1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Vlaswinkel\Lua; |
4
|
|
|
|
5
|
|
|
/** |
6
|
|
|
* Class TokenStream |
7
|
|
|
* |
8
|
|
|
* @see http://lisperator.net/pltut/parser/token-stream |
9
|
|
|
* |
10
|
|
|
* @author Koen Vlaswinkel <[email protected]> |
11
|
|
|
* @package Vlaswinkel\Lua |
12
|
|
|
*/ |
13
|
|
|
class TokenStream { |
14
|
|
|
private $current = null; |
15
|
|
|
/** |
16
|
|
|
* @var InputStream |
17
|
|
|
*/ |
18
|
|
|
private $input; |
19
|
|
|
|
20
|
|
|
/** |
21
|
|
|
* TokenStream constructor. |
22
|
|
|
* |
23
|
|
|
* @param InputStream $input |
24
|
|
|
*/ |
25
|
39 |
|
public function __construct(InputStream $input) { |
26
|
39 |
|
$this->input = $input; |
27
|
39 |
|
} |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* @return Token |
31
|
|
|
*/ |
32
|
39 |
|
public function next() { |
33
|
39 |
|
$token = $this->current; |
34
|
39 |
|
$this->current = null; |
35
|
39 |
|
if ($token) { |
36
|
23 |
|
return $token; |
37
|
|
|
} |
38
|
16 |
|
return $this->readNext(); |
39
|
|
|
} |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* @return bool |
43
|
|
|
*/ |
44
|
22 |
|
public function eof() { |
45
|
22 |
|
return $this->peek() == null; |
46
|
|
|
} |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* @return Token |
|
|
|
|
50
|
|
|
*/ |
51
|
23 |
|
public function peek() { |
52
|
23 |
|
if ($this->current) { |
53
|
23 |
|
return $this->current; |
54
|
|
|
} |
55
|
23 |
|
$this->current = $this->readNext(); |
56
|
23 |
|
return $this->current; |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* @param string $msg |
61
|
|
|
* |
62
|
|
|
* @throws ParseException |
63
|
|
|
*/ |
64
|
4 |
|
public function error($msg) { |
65
|
4 |
|
$this->input->error($msg); |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* @return Token |
|
|
|
|
70
|
|
|
* @throws ParseException |
71
|
|
|
*/ |
72
|
39 |
|
protected function readNext() { |
73
|
39 |
|
$this->readWhile([$this, 'isWhitespace']); |
74
|
39 |
|
if ($this->input->eof()) { |
75
|
19 |
|
return null; |
76
|
|
|
} |
77
|
39 |
|
$char = $this->input->peek(); |
78
|
39 |
|
if ($this->isComment()) { |
79
|
5 |
|
$this->skipComment(); |
80
|
5 |
|
return $this->readNext(); |
81
|
|
|
} |
82
|
39 |
|
if ($char == '"') { |
83
|
13 |
|
return $this->readDoubleQuotedString(); |
84
|
|
|
} |
85
|
34 |
|
if ($char == '\'') { |
86
|
1 |
|
return $this->readSingleQuotedString(); |
87
|
|
|
} |
88
|
33 |
|
if ($this->isDoubleBracketString()) { |
89
|
12 |
|
return $this->readDoubleBracketString(); |
90
|
|
|
} |
91
|
24 |
|
if ($this->isDigit($char)) { |
92
|
10 |
|
return $this->readNumber(); |
93
|
|
|
} |
94
|
20 |
|
if ($this->isStartIdentifierCharacter($char)) { |
95
|
16 |
|
return $this->readIdentifier(); |
96
|
|
|
} |
97
|
15 |
|
if ($this->isPunctuation($char)) { |
98
|
14 |
|
return $this->readPunctuation(); |
99
|
|
|
} |
100
|
1 |
|
$this->input->error('Cannot handle character: ' . $char . ' (ord: ' . ord($char) . ')'); |
101
|
|
|
} |
102
|
|
|
|
103
|
5 |
|
protected function skipComment() { |
104
|
5 |
|
$this->readWhile( |
105
|
|
|
function ($char) { |
106
|
5 |
|
return $char != "\n"; |
107
|
|
|
} |
108
|
5 |
|
); |
109
|
5 |
|
$this->input->next(); |
110
|
5 |
|
} |
111
|
|
|
|
112
|
|
|
/** |
113
|
|
|
* @return Token |
114
|
|
|
*/ |
115
|
13 |
|
protected function readDoubleQuotedString() { |
116
|
13 |
|
return new Token(Token::TYPE_STRING, $this->readEscaped('"')); |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
/** |
120
|
|
|
* @return Token |
121
|
|
|
*/ |
122
|
1 |
|
protected function readSingleQuotedString() { |
123
|
1 |
|
return new Token(Token::TYPE_STRING, $this->readEscaped('\'')); |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
