1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* Unicode UTF-8 token matcher. |
5
|
|
|
* |
6
|
|
|
* Auto-generated file, please don't edit manually. |
7
|
|
|
* Generated by UniLex. |
8
|
|
|
*/ |
9
|
|
|
|
10
|
|
|
declare(strict_types=1); |
11
|
|
|
|
12
|
|
|
namespace Remorhaz\UniLex\Unicode\Grammar; |
13
|
|
|
|
14
|
|
|
use Remorhaz\UniLex\IO\CharBufferInterface; |
15
|
|
|
use Remorhaz\UniLex\Lexer\TokenFactoryInterface; |
16
|
|
|
use Remorhaz\UniLex\Lexer\TokenMatcherTemplate; |
17
|
|
|
|
18
|
|
|
class Utf8TokenMatcher extends TokenMatcherTemplate |
19
|
|
|
{ |
20
|
|
|
|
21
|
|
|
public function match(CharBufferInterface $buffer, TokenFactoryInterface $tokenFactory): bool |
22
|
|
|
{ |
23
|
|
|
$context = $this->createContext($buffer, $tokenFactory); |
24
|
|
|
goto state1; |
25
|
|
|
|
26
|
|
|
state1: |
27
|
|
|
if ($context->getBuffer()->isEnd()) { |
28
|
|
|
goto error; |
29
|
|
|
} |
30
|
|
|
$char = $context->getBuffer()->getSymbol(); |
31
|
|
|
if (0x00 <= $char && $char <= 0x7F) { |
32
|
|
|
$context->getBuffer()->nextSymbol(); |
33
|
|
|
// [\x00-\x7F] |
34
|
|
|
$context |
35
|
|
|
->setNewToken(TokenType::SYMBOL) |
36
|
|
|
->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $char); |
37
|
|
|
|
38
|
|
|
return true; |
39
|
|
|
} |
40
|
|
|
if (0xC0 <= $char && $char <= 0xDF) { |
41
|
|
|
$context->getBuffer()->nextSymbol(); |
42
|
|
|
goto state3; |
43
|
|
|
} |
44
|
|
|
if (0xE0 <= $char && $char <= 0xEF) { |
45
|
|
|
$context->getBuffer()->nextSymbol(); |
46
|
|
|
goto state4; |
47
|
|
|
} |
48
|
|
|
if (0xF0 <= $char && $char <= 0xF7) { |
49
|
|
|
$context->getBuffer()->nextSymbol(); |
50
|
|
|
goto state5; |
51
|
|
|
} |
52
|
|
|
if (0xF8 <= $char && $char <= 0xFB) { |
53
|
|
|
$context->getBuffer()->nextSymbol(); |
54
|
|
|
goto state6; |
55
|
|
|
} |
56
|
|
|
if (0xFC == $char || 0xFD == $char) { |
57
|
|
|
$context->getBuffer()->nextSymbol(); |
58
|
|
|
goto state7; |
59
|
|
|
} |
60
|
|
|
goto error; |
61
|
|
|
|
62
|
|
|
state3: |
63
|
|
|
if ($context->getBuffer()->isEnd()) { |
64
|
|
|
goto error; |
65
|
|
|
} |
66
|
|
|
$char = $context->getBuffer()->getSymbol(); |
67
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
68
|
|
|
$context->getBuffer()->nextSymbol(); |
69
|
|
|
// [\xC0-\xDF][\x80-\xBF] |
70
|
|
|
$charList = array_slice($context->getSymbolList(), -2); |
71
|
|
|
$symbol = ($charList[0] & 0x1F) << 6; |
72
|
|
|
$symbol |= ($charList[1] & 0x3F); |
73
|
|
|
$context |
74
|
|
|
->setNewToken(TokenType::SYMBOL) |
75
|
|
|
->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol); |
76
|
|
|
|
77
|
|
|
return true; |
78
|
|
|
} |
79
|
|
|
goto error; |
80
|
|
|
|
81
|
|
|
state4: |
82
|
|
|
if ($context->getBuffer()->isEnd()) { |
83
|
|
|
goto error; |
84
|
|
|
} |
85
|
|
|
$char = $context->getBuffer()->getSymbol(); |
86
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
87
|
|
|
$context->getBuffer()->nextSymbol(); |
88
|
|
|
goto state20; |
89
|
|
|
} |
90
|
|
|
goto error; |
91
|
|
|
|
92
|
|
|
state5: |
93
|
|
|
if ($context->getBuffer()->isEnd()) { |
94
|
|
|
goto error; |
95
|
|
|
} |
96
|
|
|
$char = $context->getBuffer()->getSymbol(); |
97
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
98
|
|
|
