1
|
|
|
<?php declare(strict_types=1); |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @package s9e\RegexpBuilder |
5
|
|
|
* @copyright Copyright (c) 2016-2022 The s9e authors |
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
7
|
|
|
*/ |
8
|
|
|
namespace s9e\RegexpBuilder; |
9
|
|
|
|
10
|
|
|
use InvalidArgumentException; |
11
|
|
|
use s9e\RegexpBuilder\Input\InputInterface; |
12
|
|
|
|
13
|
|
|
class MetaCharacters |
14
|
|
|
{ |
15
|
|
|
/** |
16
|
|
|
* @const Bit value that indicates whether a meta-character represents a single character usable |
17
|
|
|
* in a character class |
18
|
|
|
*/ |
19
|
|
|
const IS_CHAR = 1; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* @const Bit value that indicates whether a meta-character represents a quantifiable expression |
23
|
|
|
*/ |
24
|
|
|
const IS_QUANTIFIABLE = 2; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* @var array Map of meta values and the expression they represent |
28
|
|
|
*/ |
29
|
|
|
protected $exprs = []; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* @var InputInterface |
33
|
|
|
*/ |
34
|
|
|
protected $input; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* @var array Map of meta-characters' codepoints and their value |
38
|
|
|
*/ |
39
|
|
|
protected $meta = []; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* @param InputInterface $input |
43
|
|
|
*/ |
44
|
17 |
|
public function __construct(InputInterface $input) |
45
|
|
|
{ |
46
|
17 |
|
$this->input = $input; |
47
|
17 |
|
} |
48
|
|
|
|
49
|
|
|
/** |
50
|
|
|
* Add a meta-character to the list |
51
|
|
|
* |
52
|
|
|
* @param string $char Meta-character |
53
|
|
|
* @param string $expr Regular expression |
54
|
|
|
* @return void |
55
|
|
|
*/ |
56
|
16 |
|
public function add(string $char, string $expr): void |
57
|
|
|
{ |
58
|
16 |
|
$split = $this->input->split($char); |
59
|
16 |
|
if (count($split) !== 1) |
60
|
|
|
{ |
61
|
1 |
|
throw new InvalidArgumentException('Meta-characters must be represented by exactly one character'); |
62
|
|
|
} |
63
|
15 |
|
if (@preg_match('(' . $expr . ')u', '') === false) |
64
|
|
|
{ |
65
|
1 |
|
throw new InvalidArgumentException("Invalid expression '" . $expr . "'"); |
66
|
|
|
} |
67
|
|
|
|
68
|
14 |
|
$inputValue = $split[0]; |
69
|
14 |
|
$metaValue = $this->computeValue($expr); |
70
|
|
|
|
71
|
14 |
|
$this->exprs[$metaValue] = $expr; |
72
|
14 |
|
$this->meta[$inputValue] = $metaValue; |
73
|
14 |
|
} |
74
|
|
|
|
75
|
|
|
/** |
76
|
|
|
* Get the expression associated with a meta value |
77
|
|
|
* |
78
|
|
|
* @param integer $metaValue |
79
|
|
|
* @return string |
80
|
|
|
*/ |
81
|
2 |
|
public function getExpression(int $metaValue): string |
82
|
|
|
{ |
83
|
2 |
|
if (!isset($this->exprs[$metaValue])) |
84
|
|
|
{ |
85
|
1 |
|
throw new InvalidArgumentException('Invalid meta value ' . $metaValue); |
86
|
|
|
} |
87
|
|
|
|
88
|
1 |
|
return $this->exprs[$metaValue]; |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* Return whether a given value represents a single character usable in a character class |
93
|
|
|
* |
94
|
|
|
* @param integer $value |
95
|
|
|
* @return bool |
96
|
|
|
*/ |
97
|
13 |
|
public static function isChar(int $value): bool |
98
|
|
|
{ |
99
|
13 |
|
return ($value >= 0 || ($value & self::IS_CHAR)); |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
/** |
103
|
|
|
* Return whether a given value represents a quantifiable expression |
104
|
|
|
* |
105
|
|
|
* @param integer $value |
106
|
|
|
* @return bool |
107
|
|
|
*/ |
108
|
13 |
|
public static function isQuantifiable(int $value): bool |
109
