1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @package s9e\SourceOptimizer |
5
|
|
|
* @copyright Copyright (c) 2014-2017 The s9e Authors |
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
7
|
|
|
*/ |
8
|
|
|
namespace s9e\SourceOptimizer; |
9
|
|
|
|
10
|
|
|
use ArrayAccess; |
11
|
|
|
use Iterator; |
12
|
|
|
|
13
|
|
|
class TokenStream implements ArrayAccess, Iterator |
14
|
|
|
{ |
15
|
|
|
/** |
16
|
|
|
* @var integer Number of tokens |
17
|
|
|
*/ |
18
|
|
|
protected $cnt; |
19
|
|
|
|
20
|
|
|
/** |
21
|
|
|
* @var string[] List of exact triplet of tokens to exclude from minification |
22
|
|
|
*/ |
23
|
|
|
protected $excludeExact = [ |
24
|
|
|
// 1 - - 1 and 1 + + 1 should not become 1--1 or 1++1 |
25
|
|
|
'- -', |
26
|
|
|
'+ +', |
27
|
|
|
// $a - --$b should not become $a---$b |
28
|
|
|
'- --', |
29
|
|
|
'+ ++' |
30
|
|
|
]; |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* @var bool Whether the source code needs to be reparsed before this stream is handed off to a |
34
|
|
|
* new pass |
35
|
|
|
*/ |
36
|
|
|
public $needsReparsing; |
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* @var integer Current token index |
40
|
|
|
*/ |
41
|
|
|
protected $offset; |
42
|
|
|
|
43
|
|
|
/** |
44
|
|
|
* @var bool Whether some tokens have been removed |
45
|
|
|
*/ |
46
|
|
|
protected $sparse; |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* @var array<array|string> List of tokens |
50
|
|
|
*/ |
51
|
|
|
protected $tokens; |
52
|
|
|
|
53
|
|
|
/** |
54
|
|
|
* Constructor |
55
|
|
|
* |
56
|
|
|
* @param string $src Original source code |
57
|
|
|
*/ |
58
|
26 |
|
public function __construct($src) |
59
|
|
|
{ |
60
|
26 |
|
$this->parse($src); |
61
|
26 |
|
} |
62
|
|
|
|
63
|
|
|
/** |
64
|
|
|
* Test whether current token can be removed safely |
65
|
|
|
* |
66
|
|
|
* @return bool |
67
|
|
|
*/ |
68
|
2 |
|
public function canRemoveCurrentToken() |
69
|
|
|
{ |
70
|
2 |
|
$prevToken = $this->lookbehind(); |
71
|
2 |
|
$nextToken = $this->lookahead(); |
72
|
|
|
|
73
|
2 |
|
if ($prevToken === false || $nextToken === false) |
74
|
|
|
{ |
75
|
|
|
return true; |
76
|
|
|
} |
77
|
|
|
|
78
|
2 |
|
if (is_array($prevToken)) |
79
|
|
|
{ |
80
|
2 |
|
if ($prevToken[0] === T_COMMENT && substr($prevToken[1], 0, 2) === '//') |
81
|
|
|
{ |
82
|
1 |
|
return false; |
83
|
|
|
} |
84
|
|
|
|
85
|
1 |
|
$prevToken = $prevToken[1]; |
86
|
|
|
} |
87
|
1 |
|
if (is_array($nextToken)) |
88
|
|
|
{ |
89
|
1 |
|
$nextToken = $nextToken[1]; |
90
|
|
|
} |
91
|
|
|
|
92
|
1 |
|
$str = $prevToken . ' ' . $nextToken; |
93
|
1 |
|
if (in_array($str, $this->excludeExact, true)) |
94
|
|
|
{ |
95
|
|
|
return false; |
96
|
|
|
} |
97
|
|
|
|
98
|
1 |
|
$delimiters = "\t\n\r !\"#$%&'()*+,-./:;<=>?@[\\]^`{|}~"; |
99
|
1 |
|
$prevChar = substr($prevToken, -1); |
100
|
