vendor/twig/twig/src/Lexer.php line 210

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;
  16. use Twig\ExpressionParser\ExpressionParsers;
  18. /**
  19. * @author Fabien Potencier <fabien@symfony.com>
  20. */
  21. class Lexer
  22. {
  23. private $isInitialized = false;
  25. private $tokens;
  26. private $code;
  27. private $cursor;
  28. private $lineno;
  29. private $end;
  30. private $state;
  31. private $states;
  32. private $brackets;
  33. private $env;
  34. private $source;
  35. private $options;
  36. private $regexes;
  37. private $position;
  38. private $positions;
  39. private $currentVarBlockLine;
  40. /** @var list<string> */
  41. private array $documentation = [];
  42. private array $openingBrackets = ['{', '(', '['];
  43. private array $closingBrackets = ['}', ')', ']'];
  45. public const STATE_DATA = 0;
  46. public const STATE_BLOCK = 1;
  47. public const STATE_VAR = 2;
  48. public const STATE_STRING = 3;
  49. public const STATE_INTERPOLATION = 4;
  51. public const REGEX_NAME = '/[a-zA-Z_\x7f-\xff][a-zA-Z0-9_\x7f-\xff]*/A';
  52. public const REGEX_STRING = '/"([^#"\\\\]*(?:\\\\.[^#"\\\\]*)*)"|\'([^\'\\\\]*(?:\\\\.[^\'\\\\]*)*)\'/As';
  54. public const REGEX_NUMBER = '/(?(DEFINE)
  55. (?<LNUM>[0-9]+(_[0-9]+)*) # Integers (with underscores) 123_456
  56. (?<FRAC>\.(?&LNUM)) # Fractional part .456
  57. (?<EXPONENT>[eE][+-]?(?&LNUM)) # Exponent part E+10
  58. (?<DNUM>(?&LNUM)(?:(?&FRAC))?) # Decimal number 123_456.456
  59. )(?:(?&DNUM)(?:(?&EXPONENT))?) # 123_456.456E+10
  60. /Ax';
  62. public const REGEX_DQ_STRING_DELIM = '/"/A';
  63. public const REGEX_DQ_STRING_PART = '/[^#"\\\\]*(?:(?:\\\\.|#(?!\{))[^#"\\\\]*)*/As';
  64. public const REGEX_INLINE_COMMENT = '/#[^\n]*/A';
  65. public const PUNCTUATION = '()[]{}?:.,|';
  67. private const REGEX_RAW_INLINE_COMMENT = '/#[^\r\n]*/A';
  69. private const SPECIAL_CHARS = [
  70. 'f' => "\f",
  71. 'n' => "\n",
  72. 'r' => "\r",
  73. 't' => "\t",
  74. 'v' => "\v",
  75. ];
  77. public function __construct(Environment $env, array $options = [])
  78. {
  79. $this->env = $env;
  81. $this->options = array_merge([
  82. 'tag_comment' => ['{#', '#}'],
  83. 'tag_block' => ['{%', '%}'],
  84. 'tag_variable' => ['{{', '}}'],
  85. 'whitespace_trim' => '-',
  86. 'whitespace_line_trim' => '~',
  87. 'whitespace_line_chars' => ' \t\0\x0B',
  88. 'interpolation' => ['#{', '}'],
  89. ], $options);
  90. }
  92. private function initialize(): void
  93. {
  94. if ($this->isInitialized) {
  95. return;
  96. }
  98. // when the comment closing tag starts with "#" (as "#}" does), the empty comment ("{##}")
  99. // starts like a documentation comment ("{##"); make sure it keeps lexing as a comment
  100. $emptyCommentLookahead = str_starts_with($this->options['tag_comment'][1], '#') ? '(?!'.preg_quote(substr($this->options['tag_comment'][1], 1), '#').')' : '';
  102. // 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
  103. $this->regexes = [
  104. // }}
  105. 'lex_var' => '{
  106. \s*
  107. (?:'.
  108. preg_quote($this->options['whitespace_trim'].$this->options['tag_variable'][1], '#').'\s*'. // -}}\s*
  109. '|'.
  110. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_variable'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~}}[ \t\0\x0B]*
  111. '|'.
  112. preg_quote($this->options['tag_variable'][1], '#'). // }}
  113. ')
  114. }Ax',
  116. // %}
  117. 'lex_block' => '{
  118. \s*
  119. (?:'.
  120. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*\n?'. // -%}\s*\n?
  121. '|'.
  122. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  123. '|'.
