|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PeacefulBit\Slate\Parser; |
|
4
|
|
|
|
|
5
|
|
|
use function Nerd\Common\Arrays\all; |
|
6
|
|
|
use function Nerd\Common\Arrays\toHeadTail; |
|
7
|
|
|
use function Nerd\Common\Arrays\append; |
|
8
|
|
|
|
|
9
|
|
|
use function Nerd\Common\Functional\tail; |
|
10
|
|
|
|
|
11
|
|
|
use PeacefulBit\Slate\Exceptions\ParserException; |
|
12
|
|
|
use PeacefulBit\Slate\Parser\Nodes; |
|
13
|
|
|
use PeacefulBit\Slate\Parser\Tokens; |
|
14
|
|
|
|
|
15
|
|
|
class Parser |
|
16
|
|
|
{ |
|
17
|
|
|
/** |
|
18
|
|
|
* Convert tokens tree to abstract syntax tree. |
|
19
|
|
|
* |
|
20
|
|
|
* @param Tokens\Token[] $tokens |
|
21
|
|
|
* @return Nodes\Node |
|
22
|
|
|
*/ |
|
23
|
|
|
public function parse(array $tokens): Nodes\Node |
|
24
|
|
|
{ |
|
25
|
|
|
$tokensTree = $this->deflate($tokens); |
|
26
|
|
|
|
|
27
|
|
|
return $this->parseProgram($tokensTree); |
|
28
|
|
|
} |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @param $token |
|
32
|
|
|
* @return Nodes\Identifier|Nodes\Literal|Nodes\Node |
|
33
|
|
|
* @throws ParserException |
|
34
|
|
|
*/ |
|
35
|
|
|
private function parseToken($token) |
|
36
|
|
|
{ |
|
37
|
|
|
if (is_array($token)) { |
|
38
|
|
|
return $this->parseExpression($token); |
|
39
|
|
|
} |
|
40
|
|
|
if ($token instanceof Tokens\StringToken) { |
|
41
|
|
|
return $this->parseString($token); |
|
42
|
|
|
} |
|
43
|
|
|
if ($token instanceof Tokens\IdentifierToken) { |
|
44
|
|
|
return $this->parseIdentifier($token); |
|
45
|
|
|
} |
|
46
|
|
|
if ($token instanceof Tokens\NumericToken) { |
|
47
|
|
|
return $this->parseNumeric($token); |
|
48
|
|
|
} |
|
49
|
|
|
throw new ParserException("Unexpected type of token - $token."); |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* @param array $tokens |
|
54
|
|
|
* @return Nodes\Node |
|
55
|
|
|
* @throws ParserException |
|
56
|
|
|
*/ |
|
57
|
|
|
private function parseExpression(array $tokens): Nodes\Node |
|
58
|
|
|
{ |
|
59
|
|
|
if (empty($tokens)) { |
|
60
|
|
|
throw new ParserException("Expression could not be empty."); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
list ($head, $tail) = toHeadTail($tokens); |
|
64
|
|
|
|
|
65
|
|
|
if ($head instanceof Tokens\IdentifierToken) { |
|
66
|
|
|
switch ($head->getValue()) { |
|
67
|
|
|
case 'def': |
|
68
|
|
|
return $this->parseDeclaration($tail); |
|
69
|
|
|
case 'lambda': |
|
70
|
|
|
return $this->parseLambdaDeclaration($tail); |
|
71
|
|
|
case 'if': |
|
72
|
|
|
return $this->parseIfExpression($tail); |
|
73
|
|
|
case 'or': |
|
74
|
|
|
return $this->parseOrExpression($tail); |
|
75
|
|
|
case 'and': |
|
76
|
|
|
return $this->parseAndExpression($tail); |
|
77
|
|
|
case 'use': |
|
78
|
|
|
case 'cond': |
|
79
|
|
|
throw new ParserException('This keyword is reserved'); |
|
80
|
|
|
} |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
return $this->parseCallExpression($tokens); |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
/** |
|
87
|
|
|
* @param array $tokens |
|
88
|
|
|
* @return Nodes\Node |
|
89
|
|
|
*/ |
|
90
|
|
|
private function parseDeclaration($tokens): Nodes\Node |
