1 <?php

     2 

     3 /*

     4  * This file is part of Twig.

     5  *

     6  * (c) 2009 Fabien Potencier

     7  * (c) 2009 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  */

    12 

    13 /**

    14  * Lexes a template string.

    15  *

    16  * @package    twig

    17  * @author     Fabien Potencier <fabien@symfony.com>

    18  */

    19 class Twig_Lexer implements Twig_LexerInterface

    20 {

    21     protected $tokens;

    22     protected $code;

    23     protected $cursor;

    24     protected $lineno;

    25     protected $end;

    26     protected $state;

    27     protected $brackets;

    28 

    29     protected $env;

    30     protected $filename;

    31     protected $options;

    32     protected $operatorRegex;

    33 

    34     const STATE_DATA  = 0;

    35     const STATE_BLOCK = 1;

    36     const STATE_VAR   = 2;

    37 

    38     const REGEX_NAME   = '/[a-zA-Z_\x7f-\xff][a-zA-Z0-9_\x7f-\xff]*/A';

    39     const REGEX_NUMBER = '/[0-9]+(?:\.[0-9]+)?/A';

    40     const REGEX_STRING = '/"([^"\\\\]*(?:\\\\.[^"\\\\]*)*)"|\'([^\'\\\\]*(?:\\\\.[^\'\\\\]*)*)\'/As';

    41     const PUNCTUATION  = '()[]{}?:.,|';

    42 

    43     public function __construct(Twig_Environment $env, array $options = array())

    44     {

    45         $this->env = $env;

    46 

    47         $this->options = array_merge(array(

    48             'tag_comment'     => array('{#', '#}'),

    49             'tag_block'       => array('{%', '%}'),

    50             'tag_variable'    => array('{{', '}}'),

    51             'whitespace_trim' => '-',

    52         ), $options);

    53     }

    54 

    55     /**

    56      * Tokenizes a source code.

    57      *

    58      * @param  string $code     The source code

    59      * @param  string $filename A unique identifier for the source code

    60      *

    61      * @return Twig_TokenStream A token stream instance

    62      */

    63     public function tokenize($code, $filename = null)

    64     {

    65         if (function_exists('mb_internal_encoding') && ((int) ini_get('mbstring.func_overload')) & 2) {

    66             $mbEncoding = mb_internal_encoding();

    67             mb_internal_encoding('ASCII');

    68         }

    69 

    70         $this->code = str_replace(array("\r\n", "\r"), "\n", $code);

    71         $this->filename = $filename;

    72         $this->cursor = 0;

    73         $this->lineno = 1;

    74         $this->end = strlen($this->code);

    75         $this->tokens = array();

    76         $this->state = self::STATE_DATA;

    77         $this->brackets = array();

    78 

    79         while ($this->cursor < $this->end) {

    80             // dispatch to the lexing functions depending

    81             // on the current state

    82             switch ($this->state) {

    83                 case self::STATE_DATA:

    84                     $this->lexData();

    85                     break;

    86 

    87                 case self::STATE_BLOCK:

    88                     $this->lexBlock();

    89                     break;

    90 

    91                 case self::STATE_VAR:

    92                     $this->lexVar();

    93                     break;

    94             }

    95         }

    96 

    97         $this->pushToken(Twig_Token::EOF_TYPE);

    98 

    99         if (!empty($this->brackets)) {

   100             list($expect, $lineno) = array_pop($this->brackets);

   101             throw new Twig_Error_Syntax(sprintf('Unclosed "%s"', $expect), $lineno, $this->filename);

   102         }

   103 

   104         if (isset($mbEncoding)) {

   105             mb_internal_encoding($mbEncoding);

   106         }

   107 

   108         return new Twig_TokenStream($this->tokens, $this->filename);

   109     }

   110 

   111     protected function lexData()

   112     {

   113         $pos = $this->end;

   114         $append = '';

   115 

   116         // Find the first token after the cursor

   117         foreach (array('tag_comment', 'tag_variable', 'tag_block') as $type) {

   118             $tmpPos = strpos($this->code, $this->options[$type][0], $this->cursor);

   119             if (false !== $tmpPos && $tmpPos < $pos) {

   120                 $trimBlock = false;

   121                 $append = '';

   122                 $pos = $tmpPos;

   123                 $token = $this->options[$type][0];

   124                 if (strpos($this->code, $this->options['whitespace_trim'], $pos) === ($pos + strlen($token))) {

   125                     $trimBlock = true;

   126                     $append = $this->options['whitespace_trim'];

   127                 }

   128             }

   129         }

   130 

   131         // if no matches are left we return the rest of the template as simple text token

