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