1 |
|
$nextChar = $nextToken[0]; |
101
|
|
|
|
102
|
1 |
|
return (strpos($delimiters, $prevChar) !== false || strpos($delimiters, $nextChar) !== false); |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
/** |
106
|
|
|
* Test whether current token is the given type of token |
107
|
|
|
* |
108
|
|
|
* @param integer $tokenValue |
109
|
|
|
* @return bool |
110
|
|
|
*/ |
111
|
|
|
public function is($tokenValue) |
112
|
|
|
{ |
113
|
|
|
return ($this->tokens[$this->offset][0] === $tokenValue); |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
/** |
117
|
|
|
* Test whether current token is any of the given types of token |
118
|
|
|
* |
119
|
|
|
* @param integer[] $tokenValues |
120
|
|
|
* @return bool |
121
|
|
|
*/ |
122
|
|
|
public function isAny(array $tokenValues) |
123
|
|
|
{ |
124
|
|
|
return in_array($this->tokens[$this->offset][0], $tokenValues, true); |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
/** |
128
|
|
|
* Test whether there's a token at given offset |
129
|
|
|
* |
130
|
|
|
* @param integer $offset |
131
|
|
|
* @return bool |
132
|
|
|
*/ |
133
|
2 |
|
public function offsetExists($offset) |
134
|
|
|
{ |
135
|
2 |
|
return isset($this->tokens[$offset]); |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
/** |
139
|
|
|
* Return the token stored at given offset |
140
|
|
|
* |
141
|
|
|
* @param integer $offset |
142
|
|
|
* @return array|string |
143
|
|
|
*/ |
144
|
2 |
|
public function offsetGet($offset) |
145
|
|
|
{ |
146
|
2 |
|
return $this->tokens[$offset]; |
147
|
|
|
} |
148
|
|
|
|
149
|
|
|
/** |
150
|
|
|
* Replace the token stored at given offset |
151
|
|
|
* |
152
|
|
|
* @param integer $offset |
153
|
|
|
* @param array|string $token |
154
|
|
|
* @return void |
155
|
|
|
*/ |
156
|
1 |
|
public function offsetSet($offset, $token) |
157
|
|
|
{ |
158
|
1 |
|
$this->tokens[$offset] = $token; |
159
|
1 |
|
} |
160
|
|
|
|
161
|
|
|
/** |
162
|
|
|
* Remove the token stored at given offset |
163
|
|
|
* |
164
|
|
|
* @return void |
165
|
|
|
*/ |
166
|
1 |
|
public function offsetUnset($offset) |
167
|
|
|
{ |
168
|
1 |
|
unset($this->tokens[$offset]); |
169
|
1 |
|
$this->sparse = true; |
170
|
1 |
|
} |
171
|
|
|
|
172
|
|
|
/** |
173
|
|
|
* Return the current token |
174
|
|
|
* |
175
|
|
|
* @return array|string |
176
|
|
|
*/ |
177
|
7 |
|
public function current() |
178
|
|
|
{ |
179
|
7 |
|
return $this->tokens[$this->offset]; |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
/** |
183
|
|
|
* Get current token's text |
184
|
|
|
* |
185
|
|
|
* @return string |
186
|
|
|
*/ |
187
|
|
|
public function currentText() |
188
|
|
|
{ |
189
|
|
|
return (is_array($this->tokens[$this->offset])) ? $this->tokens[$this->offset][1] : $this->tokens[$this->offset]; |
190
|
|
|
} |
191
|
|
|
|
192
|
|
|
/** |
193
|
|
|