|
91
|
|
|
{ |
|
92
|
|
|
list ($head, $tail) = $tokens; |
|
|
|
|
|
|
93
|
|
|
|
|
94
|
|
|
if (!is_array($head)) { |
|
95
|
|
|
return $this->parseAssignDeclaration($tokens); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
return $this->parseFunctionDeclaration($tokens); |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
/** |
|
102
|
|
|
* @param $tokens |
|
103
|
|
|
* @return Nodes\Node |
|
104
|
|
|
* @throws ParserException |
|
105
|
|
|
*/ |
|
106
|
|
|
private function parseIfExpression($tokens): Nodes\Node |
|
107
|
|
|
{ |
|
108
|
|
|
if (sizeof($tokens) != 3) { |
|
109
|
|
|
throw new ParserException("'if' requires exactly three arguments"); |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
list ($test, $cons, $alt) = array_map([$this, 'parseToken'], $tokens); |
|
113
|
|
|
|
|
114
|
|
|
return new Nodes\IfExpression($test, $cons, $alt); |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
/** |
|
118
|
|
|
* @param $tokens |
|
119
|
|
|
* @return Nodes\Node |
|
120
|
|
|
* @throws ParserException |
|
121
|
|
|
*/ |
|
122
|
|
View Code Duplication |
private function parseOrExpression($tokens): Nodes\Node |
|
|
|
|
|
|
123
|
|
|
{ |
|
124
|
|
|
if (empty($tokens)) { |
|
125
|
|
|
throw new ParserException("'or' requires at least one argument"); |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
$expressions = array_map([$this, 'parseToken'], $tokens); |
|
129
|
|
|
|
|
130
|
|
|
return new Nodes\OrExpression($expressions); |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
/** |
|
134
|
|
|
* @param $tokens |
|
135
|
|
|
* @return Nodes\Node |
|
136
|
|
|
* @throws ParserException |
|
137
|
|
|
*/ |
|
138
|
|
View Code Duplication |
private function parseAndExpression($tokens): Nodes\Node |
|
|
|
|
|
|
139
|
|
|
{ |
|
140
|
|
|
if (empty($tokens)) { |
|
141
|
|
|
throw new ParserException("'and' requires at least one argument"); |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
$expressions = array_map([$this, 'parseToken'], $tokens); |
|
145
|
|
|
|
|
146
|
|
|
return new Nodes\AndExpression($expressions); |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
/** |
|
150
|
|
|
* @param $tokens |
|
151
|
|
|
* @return Nodes\Node |
|
152
|
|
|
* @throws ParserException |
|
153
|
|
|
*/ |
|
154
|
|
|
private function parseLambdaDeclaration($tokens): Nodes\Node |
|
155
|
|
|
{ |
|
156
|
|
|
list ($head, $body) = toHeadTail($tokens); |
|
157
|
|
|
|
|
158
|
|
|
$test = all($head, function ($token) { |
|
159
|
|
|
return $token instanceof Tokens\IdentifierToken; |
|
160
|
|
|
}); |
|
161
|
|
|
|
|
162
|
|
|
if (!$test) { |
|
163
|
|
|
throw new ParserException("Lambda arguments must be valid identifiers."); |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
|
|
$params = array_map([$this, 'parseToken'], $head); |
|
167
|
|
|
|
|
168
|
|
|
return new Nodes\LambdaExpression($params, $this->parseSequence($body)); |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
|
|
/** |
|
172
|
|
|
* @param $tokens |
|
173
|
|
|
* @return Nodes\Node |
|
174
|
|
|
* @throws ParserException |
|
175
|
|
|
*/ |
|
176
|
|
|
