|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace itertools; |
|
4
|
|
|
|
|
5
|
|
|
use EmptyIterator; |
|
6
|
|
|
use Exception; |
|
7
|
|
|
use InvalidArgumentException; |
|
8
|
|
|
|
|
9
|
|
|
abstract class AbstractCsvIterator extends TakeWhileIterator |
|
10
|
|
|
{ |
|
11
|
|
|
protected $options = array(); |
|
12
|
|
|
|
|
13
|
|
|
public function __construct(array $options = array()) |
|
14
|
|
|
{ |
|
15
|
|
|
$defaultOptions = array( |
|
16
|
|
|
'delimiter' => ',', |
|
17
|
|
|
'enclosure' => '"', |
|
18
|
|
|
'escape' => '\\', |
|
19
|
|
|
'hasHeader' => true, |
|
20
|
|
|
'header' => null, |
|
21
|
|
|
'ignoreMissingRows' => false, // Deprecated: use ignoreMissingColumns |
|
22
|
|
|
'ignoreMissingColumns' => false, // columns* |
|
23
|
|
|
'combineFirstNRowsAsHeader' => 1, |
|
24
|
|
|
'skipFirstNRows' => 0 |
|
25
|
|
|
); |
|
26
|
|
|
|
|
27
|
|
|
$unknownOptions = array_diff(array_keys($options), array_keys($defaultOptions)); |
|
28
|
|
|
if(count($unknownOptions) != 0) { |
|
29
|
|
|
throw new InvalidArgumentException('Unknown options specified: ' . implode(', ', $unknownOptions)); |
|
30
|
|
|
} |
|
31
|
|
|
if(array_key_exists('header', $options) && null !== $options['header'] && ! array_key_exists('hasHeader', $options)) { |
|
32
|
|
|
$options['hasHeader'] = false; |
|
33
|
|
|
} |
|
34
|
|
|
if(array_key_exists('ignoreMissingRows', $options)) { |
|
35
|
|
|
$options['ignoreMissingColumns'] = $options['ignoreMissingRows']; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
$this->options = array_merge($this->options, $defaultOptions, $options); |
|
39
|
|
|
|
|
40
|
|
|
if($this->options['hasHeader'] || null !== $this->options['header']) { |
|
41
|
|
|
if(null === $this->options['header']) { |
|
42
|
|
|
$it = $this->getLineIteratorWithHeaderInFirstLines(); |
|
43
|
|
|
} else { |
|
44
|
|
|
$i = 0; |
|
45
|
|
|
while($this->options['skipFirstNRows'] > $i) { |
|
46
|
|
|
$i++; |
|
47
|
|
|
$this->retrieveNextCsvRow(); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
$it = $this->getLineIteratorWithHeader($this->options['header']); |
|
51
|
|
|
} |
|
52
|
|
|
} else { |
|
53
|
|
|
$it = $this->getLineIteratorWithoutHeader(); |
|
54
|
|
|
} |
|
55
|
|
|
parent::__construct($it, function($r) { return $r !== false; }); |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
abstract public function retrieveNextCsvRow(); |
|
59
|
|
|
|
|
60
|
|
|
protected function getLineIteratorWithoutHeader() |
|
61
|
|
|
{ |
|
62
|
|
|
return new SliceIterator(new CallbackIterator(array($this, 'retrieveNextCsvRow')), $this->options['skipFirstNRows']); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
protected function getLineIteratorWithHeader($header) |
|
66
|
|
|
{ |
|
67
|
|
|
if(false === $header || null === $header) { |
|
68
|
|
|
return new EmptyIterator(); |
|
69
|
|
|
} |
|
70
|
|
|
$nextRowRetriever = array($this, 'retrieveNextCsvRow'); |
|
71
|
|
|
$options = $this->options; |
|
72
|
|
|
|
|
73
|
|
|
if ($options['hasHeader'] && null !== $options['header']) { |