  124. preg_quote($this->options['tag_block'][1], '#').'(?:\r\n?|\n)?'. // %}(?:\r\n?|\n)?
  125. ')
  126. }Ax',
  128. // {% endverbatim %}
  129. 'lex_raw_data' => '{'.
  130. preg_quote($this->options['tag_block'][0], '#'). // {%
  131. '('.
  132. $this->options['whitespace_trim']. // -
  133. '|'.
  134. $this->options['whitespace_line_trim']. // ~
  135. ')?\s*endverbatim\s*'.
  136. '(?:'.
  137. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*'. // -%}
  138. '|'.
  139. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  140. '|'.
  141. preg_quote($this->options['tag_block'][1], '#'). // %}
  142. ')
  143. }sx',
  145. 'operator' => $this->getOperatorRegex(),
  147. // #}
  148. 'lex_comment' => '{
  149. (?:'.
  150. preg_quote($this->options['whitespace_trim'].$this->options['tag_comment'][1], '#').'\s*\n?'. // -#}\s*\n?
  151. '|'.
  152. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_comment'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~#}[ \t\0\x0B]*
  153. '|'.
  154. preg_quote($this->options['tag_comment'][1], '#').'(?:\r\n?|\n)?'. // #}(?:\r\n?|\n)?
  155. ')
  156. }sx',
  158. // #} or ##}
  159. 'lex_documentation_comment' => '{
  160. (?:'.
  161. preg_quote($this->options['whitespace_trim'].'#'.$this->options['tag_comment'][1], '#').'\s*\n?'. // -##}\s*\n?
  162. '|'.
  163. preg_quote($this->options['whitespace_line_trim'].'#'.$this->options['tag_comment'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~##}[ \t\0\x0B]*
  164. '|'.
  165. preg_quote($this->options['whitespace_trim'].$this->options['tag_comment'][1], '#').'\s*\n?'. // -#}\s*\n?
  166. '|'.
  167. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_comment'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~#}[ \t\0\x0B]*
  168. '|'.
  169. preg_quote($this->options['tag_comment'][1], '#').'(?:\r\n?|\n)?'. // #}(?:\r\n?|\n)?
  170. ')
  171. }sx',
  173. // verbatim %}
  174. 'lex_block_raw' => '{
  175. \s*verbatim\s*
  176. (?:'.
  177. preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*'. // -%}\s*
  178. '|'.
  179. preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  180. '|'.
  181. preg_quote($this->options['tag_block'][1], '#'). // %}
  182. ')
  183. }Asx',
  185. 'lex_block_line' => '{\s*line\s+(\d+)\s*'.preg_quote($this->options['tag_block'][1], '#').'}As',
  187. // {{ or {% or {#
  188. 'lex_tokens_start' => '{
  189. ('.
  190. preg_quote($this->options['tag_variable'][0], '#'). // {{
  191. '|'.
  192. preg_quote($this->options['tag_block'][0], '#'). // {%
  193. '|'.
  194. preg_quote($this->options['tag_comment'][0].'#', '#').$emptyCommentLookahead. // {## (but not the empty comment {##})
  195. '|'.
  196. preg_quote($this->options['tag_comment'][0], '#'). // {#
  197. ')('.
  198. preg_quote($this->options['whitespace_trim'], '#'). // -
  199. '|'.
  200. preg_quote($this->options['whitespace_line_trim'], '#'). // ~
  201. ')?
