vendor/twig/twig/src/Lexer.php line 449

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_STRING = '/"([^#"\\\\]*(?:\\\\.[^#"\\\\]*)*)"|\'([^\'\\\\]*(?:\\\\.[^\'\\\\]*)*)\'/As';
  49. public const REGEX_NUMBER = '/(?(DEFINE)
  50. (?<LNUM>[0-9]+(_[0-9]+)*) # Integers (with underscores) 123_456
  51. (?<FRAC>\.(?&LNUM)) # Fractional part .456
  52. (?<EXPONENT>[eE][+-]?(?&LNUM)) # Exponent part E+10
  53. (?<DNUM>(?&LNUM)(?:(?&FRAC))?) # Decimal number 123_456.456
  54. )(?:(?&DNUM)(?:(?&EXPONENT))?) # 123_456.456E+10
  55. /Ax';
  57. public const REGEX_DQ_STRING_DELIM = '/"/A';
  58. public const REGEX_DQ_STRING_PART = '/[^#"\\\\]*(?:(?:\\\\.|#(?!\{))[^#"\\\\]*)*/As';
  59. public const REGEX_INLINE_COMMENT = '/#[^\n]*/A';
  60. public const PUNCTUATION = '()[]{}?:.,|';
  62. private const SPECIAL_CHARS = [
  63. 'f' => "\f",
  64. 'n' => "\n",
  65. 'r' => "\r",
  66. 't' => "\t",
  67. 'v' => "\v",
  68. ];
  70. public function __construct(Environment $env, array $options = [])
  71. {
  72. $this->env = $env;
  74. $this->options = array_merge([
  75. 'tag_comment' => ['{#', '#}'],
  76. 'tag_block' => ['{%', '%}'],
  77. 'tag_variable' => ['{{', '}}'],
  78. 'whitespace_trim' => '-',
  79. 'whitespace_line_trim' => '~',
  80. 'whitespace_line_chars' => ' \t\0\x0B',
  81. 'interpolation' => ['#{', '}'],
  82. ], $options);
  83. }
  85. private function initialize(): void
  86. {
  87. if ($this->isInitialized) {
  88. return;
  89. }
  91. // 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
  92. $this->regexes = [
  93. // }}
  94. 'lex_var' => '{
  95. \s*
  96. (?:'.
  97. preg_quote($this->options['whitespace_trim'].$this->options['tag_variable'][1], '#').'\s*'. // -}}\s*
  98. '|'.
  99. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_variable'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~}}[ \t\0\x0B]*
  100. '|'.
  101. preg_quote($this->options['tag_variable'][1], '#'). // }}
  102. ')
  103. }Ax',
  105. // %}
  106. 'lex_block' => '{
  107. \s*
  108. (?:'.
  109. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*\n?'. // -%}\s*\n?
  110. '|'.
  111. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  112. '|'.
  113. preg_quote($this->options['tag_block'][1], '#').'\n?'. // %}\n?
  114. ')
  115. }Ax',
  117. // {% endverbatim %}
  118. 'lex_raw_data' => '{'.
  119. preg_quote($this->options['tag_block'][0], '#'). // {%
  120. '('.
  121. $this->options['whitespace_trim']. // -
  122. '|'.
  123. $this->options['whitespace_line_trim']. // ~
  124. ')?\s*endverbatim\s*'.
  125. '(?:'.
  126. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*'. // -%}
  127. '|'.
  128. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  129. '|'.
  130. preg_quote($this->options['tag_block'][1], '#'). // %}
  131. ')
  132. }sx',
  134. 'operator' => $this->getOperatorRegex(),
  136. // #}
  137. 'lex_comment' => '{
  138. (?:'.
  139. preg_quote($this->options['whitespace_trim'].$this->options['tag_comment'][1], '#').'\s*\n?'. // -#}\s*\n?
  140. '|'.
  141. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_comment'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~#}[ \t\0\x0B]*
  142. '|'.
  143. preg_quote($this->options['tag_comment'][1], '#').'\n?'. // #}\n?
  144. ')
  145. }sx',
  147. // verbatim %}
  148. 'lex_block_raw' => '{
  149. \s*verbatim\s*
  150. (?:'.
  151. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*'. // -%}\s*
  152. '|'.
  153. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  154. '|'.
