|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Thruster\Component\Tokenizer; |
|
4
|
|
|
|
|
5
|
|
|
use Thruster\Component\Tokenizer\Token\NewLineToken; |
|
6
|
|
|
use Thruster\Component\Tokenizer\Token\WhitespaceToken; |
|
7
|
|
|
use Thruster\Component\Tokenizer\Token\ConstantEncapsedStringToken; |
|
8
|
|
|
use Thruster\Component\Tokenizer\Exception\FileNotFoundException; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Class Parser |
|
12
|
|
|
* |
|
13
|
|
|
* @package Thruster\Component\Tokenizer |
|
14
|
|
|
* @author Aurimas Niekis <[email protected]> |
|
15
|
|
|
*/ |
|
16
|
|
|
class Parser |
|
17
|
|
|
{ |
|
18
|
|
|
public function parseFile(string $file): array |
|
19
|
|
|
{ |
|
20
|
|
|
if (false === file_exists($file)) { |
|
21
|
|
|
throw new FileNotFoundException($file); |
|
22
|
|
|
} |
|
23
|
|
|
|
|
24
|
|
|
return $this->parse(file_get_contents($file)); |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
public function parseBlock(string $blockCode): array |
|
28
|
|
|
{ |
|
29
|
|
|
$tokens = $this->parse('<?php ' . $blockCode); |
|
30
|
|
|
|
|
31
|
|
|
array_shift($tokens); |
|
32
|
|
|
array_shift($tokens); |
|
33
|
|
|
|
|
34
|
|
|
return $tokens; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
public function parse(string $sourceCode): array |
|
38
|
|
|
{ |
|
39
|
|
|
$tokens = []; |
|
40
|
|
|
|
|
41
|
|
|
$line = 0; |
|
42
|
|
|
foreach (token_get_all($sourceCode) as $data) { |
|
43
|
|
|
/** @var Token $token */ |
|
44
|
|
|
if (false === is_array($data)) { |
|
45
|
|
|
$tokenClass = TokenInterface::CHARACTER_TO_CLASS_MAP[$data]; |
|
46
|
|
|
|
|
47
|
|
|
$tokens[] = new $tokenClass($data, $line); |
|
48
|
|
|
|
|
49
|
|
|
continue; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
$tokenClass = TokenInterface::TOKEN_TO_CLASS_MAP[$data[0]]; |
|
53
|
|
|
$content = $data[1]; |
|
54
|
|
|
$line = $data[2]; |
|
55
|
|
|
|
|
56
|
|
|
if ($tokenClass === ConstantEncapsedStringToken::class || 0 === strpos($content, '/*')) { |
|
57
|
|
|
$tokens[] = new $tokenClass($content, $line); |
|
58
|
|
|
|
|
59
|
|
|
continue; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
$splitContent = preg_split( |
|
63
|
|
|
'#(\r\n|\n| +)#', |
|
64
|
|
|
$content, |
|
65
|
|
|
-1, |
|
66
|
|
|
PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY |
|
67
|
|
|
); |
|
68
|
|
|
|
|
69
|
|
|
foreach ($splitContent as $splitPart) { |
|
70
|
|
|
if ("\r\n" === $splitPart || "\n" === $splitPart) { |
|
71
|
|
|
$tokens[] = new NewLineToken($splitPart, $line); |
|
72
|
|
|
$line++; |
|
73
|
|
|
|
|
74
|
|
|
continue; |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
if (0 === strpos($splitPart, ' ')) { |
|
78
|
|
|
$tokens[] = new WhitespaceToken($splitPart, $line); |
|
79
|
|
|
|
|
80
|
|
|
continue; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
$tokens[] = new $tokenClass($splitPart, $line); |
|
84
|
|
|
} |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
return $tokens; |
|
88
|
|
|
} |
|
89
|
|
|
} |
ThrusterIO /
tokenizer