  202. }sx',
  203. 'interpolation_start' => '{'.preg_quote($this->options['interpolation'][0], '#').'\s*}A',
  204. 'interpolation_end' => '{\s*'.preg_quote($this->options['interpolation'][1], '#').'}A',
  205. ];
  207. $this->isInitialized = true;
  208. }
  210. public function tokenize(Source $source): TokenStream
  211. {
  212. $this->initialize();
  214. $this->source = $source;
  215. $this->code = $source->getCode();
  216. $this->cursor = 0;
  217. $this->lineno = 1;
  218. $this->end = \strlen($this->code);
  219. $this->tokens = [];
  220. $this->state = self::STATE_DATA;
  221. $this->states = [];
  222. $this->brackets = [];
  223. $this->position = -1;
  224. $this->documentation = [];
  226. // find all token starts in one go
  227. preg_match_all($this->regexes['lex_tokens_start'], $this->code, $matches, \PREG_OFFSET_CAPTURE);
  228. $this->positions = $matches;
  230. while ($this->cursor < $this->end) {
  231. // dispatch to the lexing functions depending
  232. // on the current state
  233. switch ($this->state) {
  234. case self::STATE_DATA:
  235. $this->lexData();
  236. break;
  238. case self::STATE_BLOCK:
  239. $this->lexBlock();
  240. break;
  242. case self::STATE_VAR:
  243. $this->lexVar();
  244. break;
  246. case self::STATE_STRING:
  247. $this->lexString();
  248. break;
  250. case self::STATE_INTERPOLATION:
  251. $this->lexInterpolation();
  252. break;
  253. }
  254. }
  256. $this->pushToken(Token::EOF_TYPE);
  258. if ($this->brackets) {
  259. [$expect, $lineno] = array_pop($this->brackets);
  260. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  261. }
  263. return new TokenStream($this->tokens, $this->source);
  264. }
  266. private function lexData(): void
  267. {
  268. // if no matches are left we return the rest of the template as simple text token
  269. if ($this->position == \count($this->positions[0]) - 1) {
  270. $text = substr($this->code, $this->cursor);
  271. $this->pushToken(Token::TEXT_TYPE, $this->normalizeNewlines($text));
  272. $this->moveCursor($text);
  274. return;
  275. }
  277. // Find the first token after the current cursor
  278. $position = $this->positions[0][++$this->position];
  279. while ($position[1] < $this->cursor) {
  280. if ($this->position == \count($this->positions[0]) - 1) {
  281. return;
  282. }
  283. $position = $this->positions[0][++$this->position];
  284. }
  286. // push the template text first
  287. $text = $textContent = substr($this->code, $this->cursor, $position[1] - $this->cursor);
  289. // trim?
  290. if (isset($this->positions[2][$this->position][0])) {
  291. if ($this->options['whitespace_trim'] === $this->positions[2][$this->position][0]) {
  292. // whitespace_trim detected ({%-, {{- or {#-)
  293. $text = rtrim($text);
  294. } elseif ($this->options['whitespace_line_trim'] === $this->positions[2][$this->position][0]) {
  295. // whitespace_line_trim detected ({%~, {{~ or {#~)
  296. // don't trim \r and \n
  297. $text = rtrim($text, " \t\0\x0B");
  298. }
  299. }
  300. $this->pushToken(Token::TEXT_TYPE, $this->normalizeNewlines($text));
  301. $this->moveCursor($textContent);
  303. switch ($this->positions[1][$this->position][0]) {
  304. case $this->options['tag_comment'][0].'#':
  305. $this->moveCursor($position[0]);
  306. $this->lexComment(true);
  307. break;
  309. case $this->options['tag_comment'][0]:
  310. $this->moveCursor($position[0]);
  311. $this->lexComment();
  312. break;
  314. case $this->options['tag_block'][0]:
  315. $lineno = $this->lineno;
  316. $cursor = $this->cursor;
  317. $this->moveCursor($position[0]);
  319. // raw data?
  320. if (preg_match($this->regexes['lex_block_raw'], $this->code, $match, 0, $this->cursor)) {
  321. $this->documentation = [];
  322. $this->moveCursor($match[0]);
  323. $this->lexRawData();
  324. // {% line \d+ %}
  325. } elseif (preg_match($this->regexes['lex_block_line'], $this->code, $match, 0, $this->cursor)) {
  326. $this->documentation = [];
  327. $this->moveCursor($match[0]);
  328. $this->lineno = (int) $match[1];
  329. } else {
  330. $this->pushToken(Token::BLOCK_START_TYPE, '', $cursor, $lineno);
  331. $this->pushState(self::STATE_BLOCK);
  332. $this->currentVarBlockLine = $this->lineno;
  333. }
  334. break;
  336. case $this->options['tag_variable'][0]:
  337. $lineno = $this->lineno;
  338. $cursor = $this->cursor;
  339. $this->moveCursor($position[0]);
  340. $this->pushToken(Token::VAR_START_TYPE, '', $cursor, $lineno);
  341. $this->pushState(self::STATE_VAR);
  342. $this->currentVarBlockLine = $this->lineno;
  343. break;
  344. }
  345. }
  347. private function lexBlock(): void
  348. {
  349. if (!$this->brackets && preg_match($this->regexes['lex_block'], $this->code, $match, 0, $this->cursor)) {
  350. $this->pushClosingToken(Token::BLOCK_END_TYPE, $match[0]);
  351. $this->popState();
  352. } else {
  353. $this->lexExpression();
  354. }
  355. }
  357. private function lexVar(): void
  358. {
  359. if (!$this->brackets && preg_match($this->regexes['lex_var'], $this->code, $match, 0, $this->cursor)) {
  360. $this->pushClosingToken(Token::VAR_END_TYPE, $match[0]);
  361. $this->popState();
  362. } else {
  363. $this->lexExpression();
  364. }
  365. }
  367. private function lexExpression(): void
  368. {
  369. // whitespace
  370. if (preg_match('/\s+/A', $this->code, $match, 0, $this->cursor)) {
  371. $this->moveCursor($match[0]);
  373. if ($this->cursor >= $this->end) {
  374. throw new SyntaxError(\sprintf('Unclosed "%s".', self::STATE_BLOCK === $this->state ? 'block' : 'variable'), $this->currentVarBlockLine, $this->source);
  375. }
  376. }
  378. // operators
  379. if (preg_match($this->regexes['operator'], $this->code, $match, 0, $this->cursor)) {
  380. $operator = preg_replace('/\s+/', ' ', $match[0]);
  381. if (\in_array($operator, $this->openingBrackets, true)) {
  382. $this->checkBrackets($operator);
  383. }
  384. $this->pushToken(Token::OPERATOR_TYPE, $operator);
  385. $this->moveCursor($match[0]);
  386. }
  387. // names
  388. elseif (preg_match(self::REGEX_NAME, $this->code, $match, 0, $this->cursor)) {
  389. $this->pushToken(Token::NAME_TYPE, $match[0]);
  390. $this->moveCursor($match[0]);
  391. }
  392. // numbers
  393. elseif (preg_match(self::REGEX_NUMBER, $this->code, $match, 0, $this->cursor)) {
  394. $this->pushToken(Token::NUMBER_TYPE, 0 + str_replace('_', '', $match[0]));
  395. $this->moveCursor($match[0]);
  396. }
  397. // punctuation
  398. elseif (str_contains(self::PUNCTUATION, $this->code[$this->cursor])) {
  399. $this->checkBrackets($this->code[$this->cursor]);
  400. $this->pushToken(Token::PUNCTUATION_TYPE, $this->code[$this->cursor]);
  401. $this->moveCursor($this->code[$this->cursor]);
  402. }
  403. // strings
  404. elseif (preg_match(self::REGEX_STRING, $this->code, $match, 0, $this->cursor)) {
  405. $this->pushToken(Token::STRING_TYPE, $this->stripcslashes($this->normalizeNewlines(substr($match[0], 1, -1)), substr($match[0], 0, 1)));
  406. $this->moveCursor($match[0]);
  407. }
  408. // opening double quoted string
  409. elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, 0, $this->cursor)) {
  410. $this->brackets[] = ['"', $this->lineno];
  411. $this->pushState(self::STATE_STRING);
  412. $this->moveCursor($match[0]);
  413. }
  414. // inline comment
  415. elseif (preg_match(self::REGEX_RAW_INLINE_COMMENT, $this->code, $match, 0, $this->cursor)) {
  416. if (str_starts_with($match[0], '##')) {
  417. $this->addDocumentation(substr($match[0], 2));
  418. }
  419. $this->moveCursor($match[0]);
  420. }
  421. // unlexable
  422. else {
  423. throw new SyntaxError(\sprintf('Unexpected character "%s".', $this->code[$this->cursor]), $this->lineno, $this->source, columnno: $this->source->getColumn($this->cursor));
  424. }
  425. }
  427. private function stripcslashes(string $str, string $quoteType): string
  428. {
  429. $result = '';
  430. $length = \strlen($str);
  432. $i = 0;
  433. while ($i < $length) {
  434. if (false === $pos = strpos($str, '\\', $i)) {
  435. $result .= substr($str, $i);
  436. break;
  437. }
  439. $result .= substr($str, $i, $pos - $i);
  440. $i = $pos + 1;
  442. if ($i >= $length) {
  443. $result .= '\\';
  444. break;
  445. }
  447. $nextChar = $str[$i];
  449. if (isset(self::SPECIAL_CHARS[$nextChar])) {
  450. $result .= self::SPECIAL_CHARS[$nextChar];
  451. } elseif ('\\' === $nextChar) {
  452. $result .= $nextChar;
  453. } elseif ("'" === $nextChar || '"' === $nextChar) {
  454. if ($nextChar !== $quoteType) {
  455. 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);
  456. }
  457. $result .= $nextChar;
  458. } elseif ('#' === $nextChar && $i + 1 < $length && '{' === $str[$i + 1]) {
  459. $result .= '#{';
  460. ++$i;
  461. } elseif ('x' === $nextChar && $i + 1 < $length && ctype_xdigit($str[$i + 1])) {
  462. $hex = $str[++$i];
  463. if ($i + 1 < $length && ctype_xdigit($str[$i + 1])) {
  464. $hex .= $str[++$i];
  465. }
  466. $result .= \chr(hexdec($hex));
  467. } elseif (ctype_digit($nextChar) && $nextChar < '8') {
  468. $octal = $nextChar;
  469. while ($i + 1 < $length && ctype_digit($str[$i + 1]) && $str[$i + 1] < '8' && \strlen($octal) < 3) {
  470. $octal .= $str[++$i];
  471. }
  472. $result .= \chr(octdec($octal) % 256);
  473. } else {
  474. 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);
  475. $result .= $nextChar;
  476. }
  478. ++$i;
  479. }
  481. return $result;
  482. }
  484. private function lexRawData(): void
  485. {
  486. if (!preg_match($this->regexes['lex_raw_data'], $this->code, $match, \PREG_OFFSET_CAPTURE, $this->cursor)) {
  487. throw new SyntaxError('Unexpected end of file: Unclosed "verbatim" block.', $this->lineno, $this->source);
  488. }
  490. $offset = $this->cursor;
  491. $text = substr($this->code, $this->cursor, $match[0][1] - $this->cursor);
  492. $this->moveCursor($text.$match[0][0]);
  494. // trim?
  495. if (isset($match[1][0])) {
  496. if ($this->options['whitespace_trim'] === $match[1][0]) {
  497. // whitespace_trim detected ({%-, {{- or {#-)
  498. $text = rtrim($text);
  499. } else {
  500. // whitespace_line_trim detected ({%~, {{~ or {#~)
  501. // don't trim \r and \n
  502. $text = rtrim($text, " \t\0\x0B");
  503. }
  504. }
  506. $this->pushToken(Token::TEXT_TYPE, $this->normalizeNewlines($text), $offset);
  507. }
  509. private function lexComment(bool $isDocumentation = false): void
  510. {
  511. $regex = $isDocumentation ? 'lex_documentation_comment' : 'lex_comment';
  512. if (!preg_match($this->regexes[$regex], $this->code, $match, \PREG_OFFSET_CAPTURE, $this->cursor)) {
  513. throw new SyntaxError('Unclosed comment.', $this->lineno, $this->source);
  514. }
  516. $comment = substr($this->code, $this->cursor, $match[0][1] - $this->cursor);
  517. if ($isDocumentation) {
  518. $documentation = $comment;
  519. // support a symmetric "##}" closing marker
  520. if (str_ends_with($documentation, '#')) {
  521. $documentation = substr($documentation, 0, -1);
  522. }
  523. $this->addDocumentation($documentation);
  524. }
  526. $this->moveCursor($comment.$match[0][0]);
  527. }
  529. private function lexString(): void
  530. {
  531. if (preg_match($this->regexes['interpolation_start'], $this->code, $match, 0, $this->cursor)) {
  532. $this->brackets[] = [$this->options['interpolation'][0], $this->lineno];
  533. $this->pushToken(Token::INTERPOLATION_START_TYPE);
  534. $this->moveCursor($match[0]);
  535. $this->pushState(self::STATE_INTERPOLATION);
  536. } elseif (preg_match(self::REGEX_DQ_STRING_PART, $this->code, $match, 0, $this->cursor) && '' !== $match[0]) {
  537. $this->pushToken(Token::STRING_TYPE, $this->stripcslashes($this->normalizeNewlines($match[0]), '"'));
  538. $this->moveCursor($match[0]);
  539. } elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, 0, $this->cursor)) {
  540. [$expect, $lineno] = array_pop($this->brackets);
  541. if ('"' != $this->code[$this->cursor]) {
  542. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  543. }
  545. $this->popState();
  546. $this->moveCursor($match[0]);
  547. } else {
  548. // unlexable
  549. throw new SyntaxError(\sprintf('Unexpected character "%s".', $this->code[$this->cursor]), $this->lineno, $this->source, columnno: $this->source->getColumn($this->cursor));
  550. }
  551. }
  553. private function lexInterpolation(): void
  554. {
  555. $bracket = end($this->brackets);
  556. if ($this->options['interpolation'][0] === $bracket[0] && preg_match($this->regexes['interpolation_end'], $this->code, $match, 0, $this->cursor)) {
  557. array_pop($this->brackets);
  558. $this->pushClosingToken(Token::INTERPOLATION_END_TYPE, $match[0]);