private function parseFunctionDeclaration($tokens): Nodes\Node |
|
177
|
|
|
{ |
|
178
|
|
|
list ($head, $body) = toHeadTail($tokens); |
|
179
|
|
|
|
|
180
|
|
|
if (sizeof($head) < 1) { |
|
181
|
|
|
throw new ParserException("Function must have identifier."); |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
$test = all($head, function ($token) { |
|
185
|
|
|
return $token instanceof Tokens\IdentifierToken; |
|
186
|
|
|
}); |
|
187
|
|
|
|
|
188
|
|
|
if (!$test) { |
|
189
|
|
|
throw new ParserException("Function name and arguments must be valid identifiers."); |
|
190
|
|
|
} |
|
191
|
|
|
|
|
192
|
|
|
list ($name, $args) = toHeadTail($head); |
|
193
|
|
|
|
|
194
|
|
|
$params = array_map([$this, 'parseToken'], $args); |
|
195
|
|
|
|
|
196
|
|
|
return new Nodes\FunctionExpression($name->getValue(), $params, $this->parseSequence($body)); |
|
|
|
|
|
|
197
|
|
|
} |
|
198
|
|
|
|
|
199
|
|
|
/** |
|
200
|
|
|
* @param array $tokens |
|
201
|
|
|
* @return Nodes\Assign |
|
202
|
|
|
* @throws ParserException |
|
203
|
|
|
*/ |
|
204
|
|
|
private function parseAssignDeclaration(array $tokens): Nodes\Assign |
|
205
|
|
|
{ |
|
206
|
|
|
if (sizeof($tokens) % 2 != 0) { |
|
207
|
|
|
throw new ParserException("Bad assign declaration"); |
|
208
|
|
|
} |
|
209
|
|
|
|
|
210
|
|
|
$assignments = array_chunk($tokens, 2); |
|
211
|
|
|
|
|
212
|
|
|
$result = array_map(function ($assign) { |
|
213
|
|
|
if (!$assign[0] instanceof Tokens\IdentifierToken) { |
|
214
|
|
|
throw new ParserException("Bad type of identifier in assign declaration"); |
|
215
|
|
|
} |
|
216
|
|
|
return array_map([$this, 'parseToken'], $assign); |
|
217
|
|
|
}, $assignments); |
|
218
|
|
|
|
|
219
|
|
|
return new Nodes\Assign($result); |
|
220
|
|
|
} |
|
221
|
|
|
|
|
222
|
|
|
/** |
|
223
|
|
|
* @param array $tokens |
|
224
|
|
|
* @return Nodes\Node |
|
225
|
|
|
*/ |
|
226
|
|
|
private function parseCallExpression(array $tokens): Nodes\Node |
|
227
|
|
|
{ |
|
228
|
|
|
list ($head, $tail) = toHeadTail($tokens); |
|
229
|
|
|
|
|
230
|
|
|
return new Nodes\CallExpression( |
|
231
|
|
|
$this->parseToken($head), |
|
232
|
|
|
array_map([$this, 'parseToken'], $tail) |
|
233
|
|
|
); |
|
234
|
|
|
} |
|
235
|
|
|
|
|
236
|
|
|
/** |
|
237
|
|
|
* @param array $tokens |
|
238
|
|
|
* @return Nodes\Program |
|
239
|
|
|
*/ |
|
240
|
|
|
private function parseProgram(array $tokens): Nodes\Program |
|
241
|
|
|
{ |
|
242
|
|
|
return new Nodes\Program(array_map([$this, 'parseToken'], $tokens)); |
|
243
|
|
|
} |
|
244
|
|
|
|
|
245
|
|
|
/** |
|
246
|
|
|
* @param array $tokens |
|
247
|
|
|
* @return Nodes\SequenceExpression |
|
248
|
|
|
*/ |
|
249
|
|
|
private function parseSequence(array $tokens): Nodes\SequenceExpression |
|
250
|
|
|
{ |
|
251
|
|
|
return new Nodes\SequenceExpression(array_map([$this, 'parseToken'], $tokens)); |
|
252
|
|
|
} |
|
253
|
|
|
|
|
254
|
|
|
/** |
|
255
|
|
|
* @param Tokens\IdentifierToken $token |
|
256
|
|
|
* @return Nodes\Identifier |
|
257
|
|
|
*/ |
|
258
|
|
|
private function parseIdentifier(Tokens\IdentifierToken $token): Nodes\Identifier |
|
259
|
|
|
{ |
|
260
|
|
|
