|
1
|
|
|
<?php declare(strict_types=1); |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\TextFormatter |
|
5
|
|
|
* @copyright Copyright (c) 2010-2023 The s9e authors |
|
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
|
7
|
|
|
*/ |
|
8
|
|
|
namespace s9e\TextFormatter\Configurator; |
|
9
|
|
|
|
|
10
|
|
|
use RuntimeException; |
|
11
|
|
|
use s9e\TextFormatter\Configurator\RecursiveParser\MatcherInterface; |
|
12
|
|
|
|
|
13
|
|
|
class RecursiveParser |
|
14
|
|
|
{ |
|
15
|
|
|
/** |
|
16
|
|
|
* @var array Callback associated with each match name |
|
17
|
|
|
*/ |
|
18
|
|
|
protected $callbacks = []; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* @var array Match names associated with each group |
|
22
|
|
|
*/ |
|
23
|
|
|
protected $groupMatches = []; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @var array Groups associated with each match name |
|
27
|
|
|
*/ |
|
28
|
|
|
protected $matchGroups = []; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @var string Regexp used to match input |
|
32
|
|
|
*/ |
|
33
|
|
|
protected $regexp; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* Parse given string |
|
37
|
|
|
* |
|
38
|
|
|
* @param string $str |
|
39
|
|
|
* @param string $name Allowed match, either match name or group name (default: allow all) |
|
40
|
|
|
* @return mixed |
|
41
|
|
|
*/ |
|
42
|
5 |
|
public function parse(string $str, string $name = '') |
|
43
|
|
|
{ |
|
44
|
5 |
|
$regexp = $this->regexp; |
|
45
|
5 |
|
if ($name !== '') |
|
46
|
|
|
{ |
|
47
|
1 |
|
$restrict = (isset($this->groupMatches[$name])) ? implode('|', $this->groupMatches[$name]) : $name; |
|
48
|
1 |
|
$regexp = preg_replace('(\\(\\?<(?!(?:' . $restrict . '|\\w+\\d+)>))', '(*F)$0', $regexp); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
5 |
|
preg_match($regexp, $str, $m); |
|
52
|
5 |
|
if (!isset($m['MARK'])) |
|
53
|
|
|
{ |
|
54
|
1 |
|
throw new RuntimeException('Cannot parse ' . var_export($str, true)); |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
4 |
|
$name = $m['MARK']; |
|
58
|
4 |
|
$args = $this->getArguments($m, $name); |
|
59
|
|
|
|
|
60
|
|
|
return [ |
|
61
|
4 |
|
'groups' => $this->matchGroups[$name] ?? [], |
|
62
|
4 |
|
'match' => $name, |
|
63
|
4 |
|
'value' => call_user_func_array($this->callbacks[$name], $args) |
|
64
|
|
|
]; |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* Set the list of matchers used by this parser |
|
69
|
|
|
* |
|
70
|
|
|
* @param MatcherInterface[] |
|
71
|
|
|
* @return void |
|
72
|
|
|
*/ |
|
73
|
5 |
|
public function setMatchers(array $matchers): void |
|
74
|
|
|
{ |
|
75
|
5 |
|
$matchRegexps = []; |
|
76
|
5 |
|
$this->groupMatches = []; |
|
77
|
5 |
|
$this->matchGroups = []; |
|
78
|
5 |
|
foreach ($this->getMatchersConfig($matchers) as $matchName => $matchConfig) |
|
79
|
|
|
{ |
|
80
|
5 |
|
foreach ($matchConfig['groups'] as $group) |
|
81
|
|
|
{ |
|
82
|
5 |
|
$this->groupMatches[$group][] = $matchName; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
5 |
|
$regexp = $matchConfig['regexp']; |
|
86
|
5 |
|
$regexp = str_replace(' ', '\\s*+', $regexp); |
|
87
|
5 |
|
$regexp = '(?<' . $matchName . '>' . $regexp . ')(*:' . $matchName . ')'; |
|
88
|
5 |
|
|
|
89
|
|
|
$matchRegexps[] = $regexp; |
|
90
|
5 |
|
$this->callbacks[$matchName] = $matchConfig['callback']; |
|
91
|
5 |
|
$this->matchGroups[$matchName] = $matchConfig['groups']; |
|
92
|
5 |
|
} |
|
93
|
|
|
|
|
94
|
|
|