   132         if ($pos === $this->end) {

   133             $this->pushToken(Twig_Token::TEXT_TYPE, substr($this->code, $this->cursor));

   134             $this->cursor = $this->end;

   135             return;

   136         }

   137 

   138         // push the template text first

   139         $text = $textContent = substr($this->code, $this->cursor, $pos - $this->cursor);

   140         if (true === $trimBlock) {

   141             $text = rtrim($text);

   142         }

   143         $this->pushToken(Twig_Token::TEXT_TYPE, $text);

   144         $this->moveCursor($textContent.$token.$append);

   145 

   146         switch ($token) {

   147             case $this->options['tag_comment'][0]:

   148                 $this->lexComment();

   149                 break;

   150 

   151             case $this->options['tag_block'][0]:

   152                 // raw data?

   153                 if (preg_match('/\s*raw\s*'.preg_quote($this->options['tag_block'][1], '/').'/As', $this->code, $match, null, $this->cursor)) {

   154                     $this->moveCursor($match[0]);

   155                     $this->lexRawData();

   156                     $this->state = self::STATE_DATA;

   157                 // {% line \d+ %}

   158                 } else if (preg_match('/\s*line\s+(\d+)\s*'.preg_quote($this->options['tag_block'][1], '/').'/As', $this->code, $match, null, $this->cursor)) {

   159                     $this->moveCursor($match[0]);

   160                     $this->lineno = (int) $match[1];

   161                     $this->state = self::STATE_DATA;

   162                 } else {

   163                     $this->pushToken(Twig_Token::BLOCK_START_TYPE);

   164                     $this->state = self::STATE_BLOCK;

   165                 }

   166                 break;

   167 

   168             case $this->options['tag_variable'][0]:

   169                 $this->pushToken(Twig_Token::VAR_START_TYPE);

   170                 $this->state = self::STATE_VAR;

   171                 break;

   172         }

   173     }

   174 

   175     protected function lexBlock()

   176     {

   177         $trimTag = preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '/');

   178         $endTag = preg_quote($this->options['tag_block'][1], '/');

   179 

   180         if (empty($this->brackets) && preg_match('/\s*(?:'.$trimTag.'\s*|\s*'.$endTag.')\n?/A', $this->code, $match, null, $this->cursor)) {

   181             $this->pushToken(Twig_Token::BLOCK_END_TYPE);

   182             $this->moveCursor($match[0]);

   183             $this->state = self::STATE_DATA;

   184         } else {

   185             $this->lexExpression();

   186         }

   187     }

   188 

   189     protected function lexVar()

   190     {

   191         $trimTag = preg_quote($this->options['whitespace_trim'].$this->options['tag_variable'][1], '/');

   192         $endTag = preg_quote($this->options['tag_variable'][1], '/');

   193 

   194         if (empty($this->brackets) && preg_match('/\s*'.$trimTag.'\s*|\s*'.$endTag.'/A', $this->code, $match, null, $this->cursor)) {

   195             $this->pushToken(Twig_Token::VAR_END_TYPE);

   196             $this->moveCursor($match[0]);

   197             $this->state = self::STATE_DATA;

   198         } else {

   199             $this->lexExpression();

   200         }

   201     }

   202 

   203     protected function lexExpression()

   204     {

   205         // whitespace

   206         if (preg_match('/\s+/A', $this->code, $match, null, $this->cursor)) {

   207             $this->moveCursor($match[0]);

   208 

   209             if ($this->cursor >= $this->end) {

   210                 throw new Twig_Error_Syntax(sprintf('Unexpected end of file: Unclosed "%s"', $this->state === self::STATE_BLOCK ? 'block' : 'variable'));

   211             }

   212         }

   213 

   214         // operators

   215         if (preg_match($this->getOperatorRegex(), $this->code, $match, null, $this->cursor)) {

   216             $this->pushToken(Twig_Token::OPERATOR_TYPE, $match[0]);

   217             $this->moveCursor($match[0]);

   218         }

   219         // names

   220         elseif (preg_match(self::REGEX_NAME, $this->code, $match, null, $this->cursor)) {

   221             $this->pushToken(Twig_Token::NAME_TYPE, $match[0]);

   222             $this->moveCursor($match[0]);

   223         }

   224         // numbers

   225         elseif (preg_match(self::REGEX_NUMBER, $this->code, $match, null, $this->cursor)) {

   226             $this->pushToken(Twig_Token::NUMBER_TYPE, ctype_digit($match[0]) ? (int) $match[0] : (float) $match[0]);

   227             $this->moveCursor($match[0]);

