | * Handle multi-byte UTF-8 characters properly. |
| 575 | * |
| 576 | * @see https://www.w3.org/TR/css-syntax-3/#delim-token-diagram |
| 577 | */ |
| 578 | if ( ord( $char ) >= 0x80 ) { |
| 579 | $new_at = $this->at; |
| 580 | $invalid_length = 0; |
| 581 | if ( 1 !== _wp_scan_utf8( $this->css, $new_at, $invalid_length, null, 1 ) ) { |
| 582 | /** |
| 583 | * Trouble ahead! |
| 584 | * Bytes at $at are not a valid UTF-8 sequence. |
| 585 | * |
| 586 | * We'll move forward by $invalid_length bytes and continue processing. |
| 587 | * Later on, during the string decoding, we'll replace the invalid bytes with U+FFFD |
| 588 | * via maximal subpart”replacement. |
| 589 | */ |
| 590 | $matched_bytes = $invalid_length; |
| 591 | } else { |
| 592 | $matched_bytes = $new_at - $this->at; |
| 593 | } |
| 594 | |
| 595 | $this->at += $matched_bytes; |
| 596 | $this->token_type = self::TOKEN_DELIM; |
| 597 | $this->token_length = $matched_bytes; |
| 598 | return true; |
| 599 | } |
| 600 | |
| 601 | // Single ASCII delim. |
| 602 | ++$this->at; |
| 603 | $this->token_type = self::TOKEN_DELIM; |
| 604 | $this->token_length = 1; |
| 605 | return true; |
| 606 | } |
| 607 | |
| 608 | /** |
| 609 | * Gets the current token type. |
| 610 | * |
| 611 | * @return string|null |
| 612 | */ |
| 613 | public function get_token_type(): ?string { |
| 614 | return $this->token_type; |
| 615 | } |
| 616 | |
| 617 | /** |
| 618 | * Gets the normalized token text from the CSS source. |
| 619 | * |
| 620 | * Returns the token with CSS normalization and escape decoding applied: |
| 621 | * - CSS escapes decoded (e.g., \6c → l, \2f → /, \A → newline) |
| 622 | * - \r\n, \r, \f → \n |
| 623 | * - \x00 → U+FFFD (�) |
| 624 | * |
| 625 | * This is different from get_token_value() which returns the semantic value |
| 626 | * (e.g., for strings: content without quotes; for numbers: numeric value). |
| 627 | * |
| 628 | * @return string|null |
| 629 | */ |
| 630 | public function get_normalized_token(): ?string { |
| 631 | if ( null === $this->token_starts_at || null === $this->token_length ) { |
| 632 | return null; |
| 633 | } |
| 634 | |
| 635 | return $this->decode_range( |
| 636 | $this->token_starts_at, |
| 637 | $this->token_length, |
| 638 | self::TOKEN_STRING === $this->token_type |
| 639 | ); |
| 640 | } |
| 641 | |
| 642 | /** |
| 643 | * Gets the raw, unnormalized token text from the CSS source. |
| 644 | * |
| 645 | * Returns the exact bytes from the source without any normalization. |
| 646 | * This preserves original line endings (\r\n, \r, \f) and null bytes. |
| 647 | * |
| 648 | * @return string|null |
| 649 | */ |
| 650 | public function get_unnormalized_token(): ?string { |
| 651 | if ( null === $this->token_starts_at || null === $this->token_length ) { |
| 652 | return null; |
| 653 | } |
| 654 | return substr( $this->css, $this->token_starts_at, $this->token_length ); |
| 655 | } |
| 656 | |
| 657 | /** |
| 658 | * Gets the current token value as a normalized and decoded string. This is |
| 659 | * a slight divergence from the CSS Syntax Level 3 spec, where all the numberic |
| 660 | * values are parsed as numbers. This processor is only concerned with their |
| 661 | * textual representation. |
| 662 | * |
| 663 | * Returns the semantic value of the token per CSS Syntax Level 3 spec: |
| 664 | * |
| 665 | * - For delimiters: the single code point |
| 666 | * - For numbers/percentages: the string representation of the number |
| 667 | * - For dimensions: the string representation of the number (use get_token_unit() for the unit) |
| 668 | * - For identifiers/functions/hash/at-keywords: the decoded identifier string |
| 669 | * - For strings/URLs: the decoded string value |
| 670 | * - For other tokens: null |
| 671 | * |
| 672 | * @see https://www.w3.org/TR/css-syntax-3/#tokenization |
| 673 | * @return string|null |
| 674 | */ |
| 675 | public function get_token_value() { |
| 676 | if ( null === $this->token_value ) { |
| 677 | if ( null === $this->token_starts_at || null === $this->token_length ) { |
| 678 | return null; |
| 679 | } |
| 680 | |
| 681 | switch ( $this->token_type ) { |
| 682 | case self::TOKEN_HASH: |
| 683 | // Hash value starts after the # character. |
| 684 | $this->token_value = $this->decode_range( $this->token_starts_at + 1, $this->token_length - 1 ); |
| 685 | break; |
| 686 | |
| 687 | case self::TOKEN_AT_KEYWORD: |
| 688 | // At-keyword value starts after the @ character. |
| 689 | $this->token_value = $this->decode_range( $this->token_starts_at + 1, $this->token_length - 1 ); |
| 690 | break; |
| 691 | |
| 692 | case self::TOKEN_FUNCTION: |
| 693 | // Function name is everything except the final (. |
| 694 | $this->token_value = $this->decode_range( $this->token_starts_at, $this->token_length - 1 ); |
| 695 | break; |
| 696 | |
| 697 | case self::TOKEN_IDENT: |
| 698 | // Identifier is the entire token. |
| 699 | $this->token_value = $this->decode_range( $this->token_starts_at, $this->token_length ); |
| 700 | break; |
| 701 | |
| 702 | case self::TOKEN_STRING: |
| 703 | if ( null !== $this->token_value_starts_at && null !== $this->token_value_length ) { |
| 704 | $this->token_value = $this->decode_range( |
| 705 | $this->token_value_starts_at, |
| 706 | $this->token_value_length, |
| 707 | true |
| 708 | ); |
| 709 | } else { |
| 710 | $this->token_value = null; |
| 711 | } |
| 712 | break; |
| 713 | |
| 714 | case self::TOKEN_URL: |
| 715 | if ( null !== $this->token_value_starts_at && null !== $this->token_value_length ) { |
| 716 | $this->token_value = $this->decode_range( |
| 717 | $this->token_value_starts_at, |
| 718 | $this->token_value_length |
| 719 | ); |
| 720 | } else { |
| 721 | $this->token_value = null; |
| 722 | } |
| 723 | break; |
| 724 | |
| 725 | case self::TOKEN_DELIM: |
| 726 | // Delim value is the single code point. |
| 727 | $this->token_value = $this->decode_range( $this->token_starts_at, $this->token_length ); |
| 728 | break; |
| 729 | |
| 730 | case self::TOKEN_NUMBER: |
| 731 | // Return the string representation of the number (not parsed to float). |
| 732 | $this->token_value = substr( $this->css, $this->token_starts_at, $this->token_length ); |
| 733 | break; |
| 734 | |
| 735 | case self::TOKEN_PERCENTAGE: |
| 736 | // Return the string representation of the number (without the %). |
