1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Cerbero\JsonParser\Tokens; |
4
|
|
|
|
5
|
|
|
use Cerbero\JsonParser\Exceptions\SyntaxException; |
6
|
|
|
use Cerbero\JsonParser\Sources\Source; |
7
|
|
|
use Cerbero\JsonParser\Tokens\Token; |
8
|
|
|
use Cerbero\JsonParser\Tokens\Tokenizer; |
9
|
|
|
use Cerbero\JsonParser\Tokens\Tokens; |
10
|
|
|
use Cerbero\JsonParser\ValueObjects\Progress; |
11
|
|
|
use IteratorAggregate; |
12
|
|
|
use Traversable; |
13
|
|
|
|
14
|
|
|
use function strlen; |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* The JSON lexer. |
18
|
|
|
* |
19
|
|
|
* @implements IteratorAggregate<int, Token> |
20
|
|
|
*/ |
21
|
|
|
final class Lexer implements IteratorAggregate |
22
|
|
|
{ |
23
|
|
|
/** |
24
|
|
|
* The parsing progress. |
25
|
|
|
* |
26
|
|
|
* @var Progress |
27
|
|
|
*/ |
28
|
|
|
private Progress $progress; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* The current position. |
32
|
|
|
* |
33
|
|
|
* @var int |
34
|
|
|
*/ |
35
|
|
|
private int $position = 1; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* Instantiate the class. |
39
|
|
|
* |
40
|
|
|
* @param Source $source |
41
|
|
|
*/ |
42
|
342 |
|
public function __construct(private Source $source) |
43
|
|
|
{ |
44
|
342 |
|
$this->progress = new Progress(); |
45
|
|
|
} |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* Retrieve the JSON fragments |
49
|
|
|
* |
50
|
|
|
* @return \Generator<int, Token> |
51
|
|
|
*/ |
52
|
330 |
|
public function getIterator(): Traversable |
53
|
|
|
{ |
54
|
330 |
|
$buffer = ''; |
55
|
330 |
|
$inString = $isEscaping = false; |
56
|
|
|
|
57
|
330 |
|
foreach ($this->source as $chunk) { |
58
|
327 |
|
for ($i = 0, $size = strlen($chunk); $i < $size; $i++, $this->position++) { |
59
|
327 |
|
$character = $chunk[$i]; |
60
|
327 |
|
$inString = ($character == '"') != $inString || $isEscaping; |
61
|
327 |
|
$isEscaping = $character == '\\' && !$isEscaping; |
62
|
327 |
|
$shouldBuffer = $inString || !isset(Tokens::BOUNDARIES[$character]); |
63
|
|
|
|
64
|
327 |
|
if ($shouldBuffer && $buffer == '' && !isset(Tokens::TYPES[$character])) { |
65
|
10 |
|
throw new SyntaxException($character); |
66
|
|
|
} |
67
|
|
|
|
68
|
326 |
|
if ($shouldBuffer) { |
69
|
274 |
|
$buffer .= $character; |
70
|
274 |
|
continue; |
71
|
|
|
} |
72
|
|
|
|
73
|
326 |
|
if ($buffer != '') { |
74
|
274 |
|
yield Tokenizer::instance()->toToken($buffer); |
75
|
270 |
|
$buffer = ''; |
76
|
|
|
} |
77
|
|
|
|
78
|
326 |
|
if (isset(Tokens::DELIMITERS[$character])) { |
79
|
326 |
|
yield Tokenizer::instance()->toToken($character); |
80
|
|
|
} |
81
|
|
|
} |
82
|
|
|
} |
83
|
|
|
} |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* Retrieve the current position |
87
|
|
|
* |
88
|
|
|
* @return int |
89
|
|
|
*/ |
90
|
11 |
|
public function position(): int |
91
|
|
|
{ |
92
|
11 |
|
return $this->position; |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* Retrieve the parsing progress |
97
|
|
|
* |
98
|
|
|
* @return Progress |
99
|
|
|
*/ |
100
|
|
|
public function progress(): Progress |
101
|
|
|
{ |
102
|
|
|
return $this->progress->setCurrent($this->position)->setTotal($this->source->size()); |
103
|
|
|
} |
104
|
|
|
} |
105
|
|
|
|