  155. preg_quote($this->options['tag_block'][1], '#'). // %}
  156. ')
  157. }Asx',
  159. 'lex_block_line' => '{\s*line\s+(\d+)\s*'.preg_quote($this->options['tag_block'][1], '#').'}As',
  161. // {{ or {% or {#
  162. 'lex_tokens_start' => '{
  163. ('.
  164. preg_quote($this->options['tag_variable'][0], '#'). // {{
  165. '|'.
  166. preg_quote($this->options['tag_block'][0], '#'). // {%
  167. '|'.
  168. preg_quote($this->options['tag_comment'][0], '#'). // {#
  169. ')('.
  170. preg_quote($this->options['whitespace_trim'], '#'). // -
  171. '|'.
  172. preg_quote($this->options['whitespace_line_trim'], '#'). // ~
  173. ')?
  174. }sx',
  175. 'interpolation_start' => '{'.preg_quote($this->options['interpolation'][0], '#').'\s*}A',
  176. 'interpolation_end' => '{\s*'.preg_quote($this->options['interpolation'][1], '#').'}A',
  177. ];
  179. $this->isInitialized = true;
  180. }
  182. public function tokenize(Source $source): TokenStream
  183. {
  184. $this->initialize();
  186. $this->source = $source;
  187. $this->code = str_replace(["\r\n", "\r"], "\n", $source->getCode());
  188. $this->cursor = 0;
  189. $this->lineno = 1;
  190. $this->end = \strlen($this->code);
  191. $this->tokens = [];
  192. $this->state = self::STATE_DATA;
  193. $this->states = [];
  194. $this->brackets = [];
  195. $this->position = -1;
  197. // find all token starts in one go
  198. preg_match_all($this->regexes['lex_tokens_start'], $this->code, $matches, \PREG_OFFSET_CAPTURE);
  199. $this->positions = $matches;
  201. while ($this->cursor < $this->end) {
  202. // dispatch to the lexing functions depending
  203. // on the current state
  204. switch ($this->state) {
  205. case self::STATE_DATA:
  206. $this->lexData();
  207. break;
  209. case self::STATE_BLOCK:
  210. $this->lexBlock();
  211. break;
  213. case self::STATE_VAR:
  214. $this->lexVar();
  215. break;
  217. case self::STATE_STRING:
  218. $this->lexString();
  219. break;
  221. case self::STATE_INTERPOLATION:
  222. $this->lexInterpolation();
  223. break;
  224. }
  225. }
  227. $this->pushToken(Token::EOF_TYPE);
  229. if ($this->brackets) {
  230. [$expect, $lineno] = array_pop($this->brackets);
  231. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  232. }
  234. return new TokenStream($this->tokens, $this->source);
  235. }
  237. private function lexData(): void
  238. {
  239. // if no matches are left we return the rest of the template as simple text token
  240. if ($this->position == \count($this->positions[0]) - 1) {
  241. $this->pushToken(Token::TEXT_TYPE, substr($this->code, $this->cursor));
  242. $this->cursor = $this->end;
  244. return;
  245. }
  247. // Find the first token after the current cursor
  248. $position = $this->positions[0][++$this->position];
  249. while ($position[1] < $this->cursor) {
  250. if ($this->position == \count($this->positions[0]) - 1) {
  251. return;
  252. }
  253. $position = $this->positions[0][++$this->position];
  254. }
  256. // push the template text first
  257. $text = $textContent = substr($this->code, $this->cursor, $position[1] - $this->cursor);
  259. // trim?
  260. if (isset($this->positions[2][$this->position][0])) {
  261. if ($this->options['whitespace_trim'] === $this->positions[2][$this->position][0]) {
  262. // whitespace_trim detected ({%-, {{- or {#-)
  263. $text = rtrim($text);
  264. } elseif ($this->options['whitespace_line_trim'] === $this->positions[2][$this->position][0]) {
  265. // whitespace_line_trim detected ({%~, {{~ or {#~)
  266. // don't trim \r and \n
  267. $text = rtrim($text, " \t\0\x0B");
  268. }
  269. }
  270. $this->pushToken(Token::TEXT_TYPE, $text);
  271. $this->moveCursor($textContent.$position[0]);
  273. switch ($this->positions[1][$this->position][0]) {
  274. case $this->options['tag_comment'][0]:
  275. $this->lexComment();
  276. break;
  278. case $this->options['tag_block'][0]:
  279. // raw data?
