|
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\BufferInterface; |
|
17
|
|
|
use Graze\CsvToken\Tokeniser\Token\Token; |
|
18
|
|
|
use Graze\CsvToken\Tokeniser\Token\TokenStoreInterface; |
|
19
|
|
|
use RuntimeException; |
|
20
|
|
|
|
|
21
|
|
|
class State |
|
22
|
|
|
{ |
|
23
|
|
|
const S_ANY = 0; |
|
24
|
|
|
const S_IN_QUOTE = 1; |
|
25
|
|
|
const S_IN_ESCAPE = 2; |
|
26
|
|
|
const S_IN_QUOTE_ESCAPE = 4; |
|
27
|
|
|
const S_INITIAL = 5; |
|
28
|
|
|
|
|
29
|
|
|
const S_INITIAL_TOKENS = Token::T_ANY & ~Token::T_DOUBLE_QUOTE; |
|
30
|
|
|
const S_ANY_TOKENS = Token::T_ANY & ~Token::T_DOUBLE_QUOTE & ~Token::T_BOM; |
|
31
|
|
|
const S_IN_QUOTE_TOKENS = Token::T_CONTENT | Token::T_QUOTE | Token::T_DOUBLE_QUOTE | Token::T_ESCAPE; |
|
32
|
|
|
const S_IN_ESCAPE_TOKENS = Token::T_CONTENT; |
|
33
|
|
|
const S_IN_QUOTE_ESCAPE_TOKENS = Token::T_CONTENT; |
|
34
|
|
|
|
|
35
|
|
|
/** @var State[] */ |
|
36
|
|
|
private $states; |
|
37
|
|
|
/** @var TokenStoreInterface */ |
|
38
|
|
|
private $tokenStore; |
|
39
|
|
|
/** @var int */ |
|
40
|
|
|
private $tokenMask; |
|
41
|
|
|
/** @var int[] */ |
|
42
|
|
|
private $tokens; |
|
43
|
|
|
/** @var string[] */ |
|
44
|
|
|
private $keys; |
|
45
|
|
|
/** @var int[] */ |
|
46
|
|
|
private $keyLengths; |
|
47
|
|
|
/** @var int */ |
|
48
|
|
|
private $maxLen; |
|
49
|
29 |
|
|
|
50
|
|
|
/** |
|
51
|
29 |
|
* TokenStoreInterface is passed in here, as the tokens can be modified by the store |
|
52
|
29 |
|
* |
|
53
|
29 |
|
* @param TokenStoreInterface $tokens |
|
54
|
|
|
* @param int $tokenMask |
|
55
|
|
|
*/ |
|
56
|
|
|
public function __construct(TokenStoreInterface $tokens, $tokenMask = Token::T_ANY) |
|
57
|
|
|
{ |
|
58
|
|
|
$this->tokenStore = $tokens; |
|
59
|
|
|
$this->tokenMask = $tokenMask; |
|
60
|
26 |
|
$this->parseTokens(); |
|
61
|
|
|
} |
|
62
|
26 |
|
|
|
63
|
26 |
|
/** |
|
64
|
26 |
|
* Parse the current set ok tokens and cache some metadata about them for speed |
|
65
|
|
|
*/ |
|
66
|
22 |
|
private function parseTokens() |
|
67
|
|
|
{ |
|
68
|
1 |
|
$this->tokens = $this->tokenStore->getTokens($this->tokenMask); |
|
69
|
|
|
$this->keys = array_keys($this->tokens); |
|
70
|
|
|
$this->keyLengths = array_unique(array_map('strlen', $this->keys)); |
|
71
|
|
|
arsort($this->keyLengths); |
|
72
|
|
|
$this->maxLen = reset($this->keyLengths); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
29 |
|
/** |
|
76
|
|
|
* @param int $token |
|
77
|
29 |
|
* |
|
78
|
29 |
|
* @return State|null |
|
79
|
|
|
*/ |
|
80
|
|
|
public function getNextState($token) |
|
81
|
|
|
{ |
|
82
|
|
|
foreach ($this->states as $mask => $state) { |
|
83
|
|
|
if ($mask & $token) { |
|
84
|
|
|
return $state; |
|
85
|
|
|
} |
|
86
|
25 |
|
} |
|
87
|
|
|
|
|
88
|
25 |
|
throw new RuntimeException("The supplied token: {$token} has no target state"); |
|
89
|
25 |
|
} |
|
90
|
24 |
|
|
|
91
|
|
|
/** |
|
92
|
25 |
|
* @param int $tokenMask |
|
93
|
|
|
* @param State $target |
|
94
|
25 |
|
*/ |
|
95
|
|
|
public function addStateTarget($tokenMask, State $target) |
|
96
|
|
|
{ |
|
97
|
|
|
$this->states[$tokenMask] = $target; |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
/** |
|
101
|
|
|
* @param BufferInterface $buffer |
|
102
|
|
|
* |
|
103
|
|
|
* @return array |
|
104
|
|
|
*/ |
|
105
|
|
|
public function match(BufferInterface $buffer) |
|
106
|
|
|
{ |
|
107
|
|
|
if ($this->tokenStore->hasChanged($this->tokenMask)) { |
|
108
|
|
|
$this->parseTokens(); |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
$contents = $buffer->getContents(); |
|
112
|
|
|
$length = $buffer->getLength(); |
|
113
|
|
|
$position = $buffer->getPosition(); |
|
114
|
|
|
if (count($this->tokens) > 0) { |
|
115
|
|
|
$totalLen = max($length - $this->maxLen, 1); |
|
116
|
|
|
for ($i = 0; $i < $totalLen; $i++) { |
|
117
|
|
|
foreach ($this->keyLengths as $len) { |
|
118
|
|
|
$buf = substr($contents, $i, $len); |
|
119
|
|
|
if (isset($this->tokens[$buf])) { |
|
120
|
|
|
if ($i > 0) { |
|
121
|
|
|
return [ |
|
122
|
|
|
[Token::T_CONTENT, substr($contents, 0, $i), $position, $i], |
|
123
|
|
|
[$this->tokens[$buf], $buf, $position + $i, $len], |
|
124
|
|
|
]; |
|
125
|
|
|
} else { |
|
126
|
|
|
return [[$this->tokens[$buf], $buf, $position, $len]]; |
|
127
|
|
|
} |
|
128
|
|
|
} |
|
129
|
|
|
} |
|
130
|
|
|
if ($totalLen !== $length && $i == $totalLen - 1) { |
|
131
|
|
|
$buffer->read(); |
|
132
|
|
|
$totalLen = $length = $buffer->getLength(); |
|
133
|
|
|
$contents = $buffer->getContents(); |
|
134
|
|
|
} |
|
135
|
|
|
} |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
|
|
return [[Token::T_CONTENT, $contents[0], $buffer->getPosition(), 1]]; |
|
139
|
|
|
} |
|
140
|
|
|
} |
|
141
|
|
|
|
graze /
csv-token
