vendor/twig/twig/src/Lexer.php line 471

  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 PUNCTUATION = '()[]{}?:.,|';
  53.     public function __construct(Environment $env, array $options = [])
  54.     {
  55.         $this->env = $env;
  57.         $this->options = array_merge([
  58.             'tag_comment' => ['{#', '#}'],
  59.             'tag_block' => ['{%', '%}'],
  60.             'tag_variable' => ['{{', '}}'],
  61.             'whitespace_trim' => '-',
  62.             'whitespace_line_trim' => '~',
  63.             'whitespace_line_chars' => ' \t\0\x0B',
  64.             'interpolation' => ['#{', '}'],
  65.         ], $options);
  66.     }
  68.     private function initialize()
  69.     {
  70.         if ($this->isInitialized) {
  71.             return;
  72.         }
  74.         // 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
  75.         $this->regexes = [
  76.             // }}
  77.             'lex_var' => '{
  78.                 \s*
  79.                 (?:'.
  80.                     preg_quote($this->options['whitespace_trim'].$this->options['tag_variable'][1], '#').'\s*'. // -}}\s*
  81.                     '|'.
  82.                     preg_quote($this->options['whitespace_line_trim'].$this->options['tag_variable'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~}}[ \t\0\x0B]*
  83.                     '|'.
  84.                     preg_quote($this->options['tag_variable'][1], '#'). // }}
  85.                 ')
  86.             }Ax',
  88.             // %}
  89.             'lex_block' => '{
  90.                 \s*
  91.                 (?:'.
  92.                     preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*\n?'. // -%}\s*\n?
  93.                     '|'.
  94.                     preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  95.                     '|'.
  96.                     preg_quote($this->options['tag_block'][1], '#').'\n?'. // %}\n?
  97.                 ')
  98.             }Ax',
  100.             // {% endverbatim %}
  101.             'lex_raw_data' => '{'.
  102.                 preg_quote($this->options['tag_block'][0], '#'). // {%
  103.                 '('.
  104.                     $this->options['whitespace_trim']. // -
  105.                     '|'.
  106.                     $this->options['whitespace_line_trim']. // ~
  107.                 ')?\s*endverbatim\s*'.
  108.                 '(?:'.
  109.                     preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*'. // -%}
  110.                     '|'.
  111.                     preg_quote($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], '#'). // %}
  114.                 ')
  115.             }sx',
  117.             'operator' => $this->getOperatorRegex(),
  119.             // #}
  120.             'lex_comment' => '{
  121.                 (?:'.
  122.                     preg_quote($this->options['whitespace_trim'].$this->options['tag_comment'][1], '#').'\s*\n?'. // -#}\s*\n?
  123.                     '|'.
  124.                     preg_quote($this->options['whitespace_line_trim'].$this->options['tag_comment'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~#}[ \t\0\x0B]*
  125.                     '|'.
  126.                     preg_quote($this->options['tag_comment'][1], '#').'\n?'. // #}\n?
  127.                 ')
  128.             }sx',
  130.             // verbatim %}
  131.             'lex_block_raw' => '{
  132.                 \s*verbatim\s*
  133.                 (?:'.
  134.                     preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '#').'\s*'. // -%}\s*
  135.                     '|'.
  136.                     preg_quote($this->options['whitespace_line_trim'].$this->options['tag_block'][1], '#').'['.$this->options['whitespace_line_chars'].']*'. // ~%}[ \t\0\x0B]*
  137.                     '|'.
  138.                     preg_quote($this->options['tag_block'][1], '#'). // %}
  139.                 ')
  140.             }Asx',
  142.             'lex_block_line' => '{\s*line\s+(\d+)\s*'.preg_quote($this->options['tag_block'][1], '#').'}As',
  144.             // {{ or {% or {#
  145.             'lex_tokens_start' => '{
  146.                 ('.
  147.                     preg_quote($this->options['tag_variable'][0], '#'). // {{
  148.                     '|'.
