1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
/** |
6
|
|
|
* This file is part of phpDocumentor. |
7
|
|
|
* |
8
|
|
|
* For the full copyright and license information, please view the LICENSE |
9
|
|
|
* file that was distributed with this source code. |
10
|
|
|
* |
11
|
|
|
* Many thanks to webmozart by providing the original code in webmozart/glob |
12
|
|
|
* |
13
|
|
|
* @link https://github.com/webmozart/glob/blob/master/src/Glob.php |
14
|
|
|
* @link http://phpdoc.org |
15
|
|
|
*/ |
16
|
|
|
|
17
|
|
|
namespace Flyfinder\Specification; |
18
|
|
|
|
19
|
|
|
use InvalidArgumentException; |
20
|
|
|
use function array_slice; |
21
|
|
|
use function count; |
22
|
|
|
use function explode; |
23
|
|
|
use function implode; |
24
|
|
|
use function max; |
25
|
|
|
use function min; |
26
|
|
|
use function preg_match; |
27
|
|
|
use function rtrim; |
28
|
|
|
use function sprintf; |
29
|
|
|
use function strlen; |
30
|
|
|
use function strpos; |
31
|
|
|
use function substr; |
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* Glob specification class |
35
|
|
|
* |
36
|
|
|
* @psalm-immutable |
37
|
|
|
*/ |
38
|
|
|
final class Glob extends CompositeSpecification |
39
|
|
|
{ |
40
|
|
|
/** @var string */ |
41
|
|
|
private $regex; |
42
|
|
|
|
43
|
|
|
/** |
44
|
|
|
* The "static prefix" is the part of the glob up to the first wildcard "*". |
45
|
|
|
* If the glob does not contain wildcards, the full glob is returned. |
46
|
|
|
* |
47
|
|
|
* @var string |
48
|
|
|
*/ |
49
|
|
|
private $staticPrefix; |
50
|
|
|
|
51
|
|
|
/** |
52
|
|
|
* The "bounded prefix" is the part of the glob up to the first recursive wildcard "**". |
53
|
|
|
* It is the longest prefix for which the number of directory segments in the partial match |
54
|
|
|
* is known. If the glob does not contain the recursive wildcard "**", the full glob is returned. |
55
|
|
|
* |
56
|
|
|
* @var string |
57
|
|
|
*/ |
58
|
|
|
private $boundedPrefix; |
59
|
|
|
|
60
|
|
|
/** |
61
|
|
|
* The "total prefix" is the part of the glob before the trailing catch-all wildcard sequence if the glob |
62
|
|
|
* ends with one, otherwise null. It is needed for implementing the A-quantifier pruning hint. |
63
|
|
|
* |
64
|
|
|
* @var string|null |
65
|
|
|
*/ |
66
|
|
|
private $totalPrefix; |
67
|
|
|
|
68
|
|
|
public function __construct(string $glob) |
69
|
|
|
{ |
70
|
|
|
$this->regex = self::toRegEx($glob); |
71
|
|
|
$this->staticPrefix = self::getStaticPrefix($glob); |
72
|
|
|
$this->boundedPrefix = self::getBoundedPrefix($glob); |
73
|
|
|
$this->totalPrefix = self::getTotalPrefix($glob); |
74
|
|
|
} |
75
|
|
|
|
76
|
|
|
/** |
77
|
|
|
* @inheritDoc |
78
|
|
|
*/ |
79
|
|
|
public function isSatisfiedBy(array $value) : bool |
80
|
|
|
{ |
81
|
|
|
//Flysystem paths are not absolute, so make it that way. |
82
|
|
|
$path = '/' . $value['path']; |
83
|
|
|
if (strpos($path, $this->staticPrefix) !== 0) { |
84
|
|
|
return false; |
85
|
|
|
} |
86
|
|
|
|
87
|
|
|
if (preg_match($this->regex, $path)) { |
88
|
|
|