|
|
|
{ |
110
|
13 |
|
return ($value >= 0 || ($value & self::IS_QUANTIFIABLE)); |
111
|
|
|
} |
112
|
|
|
|
113
|
|
|
/** |
114
|
|
|
* Replace values from meta-characters in a list of strings with their meta value |
115
|
|
|
* |
116
|
|
|
* @param array[] $strings |
117
|
|
|
* @return array[] |
118
|
|
|
*/ |
119
|
14 |
|
public function replaceMeta(array $strings): array |
120
|
|
|
{ |
121
|
14 |
|
foreach ($strings as &$string) |
122
|
|
|
{ |
123
|
14 |
|
foreach ($string as &$value) |
124
|
|
|
{ |
125
|
14 |
|
if (isset($this->meta[$value])) |
126
|
|
|
{ |
127
|
14 |
|
$value = $this->meta[$value]; |
128
|
|
|
} |
129
|
|
|
} |
130
|
|
|
} |
131
|
|
|
|
132
|
14 |
|
return $strings; |
133
|
|
|
} |
134
|
|
|
|
135
|
|
|
/** |
136
|
|
|
* Compute and return a value for given expression |
137
|
|
|
* |
138
|
|
|
* Values are meant to be a unique negative integer. The least significant bits are used to |
139
|
|
|
* store the expression's properties |
140
|
|
|
* |
141
|
|
|
* @param string $expr Regular expression |
142
|
|
|
* @return integer |
143
|
|
|
*/ |
144
|
14 |
|
protected function computeValue(string $expr): int |
145
|
|
|
{ |
146
|
|
|
$properties = [ |
147
|
14 |
|
self::IS_CHAR => 'exprIsChar', |
148
|
14 |
|
self::IS_QUANTIFIABLE => 'exprIsQuantifiable' |
149
|
|
|
]; |
150
|
14 |
|
$value = (1 + count($this->meta)) * -(2 ** count($properties)); |
151
|
14 |
|
foreach ($properties as $bitValue => $methodName) |
152
|
|
|
{ |
153
|
14 |
|
if ($this->$methodName($expr)) |
154
|
|
|
{ |
155
|
8 |
|
$value |= $bitValue; |
156
|
|
|
} |
157
|
|
|
} |
158
|
|
|
|
159
|
14 |
|
return $value; |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
/** |
163
|
|
|
* Test whether given expression represents a single character usable in a character class |
164
|
|
|
* |
165
|
|
|
* @param string $expr |
166
|
|
|
* @return bool |
167
|
|
|
*/ |
168
|
14 |
|
protected function exprIsChar(string $expr): bool |
169
|
|
|
{ |
170
|
|
|
$regexps = [ |
171
|
|
|
// Escaped literal or escape sequence such as \w but not \R |
172
|
14 |
|
'(^\\\\[adefhnrstvwDHNSVW\\W]$)D', |
173
|
|
|
|
174
|
|
|
// Unicode properties such as \pL or \p{Lu} |
175
|
|
|
'(^\\\\p(?:.|\\{[^}]+\\})$)Di', |
176
|
|
|
|
177
|
|
|
// An escape sequence such as \x1F or \x{2600} |
178
|
|
|
'(^\\\\x(?:[0-9a-f]{2}|\\{[^}]+\\})$)Di' |
179
|
|
|
]; |
180
|
|
|
|
181
|
14 |
|
return $this->matchesAny($expr, $regexps); |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
/** |
185
|
|
|
* Test whether given expression is quantifiable |
186
|
|
|
* |
187
|
|
|
* @param string $expr |
188
|
|
|
* @return bool |
189
|
|
|
*/ |
190
|
14 |
|
protected function exprIsQuantifiable(string $expr): bool |
191
|
|
|
{ |
192
|
|
|
$regexps = [ |
193
|
|
|
// A dot or \R |
194
|
14 |
|
'(^(?:\\.|\\\\R)$)D', |
195
|
|
|
|
196
|
|
|
// A character class |
197
|
|
|
'(^\\[\\^?(?:([^\\\\\\]]|\\\\.)(?:-(?-1))?)++\\]$)D' |
198
|
|
|
]; |
199
|
|
|
|
200
|
14 |
|
return $this->matchesAny($expr, $regexps) || $this->exprIsChar($expr); |
201
|
|
|
} |
202
|
|
|
|
203
|
|
|
/** |
204
|
|
|
* Test whether given expression matches any of the given regexps |
205
|
|
|
* |
206
|
|
|
* @param string $expr |
207
|
|
|
* @param string[] $regexps |
208
|
|
|
* @return bool |
209
|
|
|
*/ |
210
|
14 |
|
protected function matchesAny(string $expr, array $regexps): bool |
211
|
|
|
{ |
212
|
14 |
|
foreach ($regexps as $regexp) |
213
|
|
|
{ |
214
|
14 |
|
if (preg_match($regexp, $expr)) |
215
|
|
|
{ |
216
|
8 |
|
return true; |
217
|
|
|
} |
218
|
|
|
} |
219
|
|
|
|
220
|
14 |
|
return false; |
221
|
|
|
} |
222
|
|
|
} |