|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Graze\CsvToken\Tokeniser; |
|
4
|
|
|
|
|
5
|
|
|
use Graze\CsvToken\Csv\CsvConfigurationInterface; |
|
6
|
|
|
use Iterator; |
|
7
|
|
|
use Psr\Http\Message\StreamInterface; |
|
8
|
|
|
|
|
9
|
|
|
class StreamTokeniser implements TokeniserInterface |
|
10
|
|
|
{ |
|
11
|
|
|
use TypeBuilder; |
|
12
|
|
|
|
|
13
|
|
|
/** @var int[] */ |
|
14
|
|
|
private $types; |
|
15
|
|
|
/** @var int */ |
|
16
|
|
|
private $maxTypeLength; |
|
17
|
|
|
/** @var StreamInterface */ |
|
18
|
|
|
private $stream; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* Tokeniser constructor. |
|
22
|
|
|
* |
|
23
|
|
|
* @param CsvConfigurationInterface $config |
|
24
|
|
|
* @param StreamInterface $stream |
|
25
|
|
|
*/ |
|
26
|
16 |
|
public function __construct(CsvConfigurationInterface $config, StreamInterface $stream) |
|
27
|
|
|
{ |
|
28
|
16 |
|
$this->types = $this->getTypes($config); |
|
29
|
|
|
|
|
30
|
|
|
// sort by reverse key length |
|
31
|
16 |
|
uksort($this->types, function ($a, $b) { |
|
|
|
|
|
|
32
|
16 |
|
return strlen($b) - strlen($a); |
|
33
|
16 |
|
}); |
|
34
|
16 |
|
$this->maxTypeLength = count($this->types) > 0 ? strlen(array_keys($this->types)[0]) : 1; |
|
35
|
16 |
|
$this->stream = $stream; |
|
36
|
16 |
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Loop through the stream, pulling maximum type length each time, find the largest type that matches and create a |
|
40
|
|
|
* token, then move on length characters |
|
41
|
|
|
* |
|
42
|
|
|
* @return Iterator |
|
43
|
|
|
*/ |
|
44
|
15 |
|
public function getTokens() |
|
45
|
|
|
{ |
|
46
|
15 |
|
$this->stream->rewind(); |
|
47
|
15 |
|
$position = $this->stream->tell(); |
|
48
|
15 |
|
$buffer = $this->stream->read($this->maxTypeLength); |
|
49
|
|
|
|
|
50
|
|
|
/** @var Token $last */ |
|
51
|
15 |
|
$last = null; |
|
52
|
|
|
|
|
53
|
15 |
|
while (strlen($buffer) > 0) { |
|
54
|
13 |
|
$token = $this->match($position, $buffer); |
|
55
|
13 |
|
$len = $token->getLength(); |
|
56
|
|
|
|
|
57
|
|
|
// merge tokens together to condense T_CONTENT tokens |
|
58
|
13 |
|
if ($token->getType() == Token::T_CONTENT) { |
|
59
|
13 |
|
$last = (!is_null($last)) ? $last->addContent($token->getContent()) : $token; |
|
60
|
13 |
|
} else { |
|
61
|
12 |
|
if (!is_null($last)) { |
|
62
|
12 |
|
yield $last; |
|
63
|
12 |
|
$last = null; |
|
64
|
12 |
|
} |
|
65
|
12 |
|
yield $token; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
13 |
|
$position += $len; |
|
69
|
13 |
|
$buffer = substr($buffer, $len) . $this->stream->read($len); |
|
70
|
13 |
|
} |
|
71
|
|
|
|
|
72
|
14 |
|
if (!is_null($last)) { |
|
73
|
4 |
|
yield $last; |
|
74
|
4 |
|
} |
|
75
|
|
|
|
|
76
|
14 |
|
$this->stream->close(); |
|
77
|
14 |
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* @param int $position |
|
81
|
|
|
* @param string $buffer |
|
82
|
|
|
* |
|
83
|
|
|
* @return Token |
|
84
|
|
|
*/ |
|
85
|
13 |
|
private function match($position, $buffer) |
|
86
|
|
|
{ |
|
87
|
13 |
|
foreach ($this->types as $search => $tokenType) { |
|
88
|
13 |
|
if (substr($buffer, 0, strlen($search)) == $search) { |
|
89
|
12 |
|
return new Token($tokenType, $search, $position); |
|
90
|
|
|
} |
|
91
|
13 |
|
} |
|
92
|
|
|
|
|
93
|
13 |
|
return new Token(Token::T_CONTENT, $buffer[0], $position); |
|
94
|
|
|
} |
|
95
|
|
|
} |
|
96
|
|
|
|
graze /
csv-token
Loading history...
Short variable names may make your code harder to understand. Variable names should be self-descriptive. This check looks for variable names who are shorter than a configured minimum.