|
1
|
|
|
<?php |
|
2
|
|
|
namespace Desmond; |
|
3
|
|
|
use Exception; |
|
4
|
|
|
use Desmond\Reader; |
|
5
|
|
|
use Desmond\Tokenizer; |
|
6
|
|
|
use Desmond\data_types\ListType; |
|
7
|
|
|
use Desmond\data_types\VectorType; |
|
8
|
|
|
use Desmond\data_types\HashType; |
|
9
|
|
|
use Desmond\data_types\SymbolType; |
|
10
|
|
|
use Desmond\data_types\NumberType; |
|
11
|
|
|
use Desmond\data_types\NilType; |
|
12
|
|
|
use Desmond\data_types\TrueType; |
|
13
|
|
|
use Desmond\data_types\FalseType; |
|
14
|
|
|
use Desmond\data_types\StringType; |
|
15
|
|
|
|
|
16
|
|
|
class Lexer |
|
17
|
|
|
{ |
|
18
|
|
|
private static $CONDITON = 0; |
|
19
|
|
|
private static $VALUE = 1; |
|
20
|
|
|
|
|
21
|
354 |
|
public function readString($string) |
|
22
|
|
|
{ |
|
23
|
354 |
|
$tokens = Tokenizer::tokenize($string); |
|
24
|
354 |
|
if (empty($tokens)) { |
|
25
|
1 |
|
return null; |
|
26
|
|
|
} |
|
27
|
353 |
|
return $this->readForm(new Reader($tokens)); |
|
28
|
|
|
} |
|
29
|
|
|
|
|
30
|
353 |
|
private function readForm(Reader $reader) |
|
31
|
|
|
{ |
|
32
|
353 |
|
switch ($reader->peek()) { |
|
33
|
353 |
|
case '(': |
|
34
|
332 |
|
$reader->next(); |
|
35
|
332 |
|
$collection = new ListType(); |
|
36
|
332 |
|
return $this->readCollection($reader, $collection, ')'); |
|
37
|
353 |
|
case ')': |
|
38
|
1 |
|
throw new Exception('Unexpected )'); |
|
39
|
352 |
|
case '[': |
|
40
|
57 |
|
$reader->next(); |
|
41
|
57 |
|
$collection = new VectorType(); |
|
42
|
57 |
|
return $this->readCollection($reader, $collection, ']'); |
|
43
|
352 |
|
case ']': |
|
44
|
1 |
|
throw new Exception('Unexpected ]'); |
|
45
|
351 |
|
case '{': |
|
46
|
33 |
|
$reader->next(); |
|
47
|
33 |
|
$hash = new HashType(); |
|
48
|
33 |
|
return $this->readHash($reader, $hash); |
|
49
|
351 |
|
case '}': |
|
50
|
2 |
|
throw new Exception('Unexpected }'); |
|
51
|
|
|
default: |
|
52
|
350 |
|
$form = $this->readAtom($reader->peek()); |
|
53
|
350 |
|
$reader->next(); |
|
54
|
350 |
|
return $form; |
|
55
|
|
|
} |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
335 |
|
private function readCollection(Reader $reader, $collection, $end) |
|
59
|
|
|
{ |
|
60
|
335 |
|
while (($token = $reader->peek()) !== $end) { |
|
61
|
335 |
|
if ($token === null) { |
|
62
|
2 |
|
throw new Exception('Expected "' . $end . '", found EOF.'); |
|
63
|
|
|
} |
|
64
|
335 |
|
$collection->set($this->readForm($reader)); |
|
65
|
|
|
} |
|
66
|
333 |
|
$reader->next(); |
|
67
|
333 |
|
return $collection; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
33 |
|
private function readHash(Reader $reader, $hash) |
|
71
|
|
|
{ |
|
72
|
33 |
|
while (($token = $reader->peek()) !== '}') { |
|
73
|
26 |
|
if ($token === null) { |
|
74
|
1 |
|
throw new Exception('Expected "}", found EOF.'); |
|
75
|
|
|
} |
|
76
|
26 |
|
$key = $this->readForm($reader); |
|
77
|
|
|
try { |
|
78
|
26 |
|
$value = $this->readForm($reader); |
|
79
|
1 |
|
} catch (Exception $exeption) { |
|
80
|
1 |
|
throw new Exception('Unexpected end of hash. Every key must have a value.'); |
|
81
|
|
|
} |
|
82
|
26 |
|
$hash->set($value, $key->value()); |
|
83
|
|
|
} |
|
84
|
31 |
|
$reader->next(); |
|
85
|
31 |
|
return $hash; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
350 |
|
private function readAtom($token) |
|
89
|
|
|
{ |
|
90
|
350 |
|
$tokenLiteral = null; |
|
91
|
350 |
|
foreach ($this->tokenTestList($token) as $test) { |
|
92
|
350 |
|
if ($test[self::$CONDITON]) { |
|
93
|
350 |
|
$tokenLiteral = $test[self::$VALUE]; |
|
94
|
350 |
|
break; |
|
95
|
|
|
} |
|
96
|
|
|
} |
|
97
|
350 |
|
return $tokenLiteral; |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
350 |
|
private function tokenTestList($token) |
|
101
|
|
|
{ |
|
102
|
|
|
return [ |
|
103
|
350 |
|
[preg_match('/^-?(\.?[0-9]+)|([0-9]+.[0-9]+)$/', $token), new NumberType($token)], |
|
104
|
350 |
|
[preg_match('/".*"/s', $token), new StringType($token)], |
|
105
|
350 |
|
[$token === 'nil', new NilType()], |
|
106
|
350 |
|
[$token === 'true', new TrueType()], |
|
107
|
350 |
|
[$token === 'false', new FalseType()], |
|
108
|
350 |
|
[true, new SymbolType($token)] // Basically, if it's nothing else, it's a symbol. |
|
109
|
|
|
]; |
|
110
|
|
|
} |
|
111
|
|
|
} |
|
112
|
|
|
|
Archer70 /
desmond
