vendor/twig/twig/src/Lexer.php line 176

Open in your IDE?
  1. <?php
  3. /*
  4. * This file is part of Twig.
  5. *
  6. * (c) Fabien Potencier
  7. * (c) Armin Ronacher
  8. *
  9. * For the full copyright and license information, please view the LICENSE
  10. * file that was distributed with this source code.
  11. */
  13. namespace Twig;
  15. use Twig\Error\SyntaxError;
  17. /**
  18. * @author Fabien Potencier <fabien@symfony.com>
  19. */
  20. class Lexer
  21. {
  22. private $isInitialized = false;
  24. private $tokens;
  25. private $code;
  26. private $cursor;
  27. private $lineno;
  28. private $end;
  29. private $state;
  30. private $states;
  31. private $brackets;
  32. private $env;
  33. private $source;
  34. private $options;
  35. private $regexes;
  36. private $position;
  37. private $positions;
  38. private $currentVarBlockLine;
  40. public const STATE_DATA = 0;
  41. public const STATE_BLOCK = 1;
  42. public const STATE_VAR = 2;
  43. public const STATE_STRING = 3;
  44. public const STATE_INTERPOLATION = 4;
  46. public const REGEX_NAME = '/[a-zA-Z_\x7f-\xff][a-zA-Z0-9_\x7f-\xff]*/A';
  47. public const REGEX_NUMBER = '/[0-9]+(?:\.[0-9]+)?([Ee][\+\-][0-9]+)?/A';
  48. public const REGEX_STRING = '/"([^#"\\\\]*(?:\\\\.[^#"\\\\]*)*)"|\'([^\'\\\\]*(?:\\\\.[^\'\\\\]*)*)\'/As';
  49. public const REGEX_DQ_STRING_DELIM = '/"/A';
  50. public const REGEX_DQ_STRING_PART = '/[^#"\\\\]*(?:(?:\\\\.|#(?!\{))[^#"\\\\]*)*/As';
  51. public const REGEX_INLINE_COMMENT = '/#[^\n]*/A';
  52. public const PUNCTUATION = '()[]{}?:.,|';
  54. private const SPECIAL_CHARS = [
  55. 'f' => "\f",
  56. 'n' => "\n",
  57. 'r' => "\r",
  58. 't' => "\t",
  59. 'v' => "\v",
  60. ];
  62. public function __construct(Environment $env, array $options = [])
  63. {
  64. $this->env = $env;
  66. $this->options = array_merge([
  67. 'tag_comment' => ['{#', '#}'],
  68. 'tag_block' => ['{%', '%}'],
  69. 'tag_variable' => ['{{', '}}'],
  70. 'whitespace_trim' => '-',
  71. 'whitespace_line_trim' => '~',
  72. 'whitespace_line_chars' => ' \t\0\x0B',
  73. 'interpolation' => ['#{', '}'],
  74. ], $options);
  75. }
  77. private function initialize()
  78. {
  79. if ($this->isInitialized) {
  80. return;
  81. }
  83. // when PHP 7.3 is the min version, we will be able to remove the '#' part in preg_quote as it's part of the default
  84. $this->regexes = [
  85. // }}
  86. 'lex_var' => '{
  87. \s*
  88. (?:'.
  89. preg_quote($this->options['whitespace_trim'].$this->options['tag_variable'][1], '#').'\s*'. // -}}\s*
  90. '|'.
  91. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_variable'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~}}[ \t\0\x0B]*
  92. '|'.
  93. preg_quote($this->options['tag_variable'][1], '#'). // }}
  94. ')
  95. }Ax',
  97. // %}
  98. 'lex_block' => '{
  99. \s*
  100. (?:'.
  101. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*\n?'. // -%}\s*\n?
  102. '|'.
  103. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  104. '|'.
  105. preg_quote($this->options['tag_block'][1], '#').'\n?'. // %}\n?
  106. ')
  107. }Ax',
  109. // {% endverbatim %}
  110. 'lex_raw_data' => '{'.
  111. preg_quote($this->options['tag_block'][0], '#'). // {%
  112. '('.
  113. $this->options['whitespace_trim']. // -
  114. '|'.
  115. $this->options['whitespace_line_trim']. // ~
  116. ')?\s*endverbatim\s*'.
  117. '(?:'.
  118. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*'. // -%}
  119. '|'.
  120. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  121. '|'.
  122. preg_quote($this->options['tag_block'][1], '#'). // %}
  123. ')
  124. }sx',
  126. 'operator' => $this->getOperatorRegex(),
  128. // #}
  129. 'lex_comment' => '{
  130. (?:'.
