Project

General

Profile

1 1686 darkviper
<?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 $states;
28
    protected $brackets;
29
    protected $env;
30
    protected $filename;
31
    protected $options;
32
    protected $regexes;
33 1852 darkviper
    protected $position;
34
    protected $positions;
35 1686 darkviper
36
    const STATE_DATA            = 0;
37
    const STATE_BLOCK           = 1;
38
    const STATE_VAR             = 2;
39
    const STATE_STRING          = 3;
40
    const STATE_INTERPOLATION   = 4;
41
42
    const REGEX_NAME            = '/[a-zA-Z_\x7f-\xff][a-zA-Z0-9_\x7f-\xff]*/A';
43
    const REGEX_NUMBER          = '/[0-9]+(?:\.[0-9]+)?/A';
44
    const REGEX_STRING          = '/"([^#"\\\\]*(?:\\\\.[^#"\\\\]*)*)"|\'([^\'\\\\]*(?:\\\\.[^\'\\\\]*)*)\'/As';
45
    const REGEX_DQ_STRING_DELIM = '/"/A';
46
    const REGEX_DQ_STRING_PART  = '/[^#"\\\\]*(?:(?:\\\\.|#(?!\{))[^#"\\\\]*)*/As';
47
    const PUNCTUATION           = '()[]{}?:.,|';
48
49
    public function __construct(Twig_Environment $env, array $options = array())
50
    {
51
        $this->env = $env;
52
53
        $this->options = array_merge(array(
54
            'tag_comment'     => array('{#', '#}'),
55
            'tag_block'       => array('{%', '%}'),
56
            'tag_variable'    => array('{{', '}}'),
57
            'whitespace_trim' => '-',
58
            'interpolation'   => array('#{', '}'),
59
        ), $options);
60
61
        $this->regexes = array(
62
            'lex_var'             => '/\s*'.preg_quote($this->options['whitespace_trim'].$this->options['tag_variable'][1], '/').'\s*|\s*'.preg_quote($this->options['tag_variable'][1], '/').'/A',
63
            'lex_block'           => '/\s*(?:'.preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '/').'\s*|\s*'.preg_quote($this->options['tag_block'][1], '/').')\n?/A',
64
            'lex_raw_data'        => '/('.preg_quote($this->options['tag_block'][0].$this->options['whitespace_trim'], '/').'|'.preg_quote($this->options['tag_block'][0], '/').')\s*endraw\s*(?:'.preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '/').'\s*|\s*'.preg_quote($this->options['tag_block'][1], '/').')/s',
65
            'operator'            => $this->getOperatorRegex(),
66
            'lex_comment'         => '/(?:'.preg_quote($this->options['whitespace_trim'], '/').preg_quote($this->options['tag_comment'][1], '/').'\s*|'.preg_quote($this->options['tag_comment'][1], '/').')\n?/s',
67
            'lex_block_raw'       => '/\s*raw\s*(?:'.preg_quote($this->options['whitespace_trim'].$this->options['tag_block'][1], '/').'\s*|\s*'.preg_quote($this->options['tag_block'][1], '/').')/As',
68
            'lex_block_line'      => '/\s*line\s+(\d+)\s*'.preg_quote($this->options['tag_block'][1], '/').'/As',
69
            'lex_tokens_start'    => '/('.preg_quote($this->options['tag_variable'][0], '/').'|'.preg_quote($this->options['tag_block'][0], '/').'|'.preg_quote($this->options['tag_comment'][0], '/').')('.preg_quote($this->options['whitespace_trim'], '/').')?/s',
70
            'interpolation_start' => '/'.preg_quote($this->options['interpolation'][0], '/').'\s*/A',
71
            'interpolation_end'   => '/\s*'.preg_quote($this->options['interpolation'][1], '/').'/A',
72
        );
73
    }
74
75
    /**
76
     * Tokenizes a source code.