* Test whether current token is noise (whitespace or comment) |
194
|
|
|
* |
195
|
|
|
* @return bool |
196
|
|
|
*/ |
197
|
9 |
|
public function isNoise() |
198
|
|
|
{ |
199
|
9 |
|
return in_array($this->tokens[$this->offset][0], [T_COMMENT, T_DOC_COMMENT, T_WHITESPACE], true); |
200
|
|
|
} |
201
|
|
|
|
202
|
|
|
/** |
203
|
|
|
* Return the offset of current token |
204
|
|
|
* |
205
|
|
|
* @return integer |
206
|
|
|
*/ |
207
|
11 |
|
public function key() |
208
|
|
|
{ |
209
|
11 |
|
return $this->offset; |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
/** |
213
|
|
|
* Peek at the next token |
214
|
|
|
* |
215
|
|
|
* @return array|string|false |
216
|
|
|
*/ |
217
|
2 |
View Code Duplication |
public function lookahead() |
218
|
|
|
{ |
219
|
2 |
|
$i = $this->offset; |
220
|
2 |
|
while (++$i < $this->cnt) |
221
|
|
|
{ |
222
|
2 |
|
if (isset($this->tokens[$i])) |
223
|
|
|
{ |
224
|
2 |
|
return $this->tokens[$i]; |
225
|
|
|
} |
226
|
|
|
} |
227
|
|
|
|
228
|
|
|
return false; |
229
|
|
|
} |
230
|
|
|
|
231
|
|
|
/** |
232
|
|
|
* Peek at the previous token |
233
|
|
|
* |
234
|
|
|
* @return array|string|false |
235
|
|
|
*/ |
236
|
2 |
View Code Duplication |
public function lookbehind() |
237
|
|
|
{ |
238
|
2 |
|
$i = $this->offset; |
239
|
2 |
|
while (--$i >= 0) |
240
|
|
|
{ |
241
|
2 |
|
if (isset($this->tokens[$i])) |
242
|
|
|
{ |
243
|
2 |
|
return $this->tokens[$i]; |
244
|
|
|
} |
245
|
|
|
} |
246
|
|
|
|
247
|
|
|
return false; |
248
|
|
|
} |
249
|
|
|
|
250
|
|
|
/** |
251
|
|
|
* Move to the next token in the stream |
252
|
|
|
* |
253
|
|
|
* @return void |
254
|
|
|
*/ |
255
|
7 |
|
public function next() |
256
|
|
|
{ |
257
|
7 |
|
while (++$this->offset < $this->cnt && !isset($this->tokens[$this->offset])); |
258
|
7 |
|
} |
259
|
|
|
|
260
|
|
|
/** |
261
|
|
|
* Normalize the whitespace at the end of single-line comments |
262
|
|
|
* |
263
|
|
|
* Will remove the newline at the end of single-line comments and put it either in the next token |
264
|
|
|
* if it's a T_WHITESPACE or it will insert a T_WHITESPACE otherwise. |
265
|
|
|
* |
266
|
|
|
* @return void |
267
|
|
|
*/ |
268
|
26 |
|
protected function normalizeSingleLineComments() |
269
|
|
|
{ |
270
|
26 |
|
$keys = []; |
271
|
26 |
|
foreach ($this->tokens as $k => $token) |
272
|
|
|
{ |
273
|
|
|
// Remove the newline at the end of comments to put it in the next token |
274
|
26 |
|
if ($token[0] === T_COMMENT && $token[1][1] === '/') |
275
|
|
|
{ |
276
|
12 |
|
$this->tokens[$k][1] = rtrim($token[1], "\n"); |
277
|
26 |
|
$keys[] = $k + 1; |
278
|
|
|
} |
279
|
|
|
} |
280
|
|
|
|
281
|
26 |
|
foreach (array_reverse($keys) as $k) |
282
|
|
|
{ |
283
|
12 |
|
if (isset($this->tokens[$k]) && $this->tokens[$k][0] === T_WHITESPACE) |
284
|
|
|
{ |
285
|
5 |
|
$this->tokens[$k][1] = "\n" . $this->tokens[$k][1]; |
