|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\TextFormatter |
|
5
|
|
|
* @copyright Copyright (c) 2010-2019 The s9e Authors |
|
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
|
7
|
|
|
*/ |
|
8
|
|
|
namespace s9e\TextFormatter\Configurator\Helpers; |
|
9
|
|
|
|
|
10
|
|
|
class CharacterClassBuilder |
|
11
|
|
|
{ |
|
12
|
|
|
/** |
|
13
|
|
|
* @var string[] List of characters in the class |
|
14
|
|
|
*/ |
|
15
|
|
|
protected $chars; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* @var string Delimiter used in regexps |
|
19
|
|
|
*/ |
|
20
|
|
|
public $delimiter = '/'; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @var array[] Array of [start, end] pairs where start and end are keys from $this->chars |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $ranges; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Create a character class that matches the given list of characters |
|
29
|
|
|
* |
|
30
|
|
|
* @param string[] $chars |
|
31
|
|
|
* @return string |
|
32
|
|
|
*/ |
|
33
|
|
|
public function fromList(array $chars) |
|
34
|
|
|
{ |
|
35
|
|
|
$this->chars = $chars; |
|
36
|
|
|
|
|
37
|
|
|
$this->unescapeLiterals(); |
|
38
|
|
|
sort($this->chars); |
|
39
|
|
|
$this->storeRanges(); |
|
40
|
|
|
$this->reorderDash(); |
|
41
|
|
|
$this->fixCaret(); |
|
42
|
|
|
$this->escapeSpecialChars(); |
|
43
|
|
|
|
|
44
|
|
|
return $this->buildCharacterClass(); |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Build the character class based |
|
49
|
|
|
* |
|
50
|
|
|
* @return string |
|
51
|
|
|
*/ |
|
52
|
|
|
protected function buildCharacterClass() |
|
53
|
|
|
{ |
|
54
|
|
|
$str = '['; |
|
55
|
|
|
foreach ($this->ranges as list($start, $end)) |
|
56
|
|
|
{ |
|
57
|
|
|
if ($end > $start + 2) |
|
58
|
|
|
{ |
|
59
|
|
|
$str .= $this->chars[$start] . '-' . $this->chars[$end]; |
|
60
|
|
|
} |
|
61
|
|
|
else |
|
62
|
|
|
{ |
|
63
|
|
|
$str .= implode('', array_slice($this->chars, $start, $end + 1 - $start)); |
|
64
|
|
|
} |
|
65
|
|
|
} |
|
66
|
|
|
$str .= ']'; |
|
67
|
|
|
|
|
68
|
|
|
return $str; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Escape special characters in stored chars |
|
73
|
|
|
* |
|
74
|
|
|
* @return void |
|
75
|
|
|
*/ |
|
76
|
|
|
protected function escapeSpecialChars() |
|
77
|
|
|
{ |
|
78
|
|
|
$specialChars = ['\\', ']', $this->delimiter]; |
|
79
|
|
|
foreach (array_intersect($this->chars, $specialChars) as $k => $v) |
|
80
|
|
|
{ |
|
81
|
|
|
$this->chars[$k] = '\\' . $v; |
|
82
|
|
|
} |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
/** |
|
86
|
|
|
* Reorder or escape the caret character so that the regexp doesn't start with it |
|
87
|
|
|
* |
|
88
|
|
|
* @return void |
|
89
|
|
|
*/ |
|
90
|
|
|
protected function fixCaret() |
|
91
|
|
|
{ |
|
92
|
|
|
// Test whether the character class starts with a caret |
|
93
|
|
|
$k = array_search('^', $this->chars, true); |
|
94
|
|
|
if ($this->ranges[0][0] !== $k) |
|
95
|
|
|
{ |
|
96
|
|
|
return; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
// We swap the first two ranges if applicable, otherwise we escape the caret |
|
100
|
|
|
