|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\RegexpBuilder |
|
5
|
|
|
* @copyright Copyright (c) 2016 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\CoalesceSingleCharacterPrefix; |
|
13
|
|
|
use s9e\RegexpBuilder\Passes\GroupSingleCharacters; |
|
14
|
|
|
use s9e\RegexpBuilder\Passes\MergePrefix; |
|
15
|
|
|
use s9e\RegexpBuilder\Passes\MergeSuffix; |
|
16
|
|
|
use s9e\RegexpBuilder\Passes\PromoteSingleStrings; |
|
17
|
|
|
use s9e\RegexpBuilder\Passes\Recurse; |
|
18
|
|
|
|
|
19
|
|
|
class Builder |
|
20
|
|
|
{ |
|
21
|
|
|
/** |
|
22
|
|
|
* @var InputInterface |
|
23
|
|
|
*/ |
|
24
|
|
|
protected $input; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* @var Runner |
|
28
|
|
|
*/ |
|
29
|
|
|
protected $runner; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @var Serializer |
|
33
|
|
|
*/ |
|
34
|
|
|
protected $serializer; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @param array $config |
|
38
|
|
|
*/ |
|
39
|
9 |
|
public function __construct(array $config = []) |
|
40
|
|
|
{ |
|
41
|
9 |
|
$config = $this->getConfig($config); |
|
42
|
|
|
|
|
43
|
9 |
|
$this->input = $config['input']; |
|
44
|
9 |
|
$this->runner = $config['runner']; |
|
45
|
9 |
|
$this->serializer = new Serializer($config['output'], $config['escaper']); |
|
46
|
9 |
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* Build and return a regular expression that matches all of the given strings |
|
50
|
|
|
* |
|
51
|
|
|
* @param string[] $strings Literal strings to be matched |
|
52
|
|
|
* @return string Regular expression (without delimiters) |
|
53
|
|
|
*/ |
|
54
|
9 |
|
public function build(array $strings) |
|
55
|
|
|
{ |
|
56
|
9 |
|
$strings = array_unique($strings); |
|
57
|
9 |
|
if ($strings === ['']) |
|
58
|
|
|
{ |
|
59
|
1 |
|
return ''; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
8 |
|
$strings = $this->splitStrings($strings); |
|
63
|
8 |
|
usort($strings, __CLASS__ . '::compareStrings'); |
|
64
|
8 |
|
$strings = $this->runner->run($strings); |
|
65
|
|
|
|
|
66
|
8 |
|
return $this->serializer->serializeStrings($strings); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* Compare two split strings |
|
71
|
|
|
* |
|
72
|
|
|
* Will sort strings in ascending order |
|
73
|
|
|
* |
|
74
|
|
|
* @param integer[] $a |
|
75
|
|
|
* @param integer[] $b |
|
76
|
|
|
* @return integer |
|
77
|
|
|
*/ |
|
78
|
8 |
|
protected function compareStrings(array $a, array $b) |
|
79
|
|
|
{ |
|
80
|
8 |
|
foreach ($b as $k => $v) |
|
81
|
|
|
{ |
|
82
|
8 |
|
if (!isset($a[$k]) || $a[$k] < $b[$k]) |
|
83
|
|
|
{ |
|
84
|
|
|
// A is shorter or A[k] is less than B[k]. Sort A before B |
|
85
|
7 |
|
return -1; |
|
86
|
|
|
} |
|
87
|
4 |
|
if ($a[$k] > $b[$k]) |
|
88
|
|
|
{ |
|
89
|
|
|
// A[k] comes after B[k]. Sort A after B |
|
90
|
4 |
|
return 1; |
|
91
|
|
|
} |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
// A is longer than B. Sort A after B |
|
95
|
|
|
return 1; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* Build the full config array based on given input |
|
100
|
|
|
* |
|
101
|
|
|
* @param array $config Sparse config |
|
102
|
|
|
* @return array Full config |
|
103
|
|
|
*/ |
|
104
|
9 |
|
protected function getConfig(array $config) |
|
105
|
|
|
{ |
|
106
|
|
|
$config += [ |
|
107
|
9 |
|
'delimiter' => '/', |
|
108
|
|
|
'input' => 'Bytes', |
|
109
|
|
|
'output' => 'Bytes' |
|
110
|
|
|
]; |
|
111
|
9 |
|
$config['escaper'] = new Escaper($config['delimiter']); |
|
112
|
|
|
|
|
113
|
9 |
|
$className = __NAMESPACE__ . '\\Input\\' . $config['input']; |
|
114
|
9 |
|
$config['input'] = new $className; |
|
115
|
|
|
|
|
116
|
9 |
|
$className = __NAMESPACE__ . '\\Output\\' . $config['output']; |
|
117
|
9 |
|
$config['output'] = new $className; |
|
118
|
|
|
|
|
119
|
9 |
|
$config['runner'] = new Runner; |
|
120
|
9 |
|
$config['runner']->addPass(new MergePrefix); |
|
121
|
9 |
|
$config['runner']->addPass(new GroupSingleCharacters); |
|
122
|
9 |
|
$config['runner']->addPass(new Recurse($config['runner'])); |
|
123
|
9 |
|
$config['runner']->addPass(new PromoteSingleStrings); |
|
124
|
9 |
|
$config['runner']->addPass(new MergeSuffix); |
|
125
|
9 |
|
$config['runner']->addPass(new CoalesceSingleCharacterPrefix); |
|
126
|
|
|
|
|
127
|
9 |
|
return $config; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
/** |
|
131
|
|
|
* Split all given strings by character |
|
132
|
|
|
* |
|
133
|
|
|
* @param string[] $strings List of strings |
|
134
|
|
|
* @return array[] List of arrays |
|
135
|
|
|
*/ |
|
136
|
8 |
|
protected function splitStrings(array $strings) |
|
137
|
|
|
{ |
|
138
|
8 |
|
return array_map([$this->input, 'split'], $strings); |
|
139
|
|
|
} |
|
140
|
|
|
} |
s9e /
RegexpBuilder