| 737 | $this->token_value = substr( $this->css, $this->token_starts_at, $this->token_length - 1 ); |
| 738 | break; |
| 739 | |
| 740 | case self::TOKEN_DIMENSION: |
| 741 | // Return the string representation of the number (without the unit). |
| 742 | $this->token_value = substr( $this->get_normalized_token(), 0, -strlen( $this->token_unit ) ); |
| 743 | break; |
| 744 | |
| 745 | default: |
| 746 | $this->token_value = null; |
| 747 | break; |
| 748 | } |
| 749 | } |
| 750 | |
| 751 | return $this->token_value; |
| 752 | } |
| 753 | |
| 754 | /** |
| 755 | * Determines whether the current token is a data URI. |
| 756 | * |
| 757 | * Only meaningful for URL and STRING tokens. Returns false for all other token types. |
| 758 | * |
| 759 | * @return bool Whether the current token value starts with "data:" (case-insensitive). |
| 760 | */ |
| 761 | public function is_data_uri(): bool { |
| 762 | if ( null === $this->token_value_starts_at || null === $this->token_value_length ) { |
| 763 | return false; |
| 764 | } |
| 765 | |
| 766 | if ( $this->token_value_length < 5 ) { |
| 767 | return false; |
| 768 | } |
| 769 | |
| 770 | $offset = $this->token_value_starts_at; |
| 771 | return ( |
| 772 | ( 'd' === $this->css[ $offset ] || 'D' === $this->css[ $offset ] ) && |
| 773 | ( 'a' === $this->css[ $offset + 1 ] || 'A' === $this->css[ $offset + 1 ] ) && |
| 774 | ( 't' === $this->css[ $offset + 2 ] || 'T' === $this->css[ $offset + 2 ] ) && |
| 775 | ( 'a' === $this->css[ $offset + 3 ] || 'A' === $this->css[ $offset + 3 ] ) && |
| 776 | ':' === $this->css[ $offset + 4 ] |
| 777 | ); |
| 778 | } |
| 779 | |
| 780 | /** |
| 781 | * Gets the token start at. |
| 782 | * |
| 783 | * @return int|null |
| 784 | */ |
| 785 | public function get_token_start(): ?int { |
| 786 | return $this->token_starts_at; |
| 787 | } |
| 788 | |
| 789 | /** |
| 790 | * Gets the token length. |
| 791 | * |
| 792 | * @return int|null |
| 793 | */ |
| 794 | public function get_token_length(): ?int { |
| 795 | return $this->token_length; |
| 796 | } |
| 797 | |
| 798 | /** |
| 799 | * Gets the unit for dimension tokens. |
| 800 | * |
| 801 | * @return string|null |
| 802 | */ |
| 803 | public function get_token_unit(): ?string { |
| 804 | return $this->token_unit; |
| 805 | } |
| 806 | |
| 807 | /** |
| 808 | * Gets the byte at where the token value starts (for STRING and URL tokens). |
| 809 | * |
| 810 | * @return int|null |
| 811 | */ |
| 812 | public function get_token_value_start(): ?int { |
| 813 | return $this->token_value_starts_at; |
| 814 | } |
| 815 | |
| 816 | /** |
| 817 | * Gets the byte length of the token value (for STRING and URL tokens). |
| 818 | * |
| 819 | * @return int|null |
| 820 | */ |
| 821 | public function get_token_value_length(): ?int { |
| 822 | return $this->token_value_length; |
| 823 | } |
| 824 | |
| 825 | /** |
| 826 | * Sets the value of the current URL token. |
| 827 | * |
| 828 | * This method allows modifying the URL value in url() tokens. The new value |
| 829 | * will be properly escaped according to CSS URL syntax rules. |
| 830 | * |
| 831 | * Currently only URL tokens are supported. Attempting to set the value on |
| 832 | * other token types will return false. |
| 833 | * |
| 834 | * Example: |
| 835 | * |
| 836 | * $css = 'background: url(old.jpg);'; |
| 837 | * $processor = CSSProcessor::create( $css ); |
| 838 | * while ( $processor->next_token() ) { |
| 839 | * if ( CSSProcessor::TOKEN_URL === $processor->get_token_type() ) { |
| 840 | * $processor->set_token_value( 'new.jpg' ); |
| 841 | * } |
| 842 | * } |
| 843 | * echo $processor->get_updated_css(); |
| 844 | * // Outputs: background: url(new.jpg); |
| 845 | * |
| 846 | * @param string $new_value The new URL value (should not include url() wrapper). |
| 847 | * @return bool Whether the value was successfully updated. |
| 848 | */ |
| 849 | public function set_token_value( string $new_value ): bool { |
| 850 | // Only URL and string tokens are currently supported. |
| 851 | switch ( $this->token_type ) { |
| 852 | case self::TOKEN_URL: |
| 853 | $this->lexical_updates[] = array( |
| 854 | 'start' => $this->token_value_starts_at, |
| 855 | 'length' => $this->token_value_length, |
| 856 | 'text' => $this->escape_url_value( $new_value ), |
| 857 | ); |
| 858 | return true; |
| 859 | case self::TOKEN_STRING: |
| 860 | $this->lexical_updates[] = array( |
| 861 | 'start' => $this->token_starts_at, |
| 862 | 'length' => $this->token_length, |
| 863 | 'text' => $this->escape_url_value( $new_value ), |
| 864 | ); |
| 865 | return true; |
| 866 | default: |
| 867 | _doing_it_wrong( __METHOD__, 'set_token_value() only supports URL and string tokens. Got token type: ' . $this->token_type, '1.0.0' ); |
| 868 | return false; |
| 869 | } |
| 870 | } |
| 871 | |
| 872 | /** |
| 873 | * Escapes a URL value for use in quoted url() syntax. |
| 874 | * |
| 875 | * Always returns a quoted URL string since they're easier |
| 876 | * to escape. Quoted URLs are consumed using the string token |
| 877 | * rules, and the only values we need to escape in strings, are: |
| 878 | * |
| 879 | * * Trailing quote. |
| 880 | * * Newlines. That amounts to \n, \r, \f, \r\n when preprocessing is considered. |
| 881 | * * U+005C REVERSE SOLIDUS (\) |
| 882 | * |
| 883 | * @see https://www.w3.org/TR/css-syntax-3/#consume-url-token |
| 884 | */ |
| 885 | private function escape_url_value( string $unescaped ): string { |
| 886 | $escaped = ''; |
| 887 | $at = 0; |
| 888 | while ( $at < strlen( $unescaped ) ) { |
| 889 | $safe_len = strcspn( $unescaped, "\n\r\f\\\"", $at ); |
| 890 | if ( $safe_len > 0 ) { |
| 891 | $escaped .= substr( $unescaped, $at, $safe_len ); |
| 892 | $at += $safe_len; |
| 893 | continue; |
| 894 | } |
| 895 | |
| 896 | $unsafe_char = $unescaped[ $at ]; |
| 897 | switch ( $unsafe_char ) { |
| 898 | case "\r": |
| 899 | ++$at; |
| 900 | /** |
| 901 | * Add a trailing space to prevent accidentally creating a |
| 902 | * wrong escape sequence. This is a valid CSS syntax and |
| 903 | * CSS parsers will ignore that whitespace. |
| 904 | * |
| 905 | * Without the space, "carriage\return" would be encoded as "carriage\aeturn", |
| 906 | * making `e` a part of the escape sequence `\ae` which is not |
| 907 | * what the caller intended. |
| 908 | */ |
| 909 | $escaped .= '\\a '; |