$context->getBuffer()->nextSymbol(); |
99
|
|
|
goto state17; |
100
|
|
|
} |
101
|
|
|
goto error; |
102
|
|
|
|
103
|
|
|
state6: |
104
|
|
|
if ($context->getBuffer()->isEnd()) { |
105
|
|
|
goto error; |
106
|
|
|
} |
107
|
|
|
$char = $context->getBuffer()->getSymbol(); |
108
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
109
|
|
|
$context->getBuffer()->nextSymbol(); |
110
|
|
|
goto state13; |
111
|
|
|
} |
112
|
|
|
goto error; |
113
|
|
|
|
114
|
|
|
state7: |
115
|
|
|
if ($context->getBuffer()->isEnd()) { |
116
|
|
|
goto error; |
117
|
|
|
} |
118
|
|
|
$char = $context->getBuffer()->getSymbol(); |
119
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
120
|
|
|
$context->getBuffer()->nextSymbol(); |
121
|
|
|
goto state8; |
122
|
|
|
} |
123
|
|
|
goto error; |
124
|
|
|
|
125
|
|
|
state8: |
126
|
|
|
if ($context->getBuffer()->isEnd()) { |
127
|
|
|
goto error; |
128
|
|
|
} |
129
|
|
|
$char = $context->getBuffer()->getSymbol(); |
130
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
131
|
|
|
$context->getBuffer()->nextSymbol(); |
132
|
|
|
goto state9; |
133
|
|
|
} |
134
|
|
|
goto error; |
135
|
|
|
|
136
|
|
|
state9: |
137
|
|
|
if ($context->getBuffer()->isEnd()) { |
138
|
|
|
goto error; |
139
|
|
|
} |
140
|
|
|
$char = $context->getBuffer()->getSymbol(); |
141
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
142
|
|
|
$context->getBuffer()->nextSymbol(); |
143
|
|
|
goto state10; |
144
|
|
|
} |
145
|
|
|
goto error; |
146
|
|
|
|
147
|
|
|
state10: |
148
|
|
|
if ($context->getBuffer()->isEnd()) { |
149
|
|
|
goto error; |
150
|
|
|
} |
151
|
|
|
$char = $context->getBuffer()->getSymbol(); |
152
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
153
|
|
|
$context->getBuffer()->nextSymbol(); |
154
|
|
|
goto state11; |
155
|
|
|
} |
156
|
|
|
goto error; |
157
|
|
|
|
158
|
|
|
state11: |
159
|
|
|
if ($context->getBuffer()->isEnd()) { |
160
|
|
|
goto error; |
161
|
|
|
} |
162
|
|
|
$char = $context->getBuffer()->getSymbol(); |
163
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
164
|
|
|
$context->getBuffer()->nextSymbol(); |
165
|
|
|
// [\xFC-\xFD][\x80-\xBF]{5} |
166
|
|
|
$charList = array_slice($context->getSymbolList(), -6); |
167
|
|
|
$symbol = ($charList[0] & 0x01) << 30; |
168
|
|
|
$symbol |= ($charList[1] & 0x03) << 24; |
169
|
|
|
$symbol |= ($charList[2] & 0x3F) << 18; |
170
|
|
|
$symbol |= ($charList[3] & 0x3F) << 12; |
171
|
|
|
$symbol |= ($charList[4] & 0x3F) << 6; |
172
|
|
|
$symbol |= ($charList[5] & 0x3F); |
173
|
|
|
$context |
174
|
|
|
->setNewToken(TokenType::SYMBOL) |
175
|
|
|
->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol); |
176
|
|
|
|
177
|
|
|
return true; |
178
|
|
|
} |
179
|
|
|
goto error; |
180
|
|
|
|
181
|
|
|
state13: |
182
|
|
|
if ($context->getBuffer()->isEnd()) { |
183
|
|
|
goto error; |
184
|
|
|
} |
185
|
|
|
$char = $context->getBuffer()->getSymbol(); |
186
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
187
|
|
|
$context->getBuffer()->nextSymbol(); |
188
|
|
|
goto state14; |
189
|
|
|
} |
190
|
|
|
goto error; |
191
|
|
|
|
192
|
|
|
state14: |
193
|
|
|
if ($context->getBuffer()->isEnd()) { |
194
|
|
|
goto error; |
195
|
|
|
} |
196
|
|
|
$char = $context->getBuffer()->getSymbol(); |
