|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of graze/csv-token |
|
4
|
|
|
* |
|
5
|
|
|
* Copyright (c) 2016 Nature Delivered Ltd. <https://www.graze.com> |
|
6
|
|
|
* |
|
7
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
8
|
|
|
* file that was distributed with this source code. |
|
9
|
|
|
* |
|
10
|
|
|
* @license https://github.com/graze/csv-token/blob/master/LICENSE.md |
|
11
|
|
|
* @link https://github.com/graze/csv-token |
|
12
|
|
|
*/ |
|
13
|
|
|
|
|
14
|
|
|
namespace Graze\CsvToken\Tokeniser; |
|
15
|
|
|
|
|
16
|
|
|
use Graze\CsvToken\Csv\Bom; |
|
17
|
|
|
use Graze\CsvToken\Csv\CsvConfigurationInterface; |
|
18
|
|
|
use Graze\CsvToken\Tokeniser\Token\Token; |
|
19
|
|
|
use Graze\CsvToken\Tokeniser\Token\TokenStore; |
|
20
|
|
|
use Iterator; |
|
21
|
|
|
use Psr\Http\Message\StreamInterface; |
|
22
|
|
|
|
|
23
|
|
|
class StreamTokeniser implements TokeniserInterface |
|
24
|
|
|
{ |
|
25
|
|
|
use StateBuilder; |
|
26
|
|
|
|
|
27
|
|
|
/** @var int */ |
|
28
|
|
|
private $maxTypeLength; |
|
29
|
|
|
/** @var StreamInterface */ |
|
30
|
|
|
private $stream; |
|
31
|
|
|
/** @var State */ |
|
32
|
|
|
private $state; |
|
33
|
|
|
/** @var TokenStore */ |
|
34
|
|
|
private $tokenStore; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* Tokeniser constructor. |
|
38
|
24 |
|
* |
|
39
|
|
|
* @param CsvConfigurationInterface $config |
|
40
|
24 |
|
* @param StreamInterface $stream |
|
41
|
24 |
|
*/ |
|
42
|
24 |
|
public function __construct(CsvConfigurationInterface $config, StreamInterface $stream) |
|
43
|
24 |
|
{ |
|
44
|
24 |
|
$this->tokenStore = new TokenStore($config); |
|
45
|
|
|
$this->state = $this->buildStates($this->tokenStore); |
|
46
|
|
|
$types = $this->tokenStore->getTokens(); |
|
47
|
|
|
$this->maxTypeLength = count($types) > 0 ? strlen(array_keys($types)[0]) : 1; |
|
48
|
|
|
$this->stream = $stream; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
23 |
|
* Loop through the stream, pulling maximum type length each time, find the largest type that matches and create a |
|
53
|
|
|
* token, then move on length characters |
|
54
|
23 |
|
* |
|
55
|
23 |
|
* @return Iterator |
|
56
|
23 |
|
*/ |
|
57
|
|
|
public function getTokens() |
|
58
|
|
|
{ |
|
59
|
23 |
|
$this->stream->rewind(); |
|
60
|
|
|
$position = $this->stream->tell(); |
|
61
|
23 |
|
$buffer = $this->stream->read($this->maxTypeLength); |
|
62
|
21 |
|
|
|
63
|
21 |
|
/** @var Token $last */ |
|
64
|
|
|
$last = null; |
|
65
|
21 |
|
|
|
66
|
|
|
while (strlen($buffer) > 0) { |
|
67
|
|
|
$token = $this->state->match($position, $buffer); |
|
68
|
21 |
|
|
|
69
|
21 |
|
if ($token->getType() == Token::T_BOM) { |
|
70
|
21 |
|
$this->changeEncoding($token); |
|
71
|
20 |
|
} |
|
72
|
19 |
|
|
|
73
|
19 |
|
$this->state = $this->state->getNextState($token->getType()); |
|
74
|
19 |
|
|
|
75
|
20 |
|
$len = $token->getLength(); |
|
76
|
|
|
|
|
77
|
|
|
// merge tokens together to condense T_CONTENT tokens |
|
78
|
21 |
|
if ($token->getType() == Token::T_CONTENT) { |
|
79
|
21 |
|
$last = (!is_null($last)) ? $last->addContent($token->getContent()) : $token; |
|
80
|
21 |
|
} else { |
|
81
|
|
|
if (!is_null($last)) { |
|
82
|
22 |
|
yield $last; |
|
83
|
7 |
|
$last = null; |
|
84
|
7 |
|
} |
|
85
|
|
|
yield $token; |
|
86
|
22 |
|
} |
|
87
|
22 |
|
|
|
88
|
|
|
$position += $len; |
|
89
|
|
|
$buffer = substr($buffer, $len) . $this->stream->read($len); |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
if (!is_null($last)) { |
|
93
|
|
|
yield $last; |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
$this->stream->close(); |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
/** |
|
100
|
|
|
* @param Token $token |
|
101
|
|
|
*/ |
|
102
|
|
|
private function changeEncoding(Token $token) |
|
103
|
|
|
{ |
|
104
|
|
|
$this->tokenStore->setEncoding(Bom::getEncoding($token->getContent())); |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
graze /
csv-token