return true; |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
return false; |
92
|
|
|
} |
93
|
|
|
|
94
|
|
|
/** |
95
|
|
|
* Returns the static prefix of a glob. |
96
|
|
|
* |
97
|
|
|
* The "static prefix" is the part of the glob up to the first wildcard "*". |
98
|
|
|
* If the glob does not contain wildcards, the full glob is returned. |
99
|
|
|
* |
100
|
|
|
* @param string $glob The canonical glob. The glob should contain forward |
101
|
|
|
* slashes as directory separators only. It must not |
102
|
|
|
* contain any "." or ".." segments. |
103
|
|
|
* |
104
|
|
|
* @return string The static prefix of the glob. |
105
|
|
|
* |
106
|
|
|
* @psalm-pure |
107
|
|
|
*/ |
108
|
|
|
private static function getStaticPrefix(string $glob) : string |
109
|
|
|
{ |
110
|
|
|
self::assertValidGlob($glob); |
111
|
|
|
$prefix = ''; |
112
|
|
|
$length = strlen($glob); |
113
|
|
|
for ($i = 0; $i < $length; ++$i) { |
114
|
|
|
$c = $glob[$i]; |
115
|
|
|
switch ($c) { |
116
|
|
|
case '/': |
117
|
|
|
$prefix .= '/'; |
118
|
|
|
if (self::isRecursiveWildcard($glob, $i)) { |
119
|
|
|
break 2; |
120
|
|
|
} |
121
|
|
|
break; |
122
|
|
|
case '*': |
123
|
|
|
case '?': |
124
|
|
|
case '{': |
125
|
|
|
case '[': |
126
|
|
|
break 2; |
127
|
|
|
case '\\': |
128
|
|
|
[$unescaped, $consumedChars] = self::scanBackslashSequence($glob, $i); |
|
|
|
|
129
|
|
|
$prefix .= $unescaped; |
130
|
|
|
$i += $consumedChars; |
131
|
|
|
break; |
132
|
|
|
default: |
133
|
|
|
$prefix .= $c; |
134
|
|
|
break; |
135
|
|
|
} |
136
|
|
|
} |
137
|
|
|
return $prefix; |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
private static function getBoundedPrefix(string $glob) : string |
141
|
|
|
{ |
142
|
|
|
self::assertValidGlob($glob); |
143
|
|
|
$prefix = ''; |
144
|
|
|
$length = strlen($glob); |
145
|
|
|
|
146
|
|
|
for ($i = 0; $i < $length; ++$i) { |
147
|
|
|
$c = $glob[$i]; |
148
|
|
|
switch ($c) { |
149
|
|
|
case '/': |
150
|
|
|
$prefix .= '/'; |
151
|
|
|
if (self::isRecursiveWildcard($glob, $i)) { |
152
|
|
|
break 2; |
153
|
|
|
} |
154
|
|
|
break; |
155
|
|
|
case '\\': |
156
|
|
|
[$unescaped, $consumedChars] = self::scanBackslashSequence($glob, $i); |
|
|
|
|
157
|
|
|
$prefix .= $unescaped; |
158
|
|
|
$i += $consumedChars; |
159
|
|
|
break; |
160
|
|
|
default: |
161
|
|
|
$prefix .= $c; |
162
|
|
|
break; |
163
|
|
|
} |
164
|
|
|
} |
165
|
|
|
return $prefix; |
166
|
|
|
} |
167
|
|
|
|
168
|
|
|
private static function getTotalPrefix(string $glob) : ?string |
169
|
|
|
{ |
170
|
|
|
self::assertValidGlob($glob); |
171
|
|
|
$matches = []; |
172
|
|
|
return preg_match('~(?<!\\\\)/\\*\\*(?:/\\*\\*?)+$~', $glob, $matches) |
173
|
|
|
? substr($glob, 0, strlen($glob)-strlen($matches[0])) |
174
|
|
|
: null; |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
/** |
178
|
|
|
* @return mixed[] |
179
|
|
|
* |
180
|
|
|
* @psalm-return array{0: string, 1:int} |
181
|
|
|
* @psalm-pure |
182
|
|
|
*/ |
183
|
|
|
private static function scanBackslashSequence(string $glob, int $offset) : array |