  280. if (preg_match($this->regexes['lex_block_raw'], $this->code, $match, 0, $this->cursor)) {
  281. $this->moveCursor($match[0]);
  282. $this->lexRawData();
  283. // {% line \d+ %}
  284. } elseif (preg_match($this->regexes['lex_block_line'], $this->code, $match, 0, $this->cursor)) {
  285. $this->moveCursor($match[0]);
  286. $this->lineno = (int) $match[1];
  287. } else {
  288. $this->pushToken(Token::BLOCK_START_TYPE);
  289. $this->pushState(self::STATE_BLOCK);
  290. $this->currentVarBlockLine = $this->lineno;
  291. }
  292. break;
  294. case $this->options['tag_variable'][0]:
  295. $this->pushToken(Token::VAR_START_TYPE);
  296. $this->pushState(self::STATE_VAR);
  297. $this->currentVarBlockLine = $this->lineno;
  298. break;
  299. }
  300. }
  302. private function lexBlock(): void
  303. {
  304. if (!$this->brackets && preg_match($this->regexes['lex_block'], $this->code, $match, 0, $this->cursor)) {
  305. $this->pushToken(Token::BLOCK_END_TYPE);
  306. $this->moveCursor($match[0]);
  307. $this->popState();
  308. } else {
  309. $this->lexExpression();
  310. }
  311. }
  313. private function lexVar(): void
  314. {
  315. if (!$this->brackets && preg_match($this->regexes['lex_var'], $this->code, $match, 0, $this->cursor)) {
  316. $this->pushToken(Token::VAR_END_TYPE);
  317. $this->moveCursor($match[0]);
  318. $this->popState();
  319. } else {
  320. $this->lexExpression();
  321. }
  322. }
  324. private function lexExpression(): void
  325. {
  326. // whitespace
  327. if (preg_match('/\s+/A', $this->code, $match, 0, $this->cursor)) {
  328. $this->moveCursor($match[0]);
  330. if ($this->cursor >= $this->end) {
  331. throw new SyntaxError(\sprintf('Unclosed "%s".', self::STATE_BLOCK === $this->state ? 'block' : 'variable'), $this->currentVarBlockLine, $this->source);
  332. }
  333. }
  335. // spread operator
  336. if ('.' === $this->code[$this->cursor] && ($this->cursor + 2 < $this->end) && '.' === $this->code[$this->cursor + 1] && '.' === $this->code[$this->cursor + 2]) {
  337. $this->pushToken(Token::SPREAD_TYPE, '...');
  338. $this->moveCursor('...');
  339. }
  340. // arrow function
  341. elseif ('=' === $this->code[$this->cursor] && ($this->cursor + 1 < $this->end) && '>' === $this->code[$this->cursor + 1]) {
  342. $this->pushToken(Token::ARROW_TYPE, '=>');
  343. $this->moveCursor('=>');
  344. }
  345. // operators
  346. elseif (preg_match($this->regexes['operator'], $this->code, $match, 0, $this->cursor)) {
  347. $this->pushToken(Token::OPERATOR_TYPE, preg_replace('/\s+/', ' ', $match[0]));
  348. $this->moveCursor($match[0]);
  349. }
  350. // names
  351. elseif (preg_match(self::REGEX_NAME, $this->code, $match, 0, $this->cursor)) {
  352. $this->pushToken(Token::NAME_TYPE, $match[0]);
  353. $this->moveCursor($match[0]);
  354. }
  355. // numbers
  356. elseif (preg_match(self::REGEX_NUMBER, $this->code, $match, 0, $this->cursor)) {
  357. $this->pushToken(Token::NUMBER_TYPE, 0 + str_replace('_', '', $match[0]));
  358. $this->moveCursor($match[0]);
  359. }
  360. // punctuation
  361. elseif (str_contains(self::PUNCTUATION, $this->code[$this->cursor])) {
  362. // opening bracket
  363. if (str_contains('([{', $this->code[$this->cursor])) {
  364. $this->brackets[] = [$this->code[$this->cursor], $this->lineno];
  365. }
  366. // closing bracket
  367. elseif (str_contains(')]}', $this->code[$this->cursor])) {