286
|
|
|
} |
287
|
|
|
else |
288
|
|
|
{ |
289
|
12 |
|
array_splice($this->tokens, $k, 0, [[T_WHITESPACE, "\n"]]); |
290
|
|
|
} |
291
|
|
|
} |
292
|
26 |
|
} |
293
|
|
|
|
294
|
|
|
/** |
295
|
|
|
* Parse/tokenize given PHP source |
296
|
|
|
* |
297
|
|
|
* @param string $src |
298
|
|
|
* @return void |
299
|
|
|
*/ |
300
|
26 |
|
protected function parse($src) |
301
|
|
|
{ |
302
|
26 |
|
$this->tokens = token_get_all($src); |
303
|
26 |
|
foreach ($this->tokens as $k => &$token) |
304
|
|
|
{ |
305
|
26 |
|
if (is_array($token)) |
306
|
|
|
{ |
307
|
26 |
|
unset($token[2]); |
308
|
|
|
} |
309
|
|
|
} |
310
|
26 |
|
$this->normalizeSingleLineComments(); |
311
|
|
|
|
312
|
26 |
|
$this->cnt = count($this->tokens); |
313
|
26 |
|
$this->offset = 0; |
314
|
26 |
|
$this->needsReparsing = false; |
315
|
26 |
|
$this->sparse = false; |
316
|
26 |
|
} |
317
|
|
|
|
318
|
|
|
/** |
319
|
|
|
* Move to the previous token in the stream |
320
|
|
|
* |
321
|
|
|
* @return void |
322
|
|
|
*/ |
323
|
|
|
public function previous() |
324
|
|
|
{ |
325
|
|
|
while ($this->offset > 0) |
326
|
|
|
{ |
327
|
|
|
--$this->offset; |
328
|
|
|
if (isset($this->tokens[$this->offset])) |
329
|
|
|
{ |
330
|
|
|
break; |
331
|
|
|
} |
332
|
|
|
} |
333
|
|
|
} |
334
|
|
|
|
335
|
|
|
/** |
336
|
|
|
* Remove current token |
337
|
|
|
* |
338
|
|
|
* @return void |
339
|
|
|
*/ |
340
|
|
|
public function remove() |
341
|
|
|
{ |
342
|
|
|
$this->offsetUnset($this->offset); |
343
|
|
|
} |
344
|
|
|
|
345
|
|
|
/** |
346
|
|
|
* Replace current token |
347
|
|
|
* |
348
|
|
|
* @param array|string $token Token replacement |
349
|
|
|
* @return void |
350
|
|
|
*/ |
351
|
|
|
public function replace($token) |
352
|
|
|
{ |
353
|
|
|
$this->offsetSet($this->offset, $token); |
354
|
|
|
} |
355
|
|
|
|
356
|
|
|
/** |
357
|
|
|
* Reset this stream |
358
|
|
|
* |
359
|
|
|
* @return void |
360
|
|
|
*/ |
361
|
8 |
|
public function reset() |
362
|
|
|
{ |
363
|
8 |
|
$this->offset = 0; |
364
|
8 |
|
if ($this->needsReparsing) |
365
|
|
|
{ |
366
|
|
|
$this->parse($this->serialize()); |
367
|
|
|
} |
368
|
8 |
|
elseif ($this->sparse) |
369
|
|
|
{ |
370
|
|
|
$this->tokens = array_values($this->tokens); |
371
|
|
|
} |
372
|
8 |
|
} |
373
|
|
|
|
374
|
|
|
/** |
375
|
|
|
* Rewind/reset this stream |
376
|
|
|
* |
377
|
|
|
* @return void |
378
|
|
|
*/ |
379
|
1 |
|
public function rewind() |
380
|
|
|
{ |
381
|
1 |
|
$this->reset(); |
382
|
1 |
|
} |
383
|
|
|
|
384
|
|
|
/** |
385
|
|
|
* Move the stream to given offset |
386
|
|
|
* |
387
|
|
|
* @param integer $offset |
388
|
|
|
* @return void |
389
|
|
|
*/ |
390
|
8 |
|
public function seek($offset) |
391
|
|
|
{ |
392
|
8 |
|
$this->offset = $offset; |
393
|
8 |
|
} |
394
|
|
|
|
395
|
|
|
/** |
396
|
|
|
* Serialize these tokens back to source code |
