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\Token; |
15
|
|
|
|
16
|
|
|
use Graze\CsvToken\Csv\CsvConfigurationInterface; |
17
|
|
|
|
18
|
|
|
class TokenStore implements TokenStoreInterface |
19
|
|
|
{ |
20
|
|
|
/** @var int[][] */ |
21
|
|
|
private $maskStore = []; |
22
|
|
|
/** @var int[] */ |
23
|
|
|
private $tokens = []; |
24
|
|
|
/** @var string|null */ |
25
|
|
|
private $lastEncoding = null; |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* TokenStore constructor. |
29
|
|
|
* |
30
|
|
|
* @param CsvConfigurationInterface $config |
31
|
|
|
*/ |
32
|
30 |
|
public function __construct(CsvConfigurationInterface $config) |
33
|
|
|
{ |
34
|
30 |
|
$this->tokens = $this->buildTokens($config); |
35
|
30 |
|
$this->setEncoding($config->getEncoding()); |
36
|
30 |
|
$this->sort(); |
37
|
30 |
|
} |
38
|
|
|
|
39
|
|
|
/** |
40
|
|
|
* @param int $mask |
41
|
|
|
* |
42
|
|
|
* @return int[] |
43
|
|
|
*/ |
44
|
30 |
|
public function getTokens($mask = Token::T_ANY) |
45
|
|
|
{ |
46
|
30 |
|
if (!array_key_exists($mask, $this->maskStore)) { |
47
|
|
|
$this->maskStore[$mask] = array_filter($this->tokens, function ($type) use ($mask) { |
48
|
30 |
|
return $type & $mask; |
49
|
30 |
|
}); |
50
|
|
|
} |
51
|
|
|
|
52
|
30 |
|
return $this->maskStore[$mask]; |
53
|
|
|
} |
54
|
|
|
|
55
|
|
|
/** |
56
|
|
|
* @param CsvConfigurationInterface $config |
57
|
|
|
* |
58
|
|
|
* @return int[] |
59
|
|
|
*/ |
60
|
30 |
|
private function buildTokens(CsvConfigurationInterface $config) |
61
|
|
|
{ |
62
|
|
|
$tokens = [ |
63
|
30 |
|
$config->getDelimiter() => Token::T_DELIMITER, |
64
|
|
|
]; |
65
|
|
|
|
66
|
30 |
|
if ($config->getQuote() != '') { |
67
|
28 |
|
$tokens[$config->getQuote()] = Token::T_QUOTE; |
68
|
28 |
|
if ($config->useDoubleQuotes()) { |
69
|
5 |
|
$tokens[str_repeat($config->getQuote(), 2)] = Token::T_DOUBLE_QUOTE; |
70
|
|
|
} |
71
|
|
|
} |
72
|
30 |
|
if ($config->getEscape() != '') { |
73
|
28 |
|
$tokens[$config->getEscape()] = Token::T_ESCAPE; |
74
|
|
|
} |
75
|
|
|
|
76
|
30 |
|
foreach ($config->getNewLines() as $newLine) { |
77
|
30 |
|
$tokens[$newLine] = Token::T_NEW_LINE; |
78
|
|
|
} |
79
|
30 |
|
if (!is_null($config->getNullValue())) { |
80
|
30 |
|
$tokens[$config->getNullValue()] = Token::T_NULL; |
81
|
|
|
} |
82
|
|
|
|
83
|
30 |
|
foreach ($config->getBoms() as $bom) { |
84
|
30 |
|
$tokens[$bom] = Token::T_BOM; |
85
|
|
|
} |
86
|
|
|
|
87
|
30 |
|
return $tokens; |
88
|
|
|
} |
89
|
|
|
|
90
|
|
|
/** |
91
|
|
|
* @param string $encoding |
92
|
|
|
*/ |
93
|
30 |
|
public function setEncoding($encoding) |
94
|
|
|
{ |
95
|
30 |
|
if ($encoding != $this->lastEncoding) { |
96
|
30 |
|
if (!is_null($this->lastEncoding)) { |
97
|
|
|
$changeEncoding = function ($string) use ($encoding) { |
98
|
2 |
|
return mb_convert_encoding($string, $encoding, $this->lastEncoding); |
99
|
2 |
|
}; |
100
|
|
|
} else { |
101
|
|
|
$changeEncoding = function ($string) use ($encoding) { |
102
|
30 |
|
return mb_convert_encoding($string, $encoding); |
103
|
30 |
|
}; |
104
|
|
|
} |
105
|
30 |
|
$tokens = []; |
106
|
30 |
|
foreach ($this->tokens as $string => $token) { |
107
|
30 |
|
if ($token != Token::T_BOM) { |
108
|
30 |
|
$string = $changeEncoding($string); |
109
|
|
|
} |
110
|
30 |
|
$tokens[$string] = $token; |
111
|
|
|
} |
112
|
30 |
|
$this->tokens = $tokens; |
113
|
30 |
|
$this->sort(); |
114
|
|
|
|
115
|
30 |
|
$this->lastEncoding = $encoding; |
116
|
30 |
|
$this->maskStore = []; |
117
|
|
|
} |
118
|
30 |
|
} |
119
|
|
|
|
120
|
|
|
/** |
121
|
|
|
* Sort the tokens into reverse key length order |
122
|
|
|
*/ |
123
|
|
|
private function sort() |
124
|
|
|
{ |
125
|
|
|
// sort by reverse key length |
126
|
30 |
|
uksort($this->tokens, function ($first, $second) { |
127
|
30 |
|
return strlen($second) - strlen($first); |
128
|
30 |
|
}); |
129
|
30 |
|
} |
130
|
|
|
|
131
|
|
|
/** |
132
|
|
|
* Determine if a mask set of tokens has changed |
133
|
|
|
* |
134
|
|
|
* @param int $mask |
135
|
|
|
* |
136
|
|
|
* @return bool |
137
|
|
|
*/ |
138
|
27 |
|
public function hasChanged($mask = Token::T_ANY) |
139
|
|
|
{ |
140
|
27 |
|
return !isset($this->maskStore[$mask]); |
141
|
|
|
} |
142
|
|
|
} |
143
|
|
|
|