|
74
|
|
|
$this->retrieveNextCsvRow(); |
|
75
|
|
|
} |
|
76
|
|
|
return new CallbackIterator(function() use ($nextRowRetriever, $header, $options) { |
|
77
|
|
|
$row = call_user_func($nextRowRetriever); |
|
78
|
|
|
if(false === $row || array(null) === $row) { |
|
79
|
|
|
return false; |
|
80
|
|
|
} |
|
81
|
|
|
if(count($header) == count($row)) { |
|
82
|
|
|
return array_combine($header, $row); |
|
83
|
|
|
} |
|
84
|
|
|
if(! $options['ignoreMissingColumns']) { |
|
85
|
|
|
throw new InvalidCsvException('Your headers and columns do not match'); |
|
86
|
|
|
} |
|
87
|
|
|
if(count($header) < count($row)) { |
|
88
|
|
|
$row = array_slice($row, 0, count($header)); |
|
89
|
|
|
} else { |
|
90
|
|
|
$row = array_merge($row, array_fill(0, count($header) - count($row), null)); |
|
91
|
|
|
} |
|
92
|
|
|
return array_combine($header, $row); |
|
93
|
|
|
}); |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
protected function getLineIteratorWithHeaderInFirstLines() |
|
97
|
|
|
{ |
|
98
|
|
|
$headers = array_map(array($this, 'retrieveNextCsvRow'), range(1, $this->options['skipFirstNRows'] + $this->options['combineFirstNRowsAsHeader'])); |
|
99
|
|
|
$combinedHeader = null; |
|
100
|
|
|
$i = 0; |
|
101
|
|
|
foreach($headers as $header) { |
|
102
|
|
|
if($this->options['skipFirstNRows'] > $i) { |
|
103
|
|
|
$i++; |
|
104
|
|
|
continue; |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
if(false === $header || null === $header) { |
|
108
|
|
|
return new EmptyIterator(); |
|
109
|
|
|
} |
|
110
|
|
|
if(null === $combinedHeader) { |
|
111
|
|
|
$combinedHeader = array_fill_keys(array_keys($header), ''); |
|
112
|
|
|
} |
|
113
|
|
|
$previousNonEmptyColumnTitle = ''; |
|
114
|
|
|
foreach($header as $i => $columnTitle) { |
|
115
|
|
|
if(! array_key_exists($i, $header)) { |
|
116
|
|
|
if(! $this->options['ignoreMissingColumns']) { |
|
117
|
|
|
throw new InvalidCsvException('You provided a csv with missing columns'); |
|
118
|
|
|
} else { |
|
119
|
|
|
continue; |
|
120
|
|
|
} |
|
121
|
|
|
} |
|
122
|
|
|
$combinedHeader[$i] .= |
|
123
|
|
|
('' == trim($combinedHeader[$i]) || '' == trim($columnTitle) ? '' : ' ') . |
|
124
|
|
|
('' == trim($columnTitle) ? $previousNonEmptyColumnTitle : $columnTitle); |
|
125
|
|
|
if('' != trim($columnTitle)) { |
|
126
|
|
|
$previousNonEmptyColumnTitle = $columnTitle; |
|
127
|
|
|
} |
|
128
|
|
|
} |
|
129
|
|
|
} |
|
130
|
|
|
if(null === $combinedHeader) { |
|
131
|
|
|
throw new Exception('Unable to get header'); |
|
132
|
|
|
} |
|
133
|
|
|
// make columns unique (double columns become empty) |
|
134
|
|
|
$previousColumns = array(); |
|
135
|
|
|
foreach($combinedHeader as & $name) { |
|
136
|
|
|
if(array_key_exists($name, $previousColumns)) { |
|
137
|
|
|
$name = ''; |
|
138
|
|
|
} |
|
139
|
|
|
$previousColumns[$name] = ''; |
|
140
|
|
|
} |
|
141
|
|
|
return $this->getLineIteratorWithHeader($combinedHeader); |
|
142
|
|
|
} |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
|
SuRaMoN /
itertools
