vendor/twig/twig/src/Lexer.php line 430

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;
  39. private array $openingBrackets = ['{', '(', '['];
  40. private array $closingBrackets = ['}', ')', ']'];
  42. public const STATE_DATA = 0;
  43. public const STATE_BLOCK = 1;
  44. public const STATE_VAR = 2;
  45. public const STATE_STRING = 3;
  46. public const STATE_INTERPOLATION = 4;
  48. public const REGEX_NAME = '/[a-zA-Z_\x7f-\xff][a-zA-Z0-9_\x7f-\xff]*/A';
  49. public const REGEX_STRING = '/"([^#"\\\\]*(?:\\\\.[^#"\\\\]*)*)"|\'([^\'\\\\]*(?:\\\\.[^\'\\\\]*)*)\'/As';
  51. public const REGEX_NUMBER = '/(?(DEFINE)
  52. (?<LNUM>[0-9]+(_[0-9]+)*) # Integers (with underscores) 123_456
  53. (?<FRAC>\.(?&LNUM)) # Fractional part .456
  54. (?<EXPONENT>[eE][+-]?(?&LNUM)) # Exponent part E+10
  55. (?<DNUM>(?&LNUM)(?:(?&FRAC))?) # Decimal number 123_456.456
  56. )(?:(?&DNUM)(?:(?&EXPONENT))?) # 123_456.456E+10
  57. /Ax';
  59. public const REGEX_DQ_STRING_DELIM = '/"/A';
  60. public const REGEX_DQ_STRING_PART = '/[^#"\\\\]*(?:(?:\\\\.|#(?!\{))[^#"\\\\]*)*/As';
  61. public const REGEX_INLINE_COMMENT = '/#[^\n]*/A';
  62. public const PUNCTUATION = '()[]{}?:.,|';
  64. private const SPECIAL_CHARS = [
  65. 'f' => "\f",
  66. 'n' => "\n",
  67. 'r' => "\r",
  68. 't' => "\t",
  69. 'v' => "\v",
  70. ];
  72. public function __construct(Environment $env, array $options = [])
  73. {
  74. $this->env = $env;
  76. $this->options = array_merge([
  77. 'tag_comment' => ['{#', '#}'],
  78. 'tag_block' => ['{%', '%}'],
  79. 'tag_variable' => ['{{', '}}'],
  80. 'whitespace_trim' => '-',
  81. 'whitespace_line_trim' => '~',
  82. 'whitespace_line_chars' => ' \t\0\x0B',
  83. 'interpolation' => ['#{', '}'],
  84. ], $options);
  85. }
  87. private function initialize(): void
  88. {
  89. if ($this->isInitialized) {
  90. return;
  91. }
  93. // 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
  94. $this->regexes = [
  95. // }}
  96. 'lex_var' => '{
  97. \s*
  98. (?:'.
  99. preg_quote($this->options['whitespace_trim'].$this->options['tag_variable'][1], '#').'\s*'. // -}}\s*
  100. '|'.
  101. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_variable'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~}}[ \t\0\x0B]*
  102. '|'.
  103. preg_quote($this->options['tag_variable'][1], '#'). // }}
  104. ')
  105. }Ax',
  107. // %}
  108. 'lex_block' => '{
  109. \s*
  110. (?:'.
  111. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*\n?'. // -%}\s*\n?
  112. '|'.
  113. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  114. '|'.
  115. preg_quote($this->options['tag_block'][1], '#').'\n?'. // %}\n?
  116. ')
  117. }Ax',
  119. // {% endverbatim %}
  120. 'lex_raw_data' => '{'.
  121. preg_quote($this->options['tag_block'][0], '#'). // {%
  122. '('.
  123. $this->options['whitespace_trim']. // -
  124. '|'.
  125. $this->options['whitespace_line_trim']. // ~
  126. ')?\s*endverbatim\s*'.
  127. '(?:'.
  128. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*'. // -%}
  129. '|'.
  130. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  131. '|'.
  132. preg_quote($this->options['tag_block'][1], '#'). // %}
  133. ')
  134. }sx',
  136. 'operator' => $this->getOperatorRegex(),
  138. // #}
  139. 'lex_comment' => '{
  140. (?:'.
  141. preg_quote($this->options['whitespace_trim'].$this->options['tag_comment'][1], '#').'\s*\n?'. // -#}\s*\n?