if (isset($this->ranges[1])) |
|
101
|
|
|
{ |
|
102
|
|
|
$range = $this->ranges[0]; |
|
103
|
|
|
$this->ranges[0] = $this->ranges[1]; |
|
104
|
|
|
$this->ranges[1] = $range; |
|
105
|
|
|
} |
|
106
|
|
|
else |
|
107
|
|
|
{ |
|
108
|
|
|
$this->chars[$k] = '\\^'; |
|
109
|
|
|
} |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* Reorder the characters so that a literal dash isn't mistaken for a range |
|
114
|
|
|
* |
|
115
|
|
|
* @return void |
|
116
|
|
|
*/ |
|
117
|
|
|
protected function reorderDash() |
|
118
|
|
|
{ |
|
119
|
|
|
$dashIndex = array_search('-', $this->chars, true); |
|
120
|
|
|
if ($dashIndex === false) |
|
121
|
|
|
{ |
|
122
|
|
|
return; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
// Look for a single dash and move it to the start of the character class |
|
126
|
|
|
$k = array_search([$dashIndex, $dashIndex], $this->ranges, true); |
|
127
|
|
|
if ($k > 0) |
|
128
|
|
|
{ |
|
129
|
|
|
unset($this->ranges[$k]); |
|
130
|
|
|
array_unshift($this->ranges, [$dashIndex, $dashIndex]); |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
// Look for a comma-dash (0x2C..0x2D) range |
|
134
|
|
|
$commaIndex = array_search(',', $this->chars); |
|
135
|
|
|
$range = [$commaIndex, $dashIndex]; |
|
136
|
|
|
$k = array_search($range, $this->ranges, true); |
|
137
|
|
|
if ($k !== false) |
|
138
|
|
|
{ |
|
139
|
|
|
// Replace with a single comma and prepend a single dash |
|
140
|
|
|
$this->ranges[$k] = [$commaIndex, $commaIndex]; |
|
141
|
|
|
array_unshift($this->ranges, [$dashIndex, $dashIndex]); |
|
142
|
|
|
} |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
/** |
|
146
|
|
|
* Store the character ranges from the list of characters |
|
147
|
|
|
* |
|
148
|
|
|
* @return void |
|
149
|
|
|
*/ |
|
150
|
|
|
protected function storeRanges() |
|
151
|
|
|
{ |
|
152
|
|
|
$values = []; |
|
153
|
|
|
foreach ($this->chars as $char) |
|
154
|
|
|
{ |
|
155
|
|
|
if (strlen($char) === 1) |
|
156
|
|
|
{ |
|
157
|
|
|
$values[] = ord($char); |
|
158
|
|
|
} |
|
159
|
|
|
else |
|
160
|
|
|
{ |
|
161
|
|
|
$values[] = false; |
|
162
|
|
|
} |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
$i = count($values) - 1; |
|
166
|
|
|
$ranges = []; |
|
167
|
|
|
while ($i >= 0) |
|
168
|
|
|
{ |
|
169
|
|
|
$start = $i; |
|
170
|
|
|
$end = $i; |
|
171
|
|
|
while ($start > 0 && $values[$start - 1] === $values[$end] - ($end + 1 - $start)) |
|
172
|
|
|
{ |
|
173
|
|
|
--$start; |
|
174
|
|
|
} |
|
175
|
|
|
$ranges[] = [$start, $end]; |
|
176
|
|
|
$i = $start - 1; |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
$this->ranges = array_reverse($ranges); |
|
180
|
|
|
} |
|
181
|
|
|
|
|
182
|
|
|
/** |
|
183
|
|
|
* Unescape literals in stored chars |
|
184
|
|
|
* |
|
185
|
|
|
* @return void |
|
186
|
|
|
*/ |
|
187
|
|
|
protected function unescapeLiterals() |
|
188
|
|
|
{ |
|
189
|
|
|
foreach ($this->chars as $k => $char) |
|
190
|
|
|
{ |
|
191
|
|
|
if ($char[0] === '\\' && preg_match('(^\\\\[^a-z]$)Di', $char)) |
|
192
|
|
|
{ |
|
193
|
|
|
$this->chars[$k] = substr($char, 1); |
|
194
|
|
|
} |
|
195
|
|
|
} |
|
196
|
|
|
} |
|
197
|
|
|
} |
s9e /
TextFormatter
