|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* Highlighter |
|
4
|
|
|
* |
|
5
|
|
|
* Copyright (C) 2015, Some right reserved. |
|
6
|
|
|
* |
|
7
|
|
|
* @author Kacper "Kadet" Donat <[email protected]> |
|
8
|
|
|
* |
|
9
|
|
|
* Contact with author: |
|
10
|
|
|
* Xmpp: [email protected] |
|
11
|
|
|
* E-mail: [email protected] |
|
12
|
|
|
* |
|
13
|
|
|
* From Kadet with love. |
|
14
|
|
|
*/ |
|
15
|
|
|
|
|
16
|
|
|
namespace Kadet\Highlighter\Language; |
|
17
|
|
|
|
|
18
|
|
|
use Kadet\Highlighter\Matcher\WholeMatcher; |
|
19
|
|
|
use Kadet\Highlighter\Parser\LanguageToken; |
|
20
|
|
|
use Kadet\Highlighter\Parser\Rule; |
|
21
|
|
|
use Kadet\Highlighter\Parser\Token; |
|
22
|
|
|
use Kadet\Highlighter\Parser\TokenFactory; |
|
23
|
|
|
use Kadet\Highlighter\Parser\TokenIterator; |
|
24
|
|
|
use Kadet\Highlighter\Parser\TokenList; |
|
25
|
|
|
use Kadet\Highlighter\Utils\ArrayHelper; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Class Language |
|
29
|
|
|
* |
|
30
|
|
|
* @package Kadet\Highlighter\Language |
|
31
|
|
|
*/ |
|
32
|
|
|
abstract class Language |
|
33
|
|
|
{ |
|
34
|
|
|
/** |
|
35
|
|
|
* Tokenizer rules |
|
36
|
|
|
* |
|
37
|
|
|
* @var Rule[] |
|
38
|
|
|
*/ |
|
39
|
|
|
private $_rules; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @var array |
|
43
|
|
|
*/ |
|
44
|
|
|
protected $_options = []; |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Language constructor. |
|
48
|
|
|
* |
|
49
|
|
|
* @param array $options |
|
50
|
|
|
*/ |
|
51
|
20 |
|
public function __construct(array $options = []) { |
|
52
|
20 |
|
$this->_options = array_merge([ |
|
53
|
20 |
|
'embedded' => [], |
|
54
|
20 |
|
], $this->_options, $options); |
|
55
|
|
|
|
|
56
|
20 |
|
$this->_rules = $this->getRules(); |
|
57
|
20 |
|
} |
|
58
|
|
|
|
|
59
|
|
|
/** |
|
60
|
|
|
* Tokenization rules definition |
|
61
|
|
|
* |
|
62
|
|
|
* @return array |
|
63
|
|
|
*/ |
|
64
|
|
|
public abstract function getRules(); |
|
|
|
|
|
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* Parses source and removes wrong tokens. |
|
68
|
|
|
* |
|
69
|
|
|
* @param TokenIterator|string $tokens |
|
70
|
|
|
* |
|
71
|
|
|
* @param array $additional |
|
72
|
|
|
* @param bool $embedded |
|
73
|
|
|
* |
|
74
|
|
|
* @return TokenIterator |
|
75
|
|
|
*/ |
|
76
|
12 |
|
public function parse($tokens = null, $additional = [], $embedded = false) |
|
77
|
|
|
{ |
|
78
|
11 |
|
if (is_string($tokens)) { |
|
79
|
11 |
|
$tokens = $this->tokenize($tokens, $additional, $embedded); |
|
|
|
|
|
|
80
|
11 |
|
} elseif(!$tokens instanceof TokenIterator) { |
|
81
|
|
|
// Todo: Own Exceptions |
|
82
|
|
|
throw new \InvalidArgumentException('$tokens must be string or TokenIterator'); |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
11 |
|
$start = $tokens->current(); |
|
86
|
|
|
|
|
87
|
|
|
|
|
88
|
11 |
|
$context = []; |
|
89
|
|
|
|
|
90
|
11 |
|
/** @var Token[] $result */ $result = [$start]; |
|
91
|
11 |
|
/** @var Token[] $all */ $all = []; |
|
92
|
|
|
|
|
93
|
|
|
/** @var Token $token */ |
|
94
|
11 |
|
for($tokens->next(); $tokens->valid(); $tokens->next()) { |
|
95
|
11 |
|