  142. '|'.
  143. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_comment'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~#}[ \t\0\x0B]*
  144. '|'.
  145. preg_quote($this->options['tag_comment'][1], '#').'\n?'. // #}\n?
  146. ')
  147. }sx',
  149. // verbatim %}
  150. 'lex_block_raw' => '{
  151. \s*verbatim\s*
  152. (?:'.
  153. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*'. // -%}\s*
  154. '|'.
  155. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  156. '|'.
  157. preg_quote($this->options['tag_block'][1], '#'). // %}
  158. ')
  159. }Asx',
  161. 'lex_block_line' => '{\s*line\s+(\d+)\s*'.preg_quote($this->options['tag_block'][1], '#').'}As',
  163. // {{ or {% or {#
  164. 'lex_tokens_start' => '{
  165. ('.
  166. preg_quote($this->options['tag_variable'][0], '#'). // {{
  167. '|'.
  168. preg_quote($this->options['tag_block'][0], '#'). // {%
  169. '|'.
  170. preg_quote($this->options['tag_comment'][0], '#'). // {#
  171. ')('.
  172. preg_quote($this->options['whitespace_trim'], '#'). // -
  173. '|'.
  174. preg_quote($this->options['whitespace_line_trim'], '#'). // ~
  175. ')?
  176. }sx',
  177. 'interpolation_start' => '{'.preg_quote($this->options['interpolation'][0], '#').'\s*}A',
  178. 'interpolation_end' => '{\s*'.preg_quote($this->options['interpolation'][1], '#').'}A',
  179. ];
  181. $this->isInitialized = true;
  182. }
  184. public function tokenize(Source $source): TokenStream
  185. {
  186. $this->initialize();
  188. $this->source = $source;
  189. $this->code = str_replace(["\r\n", "\r"], "\n", $source->getCode());
  190. $this->cursor = 0;
  191. $this->lineno = 1;
  192. $this->end = \strlen($this->code);
  193. $this->tokens = [];
  194. $this->state = self::STATE_DATA;
  195. $this->states = [];
  196. $this->brackets = [];
  197. $this->position = -1;
  199. // find all token starts in one go
  200. preg_match_all($this->regexes['lex_tokens_start'], $this->code, $matches, \PREG_OFFSET_CAPTURE);
  201. $this->positions = $matches;
  203. while ($this->cursor < $this->end) {
  204. // dispatch to the lexing functions depending
  205. // on the current state
  206. switch ($this->state) {
  207. case self::STATE_DATA:
  208. $this->lexData();
  209. break;
  211. case self::STATE_BLOCK:
  212. $this->lexBlock();
  213. break;
  215. case self::STATE_VAR:
  216. $this->lexVar();
  217. break;
  219. case self::STATE_STRING:
  220. $this->lexString();
  221. break;
  223. case self::STATE_INTERPOLATION:
  224. $this->lexInterpolation();
  225. break;
  226. }
  227. }
  229. $this->pushToken(Token::EOF_TYPE);
  231. if ($this->brackets) {
  232. [$expect, $lineno] = array_pop($this->brackets);
  233. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  234. }
  236. return new TokenStream($this->tokens, $this->source);
  237. }
  239. private function lexData(): void
  240. {
  241. // if no matches are left we return the rest of the template as simple text token
  242. if ($this->position == \count($this->positions[0]) - 1) {
  243. $this->pushToken(Token::TEXT_TYPE, substr($this->code, $this->cursor));
  244. $this->cursor = $this->end;
  246. return;
  247. }
  249. // Find the first token after the current cursor
  250. $position = $this->positions[0][++$this->position];
  251. while ($position[1] < $this->cursor) {
  252. if ($this->position == \count($this->positions[0]) - 1) {
  253. return;
  254. }
  255. $position = $this->positions[0][++$this->position];
  256. }
  258. // push the template text first
  259. $text = $textContent = substr($this->code, $this->cursor, $position[1] - $this->cursor);
  261. // trim?
