1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @package s9e\RegexpBuilder |
5
|
|
|
* @copyright Copyright (c) 2016-2018 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 |
17
|
|
|
*/ |
18
|
|
|
const IS_CHAR = 1; |
19
|
|
|
|
20
|
|
|
/** |
21
|
|
|
* @const Bit value that indicates whether a meta-character represents a quantifiable expression |
22
|
|
|
*/ |
23
|
|
|
const IS_QUANTIFIABLE = 2; |
24
|
|
|
|
25
|
|
|
/** |
26
|
|
|
* @var array Map of meta values and the expression they represent |
27
|
|
|
*/ |
28
|
|
|
protected $exprs = []; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* @var InputInterface |
32
|
|
|
*/ |
33
|
|
|
protected $input; |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* @var array Map of meta-characters' codepoints and their value |
37
|
|
|
*/ |
38
|
|
|
protected $meta = []; |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* @param InputInterface $input |
42
|
|
|
*/ |
43
|
13 |
|
public function __construct(InputInterface $input) |
44
|
|
|
{ |
45
|
13 |
|
$this->input = $input; |
46
|
13 |
|
} |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* Add a meta-character to the list |
50
|
|
|
* |
51
|
|
|
* @param string $char Meta-character |
52
|
|
|
* @param string $expr Regular expression |
53
|
|
|
* @return void |
54
|
|
|
*/ |
55
|
12 |
|
public function add($char, $expr) |
56
|
|
|
{ |
57
|
12 |
|
$split = $this->input->split($char); |
58
|
12 |
|
if (count($split) !== 1) |
59
|
|
|
{ |
60
|
1 |
|
throw new InvalidArgumentException('Meta-characters must be represented by exactly one character'); |
61
|
|
|
} |
62
|
11 |
|
if (@preg_match('(' . $expr . ')u', '') === false) |
63
|
|
|
{ |
64
|
1 |
|
throw new InvalidArgumentException("Invalid expression '" . $expr . "'"); |
65
|
|
|
} |
66
|
|
|
|
67
|
10 |
|
$inputValue = $split[0]; |
68
|
10 |
|
$metaValue = $this->computeValue($expr); |
69
|
|
|
|
70
|
10 |
|
$this->exprs[$metaValue] = $expr; |
71
|
10 |
|
$this->meta[$inputValue] = $metaValue; |
72
|
10 |
|
} |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* Get the expression associated with a meta value |
76
|
|
|
* |
77
|
|
|
* @param integer $metaValue |
78
|
|
|
* @return string |
79
|
|
|
*/ |
80
|
2 |
|
public function getExpression($metaValue) |
81
|
|
|
{ |
82
|
2 |
|
if (!isset($this->exprs[$metaValue])) |
83
|
|
|
{ |
84
|
1 |
|
throw new InvalidArgumentException('Invalid meta value ' . $metaValue); |
85
|
|
|
} |
86
|
|
|
|
87
|
1 |
|
return $this->exprs[$metaValue]; |
88
|
|
|
} |
89
|
|
|
|
90
|
|
|
/** |
91
|
|
|
* Return whether a given value represents a single character |
92
|
|
|
* |
93
|
|
|
* @param integer $value |
94
|
|
|
* @return bool |
95
|
|
|
*/ |
96
|
9 |
|
public static function isChar($value) |
97
|
|
|
{ |
98
|
9 |
|
return ($value >= 0 || ($value & self::IS_CHAR)); |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* Return whether a given value represents a quantifiable expression |
103
|
|
|
* |
104
|
|
|
* @param integer $value |
105
|
|
|
* @return bool |