  559. $this->popState();
  560. } else {
  561. $this->lexExpression();
  562. }
  563. }
  565. private function pushToken($type, $value = '', ?int $offset = null, ?int $lineno = null): void
  566. {
  567. // do not push empty text tokens
  568. if (Token::TEXT_TYPE === $type && '' === $value) {
  569. return;
  570. }
  572. // by default the token starts at the current cursor; callers that
  573. // emit a token after consuming it must pass an explicit offset
  574. $documentation = null;
  575. if ($this->documentation && (Token::TEXT_TYPE !== $type || '' !== trim($value))) {
  576. $documentation = implode("\n", $this->documentation);
  577. $this->documentation = [];
  578. }
  580. $this->tokens[] = new Token($type, $value, $lineno ?? $this->lineno, $offset ?? $this->cursor, $documentation);
  581. }
  583. private function addDocumentation(string $documentation): void
  584. {
  585. if ('' !== $documentation = trim($this->normalizeNewlines($documentation))) {
  586. $this->documentation[] = $documentation;
  587. }
  588. }
  590. private function moveCursor($text): void
  591. {
  592. $this->cursor += \strlen($text);
  593. // count "\r\n" and "\r" as a single newline without allocating a
  594. // normalized copy when the chunk has no carriage return (common case)
  595. $this->lineno += str_contains($text, "\r") ? substr_count($this->normalizeNewlines($text), "\n") : substr_count($text, "\n");
  596. }
  598. private function normalizeNewlines(string $text): string
  599. {
  600. return str_replace(["\r\n", "\r"], "\n", $text);
  601. }
  603. private function pushClosingToken(int $type, string $match): void
  604. {
  605. $leadingWhitespaceLength = \strlen($match) - \strlen(ltrim($match));
  606. if ($leadingWhitespaceLength) {
  607. $this->moveCursor(substr($match, 0, $leadingWhitespaceLength));
  608. }
  610. $this->pushToken($type);
  611. $this->moveCursor(substr($match, $leadingWhitespaceLength));
  612. }
  614. private function getOperatorRegex(): string
  615. {
  616. $expressionParsers = [];
  617. foreach ($this->env->getExpressionParsers() as $expressionParser) {
  618. $expressionParsers = array_merge($expressionParsers, ExpressionParsers::getOperatorTokensFor($expressionParser));
  619. }
  621. $expressionParsers = array_combine($expressionParsers, array_map('strlen', $expressionParsers));
  622. arsort($expressionParsers);
  624. $regex = [];
  625. foreach ($expressionParsers as $expressionParser => $length) {
  626. // an operator that ends with a character must be followed by
  627. // a whitespace, a parenthesis, an opening map [ or sequence {
  628. $r = preg_quote($expressionParser, '/');
  629. if (ctype_alpha($expressionParser[$length - 1])) {
  630. $r .= '(?=[\s()\[{])';
  631. }
  633. // an operator that begins with a character must not have a dot or pipe before
  634. if (ctype_alpha($expressionParser[0])) {
  635. $r = '(?<![\.\|]\s|.[\.\|])'.$r;
  636. }
  638. // an operator with a space can be any amount of whitespaces
  639. $r = preg_replace('/\s+/', '\s+', $r);
  641. $regex[] = $r;
  642. }
  644. return '/'.implode('|', $regex).'/A';
  645. }
  647. private function pushState($state): void
  648. {
  649. $this->states[] = $this->state;
  650. $this->state = $state;
  651. }
  653. private function popState(): void
  654. {
  655. if (0 === \count($this->states)) {
  656. throw new \LogicException('Cannot pop state without a previous state.');
  657. }
  659. $this->state = array_pop($this->states);
  660. }
  662. private function checkBrackets(string $code): void
  663. {
  664. // opening bracket
  665. if (\in_array($code, $this->openingBrackets, true)) {
  666. $this->brackets[] = [$code, $this->lineno];
  667. } elseif (\in_array($code, $this->closingBrackets, true)) {
  668. // closing bracket
  669. if (!$this->brackets) {
  670. throw new SyntaxError(\sprintf('Unexpected "%s".', $code), $this->lineno, $this->source, columnno: $this->source->getColumn($this->cursor));
  671. }
  673. [$expect, $lineno] = array_pop($this->brackets);
  674. if ($code !== str_replace($this->openingBrackets, $this->closingBrackets, $expect)) {
  675. throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  676. }
  677. }
  678. }
  679. }