$token = $tokens->current(); |
|
96
|
|
|
|
|
97
|
11 |
|
if (!$token->isValid($this, $context)) { |
|
98
|
3 |
|
continue; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
11 |
|
if ($token->isStart()) { |
|
102
|
10 |
|
if ($token instanceof LanguageToken) { |
|
103
|
|
|
/** @var LanguageToken $token */ |
|
104
|
2 |
|
$result = array_merge( |
|
105
|
2 |
|
$result, |
|
106
|
2 |
|
$token->getInjected()->parse($tokens)->getTokens() |
|
107
|
2 |
|
); |
|
108
|
2 |
|
} else { |
|
109
|
10 |
|
$all[spl_object_hash($token)] = $result[] = $token; |
|
110
|
10 |
|
$context[spl_object_hash($token)] = $token->name; |
|
111
|
|
|
} |
|
112
|
10 |
|
} else { |
|
113
|
11 |
|
$start = $token->getStart(); |
|
114
|
|
|
|
|
115
|
|
|
/** @noinspection PhpUndefinedMethodInspection bug */ |
|
116
|
11 |
|
if ($token instanceof LanguageToken && $token->getLanguage() === $this) { |
|
117
|
11 |
|
$result[0]->setEnd($token); |
|
118
|
|
|
|
|
119
|
12 |
|
if($result[0]->getRule()->postProcess) { |
|
120
|
1 |
|
$source = substr($tokens->getSource(), $result[0]->pos, $result[0]->getLength()); |
|
121
|
|
|
|
|
122
|
1 |
|
$tokens = $this->tokenize($source, $result, $result[0]->pos, true); |
|
123
|
1 |
|
$result = $this->parse($tokens)->getTokens(); |
|
124
|
1 |
|
} |
|
125
|
|
|
|
|
126
|
|
|
# closing unclosed tokens |
|
127
|
11 |
|
foreach(array_reverse($context) as $hash => $name) { |
|
128
|
1 |
|
$end = new Token([$name, 'pos' => $token->pos]); |
|
129
|
1 |
|
$all[$hash]->setEnd($end); |
|
130
|
1 |
|
$result[] = $end; |
|
131
|
11 |
|
} |
|
132
|
|
|
|
|
133
|
11 |
|
$result[] = $token; |
|
134
|
11 |
|
break; |
|
135
|
|
|
} else { |
|
136
|
9 |
|
if ($start) { |
|
137
|
8 |
|
unset($context[spl_object_hash($start)]); |
|
138
|
8 |
|
} else { |
|
139
|
|
|
/** @noinspection PhpUnusedParameterInspection */ |
|
140
|
1 |
|
$start = ArrayHelper::find(array_reverse($context), function ($k, $v) use ($token) { |
|
141
|
1 |
|
return $v === $token->name; |
|
142
|
1 |
|
}); |
|
143
|
|
|
|
|
144
|
1 |
|
if ($start !== false) { |
|
145
|
1 |
|
$token->setStart($all[$start]); |
|
146
|
1 |
|
unset($context[$start]); |
|
147
|
1 |
|
} |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
9 |
|
$result[] = $token; |
|
151
|
|
|
} |
|
152
|
|
|
} |
|
153
|
10 |
|
} |
|
154
|
|
|
|
|
155
|
11 |
|
return new TokenIterator($result, $tokens->getSource()); |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
|
|
/** |
|
159
|
|
|
* Tokenize source |
|
160
|
|
|
* |
|
161
|
|
|
* @param $source |
|
162
|
|
|
* |
|
163
|
|
|
* @param int $offset |
|
164
|
|
|
* @param array $additional |
|
165
|
|
|
* |
|
166
|
|
|
* @param bool $embedded |
|
167
|
|
|
* |
|
168
|
|
|
* @return TokenList |
|
169
|
|
|
*/ |
|
170
|
12 |
|
private function _tokens($source, $offset = 0, $additional = [], $embedded = false) |
|
171
|
|
|
{ |
|
172
|
12 |
|
$result = new TokenList(); |
|
173
|
|
|
|
|
174
|
|
|
/** @var Language $language */ |
|
175
|
12 |
|
foreach($this->_rules($embedded) as $rule) { |
|
176
|
12 |
|
$rule->factory->setOffset($offset); |
|
177
|
12 |
|
foreach ($rule->match($source) as $token) { |
|
178
|
12 |
|
$result->add($token); |