/** |
127
|
|
|
* @return Token |
128
|
|
|
*/ |
129
|
12 |
|
protected function readDoubleBracketString() { |
130
|
|
|
// we cannot use readEscaped because it only supports a single char as $end |
131
|
|
|
// and we do not support escaping in double bracke strings |
132
|
12 |
|
$str = ""; |
133
|
12 |
|
$startNumberOfEqualsSigns = 0; |
134
|
|
|
// skip both |
135
|
12 |
|
$this->input->next(); |
136
|
12 |
|
while ($this->input->peek() == '=') { |
137
|
7 |
|
$startNumberOfEqualsSigns++; |
138
|
7 |
|
$this->input->next(); |
139
|
7 |
|
} |
140
|
12 |
|
if ($this->input->peek() != '[') { |
141
|
1 |
|
$this->error('Unexpected character \'' . $this->input->peek() . '\', expected \'[\''); |
142
|
|
|
} |
143
|
11 |
|
$this->input->next(); |
144
|
11 |
|
while (!$this->input->eof()) { |
145
|
11 |
|
$char = $this->input->next(); |
146
|
11 |
|
if ($char == ']') { // we might have reached the end |
147
|
11 |
|
if ($startNumberOfEqualsSigns != 0) { |
148
|
6 |
|
if ($this->input->peek() == '=') { |
149
|
5 |
|
$endNumberOfEqualsSigns = 0; |
150
|
5 |
|
while ($this->input->peek() == '=') { |
151
|
5 |
|
$endNumberOfEqualsSigns++; |
152
|
5 |
|
$this->input->next(); |
153
|
5 |
|
} |
154
|
|
|
|
155
|
|
|
// we have an equal number of equal signs |
156
|
5 |
|
if ($endNumberOfEqualsSigns == $startNumberOfEqualsSigns) { |
157
|
5 |
|
if ($this->input->peek() != ']') { |
158
|
1 |
|
$this->error('Unexpected character \'' . $this->input->peek() . '\', expected \'[\''); |
159
|
|
|
} |
160
|
4 |
|
$this->input->next(); |
161
|
4 |
|
break; |
162
|
|
|
} else { |
163
|
1 |
|
$str .= $char . str_repeat('=', $endNumberOfEqualsSigns); |
164
|
|
|
} |
165
|
1 |
|
} else { |
166
|
3 |
|
$str .= $char; |
167
|
|
|
} |
168
|
3 |
|
} else { |
169
|
5 |
|
if ($this->input->peek() == ']') { |
170
|
5 |
|
$this->input->next(); |
171
|
5 |
|
break; |
172
|
|
|
} |
173
|
|
|
} |
174
|
3 |
|
} else { |
175
|
11 |
|
$str .= $char; |
176
|
|
|
} |
177
|
11 |
|
} |
178
|
9 |
|
return new Token(Token::TYPE_STRING, $str); |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
/** |
182
|
|
|
* @param string $end |
183
|
|
|
* |
184
|
|
|
* @return string |
185
|
|
|
*/ |
186
|
14 |
|
protected function readEscaped($end) { |
187
|
14 |
|
$escaped = false; |
188
|
14 |
|
$str = ""; |
189
|
14 |
|
$this->input->next(); |
190
|
14 |
|
while (!$this->input->eof()) { |
191
|
14 |
|
$char = $this->input->next(); |
192
|
14 |
|
if ($escaped) { |
193
|
|
|
switch ($char) { |
194
|
1 |
|
case 'n': |
195
|
1 |
|
$str .= "\n"; |
196
|
1 |
|
break; |
197
|
1 |
|
case 'r': |
198
|
1 |
|
$str .= "\r"; |
199
|
1 |
|
break; |
200
|
1 |
|
case 't': |
201
|
1 |
|
$str .= "\t"; |
202
|
1 |
|
break; |
203
|
1 |
|
case 'v': |
204
|
1 |
|
$str .= "\v"; |
205
|
1 |
|
break; |
206
|
1 |
|
default: |
207
|
1 |
|
$str .= $char; |
208
|
1 |
|
break; |
209
|
1 |
|
} |
210
|
1 |
|
$escaped = false; |
211
|
1 |
|
} else { |
212
|
14 |
|
if ($char == "\\") { |
213
|
1 |
|
$escaped = true; |
214
|
1 |
|
} else { |
215
|
14 |
|
if ($char == $end) { |
216
|
14 |
|
break; |
217
|
|
|
} else { |
218
|
14 |
|
$str .= $char; |
219
|
|
|
} |
220
|
|
|
} |
221
|
|
|
} |
222
|
14 |
|
} |
223
|
14 |
|
return $str; |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
/** |
227
|
|
|
* @return Token |
228
|
|
|
*/ |
229
|
10 |
|
protected function readNumber() { |
230
|
10 |
|
$hasDot = false; |
231
|
10 |
|
$number = $this->readWhile( |
232
|
|
|
function ($char) use (&$hasDot) { |
233
|
10 |
|
if ($char == '.') { |
234
|
1 |
|
if ($hasDot) { |
235
|
|
|
return false; |
236
|
|
|
} |
237
|
1 |
|
$hasDot = true; |
238
|
1 |
|
return true; |
239
|
|
|
} |
240
|
10 |
|
