|
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\MetaToken; |
|
21
|
|
|
use Kadet\Highlighter\Parser\Result; |
|
22
|
|
|
use Kadet\Highlighter\Parser\Rule; |
|
23
|
|
|
use Kadet\Highlighter\Parser\Token; |
|
24
|
|
|
use Kadet\Highlighter\Parser\TokenFactory; |
|
25
|
|
|
use Kadet\Highlighter\Parser\TokenIterator; |
|
26
|
|
|
use Kadet\Highlighter\Parser\UnprocessedTokens; |
|
27
|
|
|
use Kadet\Highlighter\Utils\ArrayHelper; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* Class Language |
|
31
|
|
|
* |
|
32
|
|
|
* @package Kadet\Highlighter\Language |
|
33
|
|
|
*/ |
|
34
|
|
|
abstract class Language |
|
35
|
|
|
{ |
|
36
|
|
|
/** |
|
37
|
|
|
* @var array |
|
38
|
|
|
*/ |
|
39
|
|
|
protected $_options = []; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Tokenizer rules |
|
43
|
|
|
* |
|
44
|
|
|
* @var Rule[]|Rule[][] |
|
45
|
|
|
*/ |
|
46
|
|
|
private $_rules; |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* @var array |
|
50
|
|
|
*/ |
|
51
|
|
|
private $_context; |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* @var Result |
|
55
|
|
|
*/ |
|
56
|
|
|
private $_result; |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* @var TokenIterator |
|
60
|
|
|
*/ |
|
61
|
|
|
private $_iterator; |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* @var LanguageToken |
|
65
|
|
|
*/ |
|
66
|
|
|
private $_start; |
|
67
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* Language constructor. |
|
70
|
|
|
* |
|
71
|
|
|
* @param array $options |
|
72
|
|
|
*/ |
|
73
|
22 |
|
public function __construct(array $options = []) |
|
74
|
|
|
{ |
|
75
|
22 |
|
$this->_options = array_merge([ |
|
76
|
22 |
|
'embedded' => [], |
|
77
|
22 |
|
], $this->_options, $options); |
|
78
|
|
|
|
|
79
|
22 |
|
$this->_rules = $this->getRules(); |
|
80
|
22 |
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* Tokenization rules definition |
|
84
|
|
|
* |
|
85
|
|
|
* @return Rule[]|Rule[][] |
|
86
|
|
|
*/ |
|
87
|
|
|
abstract public function getRules(); |
|
88
|
|
|
|
|
89
|
|
|
/** |
|
90
|
|
|
* Parses source and removes wrong tokens. |
|
91
|
|
|
* |
|
92
|
|
|
* @param TokenIterator|string $tokens |
|
93
|
|
|
* |
|
94
|
|
|
* @param array $additional |
|
95
|
|
|
* @param bool $embedded |
|
96
|
|
|
* |
|
97
|
|
|
* @return TokenIterator |
|
98
|
|
|
*/ |
|
99
|
12 |
|
public function parse($tokens = null, $additional = [], $embedded = false) |
|
100
|
|
|
{ |
|
101
|
12 |
|
if (is_string($tokens)) { |
|
102
|
11 |
|
$tokens = $this->tokenize($tokens, $additional, $embedded); |
|
103
|
12 |
|
} elseif (!$tokens instanceof TokenIterator) { |
|
104
|
|
|
// Todo: Own Exceptions |
|
105
|
1 |
|
throw new \InvalidArgumentException('$tokens must be string or TokenIterator'); |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
// Reset variables to default state |
|
109
|
11 |
|
$this->_start = $tokens->current(); |
|
110
|
11 |
|
$this->_context = []; |
|
111
|
11 |
|
$this->_result = new Result($tokens->getSource(), [ |
|
112
|
11 |
|
$this->_start |
|
113
|
11 |
|
]); |
|
114
|
11 |
|
$this->_iterator = $tokens; |
|
115
|
|
|
|
|
116
|
|
|
/** @var Token $token */ |
|
117
|
11 |
|
for ($tokens->next(); $tokens->valid(); $tokens->next()) { |
|
118
|
11 |
|
$token = $tokens->current(); |
|
119
|
|
|
|
|
120
|
12 |
|
if ($token->isValid($this, $this->_context)) { |
|
121
|
11 |
|
if($this->_handleToken($token) === false) { |
|
122
|
11 |
|
break; |
|
123
|
|
|
}; |
|
124
|
10 |
|
} |
|
125
|
10 |
|
} |
|
126
|
|
|
|
|
127
|
11 |
|
return $this->_result; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
12 |
|
public function tokenize($source, $additional = [], $offset = 0, $embedded = false) |
|
131
|
|
|
{ |
|
132
|
12 |
|
return new TokenIterator( |
|
133
|
12 |
|
$this->_tokens($source, $offset, $additional, $embedded)->sort()->toArray(), $source |
|
134
|
12 |
|
); |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
/** |
|
138
|
|
|
* Tokenize source |
|
139
|
|
|
* |
|
140
|
|
|
* @param $source |
|
141
|
|
|
* |
|
142
|
|
|
* @param int $offset |
|
143
|
|
|
* @param array $additional |
|
144
|
|
|
* |
|
145
|
|
|
* @param bool $embedded |
|
146
|
|
|
* |
|
147
|
|
|
* @return UnprocessedTokens |
|
148
|
|
|
*/ |
|
149
|
12 |
|
private function _tokens($source, $offset = 0, $additional = [], $embedded = false) |
|
150
|
|
|
{ |
|
151
|
12 |
|
$result = new UnprocessedTokens(); |
|
152
|
|
|
|
|
153
|
|
|
/** @var Language $language */ |
|
154
|
12 |
|
foreach ($this->_rules($embedded) as $rule) { |
|
155
|
12 |
|
$rule->factory->setOffset($offset); |
|
156
|
12 |
|
foreach ($rule->match($source) as $token) { |
|
157
|
12 |
|
$result->add($token); |
|
158
|
12 |
|
} |
|
159
|
12 |
|
} |
|
160
|
|
|
|
|
161
|
12 |
|
return $result->batch($additional); |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
/** |
|
165
|
|
|
* @param bool $embedded |
|
166
|
|
|
* |
|
167
|
|
|
* @return Rule[] |
|
168
|
|
|
*/ |
|
169
|
12 |
|
private function _rules($embedded = false) |
|
170
|
|
|
{ |
|
171
|
12 |
|
$all = $this->_rules; |
|
172
|
12 |
|
if (!$embedded) { |
|
173
|
12 |
|
$all['language.' . $this->getIdentifier()] = $this->getOpenClose(); |
|
174
|
12 |
|
} |
|
175
|
|
|
|
|
176
|
|
|
// why this code sucks so much? Because RecursiveIterator performance such a lot more. |
|
177
|
12 |
|
foreach ($all as $name => $rules) { |
|
178
|
12 |
|
if (!is_array($rules)) { |
|
179
|
12 |
|
$rules = [$rules]; |
|
180
|
12 |
|
} |
|
181
|
|
|
|
|
182
|
|
|
/** @var Rule $rule */ |
|
183
|
12 |
|
foreach ($rules as $rule) { |
|
184
|
12 |
|
if ($rule->language === false) { |
|
185
|
11 |
|
$rule->language = $this; |
|
186
|
11 |
|
} |
|
187
|
|
|
|
|
188
|
12 |
|
$rule->factory->setBase($name); |
|
189
|
|
|
|
|
190
|
12 |
|
yield $rule; |
|
191
|
12 |
|
} |
|
192
|
12 |
|
} |
|
193
|
|
|
|
|
194
|
12 |
|
foreach ($this->getEmbedded() as $language) { |
|
195
|
1 |
|
foreach ($language->_rules() as $rule) { |
|
196
|
1 |
|
yield $rule; |
|
197
|
1 |
|
} |
|
198
|
12 |
|
} |
|
199
|
12 |
|
} |
|
200
|
|
|
|
|
201
|
|
|
/** |
|
202
|
|
|
* Unique language identifier, for example 'php' |
|
203
|
|
|
* |
|
204
|
|
|
* @return string |
|
205
|
|
|
*/ |
|
206
|
|
|
abstract public function getIdentifier(); |
|
207
|
|
|
|
|
208
|
|
|
/** |
|
209
|
|
|
* Language range Rule(s) |
|
210
|
|
|
* |
|
211
|
|
|
* @return Rule|Rule[] |
|
212
|
|
|
*/ |
|
213
|
16 |
|
public function getOpenClose() |
|
214
|
|
|
{ |
|
215
|
16 |
|
return new Rule( |
|
216
|
16 |
|
new WholeMatcher(), [ |
|
217
|
16 |
|
'priority' => 1000, |
|
218
|
16 |
|
'factory' => new TokenFactory(LanguageToken::class), |
|
219
|
16 |
|
'inject' => $this, |
|
220
|
16 |
|
'language' => null, |
|
221
|
16 |
|
'context' => Rule::everywhere(), |
|
222
|
|
|
] |
|
223
|
16 |
|
); |
|
224
|
|
|
} |
|
225
|
|
|
|
|
226
|
|
|
/** |
|
227
|
|
|
* @return Language[] |
|
228
|
|
|
*/ |
|
229
|
13 |
|