184
|
|
|
{ |
185
|
|
|
$startOffset = $offset; |
186
|
|
|
$result = ''; |
187
|
|
|
switch ($c = $glob[$offset + 1] ?? '') { |
188
|
|
|
case '*': |
189
|
|
|
case '?': |
190
|
|
|
case '{': |
191
|
|
|
case '}': |
192
|
|
|
case '[': |
193
|
|
|
case ']': |
194
|
|
|
case '-': |
195
|
|
|
case '^': |
196
|
|
|
case '$': |
197
|
|
|
case '~': |
198
|
|
|
case '\\': |
199
|
|
|
$result .= $c; |
200
|
|
|
++$offset; |
201
|
|
|
break; |
202
|
|
|
default: |
203
|
|
|
$result .= '\\'; |
204
|
|
|
} |
205
|
|
|
return [$result, $offset - $startOffset]; |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
/** |
209
|
|
|
* Asserts that glob is well formed |
210
|
|
|
* |
211
|
|
|
* @psalm-pure |
212
|
|
|
*/ |
213
|
|
|
private static function assertValidGlob(string $glob) : void |
214
|
|
|
{ |
215
|
|
|
if (strpos($glob, '/') !== 0 && strpos($glob, '://') === false) { |
216
|
|
|
throw new InvalidArgumentException(sprintf( |
217
|
|
|
'The glob "%s" is not absolute and not a URI.', |
218
|
|
|
$glob |
219
|
|
|
)); |
220
|
|
|
} |
221
|
|
|
} |
222
|
|
|
|
223
|
|
|
/** |
224
|
|
|
* Checks if the current position the glob is start of a Recursive directory wildcard |
225
|
|
|
* |
226
|
|
|
* @psalm-pure |
227
|
|
|
*/ |
228
|
|
|
private static function isRecursiveWildcard(string $glob, int $i) : bool |
229
|
|
|
{ |
230
|
|
|
return isset($glob[$i + 3]) && $glob[$i + 1] . $glob[$i + 2] . $glob[$i + 3] === '**/'; |
231
|
|
|
} |
232
|
|
|
|
233
|
|
|
/** |
234
|
|
|
* Converts a glob to a regular expression. |
235
|
|
|
* |
236
|
|
|
* @param string $glob The canonical glob. The glob should contain forward |
237
|
|
|
* slashes as directory separators only. It must not |
238
|
|
|
* contain any "." or ".." segments. |
239
|
|
|
* |
240
|
|
|
* @return string The regular expression for matching the glob. |
241
|
|
|
* |
242
|
|
|
* @psalm-pure |
243
|
|
|
*/ |
244
|
|
|
private static function toRegEx(string $glob) : string |
245
|
|
|
{ |
246
|
|
|
$delimiter = '~'; |
247
|
|
|
$inSquare = false; |
248
|
|
|
$curlyLevels = 0; |
249
|
|
|
$regex = ''; |
250
|
|
|
$length = strlen($glob); |
251
|
|
|
for ($i = 0; $i < $length; ++$i) { |
252
|
|
|
$c = $glob[$i]; |
253
|
|
|
switch ($c) { |
254
|
|
|
case '.': |
255
|
|
|
case '(': |
256
|
|
|
case ')': |
257
|
|
|
case '|': |
258
|
|
|
case '+': |
259
|
|
|
case '^': |
260
|
|
|
case '$': |
261
|
|
|
case $delimiter: |
262
|
|
|
$regex .= '\\' . $c; |
263
|
|
|
break; |
264
|
|
|
case '/': |
265
|
|
|
if (self::isRecursiveWildcard($glob, $i)) { |
266
|
|
|
$regex .= '/([^/]+/)*'; |
267
|
|
|
$i += 3; |
268
|
|
|
} else { |
269
|
|
|
$regex .= '/'; |
270
|
|
|
} |
271
|
|
|
break; |
272
|
|
|
case '*': |
273
|
|
|
$regex .= '[^/]*'; |
274
|
|
|
break; |
275
|
|
|
case '?': |
276
|
|
|
$regex .= '.'; |
277
|
|
|
break; |
278
|
|
|
case '{': |
279
|
|
|
$regex .= '('; |
280
|
|
|
++$curlyLevels; |
281
|
|
|
break; |
282
|
|
|
case '}': |
283
|
|
|
if ($curlyLevels > 0) { |
284
|
|
|
$regex .= ')'; |
285
|
|
|
--$curlyLevels; |
286
|
|
|
} else { |
287
|