  368. if (!$this->brackets) {
  369. throw new SyntaxError(\sprintf('Unexpected "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  370. }
  372. [$expect, $lineno] = array_pop($this->brackets);
  373. if ($this->code[$this->cursor] != strtr($expect, '([{', ')]}')) {
  374. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  375. }
  376. }
  378. $this->pushToken(Token::PUNCTUATION_TYPE, $this->code[$this->cursor]);
  379. ++$this->cursor;
  380. }
  381. // strings
  382. elseif (preg_match(self::REGEX_STRING, $this->code, $match, 0, $this->cursor)) {
  383. $this->pushToken(Token::STRING_TYPE, $this->stripcslashes(substr($match[0], 1, -1), substr($match[0], 0, 1)));
  384. $this->moveCursor($match[0]);
  385. }
  386. // opening double quoted string
  387. elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, 0, $this->cursor)) {
  388. $this->brackets[] = ['"', $this->lineno];
  389. $this->pushState(self::STATE_STRING);
  390. $this->moveCursor($match[0]);
  391. }
  392. // inline comment
  393. elseif (preg_match(self::REGEX_INLINE_COMMENT, $this->code, $match, 0, $this->cursor)) {
  394. $this->moveCursor($match[0]);
  395. }
  396. // unlexable
  397. else {
  398. throw new SyntaxError(\sprintf('Unexpected character "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  399. }
  400. }
  402. private function stripcslashes(string $str, string $quoteType): string
  403. {
  404. $result = '';
  405. $length = \strlen($str);
  407. $i = 0;
  408. while ($i < $length) {
  409. if (false === $pos = strpos($str, '\\', $i)) {
  410. $result .= substr($str, $i);
  411. break;
  412. }
  414. $result .= substr($str, $i, $pos - $i);
  415. $i = $pos + 1;
  417. if ($i >= $length) {
  418. $result .= '\\';
  419. break;
  420. }
  422. $nextChar = $str[$i];
  424. if (isset(self::SPECIAL_CHARS[$nextChar])) {
  425. $result .= self::SPECIAL_CHARS[$nextChar];
  426. } elseif ('\\' === $nextChar) {
  427. $result .= $nextChar;
  428. } elseif ("'" === $nextChar || '"' === $nextChar) {
  429. if ($nextChar !== $quoteType) {
  430. 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);
  431. }
  432. $result .= $nextChar;
  433. } elseif ('#' === $nextChar && $i + 1 < $length && '{' === $str[$i + 1]) {
  434. $result .= '#{';
  435. ++$i;
  436. } elseif ('x' === $nextChar && $i + 1 < $length && ctype_xdigit($str[$i + 1])) {
  437. $hex = $str[++$i];
  438. if ($i + 1 < $length && ctype_xdigit($str[$i + 1])) {
  439. $hex .= $str[++$i];
  440. }
  441. $result .= \chr(hexdec($hex));
  442. } elseif (ctype_digit($nextChar) && $nextChar < '8') {
  443. $octal = $nextChar;
  444. while ($i + 1 < $length && ctype_digit($str[$i + 1]) && $str[$i + 1] < '8' && \strlen($octal) < 3) {
  445. $octal .= $str[++$i];
  446. }
  447. $result .= \chr(octdec($octal));
  448. } else {
  449. 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);
  450. $result .= $nextChar;
  451. }
  453. ++$i;
  454. }
  456. return $result;
  457. }
  459. private function lexRawData(): void
  460. {
  461. if (!preg_match($this->regexes['lex_raw_data'], $this->code, $match, \PREG_OFFSET_CAPTURE, $this->cursor)) {
  462. throw new SyntaxError('Unexpected end of file: Unclosed "verbatim" block.', $this->lineno, $this->source);
  463. }
  465. $text = substr($this->code, $this->cursor, $match[0][1] - $this->cursor);
  466. $this->moveCursor($text.$match[0][0]);
  468. // trim?