  149.                     preg_quote($this->options['tag_block'][0], '#'). // {%
  150.                     '|'.
  151.                     preg_quote($this->options['tag_comment'][0], '#'). // {#
  152.                 ')('.
  153.                     preg_quote($this->options['whitespace_trim'], '#'). // -
  154.                     '|'.
  155.                     preg_quote($this->options['whitespace_line_trim'], '#'). // ~
  156.                 ')?
  157.             }sx',
  158.             'interpolation_start' => '{'.preg_quote($this->options['interpolation'][0], '#').'\s*}A',
  159.             'interpolation_end' => '{\s*'.preg_quote($this->options['interpolation'][1], '#').'}A',
  160.         ];
  162.         $this->isInitialized = true;
  163.     }
  165.     public function tokenize(Source $source): TokenStream
  166.     {
  167.         $this->initialize();
  169.         $this->source = $source;
  170.         $this->code = str_replace(["\r\n", "\r"], "\n", $source->getCode());
  171.         $this->cursor = 0;
  172.         $this->lineno = 1;
  173.         $this->end = \strlen($this->code);
  174.         $this->tokens = [];
  175.         $this->state = self::STATE_DATA;
  176.         $this->states = [];
  177.         $this->brackets = [];
  178.         $this->position = -1;
  180.         // find all token starts in one go
  181.         preg_match_all($this->regexes['lex_tokens_start'], $this->code, $matches, \PREG_OFFSET_CAPTURE);
  182.         $this->positions = $matches;
  184.         while ($this->cursor < $this->end) {
  185.             // dispatch to the lexing functions depending
  186.             // on the current state
  187.             switch ($this->state) {
  188.                 case self::STATE_DATA:
  189.                     $this->lexData();
  190.                     break;
  192.                 case self::STATE_BLOCK:
  193.                     $this->lexBlock();
  194.                     break;
  196.                 case self::STATE_VAR:
  197.                     $this->lexVar();
  198.                     break;
  200.                 case self::STATE_STRING:
  201.                     $this->lexString();
  202.                     break;
  204.                 case self::STATE_INTERPOLATION:
  205.                     $this->lexInterpolation();
  206.                     break;
  207.             }
  208.         }
  210.         $this->pushToken(/* Token::EOF_TYPE */ -1);
  212.         if (!empty($this->brackets)) {
  213.             [$expect, $lineno] = array_pop($this->brackets);
  214.             throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  215.         }
  217.         return new TokenStream($this->tokens, $this->source);
  218.     }
  220.     private function lexData(): void
  221.     {
  222.         // if no matches are left we return the rest of the template as simple text token
  223.         if ($this->position == \count($this->positions[0]) - 1) {
  224.             $this->pushToken(/* Token::TEXT_TYPE */ 0, substr($this->code, $this->cursor));
  225.             $this->cursor = $this->end;
  227.             return;
  228.         }
  230.         // Find the first token after the current cursor
  231.         $position = $this->positions[0][++$this->position];
  232.         while ($position[1] < $this->cursor) {
  233.             if ($this->position == \count($this->positions[0]) - 1) {
  234.                 return;
  235.             }
  236.             $position = $this->positions[0][++$this->position];
  237.         }
  239.         // push the template text first
  240.         $text = $textContent = substr($this->code, $this->cursor, $position[1] - $this->cursor);
  242.         // trim?
  243.         if (isset($this->positions[2][$this->position][0])) {
  244.             if ($this->options['whitespace_trim'] === $this->positions[2][$this->position][0]) {
  245.                 // whitespace_trim detected ({%-, {{- or {#-)
  246.                 $text = rtrim($text);
  247.             } elseif ($this->options['whitespace_line_trim'] === $this->positions[2][$this->position][0]) {
  248.                 // whitespace_line_trim detected ({%~, {{~ or {#~)
  249.                 // don't trim \r and \n
  250.                 $text = rtrim($text, " \t\0\x0B");
  251.             }
  252.         }
  253.         $this->pushToken(/* Token::TEXT_TYPE */ 0, $text);
  254.         $this->moveCursor($textContent.$position[0]);
  256.         switch ($this->positions[1][$this->position][0]) {
  257.             case $this->options['tag_comment'][0]:
  258.                 $this->lexComment();
  259.                 break;
  261.             case $this->options['tag_block'][0]:
  262.                 // raw data?
