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
|
27 |
|
public function __construct(InputStream $input) { |
26
|
27 |
|
$this->input = $input; |
27
|
27 |
|
} |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* @return Token |
31
|
|
|
*/ |
32
|
27 |
|
public function next() { |
33
|
27 |
|
$token = $this->current; |
34
|
27 |
|
$this->current = null; |
35
|
27 |
|
if ($token) { |
36
|
19 |
|
return $token; |
37
|
|
|
} |
38
|
8 |
|
return $this->readNext(); |
39
|
|
|
} |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* @return bool |
43
|
|
|
*/ |
44
|
9 |
|
public function eof() { |
45
|
9 |
|
return $this->peek() == null; |
46
|
|
|
} |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* @return Token |
|
|
|
|
50
|
|
|
*/ |
51
|
19 |
|
public function peek() { |
52
|
19 |
|
if ($this->current) { |
53
|
19 |
|
return $this->current; |
54
|
|
|
} |
55
|
19 |
|
$this->current = $this->readNext(); |
56
|
19 |
|
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
|
27 |
|
protected function readNext() { |
73
|
27 |
|
$this->readWhile([$this, 'isWhitespace']); |
74
|
27 |
|
if ($this->input->eof()) { |
75
|
|
|
return null; |
76
|
|
|
} |
77
|
27 |
|
$char = $this->input->peek(); |
78
|
27 |
|
if ($this->isComment()) { |
79
|
3 |
|
$this->skipComment(); |
80
|
3 |
|
return $this->readNext(); |
81
|
|
|
} |
82
|
27 |
|
if ($char == '"') { |
83
|
10 |
|
return $this->readDoubleQuotedString(); |
84
|
|
|
} |
85
|
23 |
|
if ($char == '\'') { |
86
|
1 |
|
return $this->readSingleQuotedString(); |
87
|
|
|
} |
88
|
22 |
|
if ($this->isDoubleBracketString()) { |
89
|
2 |
|
return $this->readDoubleBracketString(); |
90
|
|
|
} |
91
|
20 |
|
if ($this->isDigit($char)) { |
92
|
8 |
|
return $this->readNumber(); |
93
|
|
|
} |
94
|
16 |
|
if ($this->isStartIdentifierCharacter($char)) { |
95
|
12 |
|
return $this->readIdentifier(); |
96
|
|
|
} |
97
|
12 |
|
if ($this->isPunctuation($char)) { |
98
|
11 |
|
return $this->readPunctuation(); |
99
|
|
|
} |
100
|
1 |
|
$this->input->error('Cannot handle character: ' . $char . ' (ord: ' . ord($char) . ')'); |
101
|
|
|
} |
102
|
|
|
|
103
|
3 |
|
protected function skipComment() { |
104
|
3 |
|
$this->readWhile( |
105
|
|
|
function ($char) { |
106
|
3 |
|
return $char != "\n"; |
107
|
|
|
} |
108
|
3 |
|
); |
109
|
3 |
|
$this->input->next(); |
110
|
3 |
|
} |
111
|
|
|
|
112
|
|
|
/** |
113
|
|
|
* @return Token |
114
|
|
|
*/ |
115
|
10 |
|
protected function readDoubleQuotedString() { |
116
|
10 |
|
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
|
2 |
|
protected function readDoubleBracketString() { |
130
|
|
|
// we cannot use readEscaped because it only supports a single char as $end |
131
|
2 |
|
$escaped = false; |
132
|
2 |
|
$str = ""; |
133
|
|
|
// skip both |
134
|
2 |
|
$this->input->next(); |
135
|
2 |
|
$this->input->next(); |
136
|
2 |
|
while (!$this->input->eof()) { |
137
|
2 |
|
$char = $this->input->next(); |
138
|
2 |
|
if ($escaped) { |
139
|
|
|
$str .= $char; |
140
|
|
|
$escaped = false; |
141
|
|
|
} else { |
142
|
2 |
|
if ($char == "\\") { |
143
|
|
|
$escaped = true; |
144
|
|
|
} else { |
145
|
2 |
|
if ($char == ']' && $this->input->peek() == ']') { // we reached the end |
146
|
2 |
|
break; |
147
|
|
|
} else { |
148
|
2 |
|
$str .= $char; |
149
|
|
|
} |
150
|
|
|
} |
151
|
|
|
} |
152
|
2 |
|
} |
153
|
2 |
|
return new Token(Token::TYPE_STRING, $str); |
154
|
|
|
} |
155
|
|
|
|
156
|
|
|
/** |
157
|
|
|
* @param string $end |
158
|
|
|
* |
159
|
|
|
* @return string |
160
|
|
|
*/ |
161
|
11 |
|
protected function readEscaped($end) { |
162
|
11 |
|
$escaped = false; |
163
|
11 |
|
$str = ""; |
164
|
11 |
|
$this->input->next(); |
165
|
11 |
|
while (!$this->input->eof()) { |
166
|
11 |
|
$char = $this->input->next(); |
167
|
11 |
|
if ($escaped) { |
168
|
|
|
$str .= $char; |
169
|
|
|
$escaped = false; |
170
|
|
|
} else { |
171
|
11 |
|
if ($char == "\\") { |
172
|
|
|
$escaped = true; |
173
|
|
|
} else { |
174
|
11 |
|
if ($char == $end) { |
175
|
11 |
|
break; |
176
|
|
|
} else { |
177
|
11 |
|
$str .= $char; |
178
|
|
|
} |
179
|
|
|
} |
180
|
|
|
} |
181
|
11 |
|
} |
182
|
11 |
|
return $str; |
183
|
|
|
} |
184
|
|
|
|
185
|
|
|
/** |
186
|
|
|
* @return Token |
187
|
|
|
*/ |
188
|
8 |
|
protected function readNumber() { |
189
|
8 |
|
$hasDot = false; |
190
|
8 |
|
$number = $this->readWhile( |
191
|
|
|
function ($char) use (&$hasDot) { |
192
|
8 |
|
if ($char == '.') { |
193
|
1 |
|
if ($hasDot) { |
194
|
|
|
return false; |
195
|
|
|
} |
196
|
1 |
|
$hasDot = true; |
197
|
1 |
|
return true; |
198
|
|
|
} |
199
|
8 |
|
return $this->isDigit($char); |
200
|
|
|
} |
201
|
8 |
|
); |
202
|
8 |
|
return new Token(Token::TYPE_NUMBER, $hasDot ? floatval($number) : intval($number)); |
203
|
|
|
} |
204
|
|
|
|
205
|
|
|
/** |
206
|
|
|
* @return Token |
207
|
|
|
*/ |
208
|
12 |
|
protected function readIdentifier() { |
209
|
12 |
|
$first = false; |
210
|
12 |
|
$identifier = $this->readWhile( |
211
|
12 |
|
function ($char) use (&$first) { |
212
|
12 |
|
if ($first) { |
213
|
|
|
$first = false; |
214
|
|
|
return $this->isStartIdentifierCharacter($char); |
215
|
|
|
} |
216
|
12 |
|
return $this->isIdentifierCharacter($char); |
217
|
|
|
} |
218
|
12 |
|
); |
219
|
12 |
|
if ($this->isKeyword($identifier)) { |
220
|
3 |
|
return new Token(Token::TYPE_KEYWORD, $identifier); |
221
|
|
|
} |
222
|
9 |
|
return new Token(Token::TYPE_IDENTIFIER, $identifier); |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
/** |
226
|
|
|
* @return Token |
227
|
|
|
*/ |
228
|
11 |
|
protected function readPunctuation() { |
229
|
11 |
|
return new Token(Token::TYPE_PUNCTUATION, $this->input->next()); |
230
|
|
|
} |
231
|
|
|
|
232
|
|
|
/** |
233
|
|
|
* @param callable $predicate |
234
|
|
|
* |
235
|
|
|
* @return string |
236
|
|
|
*/ |
237
|
27 |
|
protected function readWhile(callable $predicate) { |
238
|
27 |
|
$str = ""; |
239
|
27 |
|
while (!$this->input->eof() && call_user_func($predicate, $this->input->peek())) { |
240
|
16 |
|
$str .= $this->input->next(); |
241
|
16 |
|
} |
242
|
27 |
|
return $str; |
243
|
|
|
} |
244
|
|
|
|
245
|
|
|
/** |
246
|
|
|
* @param string $char |
247
|
|
|
* |
248
|
|
|
* @return bool |
249
|
|
|
*/ |
250
|
27 |
|
protected function isWhitespace($char) { |
251
|
27 |
|
return strpos(" \t\n\r", $char) !== false; |
252
|
|
|
} |
253
|
|
|
|
254
|
|
|
/** |
255
|
|
|
* @param string $char |
256
|
|
|
* |
257
|
|
|
* @return bool |
258
|
|
|
*/ |
259
|
20 |
|
protected function isDigit($char) { |
260
|
20 |
|
return is_numeric($char); |
261
|
|
|
} |
262
|
|
|
|
263
|
|
|
/** |
264
|
|
|
* @return bool |
265
|
|
|
*/ |
266
|
22 |
|
protected function isDoubleBracketString() { |
267
|
22 |
|
return $this->input->peek() == '[' && !$this->input->eof(1) && $this->input->peek(1) == '['; |
268
|
|
|
} |
269
|
|
|
|
270
|
|
|
/** |
271
|
|
|
* @return bool |
272
|
|
|
*/ |
273
|
27 |
|
protected function isComment() { |
274
|
27 |
|
return $this->input->peek() == '-' && !$this->input->eof(1) && $this->input->peek(1) == '-'; |
275
|
|
|
} |
276
|
|
|
|
277
|
|
|
/** |
278
|
|
|
* @param string $char |
279
|
|
|
* |
280
|
|
|
* @return bool |
281
|
|
|
*/ |
282
|
16 |
|
protected function isStartIdentifierCharacter($char) { |
283
|
16 |
|
return preg_match('/[a-zA-Z_]/', $char) === 1; |
284
|
|
|
} |
285
|
|
|
|
286
|
|
|
/** |
287
|
|
|
* @param string $char |
288
|
|
|
* |
289
|
|
|
* @return bool |
290
|
|
|
*/ |
291
|
12 |
|
protected function isIdentifierCharacter($char) { |
292
|
12 |
|
return preg_match('/[a-zA-Z0-9_]/', $char) === 1; |
293
|
|
|
} |
294
|
|
|
|
295
|
|
|
/** |
296
|
|
|
* @param string $char |
297
|
|
|
* |
298
|
|
|
* @return bool |
299
|
|
|
*/ |
300
|
12 |
|
protected function isPunctuation($char) { |
301
|
12 |
|
return strpos(",{}=[]", $char) !== false; |
302
|
|
|
} |
303
|
|
|
|
304
|
|
|
/** |
305
|
|
|
* @param string $text |
306
|
|
|
* |
307
|
|
|
* @return bool |
308
|
|
|
*/ |
309
|
12 |
|
protected function isKeyword($text) { |
310
|
12 |
|
return in_array($text, Lua::$luaKeywords); |
311
|
|
|
} |
312
|
|
|
} |
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.