| 910 | if ( strlen( $unescaped ) > $at + 1 && "\n" === $unescaped[ $at + 1 ] ) { |
| 911 | ++$at; |
| 912 | } |
| 913 | break; |
| 914 | case "\f": |
| 915 | case "\n": |
| 916 | ++$at; |
| 917 | $escaped .= '\\a '; |
| 918 | break; |
| 919 | case '\\': |
| 920 | ++$at; |
| 921 | $escaped .= '\\5C '; |
| 922 | break; |
| 923 | case '"': |
| 924 | ++$at; |
| 925 | $escaped .= '\\22 '; |
| 926 | break; |
| 927 | default: |
| 928 | _doing_it_wrong( __METHOD__, 'Unexpected character in URL value: ' . $unsafe_char, '1.0.0' ); |
| 929 | break; |
| 930 | } |
| 931 | } |
| 932 | return '"' . $escaped . '"'; |
| 933 | } |
| 934 | |
| 935 | /** |
| 936 | * Returns the CSS with all modifications applied. |
| 937 | * |
| 938 | * This method applies all queued lexical updates and returns the modified CSS. |
| 939 | * If no modifications were made, returns the original CSS. |
| 940 | * |
| 941 | * Example: |
| 942 | * |
| 943 | * $css = 'background: url(old.jpg);'; |
| 944 | * $processor = CSSProcessor::create( $css ); |
| 945 | * while ( $processor->next_token() ) { |
| 946 | * if ( CSSProcessor::TOKEN_URL === $processor->get_token_type() ) { |
| 947 | * $processor->set_token_value( 'new.jpg' ); |
| 948 | * } |
| 949 | * } |
| 950 | * echo $processor->get_updated_css(); |
| 951 | * // Outputs: background: url(new.jpg); |
| 952 | * |
| 953 | * @return string The modified CSS. |
| 954 | */ |
| 955 | public function get_updated_css(): string { |
| 956 | if ( empty( $this->lexical_updates ) ) { |
| 957 | return $this->css; |
| 958 | } |
| 959 | |
| 960 | // Sort updates by start position in ascending order. |
| 961 | usort( |
| 962 | $this->lexical_updates, |
| 963 | function ( $a, $b ) { |
| 964 | return $a['start'] - $b['start']; |
| 965 | } |
| 966 | ); |
| 967 | |
| 968 | // Build the output by concatenating original CSS fragments with replacements. |
| 969 | $bytes_already_copied = 0; |
| 970 | $output = ''; |
| 971 | |
| 972 | foreach ( $this->lexical_updates as $update ) { |
| 973 | $output .= substr( $this->css, $bytes_already_copied, $update['start'] - $bytes_already_copied ); |
| 974 | $output .= $update['text']; |
| 975 | $bytes_already_copied = $update['start'] + $update['length']; |
| 976 | } |
| 977 | |
| 978 | // Copy remaining CSS after last update. |
| 979 | $output .= substr( $this->css, $bytes_already_copied ); |
| 980 | |
| 981 | return $output; |
| 982 | } |
| 983 | |
| 984 | /** |
| 985 | * Clears token state between tokens. |
| 986 | */ |
| 987 | private function after_token(): void { |
| 988 | $this->token_type = null; |
| 989 | $this->token_starts_at = null; |
| 990 | $this->token_length = null; |
| 991 | $this->token_value = null; |
| 992 | $this->token_unit = null; |
| 993 | $this->token_value_starts_at = null; |
| 994 | $this->token_value_length = null; |
| 995 | } |
| 996 | |
| 997 | /** |
| 998 | * Consumes a string token. |
| 999 | * |
| 1000 | * Strings are quoted with either " or ' and can contain escape sequences. |
| 1001 | * Newlines inside strings (without escaping) make the string invalid. |
| 1002 | * |
| 1003 | * @see https://www.w3.org/TR/css-syntax-3/#consume-string-token |
| 1004 | * |
| 1005 | * @return bool |
| 1006 | */ |
| 1007 | private function consume_string(): bool { |
| 1008 | // Initially create a <string-token> with its value set to the empty string. |
| 1009 | $this->token_starts_at = $this->at; |
| 1010 | $ending_char = $this->css[ $this->at ]; |
| 1011 | |
| 1012 | // Skip past the opening quote. |
| 1013 | ++$this->at; |
| 1014 | $value_starts_at = $this->at; |
| 1015 | |
| 1016 | // Characters that need special handling: the ending quote, newlines, backslashes. |
| 1017 | $special_chars = "'" === $ending_char ? "'\n\f\r\\" : "\"\n\f\r\\"; |
| 1018 | |
| 1019 | while ( $this->at < $this->length ) { |
| 1020 | // Consume normal characters until we hit a special character. |
| 1021 | $normal_len = strcspn( $this->css, $special_chars, $this->at ); |
| 1022 | if ( $normal_len > 0 ) { |
| 1023 | $this->at += $normal_len; |
| 1024 | } |
| 1025 | |
| 1026 | if ( $this->at >= $this->length ) { |
| 1027 | break; // EOF. |
| 1028 | } |
| 1029 | |
| 1030 | $char = $this->css[ $this->at ]; |
| 1031 | switch ( $char ) { |
| 1032 | case $ending_char: |
| 1033 | // Ending quote. |
| 1034 | // Return the <string-token>. |
| 1035 | ++$this->at; |
| 1036 | $this->token_type = self::TOKEN_STRING; |
| 1037 | $this->token_length = $this->at - $this->token_starts_at; |
| 1038 | $this->token_value_starts_at = $value_starts_at; |
| 1039 | $this->token_value_length = $this->at - $value_starts_at - 1; |
| 1040 | return true; |
| 1041 | |
| 1042 | case "\n": |
| 1043 | case "\f": |
| 1044 | case "\r": |
| 1045 | /* |
| 1046 | * Newline. |
| 1047 | * |
| 1048 | * This is a parse error. Reconsume the current input code point, |
| 1049 | * create a <bad-string-token>, and return it. |
| 1050 | * |
| 1051 | * Unescaped newlines are not allowed in strings. To include a newline, |
| 1052 | * it must be escaped as \A or the string must end and a new one begin. |
| 1053 | * |
| 1054 | * @see https://www.w3.org/TR/css-syntax-3/#consume-string-token |
| 1055 | */ |
| 1056 | $this->token_type = self::TOKEN_BAD_STRING; |
| 1057 | $this->token_length = $this->at - $this->token_starts_at; |
| 1058 | $this->token_value_starts_at = $value_starts_at; |
| 1059 | $this->token_value_length = $this->at - $value_starts_at; |
| 1060 | return true; |
| 1061 | |
| 1062 | case '\\': |
| 1063 | // U+005C REVERSE SOLIDUS (\) |
| 1064 | // If the next input code point is EOF, do nothing. |
| 1065 | ++$this->at; |
| 1066 | if ( $this->at >= $this->length ) { |
| 1067 | // Backslash-EOF: do nothing, just consume the backslash. |
| 1068 | continue 2; |
| 1069 | } |
| 1070 | |
| 1071 | // Otherwise, if the next input code point is a newline, consume it. |
| 1072 | $next = $this->css[ $this->at ]; |
| 1073 | if ( "\n" === $next || "\f" === $next ) { |
| 1074 | ++$this->at; |
| 1075 | continue 2; |
| 1076 | } elseif ( "\r" === $next ) { |
| 1077 | ++$this->at; |
| 1078 | // Handle \r\n as a single newline. |
| 1079 | if ( $this->at < $this->length && "\n" === $this->css[ $this->at ] ) { |
| 1080 | ++$this->at; |
| 1081 | } |
| 1082 | continue 2; |
| 1083 | } |
| 1084 | |
| 1085 | // Otherwise, (the stream starts with a valid escape) consume an escaped |
