1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Graze\CsvToken\Tokeniser\Token; |
4
|
|
|
|
5
|
|
|
use Graze\CsvToken\Csv\Bom; |
6
|
|
|
use Graze\CsvToken\Csv\CsvConfigurationInterface; |
7
|
|
|
|
8
|
|
|
class TokenStore implements TokenStoreInterface |
9
|
|
|
{ |
10
|
|
|
/** @var array[] */ |
11
|
|
|
private $maskStore = []; |
12
|
|
|
/** @var int[] */ |
13
|
|
|
private $tokens = []; |
14
|
|
|
/** @var string|null */ |
15
|
|
|
private $lastEncoding = null; |
16
|
|
|
|
17
|
|
|
/** |
18
|
|
|
* TokenStore constructor. |
19
|
|
|
* |
20
|
|
|
* @param CsvConfigurationInterface $config |
21
|
|
|
*/ |
22
|
28 |
|
public function __construct(CsvConfigurationInterface $config) |
23
|
|
|
{ |
24
|
28 |
|
$this->tokens = $this->buildTokens($config); |
25
|
28 |
|
$this->setEncoding($config->getEncoding()); |
26
|
28 |
|
$this->sort(); |
27
|
28 |
|
} |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* @param int $mask |
31
|
|
|
* |
32
|
|
|
* @return array |
33
|
|
|
*/ |
34
|
28 |
|
public function getTokens($mask = Token::T_ANY) |
35
|
|
|
{ |
36
|
28 |
|
if (!array_key_exists($mask, $this->maskStore)) { |
37
|
|
|
$this->maskStore[$mask] = array_filter($this->tokens, function ($type) use ($mask) { |
38
|
28 |
|
return $type & $mask; |
39
|
28 |
|
}); |
40
|
28 |
|
} |
41
|
|
|
|
42
|
28 |
|
return $this->maskStore[$mask]; |
43
|
|
|
} |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* @param CsvConfigurationInterface $config |
47
|
|
|
* |
48
|
|
|
* @return int[] |
49
|
|
|
*/ |
50
|
28 |
|
private function buildTokens(CsvConfigurationInterface $config) |
51
|
|
|
{ |
52
|
|
|
$tokens = [ |
53
|
28 |
|
$config->getDelimiter() => Token::T_DELIMITER, |
54
|
28 |
|
]; |
55
|
|
|
|
56
|
28 |
|
if ($config->getQuote() != '') { |
57
|
26 |
|
$tokens[$config->getQuote()] = Token::T_QUOTE; |
58
|
26 |
|
if ($config->useDoubleQuotes()) { |
59
|
4 |
|
$tokens[str_repeat($config->getQuote(), 2)] = Token::T_DOUBLE_QUOTE; |
60
|
4 |
|
} |
61
|
26 |
|
} |
62
|
28 |
|
if ($config->getEscape() != '') { |
63
|
26 |
|
$tokens[$config->getEscape()] = Token::T_ESCAPE; |
64
|
26 |
|
} |
65
|
|
|
|
66
|
28 |
|
$newLines = $config->getNewLine(); |
67
|
28 |
|
if (!is_array($newLines)) { |
68
|
28 |
|
$newLines = [$newLines]; |
69
|
28 |
|
} |
70
|
28 |
|
foreach ($newLines as $newLine) { |
71
|
28 |
|
$tokens[$newLine] = Token::T_NEW_LINE; |
72
|
28 |
|
} |
73
|
28 |
|
if (!is_null($config->getNullValue())) { |
74
|
28 |
|
$tokens[$config->getNullValue()] = Token::T_NULL; |
75
|
28 |
|
} |
76
|
|
|
|
77
|
28 |
|
if (is_null($config->getBom())) { |
78
|
27 |
|
$boms = [Bom::BOM_UTF8, Bom::BOM_UTF16_BE, Bom::BOM_UTF16_LE, Bom::BOM_UTF32_BE, Bom::BOM_UTF32_LE]; |
79
|
27 |
|
} else { |
80
|
1 |
|
$boms = [$config->getBom()]; |
81
|
|
|
} |
82
|
28 |
|
foreach ($boms as $bom) { |
83
|
28 |
|
$tokens[$bom] = Token::T_BOM; |
84
|
28 |
|
} |
85
|
|
|
|
86
|
28 |
|
return $tokens; |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* @param string $encoding |
91
|
|
|
*/ |
92
|
28 |
|
public function setEncoding($encoding) |
93
|
|
|
{ |
94
|
28 |
|
if ($encoding != $this->lastEncoding) { |
95
|
28 |
|
if ($this->lastEncoding) { |
|
|
|
|
96
|
|
|
$changeEncoding = function ($string) use ($encoding) { |
97
|
2 |
|
return mb_convert_encoding($string, $encoding, $this->lastEncoding); |
98
|
2 |
|
}; |
99
|
2 |
|
} else { |
100
|
|
|
$changeEncoding = function ($string) use ($encoding) { |
101
|
28 |
|
return mb_convert_encoding($string, $encoding); |
102
|
28 |
|
}; |
103
|
|
|
} |
104
|
28 |
|
$tokens = []; |
105
|
28 |
|
foreach ($this->tokens as $string => $token) { |
106
|
28 |
|
if ($token != Token::T_BOM) { |
107
|
28 |
|
$string = $changeEncoding($string); |
108
|
28 |
|
} |
109
|
28 |
|
$tokens[$string] = $token; |
110
|
28 |
|
} |
111
|
28 |
|
$this->tokens = $tokens; |
112
|
28 |
|
$this->sort(); |
113
|
|
|
|
114
|
28 |
|
$this->lastEncoding = $encoding; |
115
|
28 |
|
$this->maskStore = []; |
116
|
28 |
|
} |
117
|
28 |
|
} |
118
|
|
|
|
119
|
|
|
/** |
120
|
|
|
* Sort the tokens into reverse key length order |
121
|
|
|
*/ |
122
|
|
|
private function sort() |
123
|
|
|
{ |
124
|
|
|
// sort by reverse key length |
125
|
28 |
|
uksort($this->tokens, function ($first, $second) { |
126
|
28 |
|
return strlen($second) - strlen($first); |
127
|
28 |
|
}); |
128
|
28 |
|
} |
129
|
|
|
} |
130
|
|
|
|
In PHP, under loose comparison (like
==
, or!=
, orswitch
conditions), values of different types might be equal.For
string
values, the empty string''
is a special case, in particular the following results might be unexpected: