|
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
|
5 |
|
/** @var array */ |
|
33
|
|
|
private $headerRow = []; |
|
34
|
5 |
|
|
|
35
|
5 |
|
/** |
|
36
|
|
|
* @param CsvFormatInterface $csvFormat |
|
37
|
|
|
*/ |
|
38
|
|
|
public function __construct(CsvFormatInterface $csvFormat) |
|
39
|
|
|
{ |
|
40
|
|
|
$this->csvFormat = $csvFormat; |
|
41
|
|
|
} |
|
42
|
5 |
|
|
|
43
|
|
|
/** |
|
44
|
5 |
|
* @param StreamInterface $stream |
|
45
|
5 |
|
* |
|
46
|
5 |
|
* @return Iterator |
|
47
|
5 |
|
*/ |
|
48
|
5 |
|
public function parse(StreamInterface $stream) |
|
49
|
5 |
|
{ |
|
50
|
5 |
|
$configuration = new CsvConfiguration([ |
|
51
|
5 |
|
CsvConfiguration::OPTION_DELIMITER => $this->csvFormat->getDelimiter(), |
|
52
|
5 |
|
CsvConfiguration::OPTION_QUOTE => $this->csvFormat->getQuoteCharacter(), |
|
53
|
5 |
|
CsvConfiguration::OPTION_ESCAPE => $this->csvFormat->getEscapeCharacter(), |
|
54
|
5 |
|
CsvConfiguration::OPTION_DOUBLE_QUOTE => $this->csvFormat->isDoubleQuote(), |
|
55
|
5 |
|
CsvConfiguration::OPTION_NEW_LINE => $this->csvFormat->getLineTerminator(), |
|
56
|
5 |
|
CsvConfiguration::OPTION_NULL => $this->csvFormat->getNullOutput(), |
|
57
|
|
|
]); |
|
58
|
|
|
$tokeniser = new StreamTokeniser($configuration, $stream); |
|
59
|
|
|
$parser = new Parser(); |
|
60
|
|
|
return $this->parseIterator( |
|
61
|
|
|
$parser->parse($tokeniser->getTokens()) |
|
62
|
|
|
); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
5 |
|
* Parse a supplied iterator |
|
67
|
|
|
* |
|
68
|
5 |
|
* @param Iterator $iterator |
|
69
|
2 |
|
* |
|
70
|
2 |
|
* @return Iterator |
|
71
|
5 |
|
*/ |
|
72
|
|
|
private function parseIterator(Iterator $iterator) |
|
73
|
|
|
{ |
|
74
|
|
|
$iterator = $this->parseHeaderRow($iterator); |
|
75
|
|
|
if ($this->csvFormat->getDataStart() > 1 || $this->csvFormat->getLimit() !== -1) { |
|
76
|
|
|
$iterator = new LimitIterator( |
|
77
|
|
|
$iterator, |
|
78
|
|
|
max(0, $this->csvFormat->getDataStart() - 1), |
|
79
|
|
|
$this->csvFormat->getLimit() |
|
80
|
|
|
); |
|
81
|
|
|
} |
|
82
|
|
|
return $iterator; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
/** |
|
86
|
|
|
* @param Iterator $iterator |
|
87
|
|
|
* |
|
88
|
|
|
* @return Iterator |
|
89
|
|
|
*/ |
|
90
|
|
|
private function parseHeaderRow(Iterator $iterator) |
|
91
|
|
|
{ |
|
92
|
|
|
if ($this->csvFormat->hasHeaderRow()) { |
|
93
|
|
|
// use a callback iterator to get just the header row without having to rewind the source |
|
94
|
|
|
$iterator = new CallbackFilterIterator($iterator, function ($current, $key) { |
|
95
|
|
|
if ($key == $this->csvFormat->getHeaderRow() - 1) { |
|
96
|
|
|
$arr = iterator_to_array($current); |
|
97
|
|
|
if (count($arr) > 1 || strlen($arr[0]) > 0) { |
|
98
|
|
|
$this->headerRow = $arr; |
|
99
|
|
|
} |
|
100
|
|
|
} |
|
101
|
|
|
return true; |
|
102
|
|
|
}); |
|
103
|
|
|
|
|
104
|
|
|
// map the header row onto each row if applicable |
|
105
|
|
|
$iterator = new MapIterator($iterator, function ($current) { |
|
106
|
|
|
if (count($this->headerRow) > 0) { |
|
107
|
|
|
if (count($this->headerRow) !== count($current)) { |
|
108
|
|
|
throw new RuntimeException( |
|
109
|
|
|
"The number of entries in: " . implode(',', iterator_to_array($current)) . |
|
110
|
|
|
" does not match the header: " . implode(',', $this->headerRow) |
|
111
|
|
|
); |
|
112
|
|
|
} |
|
113
|
|
|
return new ArrayIterator(array_combine( |
|
114
|
|
|
$this->headerRow, |
|
115
|
|
|
iterator_to_array($current) |
|
116
|
|
|
)); |
|
117
|
|
|
} |
|
118
|
|
|
return $current; |
|
119
|
|
|
}); |
|
120
|
|
|
} |
|
121
|
|
|
return $iterator; |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
|