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 CallbackFilterIterator; |
17
|
|
|
use Graze\CsvToken\Parser; |
18
|
|
|
use Graze\CsvToken\Tokeniser\StreamTokeniser; |
19
|
|
|
use Graze\DataFile\Format\CsvFormatInterface; |
20
|
|
|
use Graze\DataFile\Helper\MapIterator; |
21
|
|
|
use Iterator; |
22
|
|
|
use LimitIterator; |
23
|
|
|
use RuntimeException; |
24
|
|
|
|
25
|
|
|
class CsvParser implements ParserInterface |
26
|
|
|
{ |
27
|
|
|
/** @var CsvFormatInterface */ |
28
|
|
|
private $csvFormat; |
29
|
|
|
/** @var array */ |
30
|
|
|
private $headerRow = []; |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* @param CsvFormatInterface $csvFormat |
34
|
|
|
*/ |
35
|
13 |
|
public function __construct(CsvFormatInterface $csvFormat) |
36
|
|
|
{ |
37
|
13 |
|
$this->csvFormat = $csvFormat; |
38
|
13 |
|
} |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* @param resource $stream |
42
|
|
|
* |
43
|
|
|
* @return Iterator |
44
|
|
|
*/ |
45
|
12 |
|
public function parse($stream) |
46
|
|
|
{ |
47
|
12 |
|
$tokeniser = new StreamTokeniser($this->csvFormat, $stream); |
48
|
12 |
|
$parser = new Parser(); |
49
|
12 |
|
return $this->parseIterator( |
50
|
12 |
|
$parser->parse($tokeniser->getTokens()) |
51
|
|
|
); |
52
|
|
|
} |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* Parse a supplied iterator |
56
|
|
|
* |
57
|
|
|
* @param Iterator $iterator |
58
|
|
|
* |
59
|
|
|
* @return Iterator |
60
|
|
|
*/ |
61
|
12 |
|
private function parseIterator(Iterator $iterator) |
62
|
|
|
{ |
63
|
12 |
|
$iterator = $this->parseHeaderRow($iterator); |
64
|
12 |
|
if ($this->csvFormat->getDataStart() > 1 || $this->csvFormat->getLimit() !== -1) { |
65
|
6 |
|
$iterator = new LimitIterator( |
66
|
|
|
$iterator, |
67
|
6 |
|
max(0, $this->csvFormat->getDataStart() - 1), |
68
|
6 |
|
$this->csvFormat->getLimit() |
69
|
|
|
); |
70
|
|
|
} |
71
|
12 |
|
return $iterator; |
72
|
|
|
} |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* @param Iterator $iterator |
76
|
|
|
* |
77
|
|
|
* @return Iterator |
78
|
|
|
*/ |
79
|
12 |
|
private function parseHeaderRow(Iterator $iterator) |
80
|
|
|
{ |
81
|
12 |
|
if ($this->csvFormat->hasHeaderRow()) { |
82
|
4 |
|
$iterator = new CallbackFilterIterator($iterator, [$this, 'handleHeaderRow']); |
83
|
4 |
|
$iterator = new MapIterator($iterator, [$this, 'mapHeaders']); |
84
|
|
|
} |
85
|
12 |
|
return $iterator; |
86
|
|
|
} |
87
|
|
|
|
88
|
|
|
/** |
89
|
|
|
* Parse the rows looking for the header row and storing it locally |
90
|
|
|
* |
91
|
|
|
* @param array $current |
92
|
|
|
* @param int $key |
93
|
|
|
* |
94
|
|
|
* @return bool |
95
|
|
|
*/ |
96
|
4 |
|
public function handleHeaderRow(array $current, $key) |
97
|
|
|
{ |
98
|
4 |
|
if ($key == $this->csvFormat->getHeaderRow() - 1) { |
99
|
4 |
|
if (count($current) > 1 || strlen($current[0]) > 0) { |
100
|
3 |
|
$this->headerRow = $current; |
101
|
|
|
} |
102
|
|
|
} |
103
|
4 |
|
return true; |
104
|
|
|
} |
105
|
|
|
|
106
|
|
|
/** |
107
|
|
|
* Map any headers found onto each element in the data |
108
|
|
|
* |
109
|
|
|
* @param array $current |
110
|
|
|
* |
111
|
|
|
* @return array |
112
|
|
|
*/ |
113
|
4 |
|
public function mapHeaders(array $current) |
114
|
|
|
{ |
115
|
4 |
|
if (count($this->headerRow) > 0) { |
116
|
3 |
|
if (count($this->headerRow) !== count($current)) { |
117
|
1 |
|
throw new RuntimeException( |
118
|
1 |
|
"The number of entries in: " . implode(',', $current) . |
119
|
1 |
|
" does not match the header: " . implode(',', $this->headerRow) |
120
|
|
|
); |
121
|
|
|
} |
122
|
2 |
|
return array_combine( |
123
|
2 |
|
$this->headerRow, |
124
|
|
|
$current |
125
|
|
|
); |
126
|
|
|
} |
127
|
1 |
|
return $current; |
128
|
|
|
} |
129
|
|
|
} |
130
|
|
|
|