|
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
|
21 |
|
public function __construct(InputStream $input) { |
|
26
|
21 |
|
$this->input = $input; |
|
27
|
21 |
|
} |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @return Token |
|
31
|
|
|
*/ |
|
32
|
21 |
|
public function next() { |
|
33
|
21 |
|
$token = $this->current; |
|
34
|
21 |
|
$this->current = null; |
|
35
|
21 |
|
if ($token) { |
|
36
|
13 |
|
return $token; |
|
37
|
|
|
} |
|
38
|
8 |
|
return $this->readNext(); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @return bool |
|
43
|
|
|
*/ |
|
44
|
6 |
|
public function eof() { |
|
45
|
6 |
|
return $this->peek() == null; |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* @return Token |
|
|
|
|
|
|
50
|
|
|
*/ |
|
51
|
13 |
|
public function peek() { |
|
52
|
13 |
|
if ($this->current) { |
|
53
|
13 |
|
return $this->current; |
|
54
|
|
|
} |
|
55
|
13 |
|
$this->current = $this->readNext(); |
|
56
|
13 |
|
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
|
21 |
|
protected function readNext() { |
|
73
|
21 |
|
$this->readWhile([$this, 'isWhitespace']); |
|
74
|
21 |
|
if ($this->input->eof()) { |
|
75
|
|
|
return null; |
|
76
|
|
|
} |
|
77
|
21 |
|
$char = $this->input->peek(); |
|
78
|
21 |
|
if ($char == "--") { |
|
79
|
|
|
$this->skipComment(); |
|
80
|
|
|
return $this->readNext(); |
|
81
|
|
|
} |
|
82
|
21 |
|
if ($char == '"') { |
|
83
|
8 |
|
return $this->readDoubleQuotedString(); |
|
84
|
|
|
} |
|
85
|
18 |
|
if ($char == '\'') { |
|
86
|
1 |
|
return $this->readSingleQuotedString(); |
|
87
|
|
|
} |
|
88
|
17 |
|
if ($this->isDigit($char)) { |
|
89
|
6 |
|
return $this->readNumber(); |
|
90
|
|
|
} |
|
91
|
13 |
|
if ($this->isStartIdentifierCharacter($char)) { |
|
92
|
9 |
|
return $this->readIdentifier(); |
|
93
|
|
|
} |
|
94
|
9 |
|
if ($this->isPunctuation($char)) { |
|
95
|
8 |
|
return $this->readPunctuation(); |
|
96
|
|
|
} |
|
97
|
1 |
|
$this->input->error('Cannot handle character: ' . $char . ' (ord: ' . ord($char) . ')'); |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
protected function skipComment() { |
|
101
|
|
|
$this->readWhile( |
|
102
|
|
|
function ($char) { |
|
103
|
|
|
return $char != "\n"; |
|
104
|
|
|
} |
|
105
|
|
|
); |
|
106
|
|
|
$this->input->next(); |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
/** |
|
110
|
|
|
* @return Token |
|
111
|
|
|
*/ |
|
112
|
8 |
|
protected function readDoubleQuotedString() { |
|
113
|
8 |
|
return new Token(Token::TYPE_STRING, $this->readEscaped('"')); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* @return Token |
|
118
|
|
|
*/ |
|
119
|
1 |
|
protected function readSingleQuotedString() { |
|
120
|
1 |
|
return new Token(Token::TYPE_STRING, $this->readEscaped('\'')); |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
/** |
|
124
|
|
|
* @param string $end |
|
125
|
|
|
* |
|
126
|
|
|
* @return string |
|
127
|
|
|
*/ |
|
128
|
9 |
|
protected function readEscaped($end) { |
|
129
|
9 |
|
$escaped = false; |
|
130
|
9 |
|
$str = ""; |
|
131
|
9 |
|
$this->input->next(); |
|
132
|
9 |
|
while (!$this->input->eof()) { |
|
133
|
9 |
|
$char = $this->input->next(); |
|
134
|
9 |
|
if ($escaped) { |
|
135
|
|
|
$str .= $char; |
|
136
|
|
|
$escaped = false; |
|
137
|
|
|
} else { |
|
138
|
9 |
|
if ($char == "\\") { |
|
139
|
|
|
$escaped = true; |
|
140
|
|
|
} else { |
|
141
|
9 |
|
if ($char == $end) { |
|
142
|
9 |
|
break; |
|
143
|
|
|
} else { |
|
144
|
9 |
|
$str .= $char; |
|
145
|
|
|
} |
|
146
|
|
|
} |
|
147
|
|
|
} |
|
148
|
9 |
|
} |
|
149
|
9 |
|
return $str; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
/** |
|
153
|
|
|
* @return Token |
|