| 1086 | // code point (just to advance position, don't store the result). |
| 1087 | $this->decode_escape_at( $this->at, $matched_bytes ); |
| 1088 | $this->at += $matched_bytes; |
| 1089 | continue 2; |
| 1090 | |
| 1091 | default: |
| 1092 | _doing_it_wrong( __METHOD__, 'Unexpected character in string: ' . $char, '1.0.0' ); |
| 1093 | break; |
| 1094 | } |
| 1095 | } |
| 1096 | |
| 1097 | // EOF |
| 1098 | // This is a parse error. Return the <string-token>. |
| 1099 | $this->token_type = self::TOKEN_STRING; |
| 1100 | $this->token_length = $this->at - $this->token_starts_at; |
| 1101 | $this->token_value_starts_at = $value_starts_at; |
| 1102 | $this->token_value_length = $this->at - $value_starts_at; |
| 1103 | return true; |
| 1104 | } |
| 1105 | |
| 1106 | /** |
| 1107 | * Consumes a numeric token (number, percentage, dimension). |
| 1108 | * |
| 1109 | * Numbers can be integers or decimals, with optional sign and exponent. |
| 1110 | * They can be followed by % (percentage) or an identifier (dimension). |
| 1111 | * |
| 1112 | * @TODO: Keep track of the "type" flag ("integer" or "number"). |
| 1113 | * |
| 1114 | * @see https://www.w3.org/TR/css-syntax-3/#consume-numeric-token |
| 1115 | * @see https://www.w3.org/TR/css-syntax-3/#consume-number |
| 1116 | * |
| 1117 | * @return bool |
| 1118 | */ |
| 1119 | private function consume_numeric(): bool { |
| 1120 | // Consume a number and let number be the result. |
| 1121 | |
| 1122 | // If the next input code point is U+002B PLUS SIGN (+) or U+002D HYPHEN-MINUS (-), |
| 1123 | // consume it and append it to repr. |
| 1124 | if ( '+' === $this->css[ $this->at ] || '-' === $this->css[ $this->at ] ) { |
| 1125 | ++$this->at; |
| 1126 | } |
| 1127 | |
| 1128 | // While the next input code point is a digit, consume it and append it to repr. |
| 1129 | $digits = strspn( $this->css, '0123456789', $this->at ); |
| 1130 | if ( $digits > 0 ) { |
| 1131 | $this->at += $digits; |
| 1132 | } |
| 1133 | |
| 1134 | // If the next 2 input code points are U+002E FULL STOP (.) followed by a digit, then. |
| 1135 | if ( |
| 1136 | $this->at + 1 < $this->length && |
| 1137 | '.' === $this->css[ $this->at ] && |
| 1138 | $this->css[ $this->at + 1 ] >= '0' && |
| 1139 | $this->css[ $this->at + 1 ] <= '9' |
| 1140 | ) { |
| 1141 | // Consume them. |
| 1142 | ++$this->at; |
| 1143 | // While the next input code point is a digit, consume it and append it to repr. |
| 1144 | $digits = strspn( $this->css, '0123456789', $this->at ); |
| 1145 | if ( $digits > 0 ) { |
| 1146 | $this->at += $digits; |
| 1147 | } |
| 1148 | } |
| 1149 | |
| 1150 | // If the next 2 or 3 input code points are U+0045 LATIN CAPITAL LETTER E (E) |
| 1151 | // or U+0065 LATIN SMALL LETTER E (e), optionally followed by U+002D HYPHEN-MINUS (-) |
| 1152 | // or U+002B PLUS SIGN (+), followed by a digit, then. |
| 1153 | if ( $this->at < $this->length ) { |
| 1154 | $e = $this->css[ $this->at ]; |
| 1155 | if ( 'e' === $e || 'E' === $e ) { |
| 1156 | $save_pos = $this->at; |
| 1157 | ++$this->at; |
| 1158 | $has_exp = false; |
| 1159 | |
| 1160 | if ( $this->at < $this->length ) { |
| 1161 | $next = $this->css[ $this->at ]; |
| 1162 | if ( ( '+' === $next || '-' === $next ) && $this->at + 1 < $this->length && |
| 1163 | $this->css[ $this->at + 1 ] >= '0' && $this->css[ $this->at + 1 ] <= '9' ) { |
| 1164 | // Consume them. |
| 1165 | ++$this->at; |
| 1166 | $has_exp = true; |
| 1167 | } elseif ( $next >= '0' && $next <= '9' ) { |
| 1168 | $has_exp = true; |
| 1169 | } |
| 1170 | } |
| 1171 | |
| 1172 | if ( $has_exp ) { |
| 1173 | // While the next input code point is a digit, consume it and append it to repr. |
| 1174 | $digits = strspn( $this->css, '0123456789', $this->at ); |
| 1175 | if ( $digits > 0 ) { |
| 1176 | $this->at += $digits; |
| 1177 | } |
| 1178 | } else { |
| 1179 | $this->at = $save_pos; |
| 1180 | } |
| 1181 | } |
| 1182 | } |
| 1183 | |
| 1184 | /** |
| 1185 | * This is the end of spec section 4.3.12. Consume a number. |
| 1186 | * We still have some work to do as specified in section 4.3.3. Consume a numeric token: |
| 1187 | * https://www.w3.org/TR/css-syntax-3/#consume-numeric-token |
| 1188 | */ |
| 1189 | |
| 1190 | // If the next 3 input code points would start an ident sequence, then. |
| 1191 | if ( $this->check_if_3_code_points_start_an_ident_sequence( $this->at ) ) { |
| 1192 | // Create a <dimension-token> with the same value and type flag as number, |
| 1193 | // and a unit set initially to the empty string. |
| 1194 | // Consume an ident sequence. Set the <dimension-token>'s unit to the returned value. |
| 1195 | $unit_starts_at = $this->at; |
| 1196 | $this->consume_ident_sequence(); |
| 1197 | $this->token_unit = $this->decode_range( $unit_starts_at, $this->at - $unit_starts_at ); |
| 1198 | $this->token_type = self::TOKEN_DIMENSION; |
| 1199 | $this->token_length = $this->at - $this->token_starts_at; |
| 1200 | return true; |
| 1201 | } |
| 1202 | |
| 1203 | // Otherwise, if the next input code point is U+0025 PERCENTAGE SIGN (%), consume it. |
| 1204 | // Create a <percentage-token> with the same value as number, and return it. |
| 1205 | if ( $this->at < $this->length && '%' === $this->css[ $this->at ] ) { |
| 1206 | ++$this->at; |
| 1207 | $this->token_type = self::TOKEN_PERCENTAGE; |
| 1208 | $this->token_length = $this->at - $this->token_starts_at; |
| 1209 | return true; |
| 1210 | } |
| 1211 | |
| 1212 | // Otherwise, create a <number-token> with the same value and type flag as number, and return it. |
| 1213 | $this->token_type = self::TOKEN_NUMBER; |
| 1214 | $this->token_length = $this->at - $this->token_starts_at; |
| 1215 | return true; |
| 1216 | } |
| 1217 | |
| 1218 | /** |
| 1219 | * Consumes an ident-like token (function, url, ident). |
| 1220 | * |
| 1221 | * After consuming an identifier, checks if it's followed by '(' to determine |
| 1222 | * if it's a function or url() token, otherwise it's a plain identifier. |
| 1223 | * |
| 1224 | * @see https://www.w3.org/TR/css-syntax-3/#consume-ident-like-token |
| 1225 | * |
| 1226 | * @return bool |
| 1227 | */ |
| 1228 | private function consume_ident_like(): bool { |
| 1229 | // Consume an ident sequence, and let string be the result. |
| 1230 | $ident_start = $this->at; |
| 1231 | $decoded = $this->consume_ident_sequence(); |
| 1232 | $string = $decoded ?? $this->decode_range( $ident_start, $this->at - $ident_start ); |
| 1233 | |
| 1234 | // If string's value is an ASCII case-insensitive match for "url", |
| 1235 | // and the next input code point is U+0028 LEFT PARENTHESIS ((). |
| 1236 | if ( 0 === strcasecmp( $string, 'url' ) && $this->at < $this->length && '(' === $this->css[ $this->at ] ) { |
| 1237 | // Consume it. |
| 1238 | ++$this->at; |
| 1239 | |
| 1240 | // While the next two input code points are whitespace, consume the next input code point. |
| 1241 | $ws_len = strspn( $this->css, "\t\n\f\r ", $this->at ); |
| 1242 | |
| 1243 | // If the next one or two input code points are U+0022 QUOTATION MARK ("), |
| 1244 | // U+0027 APOSTROPHE ('), or whitespace followed by U+0022 QUOTATION MARK (") |
| 1245 | // or U+0027 APOSTROPHE ('). |
| 1246 | if ( $this->at + $ws_len < $this->length ) { |
| 1247 | $next = $this->css[ $this->at + $ws_len ]; |
| 1248 | if ( '"' === $next || "'" === $next ) { |
| 1249 | // then create a <function-token> with its value set to string and return it. |
| 1250 | if ( null !== $decoded ) { |
| 1251 | $this->token_value = $decoded; |
| 1252 | } |
| 1253 | $this->token_type = self::TOKEN_FUNCTION; |
| 1254 | $this->token_length = $this->at - $this->token_starts_at; |
| 1255 | return true; |
| 1256 | } |
| 1257 | } |
| 1258 | |
| 1259 | // Otherwise, consume a url token, and return it. |
| 1260 | $this->at += $ws_len; |
| 1261 | return $this->consume_url(); |
| 1262 | } |
| 1263 | |
| 1264 | // Otherwise, if the next input code point is U+0028 LEFT PARENTHESIS ((). |
| 1265 | if ( $this->at < $this->length && '(' === $this->css[ $this->at ] ) { |
| 1266 | // Consume it. |
| 1267 | ++$this->at; |
| 1268 | // Create a <function-token> with its value set to string and return it. |
| 1269 | if ( null !== $decoded ) { |
| 1270 | $this->token_value = $decoded; |
| 1271 | } |
| 1272 | $this->token_type = self::TOKEN_FUNCTION; |
| 1273 | $this->token_length = $this->at - $this->token_starts_at; |
| 1274 | return true; |
| 1275 | } |
| 1276 | |
| 1277 | // Otherwise, create an <ident-token> with its value set to string and return it. |
| 1278 | if ( null !== $decoded ) { |
| 1279 | $this->token_value = $decoded; |
| 1280 | } |
| 1281 | $this->token_type = self::TOKEN_IDENT; |
| 1282 | $this->token_length = $this->at - $this->token_starts_at; |
| 1283 | return true; |
| 1284 | } |
| 1285 | |
| 1286 | /** |
| 1287 | * Consumes a url token. |
| 1288 | * |
| 1289 | * URL tokens can contain unquoted URLs with escape sequences but not quotes, |
| 1290 | * parentheses, or certain control characters. Invalid characters create a |
| 1291 | * bad-url token. |
| 1292 | * |
| 1293 | * @see https://www.w3.org/TR/css-syntax-3/#consume-url-token |
| 1294 | * |
| 1295 | * @return bool |
| 1296 | */ |
| 1297 | private function consume_url(): bool { |
| 1298 | // Initially create a <url-token> with its value set to the empty string. |
| 1299 | // Consume as much whitespace as possible. |
| 1300 | $this->at += strspn( $this->css, "\t\n\f\r ", $this->at ); |
| 1301 | |
| 1302 | $value_starts_at = $this->at; |
| 1303 | |
| 1304 | // Repeatedly consume the next input code point from the stream. |
| 1305 | while ( $this->at < $this->length ) { |
| 1306 | // U+0029 RIGHT PARENTHESIS ()) |
| 1307 | // Return the <url-token>. |
| 1308 | if ( ')' === $this->css[ $this->at ] ) { |
| 1309 | ++$this->at; |
| 1310 | $this->token_type = self::TOKEN_URL; |
| 1311 | $this->token_length = $this->at - $this->token_starts_at; |
| 1312 | $this->token_value_starts_at = $value_starts_at; |
| 1313 | $this->token_value_length = $this->at - $value_starts_at - 1; |
| 1314 | return true; |
| 1315 | } |
| 1316 | |
| 1317 | // whitespace |
| 1318 | // Consume as much whitespace as possible. If the next input code point is |
| 1319 | // U+0029 RIGHT PARENTHESIS ()) or EOF, consume it and return the <url-token> |
| 1320 | // (if EOF was encountered, this is a parse error); otherwise, consume the |
| 1321 | // remnants of a bad url, create a <bad-url-token>, and return it. |
| 1322 | $ws_len = strspn( $this->css, "\t\n\f\r ", $this->at ); |
| 1323 | if ( $ws_len > 0 ) { |
| 1324 | $value_ends_at = $this->at; |
| 1325 | $this->at += $ws_len; |
| 1326 | // Accept either ) or EOF after whitespace. |
| 1327 | if ( $this->at >= $this->length ) { |
| 1328 | // EOF is a parse error, but we return the <url-token> anyway. |
| 1329 | $this->token_type = self::TOKEN_URL; |
| 1330 | $this->token_length = $this->at - $this->token_starts_at; |
| 1331 | $this->token_value_starts_at = $value_starts_at; |
| 1332 | $this->token_value_length = $value_ends_at - $value_starts_at; |
| 1333 | return true; |
| 1334 | } |
| 1335 | |
| 1336 | if ( ')' === $this->css[ $this->at ] ) { |
| 1337 | // Skip the closing parenthesis and return the <url-token>. |
| 1338 | ++$this->at; |
| 1339 | $this->token_type = self::TOKEN_URL; |
| 1340 | $this->token_length = $this->at - $this->token_starts_at; |
| 1341 | $this->token_value_starts_at = $value_starts_at; |
| 1342 | $this->token_value_length = $value_ends_at - $value_starts_at; |
| 1343 | return true; |
| 1344 | } |
| 1345 | |
| 1346 | return $this->consume_remnants_of_bad_url(); |
| 1347 | } |
| 1348 | |
| 1349 | // These codepoints trigger a parse error. |
| 1350 | $byte = ord( $this->css[ $this->at ] ); |
| 1351 | if ( |
| 1352 | '"' === $this->css[ $this->at ] || |
| 1353 | "'" === $this->css[ $this->at ] || |
| 1354 | '(' === $this->css[ $this->at ] || |
| 1355 | |
| 1356 | // Non-printable code point. |
| 1357 | $byte <= 0x08 || |
| 1358 | |
| 1359 | // Line Tabulation. |
| 1360 | 0x0B === $byte || |
| 1361 | |
| 1362 | // Control characters. |
| 1363 | ( $byte >= 0x000E && $byte <= 0x001F ) || |
| 1364 | |
| 1365 | // Delete. |
| 1366 | 0x7F === $byte |
| 1367 | ) { |
| 1368 | // Consume the remnants of a bad url, |
| 1369 | // create a <bad-url-token>, and return it. |
| 1370 | return $this->consume_remnants_of_bad_url(); |
| 1371 | } |
| 1372 | |
| 1373 | // U+005C REVERSE SOLIDUS (\) |
| 1374 | // If the stream starts with a valid escape, consume an escaped code point. |
| 1375 | if ( '\\' === $this->css[ $this->at ] ) { |
| 1376 | if ( $this->is_valid_escape( $this->at ) ) { |
| 1377 | ++$this->at; |
| 1378 | $this->decode_escape_at( $this->at, $matched_bytes ); |
| 1379 | $this->at += $matched_bytes; |
| 1380 | continue; |
| 1381 | } |
| 1382 | // Otherwise, this is a parse error. Consume the remnants of a bad url, |
| 1383 | // create a <bad-url-token>, and return it. |
| 1384 | return $this->consume_remnants_of_bad_url(); |
| 1385 | } |
| 1386 | |
| 1387 | $at = $this->at; |
| 1388 | $invalid_length = 0; |
| 1389 | if ( 1 !== _wp_scan_utf8( $this->css, $at, $invalid_length, null, 1 ) ) { |
| 1390 | /** |
| 1391 | * Trouble ahead! |
| 1392 | * Bytes at $at are not a valid UTF-8 sequence. |
| 1393 | * |
| 1394 | * We'll move forward by $invalid_length bytes and continue processing. |
| 1395 | * Later on, during the string decoding, we'll replace the invalid bytes with U+FFFD |
| 1396 | * via maximal subpart”replacement. |
| 1397 | */ |
| 1398 | $this->at += $invalid_length; |
| 1399 | } else { |
| 1400 | $this->at = $at; |
| 1401 | } |
| 1402 | } |
| 1403 | |
| 1404 | // EOF |
| 1405 | // This is a parse error. Return the <url-token>. |
| 1406 | $this->token_type = self::TOKEN_URL; |
| 1407 | $this->token_length = $this->at - $this->token_starts_at; |
| 1408 | $this->token_value_starts_at = $value_starts_at; |
| 1409 | $this->token_value_length = $this->at - $value_starts_at; |
| 1410 | return true; |
| 1411 | } |
| 1412 | |
| 1413 | /** |
| 1414 | * Finishes a bad url token by consuming remnants. |
| 1415 | * |
| 1416 | * When an invalid character is encountered in a URL, we must consume |
| 1417 | * the remainder of the URL up to the closing ) or EOF. |
| 1418 | * |
| 1419 | * @see https://www.w3.org/TR/css-syntax-3/#consume-remnants-of-bad-url |
| 1420 | * |
| 1421 | * @return bool |
| 1422 | */ |
| 1423 | private function consume_remnants_of_bad_url(): bool { |
| 1424 | while ( $this->at < $this->length ) { |
| 1425 | $this->at += strcspn( $this->css, ')\\', $this->at ); |
| 1426 | |
| 1427 | if ( $this->at >= $this->length ) { |
| 1428 | break; |
| 1429 | } |
| 1430 | |
| 1431 | if ( '\\' === $this->css[ $this->at ] ) { |
| 1432 | ++$this->at; |
| 1433 | if ( $this->is_valid_escape( $this->at - 1 ) ) { |
| 1434 | $this->decode_escape_at( $this->at, $matched_bytes ); |
| 1435 | $this->at += $matched_bytes; |
| 1436 | continue; |
| 1437 | } |
| 1438 | } elseif ( ')' === $this->css[ $this->at ] ) { |
| 1439 | ++$this->at; |
| 1440 | break; |
| 1441 | } |
| 1442 | } |
| 1443 | |
| 1444 | $this->token_type = self::TOKEN_BAD_URL; |
| 1445 | $this->token_length = $this->at - $this->token_starts_at; |
| 1446 | return true; |
| 1447 | } |
| 1448 | |
| 1449 | /** |
| 1450 | * Consumes an identifier sequence. |
| 1451 | * |
| 1452 | * Identifiers can contain letters, digits, hyphens, underscores, non-ASCII |
| 1453 | * characters, and escape sequences. Null bytes are replaced with U+FFFD. |
| 1454 | * |
| 1455 | * Returns the decoded identifier string if escapes were encountered, |
| 1456 | * or null if no decoding was needed (can use raw substring). |
| 1457 | * |
| 1458 | * @see https://www.w3.org/TR/css-syntax-3/#consume-name |
| 1459 | */ |
| 1460 | private function consume_ident_sequence() { |
| 1461 | while ( $this->at < $this->length ) { |
| 1462 | $codepoint_bytes = $this->consume_ident_codepoint( $this->at ); |
| 1463 | if ( $codepoint_bytes > 0 ) { |
| 1464 | $this->at += $codepoint_bytes; |
| 1465 | continue; |
| 1466 | } |
| 1467 | |
| 1468 | if ( $this->is_valid_escape( $this->at ) ) { |
| 1469 | ++$this->at; |
| 1470 | |
| 1471 | $this->decode_escape_at( $this->at, $matched_bytes ); |
| 1472 | $this->at += $matched_bytes; |
| 1473 | continue; |
| 1474 | } |
| 1475 | |
| 1476 | break; |
| 1477 | } |
| 1478 | } |
| 1479 | |
| 1480 | /** |
| 1481 | * Ident-start code point |
| 1482 | * A letter, a non-ASCII code point, or U+005F LOW LINE (_). |
| 1483 | * |
| 1484 | * Ident code point |
| 1485 | * An ident-start code point, a digit, or U+002D HYPHEN-MINUS (-). |
| 1486 | * |
| 1487 | * @see https://www.w3.org/TR/css-syntax-3/#ident-start-code-point |
| 1488 | * @return int The number of bytes consumed. |
| 1489 | */ |
| 1490 | private function consume_ident_codepoint( $at ): int { |
| 1491 | // ident code points. |
| 1492 | if ( ( $this->css[ $at ] >= '0' && $this->css[ $at ] <= '9' ) || |
| 1493 | '-' === $this->css[ $at ] ) { |
| 1494 | return 1; |
| 1495 | } |
| 1496 | |
| 1497 | return $this->consume_ident_start_codepoint( $at ); |
| 1498 | } |
| 1499 | |
| 1500 | |
| 1501 | /** |
| 1502 | * Ident-start code point |
| 1503 | * A letter, a non-ASCII code point, or U+005F LOW LINE (_). |
| 1504 | * |
| 1505 | * Ident code point |
| 1506 | * An ident-start code point, a digit, or U+002D HYPHEN-MINUS (-). |
| 1507 | * |
| 1508 | * @see https://www.w3.org/TR/css-syntax-3/#ident-start-code-point |
| 1509 | * @return int The number of bytes consumed. |
| 1510 | */ |
| 1511 | private function consume_ident_start_codepoint( $at ): int { |
| 1512 | if ( $at >= $this->length ) { |
| 1513 | return 0; |
| 1514 | } |
| 1515 | |
| 1516 | // ASCII codepoints. |
| 1517 | if ( ( $this->css[ $at ] >= 'A' && $this->css[ $at ] <= 'Z' ) || |
| 1518 | ( $this->css[ $at ] >= 'a' && $this->css[ $at ] <= 'z' ) || |
| 1519 | '_' === $this->css[ $at ] ) { |
| 1520 | return 1; |
| 1521 | } |
| 1522 | |
| 1523 | // Special case for null bytes – they are replaced with U+FFFD during preprocessing. |
| 1524 | if ( "\x00" === $this->css[ $at ] ) { |
| 1525 | return 1; |
| 1526 | } |
| 1527 | |
| 1528 | $new_at = $at; |
| 1529 | $invalid_length = 0; |
| 1530 | if ( 1 !== _wp_scan_utf8( $this->css, $new_at, $invalid_length, null, 1 ) ) { |
| 1531 | /** |
| 1532 | * Trouble ahead! |
| 1533 | * Bytes at $at are not a valid UTF-8 sequence. |
| 1534 | * |
| 1535 | * We'll move forward by $invalid_length bytes and continue processing. |
| 1536 | * Later on, during the string decoding, we'll replace the invalid bytes with U+FFFD |
| 1537 | * via maximal subpart”replacement. |
| 1538 | */ |
| 1539 | return $invalid_length; |
| 1540 | } |
| 1541 | |
| 1542 | $codepoint_byte_length = $new_at - $at; |
| 1543 | $codepoint = utf8_ord( substr( $this->css, $at, $codepoint_byte_length ) ); |
| 1544 | if ( null !== $codepoint && $codepoint >= 0x80 ) { |
| 1545 | return $codepoint_byte_length; |
| 1546 | } |
| 1547 | return 0; |
| 1548 | } |
| 1549 | |
| 1550 | /** |
| 1551 | * Decodes and normalizes ident-like or string CSS values from a byte range. |