77
     *
78 1852 darkviper
     * @param string $code     The source code
79
     * @param string $filename A unique identifier for the source code
80 1686 darkviper
     *
81
     * @return Twig_TokenStream A token stream instance
82
     */
83
    public function tokenize($code, $filename = null)
84
    {
85
        if (function_exists('mb_internal_encoding') && ((int) ini_get('mbstring.func_overload')) & 2) {
86
            $mbEncoding = mb_internal_encoding();
87
            mb_internal_encoding('ASCII');
88
        }
89
90
        $this->code = str_replace(array("\r\n", "\r"), "\n", $code);
91
        $this->filename = $filename;
92
        $this->cursor = 0;
93
        $this->lineno = 1;
94
        $this->end = strlen($this->code);
95
        $this->tokens = array();
96
        $this->state = self::STATE_DATA;
97
        $this->states = array();
98
        $this->brackets = array();
99
        $this->position = -1;
100
101
        // find all token starts in one go
102
        preg_match_all($this->regexes['lex_tokens_start'], $this->code, $matches, PREG_OFFSET_CAPTURE);
103
        $this->positions = $matches;
104
105
        while ($this->cursor < $this->end) {
106
            // dispatch to the lexing functions depending
107
            // on the current state
108
            switch ($this->state) {
109
                case self::STATE_DATA:
110
                    $this->lexData();
111
                    break;
112
113
                case self::STATE_BLOCK:
114
                    $this->lexBlock();
115
                    break;
116
117
                case self::STATE_VAR:
118
                    $this->lexVar();
119
                    break;
120
121
                case self::STATE_STRING:
122
                    $this->lexString();
123
                    break;
124
125
                case self::STATE_INTERPOLATION:
126
                    $this->lexInterpolation();
127
                    break;
128
            }
129
        }
130
131
        $this->pushToken(Twig_Token::EOF_TYPE);
132
133
        if (!empty($this->brackets)) {
134
            list($expect, $lineno) = array_pop($this->brackets);
135
            throw new Twig_Error_Syntax(sprintf('Unclosed "%s"', $expect), $lineno, $this->filename);
136
        }
137
138
        if (isset($mbEncoding)) {
139
            mb_internal_encoding($mbEncoding);
140
        }
141
142
        return new Twig_TokenStream($this->tokens, $this->filename);
143
    }
144
145
    protected function lexData()
146
    {
147
        // if no matches are left we return the rest of the template as simple text token
148
        if ($this->position == count($this->positions[0]) - 1) {
149
            $this->pushToken(Twig_Token::TEXT_TYPE, substr($this->code, $this->cursor));
150
            $this->cursor = $this->end;
151
152
            return;
153
        }
154
155
        // Find the first token after the current cursor
156
        $position = $this->positions[0][++$this->position];
157
        while ($position[1] < $this->cursor) {
158
            if ($this->position == count($this->positions[0]) - 1) {
159
                return;
160
            }
161
            $position = $this->positions[0][++$this->position];
162
        }
163
164
        // push the template text first
165
        $text = $textContent = substr($this->code, $this->cursor, $position[1] - $this->cursor);
166
        if (isset($this->positions[2][$this->position][0])) {
167
            $text = rtrim($text);
168
        }
169
        $this->pushToken(Twig_Token::TEXT_TYPE, $text);
170
        $this->moveCursor($textContent.$position[0]);
171
172
        switch ($this->positions[1][$this->position][0]) {
173
            case $this->options['tag_comment'][0]:
174
                $this->lexComment();
175
                break;
176
177
            case $this->options['tag_block'][0]:
178
                // raw data?