   228         }

   229         // punctuation

   230         elseif (false !== strpos(self::PUNCTUATION, $this->code[$this->cursor])) {

   231             // opening bracket

   232             if (false !== strpos('([{', $this->code[$this->cursor])) {

   233                 $this->brackets[] = array($this->code[$this->cursor], $this->lineno);

   234             }

   235             // closing bracket

   236             elseif (false !== strpos(')]}', $this->code[$this->cursor])) {

   237                 if (empty($this->brackets)) {

   238                     throw new Twig_Error_Syntax(sprintf('Unexpected "%s"', $this->code[$this->cursor]), $this->lineno, $this->filename);

   239                 }

   240 

   241                 list($expect, $lineno) = array_pop($this->brackets);

   242                 if ($this->code[$this->cursor] != strtr($expect, '([{', ')]}')) {

   243                     throw new Twig_Error_Syntax(sprintf('Unclosed "%s"', $expect), $lineno, $this->filename);

   244                 }

   245             }

   246 

   247             $this->pushToken(Twig_Token::PUNCTUATION_TYPE, $this->code[$this->cursor]);

   248             ++$this->cursor;

   249         }

   250         // strings

   251         elseif (preg_match(self::REGEX_STRING, $this->code, $match, null, $this->cursor)) {

   252             $this->pushToken(Twig_Token::STRING_TYPE, stripcslashes(substr($match[0], 1, -1)));

   253             $this->moveCursor($match[0]);

   254         }

   255         // unlexable

   256         else {

   257             throw new Twig_Error_Syntax(sprintf('Unexpected character "%s"', $this->code[$this->cursor]), $this->lineno, $this->filename);

   258         }

   259     }

   260 

   261     protected function lexRawData()

   262     {

   263         if (!preg_match('/'.preg_quote($this->options['tag_block'][0], '/').'\s*endraw\s*'.preg_quote($this->options['tag_block'][1], '/').'/s', $this->code, $match, PREG_OFFSET_CAPTURE, $this->cursor)) {

   264             throw new Twig_Error_Syntax(sprintf('Unexpected end of file: Unclosed "block"'));

   265         }

   266         $text = substr($this->code, $this->cursor, $match[0][1] - $this->cursor);

   267         $this->pushToken(Twig_Token::TEXT_TYPE, $text);

   268         $this->moveCursor($text.$match[0][0]);

   269     }

   270 

   271     protected function lexComment()

   272     {

   273         $commentEndRegex = '/(?:'.preg_quote($this->options['whitespace_trim'], '/')

   274                            .preg_quote($this->options['tag_comment'][1], '/').'\s*|'

   275                            .preg_quote($this->options['tag_comment'][1], '/').')\n?/s';

   276 

   277         if (!preg_match($commentEndRegex, $this->code, $match, PREG_OFFSET_CAPTURE, $this->cursor)) {

   278             throw new Twig_Error_Syntax('Unclosed comment', $this->lineno, $this->filename);

   279         }

   280 

   281         $this->moveCursor(substr($this->code, $this->cursor, $match[0][1] - $this->cursor).$match[0][0]);

   282     }

   283 

   284     protected function pushToken($type, $value = '')

   285     {

   286         // do not push empty text tokens

   287         if (Twig_Token::TEXT_TYPE === $type && '' === $value) {

   288             return;

   289         }

   290 

   291         $this->tokens[] = new Twig_Token($type, $value, $this->lineno);

   292     }

   293 

   294     protected function moveCursor($text)

   295     {

   296         $this->cursor += strlen($text);

   297         $this->lineno += substr_count($text, "\n");

   298     }

   299 

   300     protected function getOperatorRegex()

   301     {

   302         if (null !== $this->operatorRegex) {

   303             return $this->operatorRegex;

   304         }

   305 

   306         $operators = array_merge(

   307             array('='),

   308             array_keys($this->env->getUnaryOperators()),

   309             array_keys($this->env->getBinaryOperators())

   310         );

   311 

   312         $operators = array_combine($operators, array_map('strlen', $operators));

   313         arsort($operators);

   314 

   315         $regex = array();

   316         foreach ($operators as $operator => $length) {

   317             // an operator that ends with a character must be followed by

   318             // a whitespace or a parenthesis

   319             if (ctype_alpha($operator[$length - 1])) {

   320                 $regex[] = preg_quote($operator, '/').'(?=[ ()])';

   321             } else {

   322                 $regex[] = preg_quote($operator, '/');

   323             }

   324         }

   325 

   326         return $this->operatorRegex = '/'.implode('|', $regex).'/A';

   327     }

   328 }