  262. if (isset($this->positions[2][$this->position][0])) {
  263. if ($this->options['whitespace_trim'] === $this->positions[2][$this->position][0]) {
  264. // whitespace_trim detected ({%-, {{- or {#-)
  265. $text = rtrim($text);
  266. } elseif ($this->options['whitespace_line_trim'] === $this->positions[2][$this->position][0]) {
  267. // whitespace_line_trim detected ({%~, {{~ or {#~)
  268. // don't trim \r and \n
  269. $text = rtrim($text, " \t\0\x0B");
  270. }
  271. }
  272. $this->pushToken(Token::TEXT_TYPE, $text);
  273. $this->moveCursor($textContent.$position[0]);
  275. switch ($this->positions[1][$this->position][0]) {
  276. case $this->options['tag_comment'][0]:
  277. $this->lexComment();
  278. break;
  280. case $this->options['tag_block'][0]:
  281. // raw data?
  282. if (preg_match($this->regexes['lex_block_raw'], $this->code, $match, 0, $this->cursor)) {
  283. $this->moveCursor($match[0]);
  284. $this->lexRawData();
  285. // {% line \d+ %}
  286. } elseif (preg_match($this->regexes['lex_block_line'], $this->code, $match, 0, $this->cursor)) {
  287. $this->moveCursor($match[0]);
  288. $this->lineno = (int) $match[1];
  289. } else {
  290. $this->pushToken(Token::BLOCK_START_TYPE);
  291. $this->pushState(self::STATE_BLOCK);
  292. $this->currentVarBlockLine = $this->lineno;
  293. }
  294. break;
  296. case $this->options['tag_variable'][0]:
  297. $this->pushToken(Token::VAR_START_TYPE);
  298. $this->pushState(self::STATE_VAR);
  299. $this->currentVarBlockLine = $this->lineno;
  300. break;
  301. }
  302. }
  304. private function lexBlock(): void
  305. {
  306. if (!$this->brackets && preg_match($this->regexes['lex_block'], $this->code, $match, 0, $this->cursor)) {
  307. $this->pushToken(Token::BLOCK_END_TYPE);
  308. $this->moveCursor($match[0]);
  309. $this->popState();
  310. } else {
  311. $this->lexExpression();
  312. }
  313. }
  315. private function lexVar(): void
  316. {
  317. if (!$this->brackets && preg_match($this->regexes['lex_var'], $this->code, $match, 0, $this->cursor)) {
  318. $this->pushToken(Token::VAR_END_TYPE);
  319. $this->moveCursor($match[0]);
  320. $this->popState();
  321. } else {
  322. $this->lexExpression();
  323. }
  324. }
  326. private function lexExpression(): void
  327. {
  328. // whitespace
  329. if (preg_match('/\s+/A', $this->code, $match, 0, $this->cursor)) {
  330. $this->moveCursor($match[0]);
  332. if ($this->cursor >= $this->end) {
  333. throw new SyntaxError(\sprintf('Unclosed "%s".', self::STATE_BLOCK === $this->state ? 'block' : 'variable'), $this->currentVarBlockLine, $this->source);
  334. }
  335. }
  337. // operators
  338. if (preg_match($this->regexes['operator'], $this->code, $match, 0, $this->cursor)) {
  339. $operator = preg_replace('/\s+/', ' ', $match[0]);
  340. if (\in_array($operator, $this->openingBrackets, true)) {
  341. $this->checkBrackets($operator);
  342. }
  343. $this->pushToken(Token::OPERATOR_TYPE, $operator);
  344. $this->moveCursor($match[0]);
  345. }
  346. // names
  347. elseif (preg_match(self::REGEX_NAME, $this->code, $match, 0, $this->cursor)) {
  348. $this->pushToken(Token::NAME_TYPE, $match[0]);
  349. $this->moveCursor($match[0]);
  350. }
  351. // numbers
  352. elseif (preg_match(self::REGEX_NUMBER, $this->code, $match, 0, $this->cursor)) {