return new Nodes\Identifier($token->getValue()); |
|
261
|
|
|
} |
|
262
|
|
|
|
|
263
|
|
|
/** |
|
264
|
|
|
* @param Tokens\StringToken $token |
|
265
|
|
|
* @return Nodes\Literal |
|
266
|
|
|
*/ |
|
267
|
|
|
private function parseString(Tokens\StringToken $token): Nodes\Literal |
|
268
|
|
|
{ |
|
269
|
|
|
return new Nodes\StringNode($token->getValue()); |
|
270
|
|
|
} |
|
271
|
|
|
|
|
272
|
|
|
/** |
|
273
|
|
|
* @param Tokens\NumericToken $token |
|
274
|
|
|
* @return Nodes\Literal |
|
275
|
|
|
*/ |
|
276
|
|
|
private function parseNumeric(Tokens\NumericToken $token): Nodes\Literal |
|
277
|
|
|
{ |
|
278
|
|
|
return new Nodes\Number($token->getValue()); |
|
279
|
|
|
} |
|
280
|
|
|
|
|
281
|
|
|
/** |
|
282
|
|
|
* Convert list of tokens into token tree. |
|
283
|
|
|
* |
|
284
|
|
|
* @param Tokens\Token[] $tokens |
|
285
|
|
|
* @return mixed |
|
286
|
|
|
*/ |
|
287
|
|
|
private function deflate(array $tokens) |
|
288
|
|
|
{ |
|
289
|
|
|
$iter = tail(function ($rest, $acc) use (&$iter) { |
|
290
|
|
|
if (empty($rest)) { |
|
291
|
|
|
return $acc; |
|
292
|
|
|
} |
|
293
|
|
|
list ($head, $tail) = toHeadTail($rest); |
|
294
|
|
|
switch (get_class($head)) { |
|
295
|
|
|
case Tokens\OpenBracketToken::class: |
|
296
|
|
|
$pairClosingIndex = $this->findPairClosingBracketIndex($tail); |
|
297
|
|
|
$inner = array_slice($tail, 0, $pairClosingIndex); |
|
298
|
|
|
$innerNode = $this->deflate($inner); |
|
299
|
|
|
$newTail = array_slice($tail, $pairClosingIndex + 1); |
|
300
|
|
|
return $iter($newTail, append($acc, $innerNode)); |
|
301
|
|
|
case Tokens\CloseBracketToken::class: |
|
302
|
|
|
throw new ParserException("Unpaired opening bracket found"); |
|
303
|
|
|
default: |
|
304
|
|
|
return $iter($tail, append($acc, $head)); |
|
305
|
|
|
} |
|
306
|
|
|
}); |
|
307
|
|
|
return $iter($tokens, []); |
|
308
|
|
|
} |
|
309
|
|
|
|
|
310
|
|
|
/** |
|
311
|
|
|
* @param array $tokens |
|
312
|
|
|
* @return mixed |
|
313
|
|
|
*/ |
|
314
|
|
|
private function findPairClosingBracketIndex(array $tokens) |
|
315
|
|
|
{ |
|
316
|
|
|
$iter = tail(function ($rest, $depth, $position) use (&$iter) { |
|
317
|
|
|
if (empty($rest)) { |
|
318
|
|
|
throw new ParserException("Unpaired closing bracket found."); |
|
319
|
|
|
} |
|
320
|
|
|
list ($head, $tail) = toHeadTail($rest); |
|
321
|
|
|
if ($head instanceof Tokens\CloseBracketToken) { |
|
322
|
|
|
if ($depth == 0) { |
|
323
|
|
|
return $position; |
|
324
|
|
|
} |
|
325
|
|
|
return $iter($tail, $depth - 1, $position + 1); |
|
326
|
|
|
} |
|
327
|
|
|
if ($head instanceof Tokens\OpenBracketToken) { |
|
328
|
|
|
return $iter($tail, $depth + 1, $position + 1); |
|
329
|
|
|
} |
|
330
|
|
|
return $iter($tail, $depth, $position + 1); |
|
331
|
|
|
}); |
|
332
|
|
|
return $iter($tokens, 0, 0); |
|
333
|
|
|
} |
|
334
|
|
|
} |
|
335
|
|
|
|
This checks looks for assignemnts to variables using the
list(...)function, where not all assigned variables are subsequently used.Consider the following code example.
Only the variables
$aand$care used. There was no need to assign$b.Instead, the list call could have been.