|
|
$regex .= '}'; |
288
|
|
|
} |
289
|
|
|
break; |
290
|
|
|
case ',': |
291
|
|
|
$regex .= $curlyLevels > 0 ? '|' : ','; |
292
|
|
|
break; |
293
|
|
|
case '[': |
294
|
|
|
$regex .= '['; |
295
|
|
|
$inSquare = true; |
296
|
|
|
if (isset($glob[$i + 1]) && $glob[$i + 1] === '^') { |
297
|
|
|
$regex .= '^'; |
298
|
|
|
++$i; |
299
|
|
|
} |
300
|
|
|
break; |
301
|
|
|
case ']': |
302
|
|
|
$regex .= $inSquare ? ']' : '\\]'; |
303
|
|
|
$inSquare = false; |
304
|
|
|
break; |
305
|
|
|
case '-': |
306
|
|
|
$regex .= $inSquare ? '-' : '\\-'; |
307
|
|
|
break; |
308
|
|
|
case '\\': |
309
|
|
|
if (isset($glob[$i + 1])) { |
310
|
|
|
switch ($glob[$i + 1]) { |
311
|
|
|
case '*': |
312
|
|
|
case '?': |
313
|
|
|
case '{': |
314
|
|
|
case '}': |
315
|
|
|
case '[': |
316
|
|
|
case ']': |
317
|
|
|
case '-': |
318
|
|
|
case '^': |
319
|
|
|
case '$': |
320
|
|
|
case '~': |
321
|
|
|
case '\\': |
322
|
|
|
$regex .= '\\' . $glob[$i + 1]; |
323
|
|
|
++$i; |
324
|
|
|
break; |
325
|
|
|
default: |
326
|
|
|
$regex .= '\\\\'; |
327
|
|
|
} |
328
|
|
|
} |
329
|
|
|
break; |
330
|
|
|
default: |
331
|
|
|
$regex .= $c; |
332
|
|
|
break; |
333
|
|
|
} |
334
|
|
|
} |
335
|
|
|
if ($inSquare) { |
336
|
|
|
throw new InvalidArgumentException(sprintf( |
337
|
|
|
'Invalid glob: missing ] in %s', |
338
|
|
|
$glob |
339
|
|
|
)); |
340
|
|
|
} |
341
|
|
|
if ($curlyLevels > 0) { |
342
|
|
|
throw new InvalidArgumentException(sprintf( |
343
|
|
|
'Invalid glob: missing } in %s', |
344
|
|
|
$glob |
345
|
|
|
)); |
346
|
|
|
} |
347
|
|
|
return $delimiter . '^' . $regex . '$' . $delimiter; |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
/** @inheritDoc */ |
351
|
|
|
public function canBeSatisfiedBySomethingBelow(array $value) : bool |
352
|
|
|
{ |
353
|
|
|
$valueSegments = explode('/', '/' . $value['path']); |
354
|
|
|
$boundedPrefixSegments = explode('/', rtrim($this->boundedPrefix, '/')); |
355
|
|
|
$howManySegmentsToConsider = min(count($valueSegments), count($boundedPrefixSegments)); |
356
|
|
|
$boundedPrefixGlob = implode('/', array_slice($boundedPrefixSegments, 0, $howManySegmentsToConsider)); |
357
|
|
|
$valuePathPrefix = implode('/', array_slice($valueSegments, 1, max($howManySegmentsToConsider-1, 0))); |
358
|
|
|
$prefixValue = $value; |
359
|
|
|
$prefixValue['path'] = $valuePathPrefix; |
360
|
|
|
$spec = new Glob($boundedPrefixGlob); |
361
|
|
|
return $spec->isSatisfiedBy($prefixValue); |
362
|
|
|
} |
363
|
|
|
|
364
|
|
|
/** @inheritDoc */ |
365
|
|
|
public function willBeSatisfiedByEverythingBelow(array $value) : bool |
366
|
|
|
{ |
367
|
|
|
if ($this->totalPrefix === null) { |
368
|
|
|
return false; |
369
|
|
|
} |
370
|
|
|
$spec = new Glob(rtrim($this->totalPrefix, '/') . '/**/*'); |
371
|
|
|
$terminatedValue = $value; |
372
|
|
|
$terminatedValue['path'] = rtrim($terminatedValue['path'], '/') . '/x/x'; |
373
|
|
|
return $spec->isSatisfiedBy($terminatedValue); |
374
|
|
|
} |
375
|
|
|
} |
376
|
|
|
|
This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug.