← All changes
|
back/Libraries/Sabberworm/CSS/src/Parsing/ParserState.php
+458
-0
10.10.1
→
11.1.0
View file →
| @@ -1,0 +1,458 @@ | ||
| 1 | +<?php | |
| 2 | + | |
| 3 | +declare(strict_types=1); | |
| 4 | + | |
| 5 | +namespace AcyMailing\Libraries\Sabberworm\CSS\Parsing; | |
| 6 | + | |
| 7 | +use AcyMailing\Libraries\Sabberworm\CSS\Comment\Comment; | |
| 8 | +use AcyMailing\Libraries\Sabberworm\CSS\CSSList\CSSList; | |
| 9 | +use AcyMailing\Libraries\Sabberworm\CSS\Settings; | |
| 10 | + | |
| 11 | +/** | |
| 12 | + * @internal since 8.7.0 | |
| 13 | + */ | |
| 14 | +class ParserState | |
| 15 | +{ | |
| 16 | + /** | |
| 17 | + * @var null | |
| 18 | + */ | |
| 19 | + public const EOF = null; | |
| 20 | + | |
| 21 | + /** | |
| 22 | + * @var Settings | |
| 23 | + */ | |
| 24 | + private $parserSettings; | |
| 25 | + | |
| 26 | + /** | |
| 27 | + * @var string | |
| 28 | + */ | |
| 29 | + private $text; | |
| 30 | + | |
| 31 | + /** | |
| 32 | + * @var array<int, string> | |
| 33 | + */ | |
| 34 | + private $characters; | |
| 35 | + | |
| 36 | + /** | |
| 37 | + * @var int<0, max> | |
| 38 | + */ | |
| 39 | + private $currentPosition = 0; | |
| 40 | + | |
| 41 | + /** | |
| 42 | + * will only be used if the CSS does not contain an `@charset` declaration | |
| 43 | + * | |
| 44 | + * @var string | |
| 45 | + */ | |
| 46 | + private $charset; | |
| 47 | + | |
| 48 | + /** | |
| 49 | + * @var int<1, max> $lineNumber | |
| 50 | + */ | |
| 51 | + private $lineNumber; | |
| 52 | + | |
| 53 | + /** | |
| 54 | + * @param string $text the complete CSS as text (i.e., usually the contents of a CSS file) | |
| 55 | + * @param int<1, max> $lineNumber | |
| 56 | + */ | |
| 57 | + public function __construct(string $text, Settings $parserSettings, int $lineNumber = 1) | |
| 58 | + { | |
| 59 | + $this->parserSettings = $parserSettings; | |
| 60 | + $this->text = $text; | |
| 61 | + $this->lineNumber = $lineNumber; | |
| 62 | + $this->setCharset($this->parserSettings->getDefaultCharset()); | |
| 63 | + } | |
| 64 | + | |
| 65 | + /** | |
| 66 | + * Sets the charset to be used if the CSS does not contain an `@charset` declaration. | |
| 67 | + */ | |
| 68 | + public function setCharset(string $charset): void | |
| 69 | + { | |
| 70 | + $this->charset = $charset; | |
| 71 | + $this->characters = $this->strsplit($this->text); | |
| 72 | + } | |
| 73 | + | |
| 74 | + /** | |
| 75 | + * @return int<1, max> | |
| 76 | + */ | |
| 77 | + public function currentLine(): int | |
| 78 | + { | |
| 79 | + return $this->lineNumber; | |
| 80 | + } | |
| 81 | + | |
| 82 | + /** | |
| 83 | + * @return int<0, max> | |
| 84 | + */ | |
| 85 | + public function currentColumn(): int | |
| 86 | + { | |
| 87 | + return $this->currentPosition; | |
| 88 | + } | |
| 89 | + | |
| 90 | + public function getSettings(): Settings | |
| 91 | + { | |
| 92 | + return $this->parserSettings; | |
| 93 | + } | |
| 94 | + | |
| 95 | + public function anchor(): Anchor | |
| 96 | + { | |
| 97 | + return new Anchor($this->currentPosition, $this); | |
| 98 | + } | |
| 99 | + | |
| 100 | + /** | |
| 101 | + * @param int<0, max> $position | |
| 102 | + */ | |
| 103 | + public function setPosition(int $position): void | |
| 104 | + { | |
