1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Smart\EtlBundle\Utils; |
4
|
|
|
|
5
|
|
|
use Smart\EtlBundle\Exception\Utils\ArrayUtils\MultiArrayHeaderConsistencyException; |
6
|
|
|
use Smart\EtlBundle\Exception\Utils\ArrayUtils\MultiArrayNbMaxRowsException; |
7
|
|
|
|
8
|
|
|
/** |
9
|
|
|
* @author Mathieu Ducrot <[email protected]> |
10
|
|
|
*/ |
11
|
|
|
class ArrayUtils |
12
|
|
|
{ |
13
|
|
|
/** |
14
|
|
|
* Clean data from an array |
15
|
|
|
* - remove null value and empty string |
16
|
|
|
*/ |
17
|
|
|
public static function removeEmpty(array $array): array |
18
|
|
|
{ |
19
|
|
|
return array_filter($array, function ($item) { |
20
|
|
|
return $item !== null && (!is_string($item) || $item !== ''); |
21
|
|
|
}); |
22
|
|
|
} |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* Convert and clean data from string to array |
26
|
|
|
*/ |
27
|
|
|
public static function getArrayFromString(?string $string): array |
28
|
|
|
{ |
29
|
|
|
$toReturn = explode(PHP_EOL, $string); |
|
|
|
|
30
|
|
|
$toReturn = array_map(function ($row) { |
31
|
|
|
return trim($row); |
32
|
|
|
}, $toReturn); |
33
|
|
|
|
34
|
|
|
return array_unique(self::removeEmpty($toReturn)); |
35
|
|
|
} |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* Convert and clean data from string to a multidimensional array |
39
|
|
|
* |
40
|
|
|
* @param array $options list all options to configure the transform behavior. Current available options : |
41
|
|
|
* - delimiter (string) delimiter used to explode the row data |
42
|
|
|
* - nb_max_row (integer) maximum limit number of row allowed in the data content to import |
43
|
|
|
* - fix_header_consistency (boolean) allow to fix header consistency error |
44
|
|
|
* @throws MultiArrayNbMaxRowsException |
45
|
|
|
* @throws MultiArrayHeaderConsistencyException |
46
|
|
|
*/ |
47
|
|
|
public static function getMultiArrayFromString(string $string, array $header, array $options = []): array |
48
|
|
|
{ |
49
|
|
|
$nbRows = StringUtils::countRows($string); |
50
|
|
|
if (isset($options['nb_max_row']) && $nbRows > $options['nb_max_row']) { |
51
|
|
|
throw new MultiArrayNbMaxRowsException($options['nb_max_row'], $nbRows); |
52
|
|
|
} |
53
|
|
|
|
54
|
|
|
$toReturn = self::getArrayFromString($string); |
55
|
|
|
$nbHeader = count($header); |
56
|
|
|
$headerConsistencyError = []; |
57
|
|
|
|
58
|
|
|
// options config |
59
|
|
|
$delimiterOption = $options['delimiter'] ?? ','; |
60
|
|
|
$fixHeaderConsistencyOption = (isset($options['fix_header_consistency']) && $options['fix_header_consistency']); |
61
|
|
|
|
62
|
|
|
foreach ($toReturn as $key => $row) { |
63
|
|
|
$rowData = str_getcsv($row, $delimiterOption); |
64
|
|
|
|
65
|
|
|
// @todo extract in method mapTrimNull |
66
|
|
|
$rowData = array_map(function ($data) { |
67
|
|
|
if ($data !== null) { |
68
|
|
|
$data = trim($data); |
69
|
|
|
|
70
|
|
|
if ($data === '') { |
71
|
|
|
$data = null; |
72
|
|
|
} |
73
|
|
|
} |
74
|
|
|
|
75
|
|
|
return $data; |
76
|
|
|
}, $rowData); |
77
|
|
|
|
78
|
|
|
$validRowConsistency = true; |
79
|
|
|
if ($fixHeaderConsistencyOption) { |
80
|
|
|
// @todo extract in method syncHeader |
81
|
|
|
// fill missing field on the row with null values |
82
|
|
|
$rowData = array_pad($rowData, $nbHeader, null); |
83
|
|
|
// slice extra data on the row |
84
|
|
|
$rowData = array_slice($rowData, 0, $nbHeader); |
85
|
|
|
} else { |
86
|
|
|
if (count($rowData) != $nbHeader) { |
87
|
|
|
$headerConsistencyError[] = $key + 1; |
88
|
|
|
$validRowConsistency = false; |
89
|
|
|
} |
90
|
|
|
} |
91
|
|
|
|
92
|
|
|
// combine to form the multidimensional array |
93
|
|
|
if ($validRowConsistency) { |
94
|
|
|
$toReturn[$key] = array_combine($header, $rowData); |
95
|
|
|
} |
96
|
|
|
} |
97
|
|
|
|
98
|
|
|
if (count($headerConsistencyError) > 0) { |
99
|
|
|
throw new MultiArrayHeaderConsistencyException($headerConsistencyError); |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
return $toReturn; |
103
|
|
|
} |
104
|
|
|
} |
105
|
|
|
|