|
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\Buffer\StreamBuffer; |
|
17
|
|
|
use Graze\CsvToken\Csv\Bom; |
|
18
|
|
|
use Graze\CsvToken\Csv\CsvConfigurationInterface; |
|
19
|
|
|
use Graze\CsvToken\Tokeniser\Token\Token; |
|
20
|
|
|
use Graze\CsvToken\Tokeniser\Token\TokenStore; |
|
21
|
|
|
use Iterator; |
|
22
|
|
|
|
|
23
|
|
|
class StreamTokeniser implements TokeniserInterface |
|
24
|
|
|
{ |
|
25
|
|
|
use StateBuilder; |
|
26
|
|
|
|
|
27
|
|
|
const BUFFER_SIZE = 128; |
|
28
|
|
|
|
|
29
|
|
|
/** @var int */ |
|
30
|
|
|
private $minLength; |
|
31
|
|
|
/** @var resource */ |
|
32
|
|
|
private $stream; |
|
33
|
|
|
/** @var State */ |
|
34
|
|
|
private $state; |
|
35
|
|
|
/** @var TokenStore */ |
|
36
|
|
|
private $tokenStore; |
|
37
|
|
|
/** @var StreamBuffer */ |
|
38
|
|
|
private $buffer; |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* Tokeniser constructor. |
|
42
|
|
|
* |
|
43
|
|
|
* @param CsvConfigurationInterface $config |
|
44
|
|
|
* @param resource $stream |
|
45
|
|
|
*/ |
|
46
|
30 |
|
public function __construct(CsvConfigurationInterface $config, $stream) |
|
47
|
|
|
{ |
|
48
|
30 |
|
$this->tokenStore = new TokenStore($config); |
|
49
|
30 |
|
$this->state = $this->buildStates($this->tokenStore); |
|
50
|
30 |
|
$types = $this->tokenStore->getTokens(); |
|
51
|
30 |
|
$this->minLength = count($types) > 0 ? strlen(array_keys($types)[0]) * 2 : 1; |
|
52
|
30 |
|
$this->stream = $stream; |
|
53
|
30 |
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* Loop through the stream, pulling maximum type length each time, find the largest type that matches and create a |
|
57
|
|
|
* token, then move on length characters |
|
58
|
|
|
* |
|
59
|
|
|
* @return Iterator |
|
60
|
|
|
*/ |
|
61
|
29 |
|
public function getTokens() |
|
62
|
|
|
{ |
|
63
|
29 |
|
fseek($this->stream, 0); |
|
64
|
29 |
|
$this->buffer = new StreamBuffer($this->stream, static::BUFFER_SIZE, $this->minLength); |
|
65
|
29 |
|
$this->buffer->read(); |
|
66
|
|
|
|
|
67
|
|
|
/** @var Token $last */ |
|
68
|
29 |
|
$last = null; |
|
69
|
|
|
|
|
70
|
29 |
|
while (!$this->buffer->isEof()) { |
|
71
|
27 |
|
foreach ($this->state->match($this->buffer) as $token) { |
|
72
|
27 |
|
if ($token[0] == Token::T_BOM) { |
|
73
|
3 |
|
$this->changeEncoding($token[1]); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
27 |
|
$this->state = $this->state->getNextState($token[0]); |
|
77
|
|
|
|
|
78
|
|
|
// merge tokens together to condense T_CONTENT tokens |
|
79
|
27 |
|
if ($token[0] == Token::T_CONTENT) { |
|
80
|
27 |
|
if (!is_null($last)) { |
|
81
|
14 |
|
$last[1] .= $token[1]; |
|
82
|
14 |
|
$last[3] = strlen($last[1]); |
|
83
|
|
|
} else { |
|
84
|
27 |
|
$last = $token; |
|
85
|
|
|
} |
|
86
|
|
|
} else { |
|
87
|
26 |
|
if (!is_null($last)) { |
|
88
|
25 |
|
yield $last; |
|
89
|
25 |
|
$last = null; |
|
90
|
|
|
} |
|
91
|
26 |
|
yield $token; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
27 |
|
$this->buffer->move($token[3]); |
|
95
|
27 |
|
$this->buffer->read(); |
|
96
|
|
|
} |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
28 |
|
if (!is_null($last)) { |
|
100
|
7 |
|
yield $last; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
28 |
|
fclose($this->stream); |
|
104
|
28 |
|
} |
|
105
|
|
|
|
|
106
|
|
|
/** |
|
107
|
|
|
* @param string $content |
|
108
|
|
|
*/ |
|
109
|
3 |
|
private function changeEncoding($content) |
|
110
|
|
|
{ |
|
111
|
3 |
|
$this->tokenStore->setEncoding(Bom::getEncoding($content)); |
|
112
|
3 |
|
$types = $this->tokenStore->getTokens(); |
|
113
|
3 |
|
$this->minLength = count($types) > 0 ? strlen(array_keys($types)[0]) * 2 : 1; |
|
114
|
3 |
|
$this->buffer->setMinBufferSize($this->minLength); |
|
115
|
3 |
|
} |
|
116
|
|
|
} |
|
117
|
|
|
|
graze /
csv-token
