|
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\Passes; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Replaces (?:ab?|b)? with a?b? |
|
12
|
|
|
*/ |
|
13
|
|
|
class CoalesceOptionalStrings extends AbstractPass |
|
14
|
|
|
{ |
|
15
|
|
|
/** |
|
16
|
|
|
* {@inheritdoc} |
|
17
|
|
|
*/ |
|
18
|
8 |
|
protected function canRun(array $strings): bool |
|
19
|
|
|
{ |
|
20
|
8 |
|
return ($this->isOptional && count($strings) > 1); |
|
21
|
|
|
} |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* {@inheritdoc} |
|
25
|
|
|
*/ |
|
26
|
8 |
|
protected function runPass(array $strings): array |
|
27
|
|
|
{ |
|
28
|
8 |
|
foreach ($this->getPrefixGroups($strings) as $suffix => $prefixStrings) |
|
29
|
|
|
{ |
|
30
|
8 |
|
$suffix = unserialize($suffix); |
|
31
|
8 |
|
$suffixStrings = array_diff_key($strings, $prefixStrings); |
|
32
|
8 |
|
if ($suffix === $this->buildSuffix($suffixStrings)) |
|
33
|
|
|
{ |
|
34
|
4 |
|
$this->isOptional = false; |
|
35
|
|
|
|
|
36
|
4 |
|
return $this->buildCoalescedStrings($prefixStrings, $suffix); |
|
37
|
|
|
} |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
8 |
|
return $strings; |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* Build the final list of coalesced strings |
|
45
|
|
|
* |
|
46
|
|
|
* @param array[] $prefixStrings |
|
47
|
|
|
* @param array $suffix |
|
48
|
|
|
* @return array[] |
|
49
|
|
|
*/ |
|
50
|
4 |
|
protected function buildCoalescedStrings(array $prefixStrings, array $suffix): array |
|
51
|
|
|
{ |
|
52
|
4 |
|
$strings = $this->runPass($this->buildPrefix($prefixStrings)); |
|
53
|
4 |
|
if ($this->isSingleOptionalAlternation($strings)) |
|
54
|
|
|
{ |
|
55
|
|
|
// If the prefix has been remerged into a list of strings which contains only one string |
|
56
|
|
|
// of which the first element is an optional alternation, we only need to append the |
|
57
|
|
|
// suffix |
|
58
|
2 |
|
$strings[0][] = $suffix; |
|
59
|
|
|
} |
|
60
|
|
|
else |
|
61
|
|
|
{ |
|
62
|
|
|
// Put the current list of strings that form the prefix into a new list of strings, of |
|
63
|
|
|
// which the only string is composed of our optional prefix followed by the suffix |
|
64
|
4 |
|
array_unshift($strings, []); |
|
65
|
4 |
|
$strings = [[$strings, $suffix]]; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
4 |
|
return $strings; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Build the list of strings used as prefix |
|
73
|
|
|
* |
|
74
|
|
|
* @param array[] $strings |
|
75
|
|
|
* @return array[] |
|
76
|
|
|
*/ |
|
77
|
4 |
|
protected function buildPrefix(array $strings): array |
|
78
|
|
|
{ |
|
79
|
4 |
|
$prefix = []; |
|
80
|
4 |
|
foreach ($strings as $string) |
|
81
|
|
|
{ |
|
82
|
|
|
// Remove the last element (suffix) of each string before adding it |
|
83
|
4 |
|
array_pop($string); |
|
84
|
4 |
|
$prefix[] = $string; |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
4 |
|
return $prefix; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* Build a list of strings that matches any given strings or nothing |
|
92
|
|
|
* |
|
93
|
|
|
* Will unpack groups of single characters |
|
94
|
|
|
* |
|
95
|
|
|
* @param array[] $strings |
|
96
|
|
|
* @return array[] |
|
97
|
|
|
*/ |
|
98
|
8 |
|
protected function buildSuffix(array $strings): array |
|
99
|
|
|
{ |
|
100
|
8 |
|
$suffix = [[]]; |
|
101
|
8 |
|
foreach ($strings as $string) |
|
102
|
|
|
{ |
|
103
|
8 |
|
if ($this->isCharacterClassString($string)) |
|
104
|
|
|
{ |
|
105
|
1 |
|
foreach ($string[0] as $element) |
|
106
|
|
|
{ |
|
107
|
1 |
|
$suffix[] = $element; |
|
108
|
|
|
} |
|
109
|
|
|
} |
|
110
|
|
|
else |
|
111
|
|
|
{ |
|
112
|
8 |
|
$suffix[] = $string; |
|
113
|
|
|
} |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
8 |
|
return $suffix; |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
/** |
|
120
|
|
|
* Get the list of potential prefix strings grouped by identical suffix |
|
121
|
|
|
* |
|
122
|
|
|
* @param array[] $strings |
|
123
|
|
|
* @return array |
|
124
|
|
|
*/ |
|
125
|
8 |
|
protected function getPrefixGroups(array $strings): array |
|
126
|
|
|
{ |
|
127
|
8 |
|
$groups = []; |
|
128
|
8 |
|
foreach ($strings as $k => $string) |
|
129
|
|
|
{ |
|
130
|
8 |
|
if ($this->hasOptionalSuffix($string)) |
|
131
|
|
|
{ |
|
132
|
8 |
|
$groups[serialize(end($string))][$k] = $string; |
|
133
|
|
|
} |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
8 |
|
return $groups; |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
/** |
|
140
|
|
|
* Test whether given list of strings starts with a single optional alternation |
|
141
|
|
|
* |
|
142
|
|
|
* @param array $strings |
|
143
|
|
|
* @return bool |
|
144
|
|
|
*/ |
|
145
|
4 |
|
protected function isSingleOptionalAlternation(array $strings): bool |
|
146
|
|
|
{ |
|
147
|
4 |
|
return (count($strings) === 1 && is_array($strings[0][0]) && $strings[0][0][0] === []); |
|
148
|
|
|
} |
|
149
|
|
|
} |
s9e /
RegexpBuilder