return $this->isDigit($char); |
241
|
|
|
} |
242
|
10 |
|
); |
243
|
10 |
|
return new Token(Token::TYPE_NUMBER, $hasDot ? floatval($number) : intval($number)); |
244
|
|
|
} |
245
|
|
|
|
246
|
|
|
/** |
247
|
|
|
* @return Token |
248
|
|
|
*/ |
249
|
16 |
|
protected function readIdentifier() { |
250
|
16 |
|
$first = false; |
251
|
16 |
|
$identifier = $this->readWhile( |
252
|
16 |
|
function ($char) use (&$first) { |
253
|
16 |
|
if ($first) { |
254
|
|
|
$first = false; |
255
|
|
|
return $this->isStartIdentifierCharacter($char); |
256
|
|
|
} |
257
|
16 |
|
return $this->isIdentifierCharacter($char); |
258
|
|
|
} |
259
|
16 |
|
); |
260
|
16 |
|
if ($this->isKeyword($identifier)) { |
261
|
6 |
|
return new Token(Token::TYPE_KEYWORD, $identifier); |
262
|
|
|
} |
263
|
12 |
|
return new Token(Token::TYPE_IDENTIFIER, $identifier); |
264
|
|
|
} |
265
|
|
|
|
266
|
|
|
/** |
267
|
|
|
* @return Token |
268
|
|
|
*/ |
269
|
14 |
|
protected function readPunctuation() { |
270
|
14 |
|
return new Token(Token::TYPE_PUNCTUATION, $this->input->next()); |
271
|
|
|
} |
272
|
|
|
|
273
|
|
|
/** |
274
|
|
|
* @param callable $predicate |
275
|
|
|
* |
276
|
|
|
* @return string |
277
|
|
|
*/ |
278
|
39 |
|
protected function readWhile(callable $predicate) { |
279
|
39 |
|
$str = ""; |
280
|
39 |
|
while (!$this->input->eof() && call_user_func($predicate, $this->input->peek())) { |
281
|
20 |
|
$str .= $this->input->next(); |
282
|
20 |
|
} |
283
|
39 |
|
return $str; |
284
|
|
|
} |
285
|
|
|
|
286
|
|
|
/** |
287
|
|
|
* @param string $char |
288
|
|
|
* |
289
|
|
|
* @return bool |
290
|
|
|
*/ |
291
|
39 |
|
protected function isWhitespace($char) { |
292
|
39 |
|
return strpos(" \t\n\r", $char) !== false; |
293
|
|
|
} |
294
|
|
|
|
295
|
|
|
/** |
296
|
|
|
* @param string $char |
297
|
|
|
* |
298
|
|
|
* @return bool |
299
|
|
|
*/ |
300
|
24 |
|
protected function isDigit($char) { |
301
|
24 |
|
return is_numeric($char); |
302
|
|
|
} |
303
|
|
|
|
304
|
|
|
/** |
305
|
|
|
* @return bool |
306
|
|
|
*/ |
307
|
33 |
|
protected function isDoubleBracketString() { |
308
|
33 |
|
return $this->input->peek() == '[' && !$this->input->eof(1) && ($this->input->peek(1) == '[' || $this->input->peek(1) == '='); |
309
|
|
|
} |
310
|
|
|
|
311
|
|
|
/** |
312
|
|
|
* @return bool |
313
|
|
|
*/ |
314
|
39 |
|
protected function isComment() { |
315
|
39 |
|
return $this->input->peek() == '-' && !$this->input->eof(1) && $this->input->peek(1) == '-'; |
316
|
|
|
} |
317
|
|
|
|
318
|
|
|
/** |
319
|
|
|
* @param string $char |
320
|
|
|
* |
321
|
|
|
* @return bool |
322
|
|
|
*/ |
323
|
20 |
|
protected function isStartIdentifierCharacter($char) { |
324
|
20 |
|
return preg_match('/[a-zA-Z_]/', $char) === 1; |
325
|
|
|
} |
326
|
|
|
|
327
|
|
|
/** |
328
|
|
|
* @param string $char |
329
|
|
|
* |
330
|
|
|
* @return bool |
331
|
|
|
*/ |
332
|
16 |
|
protected function isIdentifierCharacter($char) { |
333
|
16 |
|
return preg_match('/[a-zA-Z0-9_]/', $char) === 1; |
334
|
|
|
} |
335
|
|
|
|
336
|
|
|
/** |
337
|
|
|
* @param string $char |
338
|
|
|
* |
339
|
|
|
* @return bool |
340
|
|
|
*/ |
341
|
15 |
|
protected function isPunctuation($char) { |
342
|
15 |
|
return strpos(",{}=[]", $char) !== false; |
343
|
|
|
} |
344
|
|
|
|
345
|
|
|
/** |
346
|
|
|
* @param string $text |
347
|
|
|
* |
348
|
|
|
* @return bool |
349
|
|
|
*/ |
350
|
16 |
|
protected function isKeyword($text) { |
351
|
16 |
|
return in_array($text, Lua::$luaKeywords); |
352
|
|
|
} |
353
|
|
|
} |
This check compares the return type specified in the
@return
annotation of a function or method doc comment with the types returned by the function and raises an issue if they mismatch.