| 1552 | * |
| 1553 | * For example: |
| 1554 | * ┌──────────────┬────────┐ |
| 1555 | * │ Input │ Output │ |
| 1556 | * ├──────────────┼────────┤ |
| 1557 | * │ 'xyz' │ 'xyz' │ |
| 1558 | * │ '\x\y\z' │ 'xyz' │ |
| 1559 | * │ 'x\79z' │ 'xyz' │ |
| 1560 | * │ 'x\000079 z' │ 'xyz' │ |
| 1561 | * │ 'a\r\nb' │ 'a\nb' │ |
| 1562 | * │ 'a\0b' │ 'a�b' │ |
| 1563 | * └──────────────┴────────┘ |
| 1564 | * |
| 1565 | * @param int $start Start byte offset. |
| 1566 | * @param int $length Length of the substring to decode. |
| 1567 | * @param bool $string_escapes Optional, default false. When true, apply additional escape |
| 1568 | * rules that apply only to string tokens. |
| 1569 | * @return string Decoded and normalized string. |
| 1570 | */ |
| 1571 | private function decode_range( int $start, int $length, bool $string_escapes = false ): string { |
| 1572 | // Fast path: check if any processing is needed. |
| 1573 | $slice = wp_scrub_utf8( substr( $this->css, $start, $length ) ); |
| 1574 | $special_chars = "\\\r\f\x00"; |
| 1575 | if ( false === strpbrk( $slice, $special_chars ) ) { |
| 1576 | // No special chars - return raw substring (almost zero allocations). |
| 1577 | return $slice; |
| 1578 | } |
| 1579 | |
| 1580 | // Slow path: build decoded string (one allocation). |
| 1581 | $decoded = ''; |
| 1582 | $at = $start; |
| 1583 | $end = $start + $length; |
| 1584 | |
| 1585 | while ( $at < $end ) { |
| 1586 | // Find next special character. |
| 1587 | $normal_len = strcspn( $this->css, $special_chars, $at ); |
| 1588 | if ( $normal_len > 0 ) { |
| 1589 | // Clamp to not exceed the end boundary. |
| 1590 | $normal_len = min( $normal_len, $end - $at ); |
| 1591 | $decoded .= substr( $this->css, $at, $normal_len ); |
| 1592 | $at += $normal_len; |
| 1593 | } |
| 1594 | |
| 1595 | if ( $at >= $end ) { |
| 1596 | break; |
| 1597 | } |
| 1598 | |
| 1599 | $char = $this->css[ $at ]; |
| 1600 | |
| 1601 | // Handle escapes. |
| 1602 | if ( '\\' === $char ) { |
| 1603 | /** |
| 1604 | * String tokens have special escape rules: |
| 1605 | * - 0x5C (backslash) at EOF: consume the backslash, produce no value. |
| 1606 | * - 0x5C (backslash) followed by 0x0A (LF), 0x0C (FF), or 0x0D (CR): |
| 1607 | * consume both characters as a line continuation, produce no value. |
| 1608 | * - 0x5C (backslash) followed by 0x0D 0x0A (CRLF): |
| 1609 | * consume all three characters as a line continuation, produce no value. |
| 1610 | * These must be checked before the general escape path. |
| 1611 | * |
| 1612 | * @see https://www.w3.org/TR/css-syntax-3/#consume-string-token |
| 1613 | */ |
| 1614 | if ( $string_escapes ) { |
| 1615 | if ( $at + 1 >= $end ) { |
| 1616 | // 0x5C at EOF: consume the backslash and stop. |
| 1617 | ++$at; |
| 1618 | continue; |
| 1619 | } |
| 1620 | $next = $this->css[ $at + 1 ]; |
| 1621 | if ( "\n" === $next || "\f" === $next ) { |
| 1622 | // 0x5C followed by 0x0A (LF) or 0x0C (FF): line continuation. |
| 1623 | $at += 2; |
| 1624 | continue; |
| 1625 | } |
| 1626 | if ( "\r" === $next ) { |
| 1627 | // 0x5C followed by 0x0D (CR): line continuation; 0x0D 0x0A counts as one newline. |
| 1628 | $at += 2; |
| 1629 | if ( $at < $end && "\n" === $this->css[ $at ] ) { |
| 1630 | ++$at; |
| 1631 | } |
| 1632 | continue; |
| 1633 | } |
| 1634 | } |
| 1635 | |
| 1636 | if ( $this->is_valid_escape( $at ) ) { |
| 1637 | ++$at; |
| 1638 | $decoded .= $this->decode_escape_at( $at, $bytes_consumed ); |
| 1639 | $at += $bytes_consumed; |
| 1640 | continue; |
| 1641 | } |
| 1642 | // Invalid escape - consume the backslash and keep going. |
| 1643 | $decoded .= '\\'; |
| 1644 | ++$at; |
| 1645 | continue; |
| 1646 | } |
| 1647 | |
| 1648 | // CSS normalization: \r\n, \r, and \f all become \n. |
| 1649 | if ( "\r" === $char ) { |
| 1650 | $decoded .= "\n"; |
| 1651 | ++$at; |
| 1652 | // Handle \r\n as single newline. |
| 1653 | if ( $at < $end && "\n" === $this->css[ $at ] ) { |
| 1654 | ++$at; |
| 1655 | } |
| 1656 | continue; |
| 1657 | } |
| 1658 | |
| 1659 | if ( "\f" === $char ) { |
| 1660 | $decoded .= "\n"; |
| 1661 | ++$at; |
| 1662 | continue; |
| 1663 | } |
| 1664 | |
| 1665 | // Null bytes become U+FFFD. |
| 1666 | if ( "\x00" === $char ) { |
| 1667 | $decoded .= "\u{FFFD}"; |
| 1668 | ++$at; |
| 1669 | continue; |
| 1670 | } |
| 1671 | } |
| 1672 | |
| 1673 | return $decoded; |
| 1674 | } |
| 1675 | |
| 1676 | /** |
| 1677 | * Decodes an escape sequence starting at the given offset without |
| 1678 | * modifying $this->at. |
| 1679 | * |
| 1680 | * Escape sequences are backslash followed by 1-6 hex digits (with optional |
| 1681 | * trailing whitespace) or any other character. Invalid code points are |
| 1682 | * replaced with U+FFFD. |
| 1683 | * |
| 1684 | * @see https://www.w3.org/TR/css-syntax-3/#consume-escaped-code-point |
| 1685 | * |
| 1686 | * @param int $offset Byte offset (should point to the character after the backslash). |
| 1687 | * @param int &$bytes_consumed Output parameter: number of bytes consumed. |
| 1688 | * @return string The decoded character(s). |
| 1689 | */ |
| 1690 | private function decode_escape_at( int $offset, &$bytes_consumed ): string { |
| 1691 | // This method assumes the U+005C REVERSE SOLIDUS (\) has already been consumed |
| 1692 | // and the next input code point has already been verified to be part of a valid |
| 1693 | // escape sequence. |
| 1694 | $at = $offset; |
| 1695 | |
| 1696 | // EOF. |
| 1697 | if ( $at >= $this->length ) { |
| 1698 | // This is a parse error. Return U+FFFD REPLACEMENT CHARACTER (�). |
| 1699 | $bytes_consumed = 0; |
| 1700 | return "\u{FFFD}"; |
| 1701 | } |
| 1702 | |
| 1703 | // Hex digits. |
| 1704 | $hex_len = strspn( $this->css, '0123456789ABCDEFabcdef', $at ); |
| 1705 | if ( $hex_len > 0 ) { |
| 1706 | // Consume up to 6 hex digits. |
| 1707 | $hex_len = min( $hex_len, 6 ); |
| 1708 | $hex = substr( $this->css, $at, $hex_len ); |
| 1709 | $at += $hex_len; |
| 1710 | |
| 1711 | // If the next input code point is whitespace, consume it as well. |
| 1712 | if ( $at < $this->length ) { |
| 1713 | $next = $this->css[ $at ]; |
| 1714 | if ( "\t" === $next || "\n" === $next || "\f" === $next || ' ' === $next ) { |
| 1715 | ++$at; |
| 1716 | } elseif ( "\r" === $next ) { |
| 1717 | ++$at; |
| 1718 | // Handle \r\n as a single whitespace – the preprocessing phase would replace \r\n with \n. |