public function getEmbedded() |
|
230
|
|
|
{ |
|
231
|
13 |
|
return $this->_options['embedded']; |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
|
|
/** |
|
235
|
|
|
* @param Language $lang |
|
236
|
|
|
*/ |
|
237
|
1 |
|
public function embed(Language $lang) |
|
238
|
|
|
{ |
|
239
|
1 |
|
$this->_options['embedded'][] = $lang; |
|
240
|
1 |
|
} |
|
241
|
|
|
|
|
242
|
1 |
|
public function __get($name) |
|
243
|
|
|
{ |
|
244
|
1 |
|
return isset($this->_options[$name]) ? $this->_options[$name] : null; |
|
245
|
|
|
} |
|
246
|
|
|
|
|
247
|
1 |
|
public function __set($name, $value) |
|
248
|
|
|
{ |
|
249
|
1 |
|
$this->_options[$name] = $value; |
|
250
|
1 |
|
} |
|
251
|
|
|
|
|
252
|
11 |
|
protected function _handleToken(Token $token) |
|
253
|
|
|
{ |
|
254
|
11 |
|
if ($token->isStart()) { |
|
255
|
10 |
|
if ($token instanceof LanguageToken) { |
|
256
|
2 |
|
$this->_result->merge($token->getInjected()->parse($this->_iterator)); |
|
257
|
2 |
|
} else { |
|
258
|
10 |
|
$this->_result[] = $token; |
|
259
|
10 |
|
$this->_context[$this->_iterator->key()] = $token->name; |
|
260
|
|
|
} |
|
261
|
10 |
|
} else { |
|
262
|
11 |
|
$start = $token->getStart(); |
|
263
|
|
|
|
|
264
|
|
|
/** @noinspection PhpUndefinedMethodInspection bug */ |
|
265
|
11 |
|
if ($token instanceof LanguageToken && $token->getLanguage() === $this) { |
|
266
|
11 |
|
$this->_start->setEnd($token); |
|
267
|
|
|
|
|
268
|
11 |
|
if ($this->_start->postProcess) { |
|
|
|
|
|
|
269
|
1 |
|
$source = substr($this->_iterator->getSource(), $this->_start->pos, $this->_start->getLength()); |
|
270
|
|
|
|
|
271
|
1 |
|
$tokens = $this->tokenize($source, $this->_result, $this->_start->pos, true); |
|
|
|
|
|
|
272
|
1 |
|
$this->_result = $this->parse($tokens); |
|
273
|
1 |
|
} |
|
274
|
|
|
|
|
275
|
|
|
# closing unclosed tokens |
|
276
|
11 |
|
foreach (array_reverse($this->_context) as $hash => $name) { |
|
277
|
1 |
|
$end = new Token([$name, 'pos' => $token->pos]); |
|
278
|
1 |
|
$this->_iterator[$hash]->setEnd($end); |
|
279
|
1 |
|
$this->_result[] = $end; |
|
280
|
11 |
|
} |
|
281
|
|
|
|
|
282
|
11 |
|
$this->_result[] = $token; |
|
283
|
11 |
|
return false; |
|
284
|
|
|
} else { |
|
285
|
9 |
|
if ($start) { |
|
286
|
8 |
|
unset($this->_context[spl_object_hash($start)]); |
|
287
|
8 |
|
} else { |
|
288
|
1 |
|
$start = ArrayHelper::find(array_reverse($this->_context), function ($k, $v) use ($token) { |
|
289
|
1 |
|
return $v === $token->name; |
|
290
|
1 |
|
}); |
|
291
|
|
|
|
|
292
|
1 |
|
if ($start !== false) { |
|
293
|
1 |
|
$token->setStart($this->_iterator[$start]); |
|
294
|
1 |
|
unset($this->_context[$start]); |
|
295
|
1 |
|
$start = $this->_iterator[$start]; |
|
296
|
1 |
|
} |
|
297
|
|
|
} |
|
298
|
|
|
|
|
299
|
9 |
|
if (!$start instanceof MetaToken) { |
|
300
|
9 |
|
$this->_result[] = $token; |
|
301
|
9 |
|
} |
|
302
|
|
|
} |
|
303
|
|
|
} |
|
304
|
|
|
|
|
305
|
10 |
|
return true; |
|
306
|
|
|
} |
|
307
|
|
|
} |
|
308
|
|
|
|
Since your code implements the magic getter
_get, this function will be called for any read access on an undefined variable. You can add the@propertyannotation to your class or interface to document the existence of this variable.If the property has read access only, you can use the @property-read annotation instead.
Of course, you may also just have mistyped another name, in which case you should fix the error.
See also the PhpDoc documentation for @property.