| 105 | + $this->currentPosition = $position; | |
| 106 | + } | |
| 107 | + | |
| 108 | + /** | |
| 109 | + * @return non-empty-string | |
| 110 | + * | |
| 111 | + * @throws UnexpectedTokenException | |
| 112 | + */ | |
| 113 | + public function parseIdentifier(bool $ignoreCase = true): string | |
| 114 | + { | |
| 115 | + if ($this->isEnd()) { | |
| 116 | + throw new UnexpectedEOFException('', '', 'identifier', $this->lineNumber); | |
| 117 | + } | |
| 118 | + $result = $this->parseCharacter(true); | |
| 119 | + if ($result === null) { | |
| 120 | + throw new UnexpectedTokenException('', $this->peek(5), 'identifier', $this->lineNumber); | |
| 121 | + } | |
| 122 | + $character = null; | |
| 123 | + while (!$this->isEnd() && ($character = $this->parseCharacter(true)) !== null) { | |
| 124 | + if (CSSList::preg_match('/[a-zA-Z0-9\\x{00A0}-\\x{FFFF}_-]/Sux', $character) !== 0) { | |
| 125 | + $result .= $character; | |
| 126 | + } else { | |
| 127 | + $result .= '\\' . $character; | |
| 128 | + } | |
| 129 | + } | |
| 130 | + if ($ignoreCase) { | |
| 131 | + $result = $this->strtolower($result); | |
| 132 | + } | |
| 133 | + | |
| 134 | + return $result; | |
| 135 | + } | |
| 136 | + | |
| 137 | + /** | |
| 138 | + * @throws UnexpectedEOFException | |
| 139 | + * @throws UnexpectedTokenException | |
| 140 | + */ | |
| 141 | + public function parseCharacter(bool $isForIdentifier): ?string | |
| 142 | + { | |
| 143 | + if ($this->peek() === '\\') { | |
| 144 | + $this->consume('\\'); | |
| 145 | + if ($this->comes('\\n') || $this->comes('\\r')) { | |
| 146 | + return ''; | |
| 147 | + } | |
| 148 | + if (CSSList::preg_match('/[0-9a-fA-F]/Su', $this->peek()) === 0) { | |
| 149 | + return $this->consume(1); | |
| 150 | + } | |
| 151 | + $hexCodePoint = $this->consumeExpression('/^[0-9a-fA-F]{1,6}/u', 6); | |
| 152 | + if ($this->strlen($hexCodePoint) < 6) { | |
| 153 | + // Consume whitespace after incomplete unicode escape | |
| 154 | + if (CSSList::preg_match('/\\s/isSu', $this->peek()) !== 0) { | |
| 155 | + if ($this->comes('\\r\\n')) { | |
| 156 | + $this->consume(2); | |
| 157 | + } else { | |
| 158 | + $this->consume(1); | |
| 159 | + } | |
| 160 | + } | |
| 161 | + } | |
| 162 | + $codePoint = \intval($hexCodePoint, 16); | |
| 163 | + $utf32EncodedCharacter = ''; | |
| 164 | + for ($i = 0; $i < 4; ++$i) { | |
| 165 | + $utf32EncodedCharacter .= \chr($codePoint & 0xff); | |
| 166 | + $codePoint = $codePoint >> 8; | |
| 167 | + } | |
| 168 | + return CSSList::iconv('utf-32le', $this->charset, $utf32EncodedCharacter); | |
| 169 | + } | |
| 170 | + if ($isForIdentifier) { | |
| 171 | + $peek = \ord($this->peek()); | |
| 172 | + // Ranges: a-z A-Z 0-9 - _ | |
| 173 | + if ( | |
| 174 | + ($peek >= 97 && $peek <= 122) | |
| 175 | + || ($peek >= 65 && $peek <= 90) | |
| 176 | + || ($peek >= 48 && $peek <= 57) | |
| 177 | + || ($peek === 45) | |
| 178 | + || ($peek === 95) | |
| 179 | + || ($peek > 0xa1) | |
| 180 | + ) { | |
| 181 | + return $this->consume(1); | |
| 182 | + } | |
| 183 | + } else { | |
| 184 | + return $this->consume(1); | |
| 185 | + } | |
| 186 | + | |
| 187 | + return null; | |
| 188 | + } | |
| 189 | + | |
| 190 | + /** | |
| 191 | + * @return list<Comment> | |
| 192 | + * | |
