1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* This file is part of graze/data-file |
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/data-file/blob/master/LICENSE.md |
11
|
|
|
* @link https://github.com/graze/data-file |
12
|
|
|
*/ |
13
|
|
|
|
14
|
|
|
namespace Graze\DataFile\Format\Parser; |
15
|
|
|
|
16
|
|
|
use ArrayIterator; |
17
|
|
|
use CallbackFilterIterator; |
18
|
|
|
use Graze\CsvToken\Csv\CsvConfiguration; |
19
|
|
|
use Graze\CsvToken\Parser; |
20
|
|
|
use Graze\CsvToken\Tokeniser\StreamTokeniser; |
21
|
|
|
use Graze\DataFile\Format\CsvFormatInterface; |
22
|
|
|
use Graze\DataFile\Helper\MapIterator; |
23
|
|
|
use Iterator; |
24
|
|
|
use LimitIterator; |
25
|
|
|
use Psr\Http\Message\StreamInterface; |
26
|
|
|
use RuntimeException; |
27
|
|
|
|
28
|
|
|
class CsvParser implements ParserInterface |
29
|
|
|
{ |
30
|
|
|
/** @var CsvFormatInterface */ |
31
|
|
|
private $csvFormat; |
32
|
|
|
/** @var array */ |
33
|
|
|
private $headerRow = []; |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* @param CsvFormatInterface $csvFormat |
37
|
|
|
*/ |
38
|
9 |
|
public function __construct(CsvFormatInterface $csvFormat) |
39
|
|
|
{ |
40
|
9 |
|
$this->csvFormat = $csvFormat; |
41
|
9 |
|
} |
42
|
|
|
|
43
|
|
|
/** |
44
|
|
|
* @param StreamInterface $stream |
45
|
|
|
* |
46
|
|
|
* @return Iterator |
47
|
|
|
*/ |
48
|
9 |
|
public function parse(StreamInterface $stream) |
49
|
|
|
{ |
50
|
|
|
$config = [ |
51
|
9 |
|
CsvConfiguration::OPTION_DELIMITER => $this->csvFormat->getDelimiter(), |
52
|
9 |
|
CsvConfiguration::OPTION_QUOTE => $this->csvFormat->getQuoteCharacter(), |
53
|
9 |
|
CsvConfiguration::OPTION_ESCAPE => $this->csvFormat->getEscapeCharacter(), |
54
|
9 |
|
CsvConfiguration::OPTION_DOUBLE_QUOTE => $this->csvFormat->isDoubleQuote(), |
55
|
9 |
|
CsvConfiguration::OPTION_NEW_LINES => [$this->csvFormat->getLineTerminator()], |
56
|
9 |
|
CsvConfiguration::OPTION_NULL => $this->csvFormat->getNullOutput(), |
57
|
9 |
|
CsvConfiguration::OPTION_ENCODING => $this->csvFormat->getEncoding(), |
58
|
|
|
]; |
59
|
9 |
|
if (!is_null($this->csvFormat->getBom())) { |
60
|
1 |
|
$config[CsvConfiguration::OPTION_BOMS] = [$this->csvFormat->getBom()]; |
61
|
|
|
} |
62
|
|
|
|
63
|
9 |
|
$configuration = new CsvConfiguration($config); |
64
|
9 |
|
$tokeniser = new StreamTokeniser($configuration, $stream); |
65
|
9 |
|
$parser = new Parser(); |
66
|
9 |
|
return $this->parseIterator( |
67
|
9 |
|
$parser->parse($tokeniser->getTokens()) |
68
|
|
|
); |
69
|
|
|
} |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* Parse a supplied iterator |
73
|
|
|
* |
74
|
|
|
* @param Iterator $iterator |
75
|
|
|
* |
76
|
|
|
* @return Iterator |
77
|
|
|
*/ |
78
|
9 |
|
private function parseIterator(Iterator $iterator) |
79
|
|
|
{ |
80
|
9 |
|
$iterator = $this->parseHeaderRow($iterator); |
81
|
9 |
|
if ($this->csvFormat->getDataStart() > 1 || $this->csvFormat->getLimit() !== -1) { |
82
|
5 |
|
$iterator = new LimitIterator( |
83
|
|
|
$iterator, |
84
|
5 |
|
max(0, $this->csvFormat->getDataStart() - 1), |
85
|
5 |
|
$this->csvFormat->getLimit() |
86
|
|
|
); |
87
|
|
|
} |
88
|
9 |
|
return $iterator; |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* @param Iterator $iterator |
93
|
|
|
* |
94
|
|
|
* @return Iterator |
95
|
|
|
*/ |
96
|
9 |
|
private function parseHeaderRow(Iterator $iterator) |
97
|
|
|
{ |
98
|
9 |
|
if ($this->csvFormat->hasHeaderRow()) { |
99
|
3 |
|
$iterator = new CallbackFilterIterator($iterator, [$this, 'handleHeaderRow']); |
100
|
3 |
|
$iterator = new MapIterator($iterator, [$this, 'mapHeaders']); |
101
|
|
|
} |
102
|
9 |
|
return $iterator; |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
/** |
106
|
|
|
* Parse the rows looking for the header row and storing it locally |
107
|
|
|
* |
108
|
|
|
* @param ArrayIterator $current |
109
|
|
|
* @param int $key |
110
|
|
|
* |
111
|
|
|
* @return bool |
112
|
|
|
*/ |
113
|
3 |
|
public function handleHeaderRow(ArrayIterator $current, $key) |
114
|
|
|
{ |
115
|
3 |
|
if ($key == $this->csvFormat->getHeaderRow() - 1) { |
116
|
3 |
|
$arr = iterator_to_array($current); |
117
|
3 |
|
if (count($arr) > 1 || strlen($arr[0]) > 0) { |
118
|
2 |
|
$this->headerRow = $arr; |
119
|
|
|
} |
120
|
|
|
} |
121
|
3 |
|
return true; |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
/** |
125
|
|
|
* Map any headers found onto each element in the data |
126
|
|
|
* |
127
|
|
|
* @param ArrayIterator $current |
128
|
|
|
* |
129
|
|
|
* @return ArrayIterator |
130
|
|
|
*/ |
131
|
3 |
|
public function mapHeaders(ArrayIterator $current) |
132
|
|
|
{ |
133
|
3 |
|
if (count($this->headerRow) > 0) { |
134
|
2 |
|
if (count($this->headerRow) !== $current->count()) { |
135
|
1 |
|
throw new RuntimeException( |
136
|
1 |
|
"The number of entries in: " . implode(',', iterator_to_array($current)) . |
137
|
1 |
|
" does not match the header: " . implode(',', $this->headerRow) |
138
|
|
|
); |
139
|
|
|
} |
140
|
1 |
|
return new ArrayIterator(array_combine( |
141
|
1 |
|
$this->headerRow, |
142
|
|
|
iterator_to_array($current) |
143
|
|
|
)); |
144
|
|
|
} |
145
|
1 |
|
return $current; |
146
|
|
|
} |
147
|
|
|
} |
148
|
|
|
|