  353. $this->pushToken(Token::NUMBER_TYPE, 0 + str_replace('_', '', $match[0]));
  354. $this->moveCursor($match[0]);
  355. }
  356. // punctuation
  357. elseif (str_contains(self::PUNCTUATION, $this->code[$this->cursor])) {
  358. $this->checkBrackets($this->code[$this->cursor]);
  359. $this->pushToken(Token::PUNCTUATION_TYPE, $this->code[$this->cursor]);
  360. ++$this->cursor;
  361. }
  362. // strings
  363. elseif (preg_match(self::REGEX_STRING, $this->code, $match, 0, $this->cursor)) {
  364. $this->pushToken(Token::STRING_TYPE, $this->stripcslashes(substr($match[0], 1, -1), substr($match[0], 0, 1)));
  365. $this->moveCursor($match[0]);
  366. }
  367. // opening double quoted string
  368. elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, 0, $this->cursor)) {
  369. $this->brackets[] = ['"', $this->lineno];
  370. $this->pushState(self::STATE_STRING);
  371. $this->moveCursor($match[0]);
  372. }
  373. // inline comment
  374. elseif (preg_match(self::REGEX_INLINE_COMMENT, $this->code, $match, 0, $this->cursor)) {
  375. $this->moveCursor($match[0]);
  376. }
  377. // unlexable
  378. else {
  379. throw new SyntaxError(\sprintf('Unexpected character "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  380. }
  381. }
  383. private function stripcslashes(string $str, string $quoteType): string
  384. {
  385. $result = '';
  386. $length = \strlen($str);
  388. $i = 0;
  389. while ($i < $length) {
  390. if (false === $pos = strpos($str, '\\', $i)) {
  391. $result .= substr($str, $i);
  392. break;
  393. }
  395. $result .= substr($str, $i, $pos - $i);
  396. $i = $pos + 1;
  398. if ($i >= $length) {
  399. $result .= '\\';
  400. break;
  401. }
  403. $nextChar = $str[$i];
  405. if (isset(self::SPECIAL_CHARS[$nextChar])) {
  406. $result .= self::SPECIAL_CHARS[$nextChar];
  407. } elseif ('\\' === $nextChar) {
  408. $result .= $nextChar;
  409. } elseif ("'" === $nextChar || '"' === $nextChar) {
  410. if ($nextChar !== $quoteType) {
  411. 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);
  412. }
  413. $result .= $nextChar;
  414. } elseif ('#' === $nextChar && $i + 1 < $length && '{' === $str[$i + 1]) {
  415. $result .= '#{';
  416. ++$i;
  417. } elseif ('x' === $nextChar && $i + 1 < $length && ctype_xdigit($str[$i + 1])) {
  418. $hex = $str[++$i];
  419. if ($i + 1 < $length && ctype_xdigit($str[$i + 1])) {
  420. $hex .= $str[++$i];
  421. }
  422. $result .= \chr(hexdec($hex));
  423. } elseif (ctype_digit($nextChar) && $nextChar < '8') {
  424. $octal = $nextChar;
  425. while ($i + 1 < $length && ctype_digit($str[$i + 1]) && $str[$i + 1] < '8' && \strlen($octal) < 3) {
  426. $octal .= $str[++$i];
  427. }
  428. $result .= \chr(octdec($octal));
  429. } else {
  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. $result .= $nextChar;
  432. }
  434. ++$i;
  435. }
  437. return $result;
  438. }
  440. private function lexRawData(): void
  441. {
  442. if (!preg_match($this->regexes['lex_raw_data'], $this->code, $match, \PREG_OFFSET_CAPTURE, $this->cursor)) {
  443. throw new SyntaxError('Unexpected end of file: Unclosed "verbatim" block.', $this->lineno, $this->source);
  444. }
  446. $text = substr($this->code, $this->cursor, $match[0][1] - $this->cursor);
  447. $this->moveCursor($text.$match[0][0]);
  449. // trim?