| 193 | + * @throws UnexpectedEOFException | |
| 194 | + * @throws UnexpectedTokenException | |
| 195 | + */ | |
| 196 | + public function consumeWhiteSpace(): array | |
| 197 | + { | |
| 198 | + $comments = []; | |
| 199 | + do { | |
| 200 | + while (CSSList::preg_match('/\\s/isSu', $this->peek()) === 1) { | |
| 201 | + $this->consume(1); | |
| 202 | + } | |
| 203 | + if ($this->parserSettings->usesLenientParsing()) { | |
| 204 | + try { | |
| 205 | + $comment = $this->consumeComment(); | |
| 206 | + } catch (UnexpectedEOFException $e) { | |
| 207 | + $this->currentPosition = \count($this->characters); | |
| 208 | + break; | |
| 209 | + } | |
| 210 | + } else { | |
| 211 | + $comment = $this->consumeComment(); | |
| 212 | + } | |
| 213 | + if ($comment instanceof Comment) { | |
| 214 | + $comments[] = $comment; | |
| 215 | + } | |
| 216 | + } while ($comment instanceof Comment); | |
| 217 | + | |
| 218 | + return $comments; | |
| 219 | + } | |
| 220 | + | |
| 221 | + /** | |
| 222 | + * @param non-empty-string $string | |
| 223 | + */ | |
| 224 | + public function comes(string $string, bool $caseInsensitive = false): bool | |
| 225 | + { | |
| 226 | + $peek = $this->peek(\strlen($string)); | |
| 227 | + | |
| 228 | + return ($peek !== '') && $this->streql($peek, $string, $caseInsensitive); | |
| 229 | + } | |
| 230 | + | |
| 231 | + /** | |
| 232 | + * @param int<1, max> $length | |
| 233 | + * @param int<0, max> $offset | |
| 234 | + */ | |
| 235 | + public function peek(int $length = 1, int $offset = 0): string | |
| 236 | + { | |
| 237 | + $offset += $this->currentPosition; | |
| 238 | + if ($offset >= \count($this->characters)) { | |
| 239 | + return ''; | |
| 240 | + } | |
| 241 | + | |
| 242 | + return $this->substr($offset, $length); | |
| 243 | + } | |
| 244 | + | |
| 245 | + /** | |
| 246 | + * @param string|int<1, max> $value | |
| 247 | + * | |
| 248 | + * @throws UnexpectedEOFException | |
| 249 | + * @throws UnexpectedTokenException | |
| 250 | + */ | |
| 251 | + public function consume($value = 1): string | |
| 252 | + { | |
| 253 | + if (\is_string($value)) { | |
| 254 | + $numberOfLines = \substr_count($value, "\n"); | |
| 255 | + $length = $this->strlen($value); | |
| 256 | + if (!$this->streql($this->substr($this->currentPosition, $length), $value)) { | |
| 257 | + throw new UnexpectedTokenException( | |
| 258 | + $value, | |
| 259 | + $this->peek(\max($length, 5)), | |
| 260 | + 'literal', | |
| 261 | + $this->lineNumber | |
| 262 | + ); | |
| 263 | + } | |
| 264 | + | |
| 265 | + $this->lineNumber += $numberOfLines; | |
| 266 | + $this->currentPosition += $this->strlen($value); | |
| 267 | + $result = $value; | |
| 268 | + } else { | |
| 269 | + if ($this->currentPosition + $value > \count($this->characters)) { | |
| 270 | + throw new UnexpectedEOFException((string) $value, $this->peek(5), 'count', $this->lineNumber); | |
| 271 | + } | |
| 272 | + | |
| 273 | + $result = $this->substr($this->currentPosition, $value); | |
| 274 | + $numberOfLines = \substr_count($result, "\n"); | |
| 275 | + $this->lineNumber += $numberOfLines; | |
| 276 | + $this->currentPosition += $value; | |
| 277 | + } | |
| 278 | + | |
| 279 | + return $result; | |
| 280 | + } | |
| 281 | + | |