  263.                 if (preg_match($this->regexes['lex_block_raw'], $this->code, $match, 0, $this->cursor)) {
  264.                     $this->moveCursor($match[0]);
  265.                     $this->lexRawData();
  266.                 // {% line \d+ %}
  267.                 } elseif (preg_match($this->regexes['lex_block_line'], $this->code, $match, 0, $this->cursor)) {
  268.                     $this->moveCursor($match[0]);
  269.                     $this->lineno = (int) $match[1];
  270.                 } else {
  271.                     $this->pushToken(/* Token::BLOCK_START_TYPE */ 1);
  272.                     $this->pushState(self::STATE_BLOCK);
  273.                     $this->currentVarBlockLine = $this->lineno;
  274.                 }
  275.                 break;
  277.             case $this->options['tag_variable'][0]:
  278.                 $this->pushToken(/* Token::VAR_START_TYPE */ 2);
  279.                 $this->pushState(self::STATE_VAR);
  280.                 $this->currentVarBlockLine = $this->lineno;
  281.                 break;
  282.         }
  283.     }
  285.     private function lexBlock(): void
  286.     {
  287.         if (empty($this->brackets) && preg_match($this->regexes['lex_block'], $this->code, $match, 0, $this->cursor)) {
  288.             $this->pushToken(/* Token::BLOCK_END_TYPE */ 3);
  289.             $this->moveCursor($match[0]);
  290.             $this->popState();
  291.         } else {
  292.             $this->lexExpression();
  293.         }
  294.     }
  296.     private function lexVar(): void
  297.     {
  298.         if (empty($this->brackets) && preg_match($this->regexes['lex_var'], $this->code, $match, 0, $this->cursor)) {
  299.             $this->pushToken(/* Token::VAR_END_TYPE */ 4);
  300.             $this->moveCursor($match[0]);
  301.             $this->popState();
  302.         } else {
  303.             $this->lexExpression();
  304.         }
  305.     }
  307.     private function lexExpression(): void
  308.     {
  309.         // whitespace
  310.         if (preg_match('/\s+/A', $this->code, $match, 0, $this->cursor)) {
  311.             $this->moveCursor($match[0]);
  313.             if ($this->cursor >= $this->end) {
  314.                 throw new SyntaxError(\sprintf('Unclosed "%s".', self::STATE_BLOCK === $this->state ? 'block' : 'variable'), $this->currentVarBlockLine, $this->source);
  315.             }
  316.         }
  318.         // spread operator
  319.         if ('.' === $this->code[$this->cursor] && ($this->cursor + 2 < $this->end) && '.' === $this->code[$this->cursor + 1] && '.' === $this->code[$this->cursor + 2]) {
  320.             $this->pushToken(Token::SPREAD_TYPE, '...');
  321.             $this->moveCursor('...');
  322.         }
  323.         // arrow function
  324.         elseif ('=' === $this->code[$this->cursor] && ($this->cursor + 1 < $this->end) && '>' === $this->code[$this->cursor + 1]) {
  325.             $this->pushToken(Token::ARROW_TYPE, '=>');
  326.             $this->moveCursor('=>');
  327.         }
  328.         // operators
  329.         elseif (preg_match($this->regexes['operator'], $this->code, $match, 0, $this->cursor)) {
  330.             $this->pushToken(/* Token::OPERATOR_TYPE */ 8, preg_replace('/\s+/', ' ', $match[0]));
