|
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 s9e\RegexpBuilder\Input\InputInterface; |
|
11
|
|
|
use s9e\RegexpBuilder\Output\OutputInterface; |
|
12
|
|
|
use s9e\RegexpBuilder\Passes\CoalesceOptionalStrings; |
|
13
|
|
|
use s9e\RegexpBuilder\Passes\CoalesceSingleCharacterPrefix; |
|
14
|
|
|
use s9e\RegexpBuilder\Passes\GroupSingleCharacters; |
|
15
|
|
|
use s9e\RegexpBuilder\Passes\MergePrefix; |
|
16
|
|
|
use s9e\RegexpBuilder\Passes\MergeSuffix; |
|
17
|
|
|
use s9e\RegexpBuilder\Passes\PromoteSingleStrings; |
|
18
|
|
|
use s9e\RegexpBuilder\Passes\Recurse; |
|
19
|
|
|
|
|
20
|
|
|
class Builder |
|
21
|
|
|
{ |
|
22
|
|
|
/** |
|
23
|
|
|
* @var InputInterface |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $input; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @var MetaCharacters |
|
29
|
|
|
*/ |
|
30
|
|
|
protected $meta; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* @var Runner |
|
34
|
|
|
*/ |
|
35
|
|
|
protected $runner; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @var Serializer |
|
39
|
|
|
*/ |
|
40
|
10 |
|
protected $serializer; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
10 |
|
* @param array $config |
|
44
|
|
|
*/ |
|
45
|
|
|
public function __construct(array $config = []) |
|
46
|
|
|
{ |
|
47
|
|
|
$config += [ |
|
48
|
|
|
'delimiter' => '/', |
|
49
|
|
|
'input' => 'Bytes', |
|
50
|
10 |
|
'inputOptions' => [], |
|
51
|
10 |
|
'meta' => [], |
|
52
|
10 |
|
'output' => 'Bytes', |
|
53
|
10 |
|
'outputOptions' => [] |
|
54
|
|
|
]; |
|
55
|
|
|
|
|
56
|
|
|
$this->setInput($config['input'], $config['inputOptions']); |
|
57
|
|
|
$this->setMeta($config['meta']); |
|
58
|
|
|
$this->setSerializer($config['output'], $config['outputOptions'], $config['delimiter']); |
|
59
|
|
|
$this->setRunner(); |
|
60
|
|
|
} |
|
61
|
10 |
|
|
|
62
|
|
|
/** |
|
63
|
10 |
|
* Build and return a regular expression that matches all of the given strings |
|
64
|
10 |
|
* |
|
65
|
|
|
* @param string[] $strings Literal strings to be matched |
|
66
|
1 |
|
* @return string Regular expression (without delimiters) |
|
67
|
|
|
*/ |
|
68
|
|
|
public function build(array $strings) |
|
69
|
9 |
|
{ |
|
70
|
9 |
|
$strings = array_unique($strings); |
|
71
|
9 |
|
if ($this->isEmpty($strings)) |
|
72
|
|
|
{ |
|
73
|
9 |
|
return ''; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
$strings = $this->splitStrings($strings); |
|
77
|
|
|
usort($strings, __CLASS__ . '::compareStrings'); |
|
78
|
|
|
$strings = $this->meta->replaceMeta($strings); |
|
79
|
|
|
$strings = $this->runner->run($strings); |
|
80
|
|
|
|
|
81
|
|
|
return $this->serializer->serializeStrings($strings); |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
9 |
|
* Compare two split strings |
|
86
|
|
|
* |
|
87
|
9 |
|
* Will sort strings in ascending order |
|
88
|
9 |
|
* |
|
89
|
9 |
|
* @param integer[] $a |
|
90
|
|
|
* @param integer[] $b |
|
91
|
9 |
|
* @return integer |
|
92
|
|
|
*/ |
|
93
|
9 |
|
protected function compareStrings(array $a, array $b) |
|
94
|
|
|
{ |
|
95
|
|
|
$i = -1; |
|
96
|
|
|
$cnt = min(count($a), count($b)); |
|
97
|
1 |
|
while (++$i < $cnt) |
|
98
|
|
|
{ |
|
99
|
|
|
if ($a[$i] !== $b[$i]) |
|
100
|
|
|
{ |
|
101
|
|
|
return $a[$i] - $b[$i]; |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