| 282 | + /** | |
| 283 | + * @param string $expression | |
| 284 | + * @param int<1, max>|null $maximumLength | |
| 285 | + * | |
| 286 | + * @throws UnexpectedEOFException | |
| 287 | + * @throws UnexpectedTokenException | |
| 288 | + */ | |
| 289 | + public function consumeExpression(string $expression, ?int $maximumLength = null): string | |
| 290 | + { | |
| 291 | + $matches = null; | |
| 292 | + $input = ($maximumLength !== null) ? $this->peek($maximumLength) : $this->inputLeft(); | |
| 293 | + if (CSSList::preg_match($expression, $input, $matches, PREG_OFFSET_CAPTURE) !== 1) { | |
| 294 | + throw new UnexpectedTokenException($expression, $this->peek(5), 'expression', $this->lineNumber); | |
| 295 | + } | |
| 296 | + | |
| 297 | + return $this->consume($matches[0][0]); | |
| 298 | + } | |
| 299 | + | |
| 300 | + /** | |
| 301 | + * @return Comment|false | |
| 302 | + */ | |
| 303 | + public function consumeComment() | |
| 304 | + { | |
| 305 | + $lineNumber = $this->lineNumber; | |
| 306 | + $comment = null; | |
| 307 | + | |
| 308 | + if ($this->comes('/*')) { | |
| 309 | + $this->consume(1); | |
| 310 | + $comment = ''; | |
| 311 | + while (($char = $this->consume(1)) !== '') { | |
| 312 | + $comment .= $char; | |
| 313 | + if ($this->comes('*/')) { | |
| 314 | + $this->consume(2); | |
| 315 | + break; | |
| 316 | + } | |
| 317 | + } | |
| 318 | + } | |
| 319 | + | |
| 320 | + // We skip the * which was included in the comment. | |
| 321 | + return \is_string($comment) ? new Comment(\substr($comment, 1), $lineNumber) : false; | |
| 322 | + } | |
| 323 | + | |
| 324 | + public function isEnd(): bool | |
| 325 | + { | |
| 326 | + return $this->currentPosition >= \count($this->characters); | |
| 327 | + } | |
| 328 | + | |
| 329 | + /** | |
| 330 | + * @param list<string|self::EOF>|string|self::EOF $stopCharacters | |
| 331 | + * @param array<int, Comment> $comments | |
| 332 | + * | |
| 333 | + * @throws UnexpectedEOFException | |
| 334 | + * @throws UnexpectedTokenException | |
| 335 | + */ | |
| 336 | + public function consumeUntil( | |
| 337 | + $stopCharacters, | |
| 338 | + bool $includeEnd = false, | |
| 339 | + bool $consumeEnd = false, | |
| 340 | + array &$comments = [] | |
| 341 | + ): string { | |
| 342 | + $stopCharacters = \is_array($stopCharacters) ? $stopCharacters : [$stopCharacters]; | |
| 343 | + $consumedCharacters = ''; | |
| 344 | + $start = $this->currentPosition; | |
| 345 | + | |
| 346 | + while (!$this->isEnd()) { | |
| 347 | + $character = $this->consume(1); | |
| 348 | + if (\in_array($character, $stopCharacters, true)) { | |
| 349 | + if ($includeEnd) { | |
| 350 | + $consumedCharacters .= $character; | |
| 351 | + } elseif (!$consumeEnd) { | |
| 352 | + $this->currentPosition -= $this->strlen($character); | |
| 353 | + } | |
| 354 | + return $consumedCharacters; | |
| 355 | + } | |
| 356 | + $consumedCharacters .= $character; | |
| 357 | + $comment = $this->consumeComment(); | |
| 358 | + if ($comment instanceof Comment) { | |
| 359 | + $comments[] = $comment; | |
| 360 | + } | |
| 361 | + } | |
| 362 | + | |
| 363 | + if (\in_array(self::EOF, $stopCharacters, true)) { | |
| 364 | + return $consumedCharacters; | |
| 365 | + } | |
| 366 | + | |