179
                if (preg_match($this->regexes['lex_block_raw'], $this->code, $match, null, $this->cursor)) {
180
                    $this->moveCursor($match[0]);
181
                    $this->lexRawData();
182
                // {% line \d+ %}
183
                } elseif (preg_match($this->regexes['lex_block_line'], $this->code, $match, null, $this->cursor)) {
184
                    $this->moveCursor($match[0]);
185
                    $this->lineno = (int) $match[1];
186
                } else {
187
                    $this->pushToken(Twig_Token::BLOCK_START_TYPE);
188
                    $this->pushState(self::STATE_BLOCK);
189
                }
190
                break;
191
192
            case $this->options['tag_variable'][0]:
193
                $this->pushToken(Twig_Token::VAR_START_TYPE);
194
                $this->pushState(self::STATE_VAR);
195
                break;
196
        }
197
    }
198
199
    protected function lexBlock()
200
    {
201
        if (empty($this->brackets) && preg_match($this->regexes['lex_block'], $this->code, $match, null, $this->cursor)) {
202
            $this->pushToken(Twig_Token::BLOCK_END_TYPE);
203
            $this->moveCursor($match[0]);
204
            $this->popState();
205
        } else {
206
            $this->lexExpression();
207
        }
208
    }
209
210
    protected function lexVar()
211
    {
212
        if (empty($this->brackets) && preg_match($this->regexes['lex_var'], $this->code, $match, null, $this->cursor)) {
213
            $this->pushToken(Twig_Token::VAR_END_TYPE);
214
            $this->moveCursor($match[0]);
215
            $this->popState();
216
        } else {
217
            $this->lexExpression();
218
        }
219
    }
220
221
    protected function lexExpression()
222
    {
223
        // whitespace
224
        if (preg_match('/\s+/A', $this->code, $match, null, $this->cursor)) {
225
            $this->moveCursor($match[0]);
226
227
            if ($this->cursor >= $this->end) {
228
                throw new Twig_Error_Syntax(sprintf('Unexpected end of file: Unclosed "%s"', $this->state === self::STATE_BLOCK ? 'block' : 'variable'), $this->lineno, $this->filename);
229
            }
230
        }
231
232
        // operators
233
        if (preg_match($this->regexes['operator'], $this->code, $match, null, $this->cursor)) {
234
            $this->pushToken(Twig_Token::OPERATOR_TYPE, $match[0]);
235
            $this->moveCursor($match[0]);
236
        }
237
        // names
238
        elseif (preg_match(self::REGEX_NAME, $this->code, $match, null, $this->cursor)) {
239
            $this->pushToken(Twig_Token::NAME_TYPE, $match[0]);
240
            $this->moveCursor($match[0]);
241
        }
242
        // numbers
243
        elseif (preg_match(self::REGEX_NUMBER, $this->code, $match, null, $this->cursor)) {
244
            $number = (float) $match[0];  // floats
245
            if (ctype_digit($match[0]) && $number <= PHP_INT_MAX) {
246
                $number = (int) $match[0]; // integers lower than the maximum
247
            }
248
            $this->pushToken(Twig_Token::NUMBER_TYPE, $number);
249
            $this->moveCursor($match[0]);
250
        }
251
        // punctuation
252
        elseif (false !== strpos(self::PUNCTUATION, $this->code[$this->cursor])) {
253
            // opening bracket
254
            if (false !== strpos('([{', $this->code[$this->cursor])) {
255
                $this->brackets[] = array($this->code[$this->cursor], $this->lineno);
256
            }
257
            // closing bracket
258
            elseif (false !== strpos(')]}', $this->code[$this->cursor])) {
259
                if (empty($this->brackets)) {
260
                    throw new Twig_Error_Syntax(sprintf('Unexpected "%s"', $this->code[$this->cursor]), $this->lineno, $this->filename);
261
                }
262
263
                list($expect, $lineno) = array_pop($this->brackets);
264
                if ($this->code[$this->cursor] != strtr($expect, '([{', ')]}')) {
265
                    throw new Twig_Error_Syntax(sprintf('Unclosed "%s"', $expect), $lineno, $this->filename);
266
                }
267
            }
268
269
            $this->pushToken(Twig_Token::PUNCTUATION_TYPE, $this->code[$this->cursor]);
270
            ++$this->cursor;
271
        }
272
        // strings
273
        elseif (preg_match(self::REGEX_STRING, $this->code, $match, null, $this->cursor)) {
274
            $this->pushToken(Twig_Token::STRING_TYPE, stripcslashes(substr($match[0], 1, -1)));
275
            $this->moveCursor($match[0]);
276
        }
277
        // opening double quoted string
278
        elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, null, $this->cursor)) {
279
            $this->brackets[] = array('"', $this->lineno);
280
            $this->pushState(self::STATE_STRING);
281