  450. if (isset($match[1][0])) {
  451. if ($this->options['whitespace_trim'] === $match[1][0]) {
  452. // whitespace_trim detected ({%-, {{- or {#-)
  453. $text = rtrim($text);
  454. } else {
  455. // whitespace_line_trim detected ({%~, {{~ or {#~)
  456. // don't trim \r and \n
  457. $text = rtrim($text, " \t\0\x0B");
  458. }
  459. }
  461. $this->pushToken(Token::TEXT_TYPE, $text);
  462. }
  464. private function lexComment(): void
  465. {
  466. if (!preg_match($this->regexes['lex_comment'], $this->code, $match, \PREG_OFFSET_CAPTURE, $this->cursor)) {
  467. throw new SyntaxError('Unclosed comment.', $this->lineno, $this->source);
  468. }
  470. $this->moveCursor(substr($this->code, $this->cursor, $match[0][1] - $this->cursor).$match[0][0]);
  471. }
  473. private function lexString(): void
  474. {
  475. if (preg_match($this->regexes['interpolation_start'], $this->code, $match, 0, $this->cursor)) {
  476. $this->brackets[] = [$this->options['interpolation'][0], $this->lineno];
  477. $this->pushToken(Token::INTERPOLATION_START_TYPE);
  478. $this->moveCursor($match[0]);
  479. $this->pushState(self::STATE_INTERPOLATION);
  480. } elseif (preg_match(self::REGEX_DQ_STRING_PART, $this->code, $match, 0, $this->cursor) && '' !== $match[0]) {
  481. $this->pushToken(Token::STRING_TYPE, $this->stripcslashes($match[0], '"'));
  482. $this->moveCursor($match[0]);
  483. } elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, 0, $this->cursor)) {
  484. [$expect, $lineno] = array_pop($this->brackets);
  485. if ('"' != $this->code[$this->cursor]) {
  486. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  487. }
  489. $this->popState();
  490. ++$this->cursor;
  491. } else {
  492. // unlexable
  493. throw new SyntaxError(\sprintf('Unexpected character "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  494. }
  495. }
  497. private function lexInterpolation(): void
  498. {
  499. $bracket = end($this->brackets);
  500. if ($this->options['interpolation'][0] === $bracket[0] && preg_match($this->regexes['interpolation_end'], $this->code, $match, 0, $this->cursor)) {
  501. array_pop($this->brackets);
  502. $this->pushToken(Token::INTERPOLATION_END_TYPE);
  503. $this->moveCursor($match[0]);
  504. $this->popState();
  505. } else {
  506. $this->lexExpression();
  507. }
  508. }
  510. private function pushToken($type, $value = ''): void
  511. {
  512. // do not push empty text tokens
  513. if (Token::TEXT_TYPE === $type && '' === $value) {
  514. return;
  515. }
  517. $this->tokens[] = new Token($type, $value, $this->lineno);
  518. }
  520. private function moveCursor($text): void
  521. {
  522. $this->cursor += \strlen($text);
  523. $this->lineno += substr_count($text, "\n");
  524. }
  526. private function getOperatorRegex(): string
  527. {
  528. $expressionParsers = ['='];
  529. foreach ($this->env->getExpressionParsers() as $expressionParser) {
  530. $expressionParsers = array_merge($expressionParsers, [$expressionParser->getName()], $expressionParser->getAliases());
  531. }
  533. $expressionParsers = array_combine($expressionParsers, array_map('strlen', $expressionParsers));
  534. arsort($expressionParsers);
  536. $regex = [];
  537. foreach ($expressionParsers as $expressionParser => $length) {
  538. // an operator that ends with a character must be followed by
  539. // a whitespace, a parenthesis, an opening map [ or sequence {
  540. $r = preg_quote($expressionParser, '/');
  541. if (ctype_alpha($expressionParser[$length - 1])) {
  542. $r .= '(?=[\s()\[{])';
  543. }
  545. // an operator that begins with a character must not have a dot or pipe before
  546. if (ctype_alpha($expressionParser[0])) {
  547. $r = '(?<![\.\|])'.$r;
  548. }
  550. // an operator with a space can be any amount of whitespaces
  551. $r = preg_replace('/\s+/', '\s+', $r);
  553. $regex[] = $r;
  554. }
  556. return '/'.implode('|', $regex).'/A';
  557. }
  559. private function pushState($state): void
  560. {
  561. $this->states[] = $this->state;
  562. $this->state = $state;
  563. }
  565. private function popState(): void
  566. {
  567. if (0 === \count($this->states)) {
  568. throw new \LogicException('Cannot pop state without a previous state.');
  569. }
  571. $this->state = array_pop($this->states);
  572. }
  574. private function checkBrackets(string $code): void
  575. {
  576. // opening bracket
  577. if (\in_array($code, $this->openingBrackets, true)) {
  578. $this->brackets[] = [$code, $this->lineno];
  579. } elseif (\in_array($code, $this->closingBrackets, true)) {
  580. // closing bracket
  581. if (!$this->brackets) {
  582. throw new SyntaxError(\sprintf('Unexpected "%s".', $code), $this->lineno, $this->source);
  583. }
  585. [$expect, $lineno] = array_pop($this->brackets);
  586. if ($code !== str_replace($this->openingBrackets, $this->closingBrackets, $expect)) {
  587. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  588. }
  589. }
  590. }
  591. }