| 367 | + $this->currentPosition = $start; | |
| 368 | + throw new UnexpectedEOFException( | |
| 369 | + 'One of ("' . \implode('","', $stopCharacters) . '")', | |
| 370 | + $this->peek(5), | |
| 371 | + 'search', | |
| 372 | + $this->lineNumber | |
| 373 | + ); | |
| 374 | + } | |
| 375 | + | |
| 376 | + private function inputLeft(): string | |
| 377 | + { | |
| 378 | + return $this->substr($this->currentPosition, -1); | |
| 379 | + } | |
| 380 | + | |
| 381 | + public function streql(string $string1, string $string2, bool $caseInsensitive = true): bool | |
| 382 | + { | |
| 383 | + return $caseInsensitive | |
| 384 | + ? ($this->strtolower($string1) === $this->strtolower($string2)) | |
| 385 | + : ($string1 === $string2); | |
| 386 | + } | |
| 387 | + | |
| 388 | + /** | |
| 389 | + * @param int<1, max> $numberOfCharacters | |
| 390 | + */ | |
| 391 | + public function backtrack(int $numberOfCharacters): void | |
| 392 | + { | |
| 393 | + $this->currentPosition -= $numberOfCharacters; | |
| 394 | + } | |
| 395 | + | |
| 396 | + /** | |
| 397 | + * @return int<0, max> | |
| 398 | + */ | |
| 399 | + public function strlen(string $string): int | |
| 400 | + { | |
| 401 | + return $this->parserSettings->hasMultibyteSupport() | |
| 402 | + ? \mb_strlen($string, $this->charset) | |
| 403 | + : \strlen($string); | |
| 404 | + } | |
| 405 | + | |
| 406 | + /** | |
| 407 | + * @param int<0, max> $offset | |
| 408 | + */ | |
| 409 | + private function substr(int $offset, int $length): string | |
| 410 | + { | |
| 411 | + if ($length < 0) { | |
| 412 | + $length = \count($this->characters) - $offset + $length; | |
| 413 | + } | |
| 414 | + if ($offset + $length > \count($this->characters)) { | |
| 415 | + $length = \count($this->characters) - $offset; | |
| 416 | + } | |
| 417 | + $result = ''; | |
| 418 | + while ($length > 0) { | |
| 419 | + $result .= $this->characters[$offset]; | |
| 420 | + $offset++; | |
| 421 | + $length--; | |
| 422 | + } | |
| 423 | + | |
| 424 | + return $result; | |
| 425 | + } | |
| 426 | + | |
| 427 | + /** | |
| 428 | + * @return ($string is non-empty-string ? non-empty-string : string) | |
| 429 | + */ | |
| 430 | + private function strtolower(string $string): string | |
| 431 | + { | |
| 432 | + return $this->parserSettings->hasMultibyteSupport() | |
| 433 | + ? \mb_strtolower($string, $this->charset) | |
| 434 | + : \strtolower($string); | |
| 435 | + } | |
| 436 | + | |
| 437 | + /** | |
| 438 | + * @return list<string> | |
| 439 | + */ | |
| 440 | + private function strsplit(string $string): array | |
| 441 | + { | |
| 442 | + if ($this->parserSettings->hasMultibyteSupport()) { | |
| 443 | + if ($this->streql($this->charset, 'utf-8')) { | |
| 444 | + $result = CSSList::preg_split('//u', $string, -1, PREG_SPLIT_NO_EMPTY); | |
| 445 | + } else { | |
| 446 | + $length = \mb_strlen($string, $this->charset); | |
| 447 | + $result = []; | |
| 448 | + for ($i = 0; $i < $length; ++$i) { | |
| 449 | + $result[] = \mb_substr($string, $i, 1, $this->charset); | |
| 450 | + } | |
| 451 | + } | |
| 452 | + } else { | |
| 453 | + $result = ($string !== '') ? \str_split($string) : []; | |
| 454 | + } | |
| 455 | + | |
| 456 | + return $result; | |
| 457 | + } | |
| 458 | +} | |