| 1719 | if ( $at < $this->length && "\n" === $this->css[ $at ] ) { |
| 1720 | ++$at; |
| 1721 | } |
| 1722 | } |
| 1723 | } |
| 1724 | |
| 1725 | $bytes_consumed = $at - $offset; |
| 1726 | // Convert the hex digits to a UTF-8 string. |
| 1727 | return codepoint_to_utf8_bytes( hexdec( $hex ) ); |
| 1728 | } |
| 1729 | |
| 1730 | // Anything else. |
| 1731 | // Return the current input code point. |
| 1732 | // Null bytes are replaced with U+FFFD during preprocessing. |
| 1733 | if ( "\x00" === $this->css[ $at ] ) { |
| 1734 | $bytes_consumed = 1; |
| 1735 | return "\u{FFFD}"; |
| 1736 | } |
| 1737 | |
| 1738 | $new_at = $at; |
| 1739 | $invalid_length = 0; |
| 1740 | if ( 1 !== _wp_scan_utf8( $this->css, $new_at, $invalid_length, null, 1 ) ) { |
| 1741 | /** |
| 1742 | * Trouble ahead! |
| 1743 | * Bytes at $at are not a valid UTF-8 sequence. |
| 1744 | * |
| 1745 | * We'll move forward by $invalid_length bytes and continue processing. |
| 1746 | * Later on, during the string decoding, we'll replace the invalid bytes with U+FFFD |
| 1747 | * via maximal subpart”replacement. |
| 1748 | */ |
| 1749 | $matched_bytes = $invalid_length; |
| 1750 | } else { |
| 1751 | $matched_bytes = $new_at - $at; |
| 1752 | } |
| 1753 | |
| 1754 | $bytes_consumed = $matched_bytes; |
| 1755 | return substr( $this->css, $at, $matched_bytes ); |
| 1756 | } |
| 1757 | |
| 1758 | /** |
| 1759 | * Checks if current position starts a valid escape sequence. |
| 1760 | * |
| 1761 | * A valid escape is a backslash not followed by a newline or EOF. |
| 1762 | * |
| 1763 | * @see https://www.w3.org/TR/css-syntax-3/#starts-with-a-valid-escape |
| 1764 | * |
| 1765 | * @param int $offset Byte offset. |
| 1766 | * @return bool |
| 1767 | */ |
| 1768 | private function is_valid_escape( int $offset ): bool { |
| 1769 | // If the first code point is not U+005C REVERSE SOLIDUS (\), return false. |
| 1770 | if ( $offset >= $this->length || '\\' !== $this->css[ $offset ] ) { |
| 1771 | return false; |
| 1772 | } |
| 1773 | // Otherwise, if the second code point is a newline, return false. |
| 1774 | if ( $offset + 1 >= $this->length ) { |
| 1775 | // Second code point is EOF - this is a valid escape per spec (weird!) |
| 1776 | // Are we sure we're interpreting the spec correctly? |
| 1777 | return true; |
| 1778 | } |
| 1779 | |
| 1780 | // Otherwise, if the second code point is not a newline, return true. |
| 1781 | return ( |
| 1782 | "\n" !== $this->css[ $offset + 1 ] && |
| 1783 | |
| 1784 | // Form feed is normalized to newline during preprocessing. |
| 1785 | "\f" !== $this->css[ $offset + 1 ] && |
| 1786 | |
| 1787 | // Carriage return is normalized to newline during preprocessing. |
| 1788 | "\r" !== $this->css[ $offset + 1 ] |
| 1789 | |
| 1790 | // We don't need to check for \r\n separately here. The \r check alone covers |
| 1791 | // that scenario. |
| 1792 | ); |
| 1793 | } |
| 1794 | |
| 1795 | /** |
| 1796 | * Checks if the next 3 code points would start a number. |
| 1797 | * |
| 1798 | * @see https://www.w3.org/TR/css-syntax-3/#starts-with-a-number |
| 1799 | * |
| 1800 | * @return bool |
| 1801 | */ |
| 1802 | private function would_next_3_code_points_start_a_number(): bool { |
| 1803 | if ( $this->at >= $this->length ) { |
| 1804 | return false; |
| 1805 | } |
| 1806 | |
| 1807 | // Look at the first code point. |
| 1808 | |
| 1809 | // U+002B PLUS SIGN (+) or U+002D HYPHEN-MINUS (-). |
| 1810 | if ( '+' === $this->css[ $this->at ] || '-' === $this->css[ $this->at ] ) { |
| 1811 | if ( $this->at + 1 >= $this->length ) { |
| 1812 | return false; |
| 1813 | } |
| 1814 | // If the second code point is a digit, return true. |
| 1815 | if ( $this->css[ $this->at + 1 ] >= '0' && $this->css[ $this->at + 1 ] <= '9' ) { |
| 1816 | return true; |
| 1817 | } |
| 1818 | // Otherwise, the second code point must be a full stop (.) and the third code point must be a digit. |
| 1819 | if ( '.' === $this->css[ $this->at + 1 ] && $this->at + 2 < $this->length ) { |
| 1820 | return $this->css[ $this->at + 2 ] >= '0' && $this->css[ $this->at + 2 ] <= '9'; |
| 1821 | } |
| 1822 | |
| 1823 | // Otherwise, return false. |
| 1824 | return false; |
| 1825 | } |
| 1826 | |
| 1827 | // U+002E FULL STOP (.). |
| 1828 | if ( '.' === $this->css[ $this->at ] ) { |
| 1829 | if ( $this->at + 1 >= $this->length ) { |
| 1830 | return false; |
| 1831 | } |
| 1832 | return $this->css[ $this->at + 1 ] >= '0' && $this->css[ $this->at + 1 ] <= '9'; |
| 1833 | } |
| 1834 | |
| 1835 | // Digit. |
| 1836 | if ( $this->css[ $this->at ] >= '0' && $this->css[ $this->at ] <= '9' ) { |
| 1837 | return true; |
| 1838 | } |
| 1839 | |
| 1840 | // Anything else – return false. |
| 1841 | return false; |
| 1842 | } |
| 1843 | |
| 1844 | /** |
| 1845 | * Checks if three code points would start an identifier sequence. |
| 1846 | * |
| 1847 | * This implements the CSS spec's "Check if three code points would start an ident sequence" |
| 1848 | * algorithm, which checks the code point at $offset and the following two code points. |
| 1849 | * |
| 1850 | * NOTE: "Three code points" means three Unicode code points, not three bytes. |
| 1851 | * Multi-byte UTF-8 sequences count as single code points. |
| 1852 | * |
| 1853 | * @see https://www.w3.org/TR/css-syntax-3/#would-start-an-identifier |
| 1854 | * |
| 1855 | * @param int $offset Byte offset of the first code point to check. |
| 1856 | * @return bool |
| 1857 | */ |
| 1858 | private function check_if_3_code_points_start_an_ident_sequence( int $offset ): bool { |
| 1859 | if ( $offset >= $this->length ) { |
| 1860 | return false; |
| 1861 | } |
| 1862 | |
| 1863 | if ( '-' === $this->css[ $offset ] ) { |
| 1864 | // If the second code point is a U+002D HYPHEN-MINUS (-), return true. |
| 1865 | // e.g. --custom-property. |
| 1866 | if ( $offset + 1 < $this->length && '-' === $this->css[ $offset + 1 ] ) { |
| 1867 | return true; |
| 1868 | } |
| 1869 | // Otherwise, check if the second code point is an ident-START code point or valid escape. |
| 1870 | // Note: After a hyphen, only ident-START code points are valid, NOT digits or hyphens. |
| 1871 | ++$offset; |
| 1872 | } |
| 1873 | |
| 1874 | return $this->consume_ident_start_codepoint( $offset ) > 0 || $this->is_valid_escape( $offset ); |
| 1875 | } |
| 1876 | } |
| 1877 |