106
|
|
|
*/ |
107
|
9 |
|
public static function isQuantifiable($value) |
108
|
|
|
{ |
109
|
9 |
|
return ($value >= 0 || ($value & self::IS_QUANTIFIABLE)); |
110
|
|
|
} |
111
|
|
|
|
112
|
|
|
/** |
113
|
|
|
* Replace values from meta-characters in a list of strings with their meta value |
114
|
|
|
* |
115
|
|
|
* @param array[] $strings |
116
|
|
|
* @return array[] |
117
|
|
|
*/ |
118
|
10 |
|
public function replaceMeta(array $strings) |
119
|
|
|
{ |
120
|
10 |
|
foreach ($strings as &$string) |
121
|
|
|
{ |
122
|
10 |
|
foreach ($string as &$value) |
123
|
|
|
{ |
124
|
10 |
|
if (isset($this->meta[$value])) |
125
|
|
|
{ |
126
|
10 |
|
$value = $this->meta[$value]; |
127
|
|
|
} |
128
|
|
|
} |
129
|
|
|
} |
130
|
|
|
|
131
|
10 |
|
return $strings; |
132
|
|
|
} |
133
|
|
|
|
134
|
|
|
/** |
135
|
|
|
* Compute and return a value for given expression |
136
|
|
|
* |
137
|
|
|
* Values are meant to be a unique negative integer. The last 2 bits indicate whether the |
138
|
|
|
* expression is quantifiable and/or represents a single character. |
139
|
|
|
* |
140
|
|
|
* @param string $expr Regular expression |
141
|
|
|
* @return integer |
142
|
|
|
*/ |
143
|
10 |
|
protected function computeValue($expr) |
144
|
|
|
{ |
145
|
10 |
|
$value = (1 + count($this->meta)) * -4; |
146
|
10 |
|
if ($this->exprIsChar($expr)) |
147
|
|
|
{ |
148
|
5 |
|
$value |= self::IS_CHAR; |
149
|
|
|
} |
150
|
10 |
|
if ($this->exprIsQuantifiable($expr)) |
151
|
|
|
{ |
152
|
5 |
|
$value |= self::IS_QUANTIFIABLE; |
153
|
|
|
} |
154
|
|
|
|
155
|
10 |
|
return $value; |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
/** |
159
|
|
|
* Test whether given expression represents a single character usable in a character class |
160
|
|
|
* |
161
|
|
|
* @param string $expr |
162
|
|
|
* @return bool |
163
|
|
|
*/ |
164
|
10 |
|
protected function exprIsChar($expr) |
165
|
|
|
{ |
166
|
|
|
$regexps = [ |
167
|
|
|
// A dot |
168
|
10 |
|
'(^\\.$)D', |
169
|
|
|
|
170
|
|
|
// Escaped literal or escape sequence such as \w but not \R |
171
|
|
|
'(^\\\\[adefhnrstvwDHNSVW\\W]$)D', |
172
|
|
|
|
173
|
|
|
// Unicode properties such as \pL or \p{Lu} |
174
|
|
|
'(^\\\\p(?:.|\\{[^}]+\\})$)Di', |
175
|
|
|
|
176
|
|
|
// An escape sequence such as \x1F or \x{2600} |
177
|
|
|
'(^\\\\x(?:[0-9a-f]{2}|\\{[^}]+\\})$)Di', |
178
|
|
|
|
179
|
|
|
// A character class |
180
|
|
|
'(^\\[\\^?(?:([^\\\\\\]]|\\\\.)(?:-(?-1))?)++\\]$)D' |
181
|
|
|
]; |
182
|
10 |
|
foreach ($regexps as $regexp) |
183
|
|
|
{ |
184
|
10 |
|
if (preg_match($regexp, $expr)) |
185
|
|
|
{ |
186
|
10 |
|
return true; |
187
|
|
|
} |
188
|
|
|
} |
189
|
|
|
|
190
|
5 |
|
return false; |
191
|
|
|
} |
192
|
|
|
|
193
|
|
|
/** |
194
|
|
|
* Test whether given expression is quantifiable |
195
|
|
|
* |
196
|
|
|
* @param string $expr |
197
|
|
|
* @return bool |
198
|
|
|
*/ |
199
|
10 |
|
protected function exprIsQuantifiable($expr) |
200
|
|
|
{ |
201
|
10 |
|
return $this->exprIsChar($expr); |
202
|
|
|
} |
203
|
|
|
} |