  469. if (isset($match[1][0])) {
  470. if ($this->options['whitespace_trim'] === $match[1][0]) {
  471. // whitespace_trim detected ({%-, {{- or {#-)
  472. $text = rtrim($text);
  473. } else {
  474. // whitespace_line_trim detected ({%~, {{~ or {#~)
  475. // don't trim \r and \n
  476. $text = rtrim($text, " \t\0\x0B");
  477. }
  478. }
  480. $this->pushToken(Token::TEXT_TYPE, $text);
  481. }
  483. private function lexComment(): void
  484. {
  485. if (!preg_match($this->regexes['lex_comment'], $this->code, $match, \PREG_OFFSET_CAPTURE, $this->cursor)) {
  486. throw new SyntaxError('Unclosed comment.', $this->lineno, $this->source);
  487. }
  489. $this->moveCursor(substr($this->code, $this->cursor, $match[0][1] - $this->cursor).$match[0][0]);
  490. }
  492. private function lexString(): void
  493. {
  494. if (preg_match($this->regexes['interpolation_start'], $this->code, $match, 0, $this->cursor)) {
  495. $this->brackets[] = [$this->options['interpolation'][0], $this->lineno];
  496. $this->pushToken(Token::INTERPOLATION_START_TYPE);
  497. $this->moveCursor($match[0]);
  498. $this->pushState(self::STATE_INTERPOLATION);
  499. } elseif (preg_match(self::REGEX_DQ_STRING_PART, $this->code, $match, 0, $this->cursor) && '' !== $match[0]) {
  500. $this->pushToken(Token::STRING_TYPE, $this->stripcslashes($match[0], '"'));
  501. $this->moveCursor($match[0]);
  502. } elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, 0, $this->cursor)) {
  503. [$expect, $lineno] = array_pop($this->brackets);
  504. if ('"' != $this->code[$this->cursor]) {
  505. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  506. }
  508. $this->popState();
  509. ++$this->cursor;
  510. } else {
  511. // unlexable
  512. throw new SyntaxError(\sprintf('Unexpected character "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  513. }
  514. }
  516. private function lexInterpolation(): void
  517. {
  518. $bracket = end($this->brackets);
  519. if ($this->options['interpolation'][0] === $bracket[0] && preg_match($this->regexes['interpolation_end'], $this->code, $match, 0, $this->cursor)) {
  520. array_pop($this->brackets);
  521. $this->pushToken(Token::INTERPOLATION_END_TYPE);
  522. $this->moveCursor($match[0]);
  523. $this->popState();
  524. } else {
  525. $this->lexExpression();
  526. }
  527. }
  529. private function pushToken($type, $value = ''): void
  530. {
  531. // do not push empty text tokens
  532. if (Token::TEXT_TYPE === $type && '' === $value) {
  533. return;
  534. }
  536. $this->tokens[] = new Token($type, $value, $this->lineno);
  537. }
  539. private function moveCursor($text): void
  540. {
  541. $this->cursor += \strlen($text);
  542. $this->lineno += substr_count($text, "\n");
  543. }
  545. private function getOperatorRegex(): string
  546. {
  547. $operators = array_merge(
  548. ['='],
  549. array_keys($this->env->getUnaryOperators()),
  550. array_keys($this->env->getBinaryOperators())
  551. );
  553. $operators = array_combine($operators, array_map('strlen', $operators));
  554. arsort($operators);
  556. $regex = [];
  557. foreach ($operators as $operator => $length) {
  558. // an operator that ends with a character must be followed by
  559. // a whitespace, a parenthesis, an opening map [ or sequence {
  560. $r = preg_quote($operator, '/');
  561. if (ctype_alpha($operator[$length - 1])) {
  562. $r .= '(?=[\s()\[{])';
  563. }
  565. // an operator that begins with a character must not have a dot or pipe before
  566. if (ctype_alpha($operator[0])) {
  567. $r = '(?<![\.\|])'.$r;
  568. }
  570. // an operator with a space can be any amount of whitespaces
  571. $r = preg_replace('/\s+/', '\s+', $r);
  573. $regex[] = $r;
  574. }
  576. return '/'.implode('|', $regex).'/A';
  577. }
  579. private function pushState($state): void
  580. {
  581. $this->states[] = $this->state;
  582. $this->state = $state;
  583. }
  585. private function popState(): void
  586. {
  587. if (0 === \count($this->states)) {
  588. throw new \LogicException('Cannot pop state without a previous state.');
  589. }
  591. $this->state = array_pop($this->states);
  592. }
  593. }