$groupRegexps = []; |
|
95
|
5 |
|
foreach ($this->groupMatches as $group => $names) |
|
96
|
5 |
|
{ |
|
97
|
|
|
$groupRegexps[] = '(?<' . $group . '>(?&' . implode(')|(?&', $names) . '))'; |
|
98
|
5 |
|
} |
|
99
|
|
|
|
|
100
|
|
|
$this->regexp = '((?(DEFINE)' . implode('', $groupRegexps). ')' |
|
101
|
5 |
|
. '^(?:' . implode('|', $matchRegexps) . ')$)s'; |
|
102
|
5 |
|
} |
|
103
|
|
|
|
|
104
|
|
|
/** |
|
105
|
|
|
* Get the list of arguments produced by a regexp's match |
|
106
|
|
|
* |
|
107
|
|
|
* @param string[] $matches Regexp matches |
|
108
|
|
|
* @param string $name Regexp name |
|
109
|
|
|
* @return string[] |
|
110
|
|
|
*/ |
|
111
|
|
|
protected function getArguments(array $matches, string $name): array |
|
112
|
4 |
|
{ |
|
113
|
|
|
$args = []; |
|
114
|
4 |
|
$collect = false; |
|
115
|
4 |
|
foreach ($matches as $k => $v) |
|
116
|
4 |
|
{ |
|
117
|
|
|
if ($k === $name) |
|
118
|
4 |
|
{ |
|
119
|
4 |
|
// Start collecting matches once we reach the target capture |
|
120
|
|
|
$collect = true; |
|
121
|
|
|
} |
|
122
|
4 |
|
elseif ($collect) |
|
123
|
|
|
{ |
|
124
|
|
|
if (!is_int($k)) |
|
125
|
|
|
{ |
|
126
|
|
|
// Stop collecting when we reach the next capture |
|
127
|
|
|
break; |
|
128
|
|
|
} |
|
129
|
|
|
$args[] = $v; |
|
130
|
|
|
} |
|
131
|
5 |
|
} |
|
132
|
|
|
|
|
133
|
5 |
|
// Remove the first entry, which contains the whole string that was matched |
|
134
|
5 |
|
array_shift($args); |
|
135
|
|
|
|
|
136
|
5 |
|
return $args; |
|
137
|
|
|
} |
|
138
|
5 |
|
|
|
139
|
|
|
/** |
|
140
|
5 |
|
* Collect, normalize, sort and return the config for all matchers |
|
141
|
|
|
* |
|
142
|
5 |
|
* @param MatcherInterface[] $matchers |
|
143
|
5 |
|
* @return array |
|
144
|
|
|
*/ |
|
145
|
5 |
|
protected function getMatchersConfig(array $matchers): array |
|
146
|
|
|
{ |
|
147
|
5 |
|
$matchersConfig = []; |
|
148
|
|
|
foreach ($matchers as $matcher) |
|
149
|
5 |
|
{ |
|
150
|
5 |
|
foreach ($matcher->getMatchers() as $matchName => $matchConfig) |
|
151
|
5 |
|
{ |
|
152
|
|
|
if (is_string($matchConfig)) |
|
153
|
5 |
|
{ |
|
154
|
|
|
$matchConfig = ['regexp' => $matchConfig]; |
|
155
|
|
|
} |
|
156
|
5 |
|
$parts = explode(':', $matchName); |
|
157
|
|
|
$matchName = array_pop($parts); |
|
158
|
5 |
|
$matchConfig += [ |
|
159
|
|
|
'callback' => [$matcher, 'parse' . $matchName], |
|
160
|
|
|
'groups' => [], |
|
161
|
|
|
'order' => 0 |
|
162
|
|
|
]; |
|
163
|
|
|
$matchConfig['name'] = $matchName; |
|
164
|
|
|
$matchConfig['groups'] = array_unique(array_merge($matchConfig['groups'], $parts)); |
|
165
|
|
|
sort($matchConfig['groups']); |
|
166
|
|
|
|
|
167
|
|
|
$matchersConfig[$matchName] = $matchConfig; |
|
168
|
5 |
|
} |
|
169
|
|
|
} |
|
170
|
5 |
|
uasort($matchersConfig, static::class . '::sortMatcherConfig'); |
|
171
|
|
|
|
|
172
|
5 |
|
return $matchersConfig; |
|
173
|
5 |
|
} |
|
174
|
5 |
|
|
|
175
|
|
|
/** |
|
176
|
5 |
|
* Compare two matchers' config |
|
177
|
5 |
|
* |
|
178
|
|
|
* @param array $a |
|
179
|
|
|
* @param array $b |
|
180
|
|
|
* @return integer |
|
181
|
|
|
*/ |
|
182
|
|
|
protected static function sortMatcherConfig(array $a, array $b): int |
|
183
|
|
|
{ |
|
184
|
|
|
if ($a['order'] !== $b['order']) |
|
185
|
|
|
{ |
|
186
|
|
|
return $a['order'] - $b['order']; |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
5 |
|
return strcmp($a['name'], $b['name']); |
|
190
|
|
|
} |
|
191
|
|
|
} |
s9e /
TextFormatter