397
|
|
|
* |
398
|
|
|
* @return string |
399
|
|
|
*/ |
400
|
|
|
public function serialize() |
401
|
|
|
{ |
402
|
|
|
$src = ''; |
403
|
|
|
foreach ($this->tokens as $token) |
404
|
|
|
{ |
405
|
|
|
$src .= (is_array($token)) ? $token[1] : $token; |
406
|
|
|
} |
407
|
|
|
|
408
|
|
|
return $src; |
409
|
|
|
} |
410
|
|
|
|
411
|
|
|
/** |
412
|
|
|
* Skip all whitespace, comments and docblocks starting at current offset |
413
|
|
|
* |
414
|
|
|
* @return void |
415
|
|
|
*/ |
416
|
5 |
|
public function skipNoise() |
417
|
|
|
{ |
418
|
5 |
|
while ($this->valid()) |
419
|
|
|
{ |
420
|
5 |
|
if (!$this->isNoise()) |
421
|
|
|
{ |
422
|
5 |
|
break; |
423
|
|
|
} |
424
|
4 |
|
$this->next(); |
425
|
|
|
} |
426
|
5 |
|
} |
427
|
|
|
|
428
|
|
|
/** |
429
|
|
|
* Skip all whitespace starting at current offset |
430
|
|
|
* |
431
|
|
|
* @return void |
432
|
|
|
*/ |
433
|
2 |
|
public function skipWhitespace() |
434
|
|
|
{ |
435
|
2 |
|
while ($this->offset < $this->cnt) |
436
|
|
|
{ |
437
|
2 |
|
if ($this->tokens[$this->offset][0] !== T_WHITESPACE) |
438
|
|
|
{ |
439
|
2 |
|
break; |
440
|
|
|
} |
441
|
1 |
|
++$this->offset; |
442
|
|
|
} |
443
|
2 |
|
} |
444
|
|
|
|
445
|
|
|
/** |
446
|
|
|
* Iterate through tokens until the stream reaches a token of given value or the end of stream |
447
|
|
|
* |
448
|
|
|
* @param integer $tokenValue The target value, e.g. T_ELSE |
449
|
|
|
* @return bool Whether a matching token was found |
450
|
|
|
*/ |
451
|
10 |
View Code Duplication |
public function skipTo($tokenValue) |
452
|
|
|
{ |
453
|
10 |
|
while (++$this->offset < $this->cnt) |
454
|
|
|
{ |
455
|
10 |
|
if (isset($this->tokens[$this->offset][0]) && $this->tokens[$this->offset][0] === $tokenValue) |
456
|
|
|
{ |
457
|
7 |
|
return true; |
458
|
|
|
} |
459
|
|
|
} |
460
|
|
|
|
461
|
9 |
|
return false; |
462
|
|
|
} |
463
|
|
|
|
464
|
|
|
/** |
465
|
|
|
* Iterate through tokens until the stream reaches given token or the end of stream |
466
|
|
|
* |
467
|
|
|
* @param array|string $token The target token, either a string or a [tokenValue, string] pair |
468
|
|
|
* @return bool Whether a matching token was found |
469
|
|
|
*/ |
470
|
2 |
View Code Duplication |
public function skipToToken($token) |
471
|
|
|
{ |
472
|
2 |
|
while (++$this->offset < $this->cnt) |
473
|
|
|
{ |
474
|
2 |
|
if (isset($this->tokens[$this->offset]) && $this->tokens[$this->offset] === $token) |
475
|
|
|
{ |
476
|
2 |
|
return true; |
477
|
|
|
} |
478
|
|
|
} |
479
|
|
|
|
480
|
|
|
return false; |
481
|
|
|
} |
482
|
|
|
|
483
|
|
|
/** |
484
|
|
|
* Test whether the stream is at a valid offset |
485
|
|
|
* |
486
|
|
|
* @return bool |
487
|
|
|
*/ |
488
|
8 |
|
public function valid() |
489
|
|
|
{ |
490
|
8 |
|
return ($this->offset < $this->cnt); |
491
|
|
|
} |
492
|
|
|
} |