  331.             $this->moveCursor($match[0]);
  332.         }
  333.         // names
  334.         elseif (preg_match(self::REGEX_NAME, $this->code, $match, 0, $this->cursor)) {
  335.             $this->pushToken(/* Token::NAME_TYPE */ 5, $match[0]);
  336.             $this->moveCursor($match[0]);
  337.         }
  338.         // numbers
  339.         elseif (preg_match(self::REGEX_NUMBER, $this->code, $match, 0, $this->cursor)) {
  340.             $number = (float) $match[0];  // floats
  341.             if (ctype_digit($match[0]) && $number <= \PHP_INT_MAX) {
  342.                 $number = (int) $match[0]; // integers lower than the maximum
  343.             }
  344.             $this->pushToken(/* Token::NUMBER_TYPE */ 6, $number);
  345.             $this->moveCursor($match[0]);
  346.         }
  347.         // punctuation
  348.         elseif (str_contains(self::PUNCTUATION, $this->code[$this->cursor])) {
  349.             // opening bracket
  350.             if (str_contains('([{', $this->code[$this->cursor])) {
  351.                 $this->brackets[] = [$this->code[$this->cursor], $this->lineno];
  352.             }
  353.             // closing bracket
  354.             elseif (str_contains(')]}', $this->code[$this->cursor])) {
  355.                 if (empty($this->brackets)) {
  356.                     throw new SyntaxError(\sprintf('Unexpected "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  357.                 }
  359.                 [$expect, $lineno] = array_pop($this->brackets);
  360.                 if ($this->code[$this->cursor] != strtr($expect, '([{', ')]}')) {
  361.                     throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  362.                 }
  363.             }
  365.             $this->pushToken(/* Token::PUNCTUATION_TYPE */ 9, $this->code[$this->cursor]);
  366.             ++$this->cursor;
  367.         }
  368.         // strings
  369.         elseif (preg_match(self::REGEX_STRING, $this->code, $match, 0, $this->cursor)) {
  370.             $this->pushToken(/* Token::STRING_TYPE */ 7, stripcslashes(substr($match[0], 1, -1)));
  371.             $this->moveCursor($match[0]);
  372.         }
  373.         // opening double quoted string
  374.         elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, 0, $this->cursor)) {
  375.             $this->brackets[] = ['"', $this->lineno];
  376.             $this->pushState(self::STATE_STRING);
  377.             $this->moveCursor($match[0]);
  378.         }
  379.         // unlexable
  380.         else {
  381.             throw new SyntaxError(\sprintf('Unexpected character "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  382.         }
  383.     }
  385.     private function lexRawData(): void
  386.     {
  387.         if (!preg_match($this->regexes['lex_raw_data'], $this->code, $match, \PREG_OFFSET_CAPTURE, $this->cursor)) {
  388.             throw new SyntaxError('Unexpected end of file: Unclosed "verbatim" block.', $this->lineno, $this->source);
  389.         }
  391.         $text = substr($this->code, $this->cursor, $match[0][1] - $this->cursor);
  392.         $this->moveCursor($text.$match[0][0]);
  394.         // trim?
  395.         if (isset($match[1][0])) {
  396.             if ($this->options['whitespace_trim'] === $match[1][0]) {
  397.                 // whitespace_trim detected ({%-, {{- or {#-)
  398.                 $text = rtrim($text);
  399.             } else {
  400.                 // whitespace_line_trim detected ({%~, {{~ or {#~)
  401.                 // don't trim \r and \n
  402.                 $text = rtrim($text, " \t\0\x0B");
  403.             }
  404.         }
  406.         $this->pushToken(/* Token::TEXT_TYPE */ 0, $text);
  407.     }
  409.     private function lexComment(): void
  410.     {
  411.         if (!preg_match($this->regexes['lex_comment'], $this->code, $match, \PREG_OFFSET_CAPTURE, $this->cursor)) {
  412.             throw new SyntaxError('Unclosed comment.', $this->lineno, $this->source);
  413.         }
  415.         $this->moveCursor(substr($this->code, $this->cursor, $match[0][1] - $this->cursor).$match[0][0]);
  416.     }
  418.     private function lexString(): void
  419.     {
  420.         if (preg_match($this->regexes['interpolation_start'], $this->code, $match, 0, $this->cursor)) {
  421.             $this->brackets[] = [$this->options['interpolation'][0], $this->lineno];
  422.             $this->pushToken(/* Token::INTERPOLATION_START_TYPE */ 10);
  423.             $this->moveCursor($match[0]);
  424.             $this->pushState(self::STATE_INTERPOLATION);
  425.         } elseif (preg_match(self::REGEX_DQ_STRING_PART, $this->code, $match, 0, $this->cursor) && '' !== $match[0]) {
  426.             $this->pushToken(/* Token::STRING_TYPE */ 7, stripcslashes($match[0]));
  427.             $this->moveCursor($match[0]);
  428.         } elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, 0, $this->cursor)) {
  429.             [$expect, $lineno] = array_pop($this->brackets);
  430.             if ('"' != $this->code[$this->cursor]) {
  431.                 throw new SyntaxError(\sprintf('Unclosed "%s".', $expect), $lineno, $this->source);
  432.             }
  434.             $this->popState();
  435.             ++$this->cursor;
  436.         } else {
  437.             // unlexable
  438.             throw new SyntaxError(\sprintf('Unexpected character "%s".', $this->code[$this->cursor]), $this->lineno, $this->source);
  439.         }
  440.     }
  442.     private function lexInterpolation(): void
  443.     {
  444.         $bracket = end($this->brackets);
  445.         if ($this->options['interpolation'][0] === $bracket[0] && preg_match($this->regexes['interpolation_end'], $this->code, $match, 0, $this->cursor)) {
  446.             array_pop($this->brackets);
  447.             $this->pushToken(/* Token::INTERPOLATION_END_TYPE */ 11);
  448.             $this->moveCursor($match[0]);
  449.             $this->popState();
  450.         } else {
  451.             $this->lexExpression();
  452.         }
  453.     }
  455.     private function pushToken($type, $value = ''): void
  456.     {
  457.         // do not push empty text tokens
  458.         if (/* Token::TEXT_TYPE */ 0 === $type && '' === $value) {
  459.             return;
  460.         }
  462.         $this->tokens[] = new Token($type, $value, $this->lineno);
  463.     }
  465.     private function moveCursor($text): void
  466.     {
  467.         $this->cursor += \strlen($text);
  468.         $this->lineno += substr_count($text, "\n");
  469.     }
  471.     private function getOperatorRegex(): string
  472.     {
  473.         $operators = array_merge(
  474.             ['='],
  475.             array_keys($this->env->getUnaryOperators()),
  476.             array_keys($this->env->getBinaryOperators())
  477.         );
  479.         $operators = array_combine($operators, array_map('strlen', $operators));
  480.         arsort($operators);
  482.         $regex = [];
  483.         foreach ($operators as $operator => $length) {
  484.             // an operator that ends with a character must be followed by
  485.             // a whitespace, a parenthesis, an opening map [ or sequence {
  486.             $r = preg_quote($operator, '/');
  487.             if (ctype_alpha($operator[$length - 1])) {
  488.                 $r .= '(?=[\s()\[{])';
  489.             }
  491.             // an operator that begins with a character must not have a dot or pipe before
  492.             if (ctype_alpha($operator[0])) {
  493.                 $r = '(?<![\.\|])'.$r;
  494.             }
  496.             // an operator with a space can be any amount of whitespaces
  497.             $r = preg_replace('/\s+/', '\s+', $r);
  499.             $regex[] = $r;
  500.         }
  502.         return '/'.implode('|', $regex).'/A';
  503.     }
  505.     private function pushState($state): void
  506.     {
  507.         $this->states[] = $this->state;
  508.         $this->state = $state;
  509.     }
  511.     private function popState(): void
  512.     {
  513.         if (0 === \count($this->states)) {
  514.             throw new \LogicException('Cannot pop state without a previous state.');
  515.         }
  517.         $this->state = array_pop($this->states);
  518.     }
  519. }