            $this->moveCursor($match[0]);
282
        }
283
        // unlexable
284
        else {
285
            throw new Twig_Error_Syntax(sprintf('Unexpected character "%s"', $this->code[$this->cursor]), $this->lineno, $this->filename);
286
        }
287
    }
288
289
    protected function lexRawData()
290
    {
291
        if (!preg_match($this->regexes['lex_raw_data'], $this->code, $match, PREG_OFFSET_CAPTURE, $this->cursor)) {
292
            throw new Twig_Error_Syntax(sprintf('Unexpected end of file: Unclosed "block"'), $this->lineno, $this->filename);
293
        }
294
295
        $text = substr($this->code, $this->cursor, $match[0][1] - $this->cursor);
296
        $this->moveCursor($text.$match[0][0]);
297
298
        if (false !== strpos($match[1][0], $this->options['whitespace_trim'])) {
299
            $text = rtrim($text);
300
        }
301
302
        $this->pushToken(Twig_Token::TEXT_TYPE, $text);
303
    }
304
305
    protected function lexComment()
306
    {
307
        if (!preg_match($this->regexes['lex_comment'], $this->code, $match, PREG_OFFSET_CAPTURE, $this->cursor)) {
308
            throw new Twig_Error_Syntax('Unclosed comment', $this->lineno, $this->filename);
309
        }
310
311
        $this->moveCursor(substr($this->code, $this->cursor, $match[0][1] - $this->cursor).$match[0][0]);
312
    }
313
314
    protected function lexString()
315
    {
316
        if (preg_match($this->regexes['interpolation_start'], $this->code, $match, null, $this->cursor)) {
317
            $this->brackets[] = array($this->options['interpolation'][0], $this->lineno);
318
            $this->pushToken(Twig_Token::INTERPOLATION_START_TYPE);
319
            $this->moveCursor($match[0]);
320
            $this->pushState(self::STATE_INTERPOLATION);
321
322
        } elseif (preg_match(self::REGEX_DQ_STRING_PART, $this->code, $match, null, $this->cursor) && strlen($match[0]) > 0) {
323
            $this->pushToken(Twig_Token::STRING_TYPE, stripcslashes($match[0]));
324
            $this->moveCursor($match[0]);
325
326
        } elseif (preg_match(self::REGEX_DQ_STRING_DELIM, $this->code, $match, null, $this->cursor)) {
327
328
            list($expect, $lineno) = array_pop($this->brackets);
329
            if ($this->code[$this->cursor] != '"') {
330
                throw new Twig_Error_Syntax(sprintf('Unclosed "%s"', $expect), $lineno, $this->filename);
331
            }
332
333
            $this->popState();
334
            ++$this->cursor;
335
        }
336
    }
337
338
    protected function lexInterpolation()
339
    {
340
        $bracket = end($this->brackets);
341
        if ($this->options['interpolation'][0] === $bracket[0] && preg_match($this->regexes['interpolation_end'], $this->code, $match, null, $this->cursor)) {
342
            array_pop($this->brackets);
343
            $this->pushToken(Twig_Token::INTERPOLATION_END_TYPE);
344
            $this->moveCursor($match[0]);
345
            $this->popState();
346
        } else {
347
            $this->lexExpression();
348
        }
349
    }
350
351
    protected function pushToken($type, $value = '')
352
    {
353
        // do not push empty text tokens
354
        if (Twig_Token::TEXT_TYPE === $type && '' === $value) {
355
            return;
356
        }
357
358
        $this->tokens[] = new Twig_Token($type, $value, $this->lineno);
359
    }
360
361
    protected function moveCursor($text)
362
    {
363
        $this->cursor += strlen($text);
364
        $this->lineno += substr_count($text, "\n");
365
    }
366
367
    protected function getOperatorRegex()
368
    {
369
        $operators = array_merge(
370
            array('='),
371
            array_keys($this->env->getUnaryOperators()),
372
            array_keys($this->env->getBinaryOperators())
373
        );
374
375
        $operators = array_combine($operators, array_map('strlen', $operators));
376
        arsort($operators);
377
378
        $regex = array();
379
        foreach ($operators as $operator => $length) {
380
            // an operator that ends with a character must be followed by
381
            // a whitespace or a parenthesis
382
            if (ctype_alpha($operator[$length - 1])) {
383
                $regex[] = preg_quote($operator, '/').'(?=[\s()])';
384
            } else {
385
                $regex[] = preg_quote($operator, '/');
386
            }
387
        }
388
389
        return '/'.implode('|', $regex).'/A';
390
    }
391
392
    protected function pushState($state)
393
    {
394
        $this->states[] = $this->state;
395
        $this->state = $state;
396
    }
397
398
    protected function popState()
399
    {
400
        if (0 === count($this->states)) {
401
            throw new Exception('Cannot pop state without a previous state');
402
        }
403
404
        $this->state = array_pop($this->states);
405
    }
406
}