|
179
|
12 |
|
} |
|
180
|
12 |
|
} |
|
181
|
|
|
|
|
182
|
12 |
|
return $result->batch($additional); |
|
183
|
|
|
} |
|
184
|
|
|
|
|
185
|
12 |
|
public function tokenize($source, $additional = [], $offset = 0, $embedded = false) |
|
186
|
|
|
{ |
|
187
|
12 |
|
$iterator = new TokenIterator($this->_tokens($source, $offset, $additional, $embedded)->sort()->toArray(), $source); |
|
188
|
12 |
|
return $iterator; |
|
189
|
|
|
} |
|
190
|
|
|
|
|
191
|
|
|
/** |
|
192
|
|
|
* @param bool $embedded |
|
193
|
|
|
* |
|
194
|
|
|
* @return Rule[] |
|
195
|
|
|
*/ |
|
196
|
12 |
|
private function _rules($embedded = false) { |
|
197
|
12 |
|
$all = $this->_rules; |
|
198
|
12 |
|
if(!$embedded) { |
|
199
|
12 |
|
$all['language.' . $this->getIdentifier()] = $this->getOpenClose(); |
|
200
|
12 |
|
} |
|
201
|
|
|
|
|
202
|
|
|
// why this code sucks so much? Because RecursiveIterator performance such a lot more. |
|
203
|
12 |
|
foreach ($all as $name => $rules) { |
|
204
|
12 |
|
if (!is_array($rules)) { |
|
205
|
12 |
|
$rules = [$rules]; |
|
206
|
12 |
|
} |
|
207
|
|
|
|
|
208
|
|
|
/** @var Rule $rule */ |
|
209
|
12 |
|
foreach ($rules as $rule) { |
|
210
|
12 |
|
if($rule->language === false) { |
|
211
|
11 |
|
$rule->language = $this; |
|
212
|
11 |
|
} |
|
213
|
|
|
|
|
214
|
12 |
|
$rule->factory->setBase($name); |
|
215
|
|
|
|
|
216
|
12 |
|
yield $rule; |
|
217
|
12 |
|
} |
|
218
|
12 |
|
} |
|
219
|
|
|
|
|
220
|
12 |
|
foreach($this->getEmbedded() as $language) { |
|
221
|
1 |
|
foreach($language->_rules() as $rule) { |
|
222
|
1 |
|
yield $rule; |
|
223
|
1 |
|
} |
|
224
|
12 |
|
} |
|
225
|
12 |
|
} |
|
226
|
|
|
|
|
227
|
|
|
/** |
|
228
|
|
|
* Unique language identifier, for example 'php' |
|
229
|
|
|
* |
|
230
|
|
|
* @return string |
|
231
|
|
|
*/ |
|
232
|
|
|
public abstract function getIdentifier(); |
|
|
|
|
|
|
233
|
|
|
|
|
234
|
|
|
/** |
|
235
|
|
|
* Language range Rule(s) |
|
236
|
|
|
* |
|
237
|
|
|
* @return Rule|Rule[] |
|
238
|
|
|
*/ |
|
239
|
15 |
|
public function getOpenClose() |
|
240
|
|
|
{ |
|
241
|
15 |
|
return new Rule( |
|
242
|
15 |
|
new WholeMatcher(), [ |
|
243
|
15 |
|
'priority' => 1000, |
|
244
|
15 |
|
'factory' => new TokenFactory(LanguageToken::class), |
|
245
|
15 |
|
'inject' => $this, |
|
246
|
15 |
|
'language' => null, |
|
247
|
15 |
|
'context' => ['!!'], |
|
248
|
|
|
] |
|
249
|
15 |
|
); |
|
250
|
|
|
} |
|
251
|
|
|
|
|
252
|
|
|
/** |
|
253
|
|
|
* @return Language[] |
|
254
|
|
|
*/ |
|
255
|
13 |
|
public function getEmbedded() { |
|
256
|
13 |
|
return $this->_options['embedded']; |
|
257
|
|
|
} |
|
258
|
|
|
|
|
259
|
|
|
/** |
|
260
|
|
|
* @param Language $lang |
|
261
|
|
|
*/ |
|
262
|
1 |
|
public function embed(Language $lang) { |
|
263
|
1 |
|
$this->_options['embedded'][] = $lang; |
|
264
|
1 |
|
} |
|
265
|
|
|
|
|
266
|
1 |
|
public function __get($name) { |
|
267
|
1 |
|
return isset($this->_options[$name]) ? $this->_options[$name] : null; |
|
268
|
|
|
} |
|
269
|
|
|
|
|
270
|
1 |
|
public function __set($name, $value) { |
|
271
|
1 |
|
$this->_options[$name] = $value; |
|
272
|
1 |
|
} |
|
273
|
|
|
} |
|
274
|
|
|
|