  131. preg_quote($this->options['whitespace_trim'].$this->options['tag_comment'][1], '#').'\s*\n?'. // -#}\s*\n?
  132. '|'.
  133. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_comment'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~#}[ \t\0\x0B]*
  134. '|'.
  135. preg_quote($this->options['tag_comment'][1], '#').'\n?'. // #}\n?
  136. ')
  137. }sx',
  139. // verbatim %}
  140. 'lex_block_raw' => '{
  141. \s*verbatim\s*
  142. (?:'.
  143. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*'. // -%}\s*
  144. '|'.
  145. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  146. '|'.
  147. preg_quote($this->options['tag_block'][1], '#'). // %}
  148. ')
  149. }Asx',
  151. 'lex_block_line' => '{\s*line\s+(\d+)\s*'.preg_quote($this->options['tag_block'][1], '#').'}As',
  153. // {{ or {% or {#
  154. 'lex_tokens_start' => '{
  155. ('.
  156. preg_quote($this->options['tag_variable'][0], '#'). // {{
  157. '|'.
  158. preg_quote($this->options['tag_block'][0], '#'). // {%
  159. '|'.
  160. preg_quote($this->options['tag_comment'][0], '#'). // {#
  161. ')('.
  162. preg_quote($this->options['whitespace_trim'], '#'). // -
  163. '|'.
  164. preg_quote($this->options['whitespace_line_trim'], '#'). // ~
  165. ')?
  166. }sx',
  167. 'interpolation_start' => '{'.preg_quote($this->options['interpolation'][0], '#').'\s*}A',
  168. 'interpolation_end' => '{\s*'.preg_quote($this->options['interpolation'][1], '#').'}A',
  169. ];
  171. $this->isInitialized = true;
  172. }
  174. public function tokenize(Source $source): TokenStream
  175. {
  176. $this->initialize();
  178. $this->source = $source;
  179. $this->code = str_replace(["\r\n", "\r"], "\n", $source->getCode());
  180. $this->cursor = 0;
  181. $this->lineno = 1;
  182. $this->end = \strlen($this->code);
  183. $this->tokens = [];
  184. $this->state = self::STATE_DATA;
  185. $this->states = [];
  186. $this->brackets = [];
  187. $this->position = -1;
  189. // find all token starts in one go
  190. preg_match_all($this->regexes['lex_tokens_start'], $this->code, $matches, \PREG_OFFSET_CAPTURE);
  191. $this->positions = $matches;
  193. while ($this->cursor < $this->end) {
  194. // dispatch to the lexing functions depending
  195. // on the current state
  196. switch ($this->state) {
  197. case self::STATE_DATA:
  198. $this->lexData();
  199. break;
  201. case self::STATE_BLOCK:
  202. $this->lexBlock();
  203. break;
  205. case self::STATE_VAR:
  206. $this->lexVar();
  207. break;
  209. case self::STATE_STRING:
  210. $this->lexString();
  211. break;
  213. case self::STATE_INTERPOLATION:
  214. $this->lexInterpolation();
  215. break;
  216. }
  217. }
  219. $this->pushToken(Token::EOF_TYPE);
  221. if ($this->brackets) {
  222. [$expect, $lineno] = array_pop($this->brackets);
  223. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  224. }
  226. return new TokenStream($this->tokens, $this->source);
  227. }
  229. private function lexData(): void
  230. {
  231. // if no matches are left we return the rest of the template as simple text token
  232. if ($this->position == \count($this->positions[0]) - 1) {
  233. $this->pushToken(Token::TEXT_TYPE, substr($this->code, $this->cursor));
  234. $this->cursor = $this->end;
  236. return;
  237. }
  239. // Find the first token after the current cursor
  240. $position = $this->positions[0][++$this->position];
  241. while ($position[1] < $this->cursor) {
  242. if ($this->position == \count($this->positions[0]) - 1) {
  243. return;
  244. }
  245. $position = $this->positions[0][++$this->position];
  246. }
  248. // push the template text first
  249. $text = $textContent = substr($this->code, $this->cursor, $position[1] - $this->cursor);
  251. // trim?
  252. if (isset($this->positions[2][$this->position][0])) {
  253. if ($this->options['whitespace_trim'] === $this->positions[2][$this->position][0]) {
  254. // whitespace_trim detected ({%-, {{- or {#-)
  255. $text = rtrim($text);
  256. } elseif ($this->options['whitespace_line_trim'] === $this->positions[2][$this->position][0]) {
  257. // whitespace_line_trim detected ({%~, {{~ or {#~)
  258. // don't trim \r and \n
  259. $text = rtrim($text, " \t\0\x0B");
  260. }
  261. }
  262. $this->pushToken(Token::TEXT_TYPE, $text);
  263. $this->moveCursor($textContent.$position[0]);
  265. switch ($this->positions[1][$this->position][0]) {
  266. case $this->options['tag_comment'][0]:
  267. $this->lexComment();
  268. break;
  270. case $this->options['tag_block'][0]:
  271. // raw data?
  272. if (preg_match($this->regexes['lex_block_raw'], $this->code, $match, 0, $this->cursor)) {
  273. $this->moveCursor($match[0]);
  274. $this->lexRawData();
  275. // {% line \d+ %}
  276. } elseif (preg_match($this->regexes['lex_block_line'], $this->code, $match, 0, $this->cursor)) {
  277. $this->moveCursor($match[0]);
  278. $this->lineno = (int) $match[1];
  279. } else {
  280. $this->pushToken(Token::BLOCK_START_TYPE);
  281. $this->pushState(self::STATE_BLOCK);
  282. $this->currentVarBlockLine = $this->lineno;
  283. }
  284. break;
  286. case $this->options['tag_variable'][0]:
  287. $this->pushToken(Token::VAR_START_TYPE);
  288. $this->pushState(self::STATE_VAR);
  289. $this->currentVarBlockLine = $this->lineno;
  290. break;
  291. }
  292. }
  294. private function lexBlock(): void
  295. {
  296. if (!$this->brackets && preg_match($this->regexes['lex_block'], $this->code, $match, 0, $this->cursor)) {
  297. $this->pushToken(Token::BLOCK_END_TYPE);
  298. $this->moveCursor($match[0]);
  299. $this->popState();
  300. } else {
  301. $this->lexExpression();
  302. }
  303. }
  305. private function lexVar(): void
  306. {
  307. if (!$this->brackets && preg_match($this->regexes['lex_var'], $this->code, $match, 0, $this->cursor)) {
  308. $this->pushToken(Token::VAR_END_TYPE);
  309. $this->moveCursor($match[0]);
  310. $this->popState();
  311. } else {
  312. $this->lexExpression();
  313. }
  314. }
  316. private function lexExpression(): void
  317. {
  318. // whitespace
  319. if (preg_match('/\s+/A', $this->code, $match, 0, $this->cursor)) {
  320. $this->moveCursor($match[0]);
  322. if ($this->cursor >= $this->end) {
  323. throw new SyntaxError(\sprintf('Unclosed "%s".', self::STATE_BLOCK === $this->state ? 'block' : 'variable'), $this->currentVarBlockLine, $this->source);
  324. }
  325. }
  327. // spread operator
  328. if ('.' === $this->code[$this->cursor] && ($this->cursor + 2 < $this->end) && '.' === $this->code[$this->cursor + 1] && '.' === $this->code[$this->cursor + 2]) {
  329. $this->pushToken(Token::SPREAD_TYPE, '...');
  330. $this->moveCursor('...');
  331. }
  332. // arrow function
  333. elseif ('=' === $this->code[$this->cursor] && ($this->cursor + 1 < $this->end) && '>' === $this->code[$this->cursor + 1]) {
  334. $this->pushToken(Token::ARROW_TYPE, '=>');
  335. $this->moveCursor('=>');
  336. }
  337. // operators
  338. elseif (preg_match($this->regexes['operator'], $this->code, $match, 0, $this->cursor)) {
  339. $this->pushToken(Token::OPERATOR_TYPE, preg_replace('/\s+/', ' ', $match[0]));
  340. $this->moveCursor($match[0]);
  341. }
  342. // names
  343. elseif (preg_match(self::REGEX_NAME, $this->code, $match, 0, $this->cursor)) {
  344. $this->pushToken(Token::NAME_TYPE, $match[0]);
  345. $this->moveCursor($match[0]);
  346. }
  347. // numbers
  348. elseif (preg_match(self::REGEX_NUMBER, $this->code, $match, 0, $this->cursor)) {
  349. $number = (float) $match[0]; // floats
  350. if (ctype_digit($match[0]) && $number <= \PHP_INT_MAX) {
  351. $number = (int) $match[0]; // integers lower than the maximum
  352. }
  353. $this->pushToken(Token::NUMBER_TYPE, $number);
  354. $this->moveCursor($match[0]);
  355. }
  356. // punctuation
  357. elseif (str_contains(self::PUNCTUATION, $this->code[$this->cursor])) {
  358. // opening bracket
  359. if (str_contains('([{', $this->code[$this->cursor])) {
  360. $this->brackets[] = [$this->code[$this->cursor], $this->lineno];
  361. }
  362. // closing bracket
  363. elseif (str_contains(')]}', $this->code[$this->cursor])) {
  364. if (!$this->brackets) {
  365. throw new SyntaxError(\sprintf('Unexpected "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  366. }
  368. [$expect, $lineno] = array_pop($this->brackets);
  369. if ($this->code[$this->cursor] != strtr($expect, '([{', ')]}')) {
  370. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  371. }
  372. }
  374. $this->pushToken(Token::PUNCTUATION_TYPE, $this->code[$this->cursor]);
  375. ++$this->cursor;
  376. }
  377. // strings
  378. elseif (preg_match(self::REGEX_STRING, $this->code, $match, 0, $this->cursor)) {
  379. $this->pushToken(Token::STRING_TYPE, $this->stripcslashes(substr($match[0], 1, -1), substr($match[0], 0, 1)));
  380. $this->moveCursor($match[0]);
  381. }
  382. // opening double quoted string
  383. elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, 0, $this->cursor)) {
  384. $this->brackets[] = ['"', $this->lineno];
  385. $this->pushState(self::STATE_STRING);
  386. $this->moveCursor($match[0]);
  387. }
  388. // inline comment
  389. elseif (preg_match(self::REGEX_INLINE_COMMENT, $this->code, $match, 0, $this->cursor)) {
  390. $this->moveCursor($match[0]);
  391. }
  392. // unlexable
  393. else {
  394. throw new SyntaxError(\sprintf('Unexpected character "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  395. }
  396. }
  398. private function stripcslashes(string $str, string $quoteType): string
  399. {
  400. $result = '';
  401. $length = \strlen($str);
  403. $i = 0;
  404. while ($i < $length) {
  405. if (false === $pos = strpos($str, '\\', $i)) {
  406. $result .= substr($str, $i);
  407. break;
  408. }
  410. $result .= substr($str, $i, $pos - $i);
  411. $i = $pos + 1;
  413. if ($i >= $length) {
  414. $result .= '\\';
  415. break;
  416. }
  418. $nextChar = $str[$i];
  420. if (isset(self::SPECIAL_CHARS[$nextChar])) {
  421. $result .= self::SPECIAL_CHARS[$nextChar];
  422. } elseif ('\\' === $nextChar) {
  423. $result .= $nextChar;
  424. } elseif ("'" === $nextChar || '"' === $nextChar) {
  425. if ($nextChar !== $quoteType) {
  426. trigger_deprecation('twig/twig', '3.12', 'Character "%s" should not be escaped; the "\" character is ignored in Twig 3 but will not be in Twig 4. Please remove the extra "\" character at position %d in "%s" at line %d.', $nextChar, $i + 1, $this->source->getName(), $this->lineno);
  427. }
  428. $result .= $nextChar;
  429. } elseif ('#' === $nextChar && $i + 1 < $length && '{' === $str[$i + 1]) {
  430. $result .= '#{';
  431. ++$i;
  432. } elseif ('x' === $nextChar && $i + 1 < $length && ctype_xdigit($str[$i + 1])) {
  433. $hex = $str[++$i];
  434. if ($i + 1 < $length && ctype_xdigit($str[$i + 1])) {
  435. $hex .= $str[++$i];
  436. }
  437. $result .= \chr(hexdec($hex));
  438. } elseif (ctype_digit($nextChar) && $nextChar < '8') {
  439. $octal = $nextChar;
  440. while ($i + 1 < $length && ctype_digit($str[$i + 1]) && $str[$i + 1] < '8' && \strlen($octal) < 3) {
  441. $octal .= $str[++$i];
  442. }
  443. $result .= \chr(octdec($octal));
  444. } else {
  445. trigger_deprecation('twig/twig', '3.12', 'Character "%s" should not be escaped; the "\" character is ignored in Twig 3 but will not be in Twig 4. Please remove the extra "\" character at position %d in "%s" at line %d.', $nextChar, $i + 1, $this->source->getName(), $this->lineno);
  446. $result .= $nextChar;
  447. }
  449. ++$i;
  450. }
  452. return $result;
  453. }
  455. private function lexRawData(): void
  456. {
  457. if (!preg_match($this->regexes['lex_raw_data'], $this->code, $match, \PREG_OFFSET_CAPTURE, $this->cursor)) {
  458. throw new SyntaxError('Unexpected end of file: Unclosed "verbatim" block.', $this->lineno, $this->source);
  459. }
  461. $text = substr($this->code, $this->cursor, $match[0][1] - $this->cursor);
  462. $this->moveCursor($text.$match[0][0]);
  464. // trim?
  465. if (isset($match[1][0])) {
  466. if ($this->options['whitespace_trim'] === $match[1][0]) {
  467. // whitespace_trim detected ({%-, {{- or {#-)
  468. $text = rtrim($text);
  469. } else {
  470. // whitespace_line_trim detected ({%~, {{~ or {#~)
  471. // don't trim \r and \n
  472. $text = rtrim($text, " \t\0\x0B");
  473. }
  474. }
  476. $this->pushToken(Token::TEXT_TYPE, $text);
  477. }
  479. private function lexComment(): void
  480. {
  481. if (!preg_match($this->regexes['lex_comment'], $this->code, $match, \PREG_OFFSET_CAPTURE, $this->cursor)) {
  482. throw new SyntaxError('Unclosed comment.', $this->lineno, $this->source);
  483. }
  485. $this->moveCursor(substr($this->code, $this->cursor, $match[0][1] - $this->cursor).$match[0][0]);
  486. }
  488. private function lexString(): void
  489. {
  490. if (preg_match($this->regexes['interpolation_start'], $this->code, $match, 0, $this->cursor)) {
  491. $this->brackets[] = [$this->options['interpolation'][0], $this->lineno];
  492. $this->pushToken(Token::INTERPOLATION_START_TYPE);
  493. $this->moveCursor($match[0]);
  494. $this->pushState(self::STATE_INTERPOLATION);
  495. } elseif (preg_match(self::REGEX_DQ_STRING_PART, $this->code, $match, 0, $this->cursor) && '' !== $match[0]) {
  496. $this->pushToken(Token::STRING_TYPE, $this->stripcslashes($match[0], '"'));
  497. $this->moveCursor($match[0]);
  498. } elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, 0, $this->cursor)) {
  499. [$expect, $lineno] = array_pop($this->brackets);
  500. if ('"' != $this->code[$this->cursor]) {
  501. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  502. }
  504. $this->popState();
  505. ++$this->cursor;
  506. } else {
  507. // unlexable
  508. throw new SyntaxError(\sprintf('Unexpected character "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  509. }
  510. }
  512. private function lexInterpolation(): void
  513. {
  514. $bracket = end($this->brackets);
  515. if ($this->options['interpolation'][0] === $bracket[0] && preg_match($this->regexes['interpolation_end'], $this->code, $match, 0, $this->cursor)) {
  516. array_pop($this->brackets);
  517. $this->pushToken(Token::INTERPOLATION_END_TYPE);
  518. $this->moveCursor($match[0]);
  519. $this->popState();
  520. } else {
  521. $this->lexExpression();
  522. }
  523. }
  525. private function pushToken($type, $value = ''): void
  526. {
  527. // do not push empty text tokens
  528. if (Token::TEXT_TYPE === $type && '' === $value) {
  529. return;
  530. }
  532. $this->tokens[] = new Token($type, $value, $this->lineno);
  533. }
  535. private function moveCursor($text): void
  536. {
  537. $this->cursor += \strlen($text);
  538. $this->lineno += substr_count($text, "\n");
  539. }
  541. private function getOperatorRegex(): string
  542. {
  543. $operators = array_merge(
  544. ['='],
  545. array_keys($this->env->getUnaryOperators()),
  546. array_keys($this->env->getBinaryOperators())
  547. );
  549. $operators = array_combine($operators, array_map('strlen', $operators));
  550. arsort($operators);
  552. $regex = [];
  553. foreach ($operators as $operator => $length) {
  554. // an operator that ends with a character must be followed by
  555. // a whitespace, a parenthesis, an opening map [ or sequence {
  556. $r = preg_quote($operator, '/');
  557. if (ctype_alpha($operator[$length - 1])) {
  558. $r .= '(?=[\s()\[{])';
  559. }
  561. // an operator that begins with a character must not have a dot or pipe before
  562. if (ctype_alpha($operator[0])) {
  563. $r = '(?<![\.\|])'.$r;
  564. }
  566. // an operator with a space can be any amount of whitespaces
  567. $r = preg_replace('/\s+/', '\s+', $r);
  569. $regex[] = $r;
  570. }
  572. return '/'.implode('|', $regex).'/A';
  573. }
  575. private function pushState($state): void
  576. {
  577. $this->states[] = $this->state;
  578. $this->state = $state;
  579. }
  581. private function popState(): void
  582. {
  583. if (0 === \count($this->states)) {
  584. throw new \LogicException('Cannot pop state without a previous state.');
  585. }
  587. $this->state = array_pop($this->states);
  588. }
  589. }