return count($a) - count($b); |
|
106
|
10 |
|
} |
|
107
|
|
|
|
|
108
|
10 |
|
/** |
|
109
|
|
|
* Test whether the list of strings is empty |
|
110
|
|
|
* |
|
111
|
|
|
* @param string[] $strings |
|
112
|
|
|
* @return bool |
|
113
|
|
|
*/ |
|
114
|
|
|
protected function isEmpty(array $strings) |
|
115
|
|
|
{ |
|
116
|
|
|
return (empty($strings) || $strings === ['']); |
|
117
|
|
|
} |
|
118
|
10 |
|
|
|
119
|
|
|
/** |
|
120
|
10 |
|
* Set the InputInterface instance in $this->input |
|
121
|
10 |
|
* |
|
122
|
10 |
|
* @param string $inputType |
|
123
|
|
|
* @param array $inputOptions |
|
124
|
|
|
* @return void |
|
125
|
|
|
*/ |
|
126
|
|
|
protected function setInput($inputType, array $inputOptions) |
|
127
|
|
|
{ |
|
128
|
|
|
$className = __NAMESPACE__ . '\\Input\\' . $inputType; |
|
129
|
10 |
|
$this->input = new $className($inputOptions); |
|
130
|
|
|
} |
|
131
|
10 |
|
|
|
132
|
10 |
|
/** |
|
133
|
10 |
|
* Set the MetaCharacters instance in $this->meta |
|
134
|
10 |
|
* |
|
135
|
10 |
|
* @param array $map |
|
136
|
10 |
|
* @return void |
|
137
|
10 |
|
*/ |
|
138
|
10 |
|
protected function setMeta(array $map) |
|
139
|
10 |
|
{ |
|
140
|
|
|
$this->meta = new MetaCharacters($this->input); |
|
141
|
|
|
foreach ($map as $char => $expr) |
|
142
|
|
|
{ |
|
143
|
|
|
$this->meta->add($char, $expr); |
|
144
|
|
|
} |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
/** |
|
148
|
|
|
* Set the Runner instance $in this->runner |
|
149
|
10 |
|
* |
|
150
|
|
|
* @return void |
|
151
|
10 |
|
*/ |
|
152
|
10 |
|
protected function setRunner() |
|
153
|
10 |
|
{ |
|
154
|
|
|
$this->runner = new Runner; |
|
155
|
10 |
|
$this->runner->addPass(new MergePrefix); |
|
156
|
10 |
|
$this->runner->addPass(new GroupSingleCharacters); |
|
157
|
|
|
$this->runner->addPass(new Recurse($this->runner)); |
|
158
|
|
|
$this->runner->addPass(new PromoteSingleStrings); |
|
159
|
|
|
$this->runner->addPass(new CoalesceOptionalStrings); |
|
160
|
|
|
$this->runner->addPass(new MergeSuffix); |
|
161
|
|
|
$this->runner->addPass(new CoalesceSingleCharacterPrefix); |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
9 |
|
/** |
|
165
|
|
|
* Set the Serializer instance in $this->serializer |
|
166
|
9 |
|
* |
|
167
|
|
|
* @param string $outputType |
|
168
|
|
|
* @param array $outputOptions |
|
169
|
|
|
* @param string $delimiter |
|
170
|
|
|
* @return void |
|
171
|
|
|
*/ |
|
172
|
|
|
protected function setSerializer($outputType, array $outputOptions, $delimiter) |
|
173
|
|
|
{ |
|
174
|
|
|
$className = __NAMESPACE__ . '\\Output\\' . $outputType; |
|
175
|
|
|
$output = new $className($outputOptions); |
|
176
|
|
|
$escaper = new Escaper($delimiter); |
|
177
|
|
|
|
|
178
|
|
|
$this->serializer = new Serializer($output, $this->meta, $escaper); |
|
179
|
|
|
} |
|
180
|
|
|
|
|
181
|
|
|
/** |
|
182
|
|
|
* Split all given strings by character |
|
183
|
|
|
* |
|
184
|
|
|
* @param string[] $strings List of strings |
|
185
|
|
|
* @return array[] List of arrays |
|
186
|
|
|
*/ |
|
187
|
|
|
protected function splitStrings(array $strings) |
|
188
|
|
|
{ |
|
189
|
|
|
return array_map([$this->input, 'split'], $strings); |
|
190
|
|
|
} |
|
191
|
|
|
} |
s9e /
RegexpBuilder
