|
1
|
|
|
<?php declare(strict_types=1); |
|
2
|
|
|
|
|
3
|
|
|
namespace Diezz\YamlToObjectMapper\Resolver\Parser; |
|
4
|
|
|
|
|
5
|
|
|
class Tokenizer |
|
6
|
|
|
{ |
|
7
|
|
|
public const T_BEGIN_OF_EXPRESSION = 1; |
|
8
|
|
|
public const T_END_OF_EXPRESSION = 2; |
|
9
|
|
|
public const T_SEMICOLON = 3; |
|
10
|
|
|
public const T_SINGLE_QUOTE = 4; |
|
11
|
|
|
public const T_DOUBLE_QUOTE = 5; |
|
12
|
|
|
public const T_STRING_LITERAL = 6; |
|
13
|
|
|
public const T_QUOTED_STRING_LITERAL = 7; |
|
14
|
|
|
public const T_NUMERIC_LITERAL = 8; |
|
15
|
|
|
public const T_DASH = 9; |
|
16
|
|
|
public const T_UNDERSCORE = 10; |
|
17
|
|
|
public const T_ARRAY_BEGIN = 11; |
|
18
|
|
|
public const T_ARRAY_END = 12; |
|
19
|
|
|
public const T_COMMA = 13; |
|
20
|
|
|
public const T_SPACE = 14; |
|
21
|
|
|
public const T_DOT = 15; |
|
22
|
|
|
|
|
23
|
|
|
|
|
24
|
|
|
private string $string; |
|
25
|
|
|
private int $cursor; |
|
26
|
|
|
|
|
27
|
17 |
|
public function __construct(string $string) |
|
28
|
|
|
{ |
|
29
|
17 |
|
$this->string = $string; |
|
30
|
17 |
|
$this->cursor = 0; |
|
31
|
17 |
|
} |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* @throws SyntaxException |
|
35
|
|
|
*/ |
|
36
|
17 |
|
public function getNextToken(): ?Token |
|
37
|
|
|
{ |
|
38
|
17 |
|
if (!$this->hasMoreToken()) { |
|
39
|
13 |
|
return null; |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
17 |
|
$string = substr($this->string, $this->cursor); |
|
43
|
|
|
|
|
44
|
17 |
|
foreach ($this->getSpec() as $specItem) { |
|
45
|
17 |
|
$tokenType = $specItem->getTokenType(); |
|
46
|
17 |
|
$tokenValue = $this->match($specItem->getPattern(), $string); |
|
47
|
17 |
|
if ($tokenValue === null) { |
|
48
|
17 |
|
continue; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
17 |
|
if ($tokenType === null) { |
|
52
|
|
|
return $this->getNextToken(); |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
17 |
|
return Token::of($tokenType, $tokenValue); |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
throw new SyntaxException("Unexpected token ${string[0]}"); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
public function isEOF(): bool |
|
62
|
|
|
{ |
|
63
|
|
|
return $this->cursor === strlen($this->string); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
17 |
|
public function isScalar(): bool |
|
67
|
|
|
{ |
|
68
|
17 |
|
return preg_match("/\\$\{.*\}/", $this->string) === 0; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
4 |
|
public function getString(): string |
|
72
|
|
|
{ |
|
73
|
4 |
|
return $this->string; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* Match a token for a regular expression |
|
78
|
|
|
*/ |
|
79
|
17 |
|
private function match(string $pattern, string $string): ?string |
|
80
|
|
|
{ |
|
81
|
17 |
|
preg_match($pattern, $string, $matches); |
|
82
|
17 |
|
if (empty($matches)) { |
|
83
|
17 |
|
return null; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
17 |
|
$result = $matches[0]; |
|
87
|
17 |
|
$this->cursor += strlen($result); |
|
88
|
|
|
|
|
89
|
17 |
|
return $result; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
17 |
|
private function hasMoreToken(): bool |
|
93
|
|
|
{ |
|
94
|
17 |
|
return $this->cursor < strlen($this->string); |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
/** |
|
98
|
|
|
* @return SpecItem[] |
|
99
|
|
|
*/ |
|
100
|
17 |
|
private function getSpec(): array |
|
101
|
|
|
{ |
|
102
|
|
|
return [ |
|
103
|
|
|
//Skip spaces |
|
104
|
17 |
|
SpecItem::of("/^\s+/", self::T_SPACE), |
|
105
|
|
|
|
|
106
|
|
|
//Special chars |
|
107
|
17 |
|
SpecItem::of("/^\\$\{/", self::T_BEGIN_OF_EXPRESSION), |
|
108
|
17 |
|
SpecItem::of("/^\}/", self::T_END_OF_EXPRESSION), |
|
109
|
17 |
|
SpecItem::of("/^\:/", self::T_SEMICOLON), |
|
110
|
17 |
|
SpecItem::of("/^-/", self::T_DASH), |
|
111
|
17 |
|
SpecItem::of("/^\[/", self::T_ARRAY_BEGIN), |
|
112
|
17 |
|
SpecItem::of("/^\]/", self::T_ARRAY_END), |
|
113
|
17 |
|
SpecItem::of("/^\,/", self::T_COMMA), |
|
114
|
17 |
|
SpecItem::of("/^\./", self::T_DOT), |
|
115
|
|
|
|
|
116
|
|
|
// Literals |
|
117
|
17 |
|
SpecItem::of("/^\"[^\"]*\"/", self::T_QUOTED_STRING_LITERAL), |
|
118
|
17 |
|
SpecItem::of("/^'[^']*'/", self::T_QUOTED_STRING_LITERAL), |
|
119
|
17 |
|
SpecItem::of("/^\w+/", self::T_STRING_LITERAL), |
|
120
|
17 |
|
SpecItem::of("/^\d+/", self::T_NUMERIC_LITERAL), |
|
121
|
|
|
]; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
} |
|
125
|
|
|
|
ww-stas /
yaml-to-object-mapper