154
|
|
|
*/ |
|
155
|
6 |
|
protected function readNumber() { |
|
156
|
6 |
|
$hasDot = false; |
|
157
|
6 |
|
$number = $this->readWhile( |
|
158
|
|
|
function ($char) use (&$hasDot) { |
|
159
|
6 |
|
if ($char == '.') { |
|
160
|
1 |
|
if ($hasDot) { |
|
161
|
|
|
return false; |
|
162
|
|
|
} |
|
163
|
1 |
|
$hasDot = true; |
|
164
|
1 |
|
return true; |
|
165
|
|
|
} |
|
166
|
6 |
|
return $this->isDigit($char); |
|
167
|
|
|
} |
|
168
|
6 |
|
); |
|
169
|
6 |
|
return new Token(Token::TYPE_NUMBER, $hasDot ? floatval($number) : intval($number)); |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
/** |
|
173
|
|
|
* @return Token |
|
174
|
|
|
*/ |
|
175
|
9 |
|
protected function readIdentifier() { |
|
176
|
9 |
|
$first = false; |
|
177
|
9 |
|
$identifier = $this->readWhile( |
|
178
|
9 |
|
function ($char) use (&$first) { |
|
179
|
9 |
|
if ($first) { |
|
180
|
|
|
$first = false; |
|
181
|
|
|
return $this->isStartIdentifierCharacter($char); |
|
182
|
|
|
} |
|
183
|
9 |
|
return $this->isIdentifierCharacter($char); |
|
184
|
|
|
} |
|
185
|
9 |
|
); |
|
186
|
9 |
|
if ($this->isKeyword($identifier)) { |
|
187
|
3 |
|
return new Token(Token::TYPE_KEYWORD, $identifier); |
|
188
|
|
|
} |
|
189
|
6 |
|
return new Token(Token::TYPE_IDENTIFIER, $identifier); |
|
190
|
|
|
} |
|
191
|
|
|
|
|
192
|
|
|
/** |
|
193
|
|
|
* @return Token |
|
194
|
|
|
*/ |
|
195
|
8 |
|
protected function readPunctuation() { |
|
196
|
8 |
|
return new Token(Token::TYPE_PUNCTUATION, $this->input->next()); |
|
197
|
|
|
} |
|
198
|
|
|
|
|
199
|
|
|
/** |
|
200
|
|
|
* @param callable $predicate |
|
201
|
|
|
* |
|
202
|
|
|
* @return string |
|
203
|
|
|
*/ |
|
204
|
21 |
|
protected function readWhile(callable $predicate) { |
|
205
|
21 |
|
$str = ""; |
|
206
|
21 |
|
while (!$this->input->eof() && call_user_func($predicate, $this->input->peek())) { |
|
207
|
13 |
|
$str .= $this->input->next(); |
|
208
|
13 |
|
} |
|
209
|
21 |
|
return $str; |
|
210
|
|
|
} |
|
211
|
|
|
|
|
212
|
21 |
|
protected function isWhitespace($char) { |
|
213
|
21 |
|
return strpos(" \t\n\r", $char) !== false; |
|
214
|
|
|
} |
|
215
|
|
|
|
|
216
|
|
|
/** |
|
217
|
|
|
* @param string $char |
|
218
|
|
|
* |
|
219
|
|
|
* @return bool |
|
220
|
|
|
*/ |
|
221
|
17 |
|
protected function isDigit($char) { |
|
222
|
17 |
|
return is_numeric($char); |
|
223
|
|
|
} |
|
224
|
|
|
|
|
225
|
|
|
/** |
|
226
|
|
|
* @param string $char |
|
227
|
|
|
* |
|
228
|
|
|
* @return bool |
|
229
|
|
|
*/ |
|
230
|
13 |
|
protected function isStartIdentifierCharacter($char) { |
|
231
|
13 |
|
return preg_match('/[a-zA-Z_]/', $char) === 1; |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
|
|
/** |
|
235
|
|
|
* @param string $char |
|
236
|
|
|
* |
|
237
|
|
|
* @return bool |
|
238
|
|
|
*/ |
|
239
|
9 |
|
protected function isIdentifierCharacter($char) { |
|
240
|
9 |
|
return preg_match('/[a-zA-Z0-9_]/', $char) === 1; |
|
241
|
|
|
} |
|
242
|
|
|
|
|
243
|
|
|
/** |
|
244
|
|
|
* @param string $char |
|
245
|
|
|
* |
|
246
|
|
|
* @return bool |
|
247
|
|
|
*/ |
|
248
|
9 |
|
protected function isPunctuation($char) { |
|
249
|
9 |
|
return strpos(",{}=[]", $char) !== false; |
|
250
|
|
|
} |
|
251
|
|
|
|
|
252
|
|
|
/** |
|
253
|
|
|
* @param string $text |
|
254
|
|
|
* |
|
255
|
|
|
* @return bool |
|
256
|
|
|
*/ |
|
257
|
9 |
|
protected function isKeyword($text) { |
|
258
|
9 |
|
return in_array($text, Lua::$luaKeywords); |
|
259
|
|
|
} |
|
260
|
|
|
} |
This check compares the return type specified in the
@returnannotation of a function or method doc comment with the types returned by the function and raises an issue if they mismatch.