|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* League.Csv (https://csv.thephpleague.com) |
|
5
|
|
|
* |
|
6
|
|
|
* (c) Ignace Nyamagana Butera <[email protected]> |
|
7
|
|
|
* |
|
8
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
9
|
|
|
* file that was distributed with this source code. |
|
10
|
|
|
*/ |
|
11
|
|
|
|
|
12
|
|
|
declare(strict_types=1); |
|
13
|
|
|
|
|
14
|
|
|
namespace League\Csv { |
|
15
|
|
|
|
|
16
|
|
|
use ReflectionClass; |
|
17
|
|
|
use function array_fill_keys; |
|
18
|
|
|
use function array_filter; |
|
19
|
|
|
use function array_reduce; |
|
20
|
|
|
use function array_unique; |
|
21
|
|
|
use function count; |
|
22
|
|
|
use function iterator_to_array; |
|
23
|
|
|
use function rsort; |
|
24
|
|
|
use function strlen; |
|
25
|
|
|
use function strpos; |
|
26
|
|
|
use const COUNT_RECURSIVE; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Returns the BOM sequence found at the start of the string. |
|
30
|
|
|
* |
|
31
|
|
|
* If no valid BOM sequence is found an empty string is returned |
|
32
|
|
|
*/ |
|
33
|
|
|
function bom_match(string $str): string |
|
34
|
|
|
{ |
|
35
|
42 |
|
static $list; |
|
36
|
42 |
|
if (null === $list) { |
|
37
|
3 |
|
$list = (new ReflectionClass(ByteSequence::class))->getConstants(); |
|
38
|
|
|
|
|
39
|
3 |
|
rsort($list); |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
42 |
|
foreach ($list as $sequence) { |
|
43
|
42 |
|
if (0 === strpos($str, $sequence)) { |
|
44
|
12 |
|
return $sequence; |
|
45
|
|
|
} |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
30 |
|
return ''; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Detect Delimiters usage in a {@link Reader} object. |
|
53
|
|
|
* |
|
54
|
|
|
* Returns a associative array where each key represents |
|
55
|
|
|
* a submitted delimiter and each value the number CSV fields found |
|
56
|
|
|
* when processing at most $limit CSV records with the given delimiter |
|
57
|
|
|
* |
|
58
|
|
|
* @param string[] $delimiters |
|
59
|
|
|
* |
|
60
|
|
|
* @return int[] |
|
61
|
|
|
*/ |
|
62
|
|
|
function delimiter_detect(Reader $csv, array $delimiters, int $limit = 1): array |
|
63
|
|
|
{ |
|
64
|
|
|
$delimiter_filter = static function (string $value): bool { |
|
65
|
18 |
|
return 1 === strlen($value); |
|
66
|
21 |
|
}; |
|
67
|
|
|
|
|
68
|
|
|
$record_filter = static function (array $record): bool { |
|
69
|
12 |
|
return count($record) > 1; |
|
70
|
21 |
|
}; |
|
71
|
|
|
|
|
72
|
21 |
|
$stmt = (new Statement())->limit($limit); |
|
73
|
|
|
|
|
74
|
|
|
$delimiter_stats = static function (array $stats, string $delimiter) use ($csv, $stmt, $record_filter): array { |
|
75
|
12 |
|
$csv->setDelimiter($delimiter); |
|
76
|
12 |
|
$found_records = array_filter( |
|
77
|
12 |
|
iterator_to_array($stmt->process($csv), false), |
|
78
|
12 |
|
$record_filter |
|
79
|
|
|
); |
|
80
|
|
|
|
|
81
|
12 |
|
$stats[$delimiter] = count($found_records, COUNT_RECURSIVE); |
|
82
|
|
|
|
|
83
|
12 |
|
return $stats; |
|
84
|
18 |
|
}; |
|
85
|
|
|
|
|
86
|
18 |
|
$current_delimiter = $csv->getDelimiter(); |
|
87
|
18 |
|
$current_header_offset = $csv->getHeaderOffset(); |
|
88
|
18 |
|
$csv->setHeaderOffset(null); |
|
89
|
|
|
|
|
90
|
18 |
|
$stats = array_reduce( |
|
91
|
18 |
|
array_unique(array_filter($delimiters, $delimiter_filter)), |
|
92
|
15 |
|
$delimiter_stats, |
|
93
|
15 |
|
array_fill_keys($delimiters, 0) |
|
94
|
|
|
); |
|
95
|
|
|
|
|
96
|
15 |
|
$csv->setHeaderOffset($current_header_offset); |
|
97
|
15 |
|
$csv->setDelimiter($current_delimiter); |
|
98
|
|
|
|
|
99
|
15 |
|
return $stats; |
|
100
|
|
|
} |
|
101
|
|
|
} |
|
102
|
|
|
|
thephpleague /
csv