197
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
198
|
|
|
$context->getBuffer()->nextSymbol(); |
199
|
|
|
goto state15; |
200
|
|
|
} |
201
|
|
|
goto error; |
202
|
|
|
|
203
|
|
|
state15: |
204
|
|
|
if ($context->getBuffer()->isEnd()) { |
205
|
|
|
goto error; |
206
|
|
|
} |
207
|
|
|
$char = $context->getBuffer()->getSymbol(); |
208
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
209
|
|
|
$context->getBuffer()->nextSymbol(); |
210
|
|
|
// [\xF8-\xFB][\x80-\xBF]{4} |
211
|
|
|
$charList = array_slice($context->getSymbolList(), -5); |
212
|
|
|
$symbol = ($charList[0] & 0x03) << 24; |
213
|
|
|
$symbol |= ($charList[1] & 0x3F) << 18; |
214
|
|
|
$symbol |= ($charList[2] & 0x3F) << 12; |
215
|
|
|
$symbol |= ($charList[3] & 0x3F) << 6; |
216
|
|
|
$symbol |= ($charList[4] & 0x3F); |
217
|
|
|
$context |
218
|
|
|
->setNewToken(TokenType::SYMBOL) |
219
|
|
|
->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol); |
220
|
|
|
|
221
|
|
|
return true; |
222
|
|
|
} |
223
|
|
|
goto error; |
224
|
|
|
|
225
|
|
|
state17: |
226
|
|
|
if ($context->getBuffer()->isEnd()) { |
227
|
|
|
goto error; |
228
|
|
|
} |
229
|
|
|
$char = $context->getBuffer()->getSymbol(); |
230
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
231
|
|
|
$context->getBuffer()->nextSymbol(); |
232
|
|
|
goto state18; |
233
|
|
|
} |
234
|
|
|
goto error; |
235
|
|
|
|
236
|
|
|
state18: |
237
|
|
|
if ($context->getBuffer()->isEnd()) { |
238
|
|
|
goto error; |
239
|
|
|
} |
240
|
|
|
$char = $context->getBuffer()->getSymbol(); |
241
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
242
|
|
|
$context->getBuffer()->nextSymbol(); |
243
|
|
|
// [\xF0-\xF7][\x80-\xBF]{3} |
244
|
|
|
$charList = array_slice($context->getSymbolList(), -4); |
245
|
|
|
$symbol = ($charList[0] & 0x07) << 18; |
246
|
|
|
$symbol |= ($charList[1] & 0x3F) << 12; |
247
|
|
|
$symbol |= ($charList[2] & 0x3F) << 6; |
248
|
|
|
$symbol |= ($charList[3] & 0x3F); |
249
|
|
|
$context |
250
|
|
|
->setNewToken(TokenType::SYMBOL) |
251
|
|
|
->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol); |
252
|
|
|
|
253
|
|
|
return true; |
254
|
|
|
} |
255
|
|
|
goto error; |
256
|
|
|
|
257
|
|
|
state20: |
258
|
|
|
if ($context->getBuffer()->isEnd()) { |
259
|
|
|
goto error; |
260
|
|
|
} |
261
|
|
|
$char = $context->getBuffer()->getSymbol(); |
262
|
|
|
if (0x80 <= $char && $char <= 0xBF) { |
263
|
|
|
$context->getBuffer()->nextSymbol(); |
264
|
|
|
// [\xE0-\xEF][\x80-\xBF]{2} |
265
|
|
|
$charList = array_slice($context->getSymbolList(), -3); |
266
|
|
|
$symbol = ($charList[0] & 0x0F) << 12; |
267
|
|
|
$symbol |= ($charList[1] & 0x3F) << 6; |
268
|
|
|
$symbol |= ($charList[2] & 0x3F); |
269
|
|
|
$context |
270
|
|
|
->setNewToken(TokenType::SYMBOL) |
271
|
|
|
->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol); |
272
|
|
|
|
273
|
|
|
return true; |
274
|
|
|
} |
275
|
|
|
goto error; |
276
|
|
|
|
277
|
|
|
error: |
278
|
|
|
if ($context->getBuffer()->isEnd()) { |
279
|
|
|
return false; |
280
|
|
|
} |
281
|
|
|
$context->getBuffer()->nextSymbol(); |
282
|
|
|
$context->setNewToken(TokenType::INVALID_BYTES); |
283
|
|
|
return true; |
284
|
|
|
} |
285
|
|
|
} |
286
|
|
|
|