1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* CSVelte: Slender, elegant CSV for PHP |
4
|
|
|
* |
5
|
|
|
* Inspired by Python's CSV module and Frictionless Data and the W3C's CSV |
6
|
|
|
* standardization efforts, CSVelte was written in an effort to take all the |
7
|
|
|
* suck out of working with CSV. |
8
|
|
|
* |
9
|
|
|
* @copyright Copyright (c) 2018 Luke Visinoni |
10
|
|
|
* @author Luke Visinoni <[email protected]> |
11
|
|
|
* @license See LICENSE file (MIT license) |
12
|
|
|
*/ |
13
|
|
|
namespace CSVelte\Sniffer; |
14
|
|
|
|
15
|
|
|
use function Noz\collect; |
16
|
|
|
use Noz\Collection\Collection; |
17
|
|
|
use function Stringy\create as s; |
18
|
|
|
|
19
|
|
|
class SniffDelimiterByDistribution extends AbstractSniffer |
20
|
|
|
{ |
21
|
|
|
/** |
22
|
|
|
* Guess delimiter in a string of data |
23
|
|
|
* |
24
|
|
|
* Guesses the delimiter in a data set by analyzing which of the provided possible delimiter characters is most |
25
|
|
|
* evenly distributed (horizontally) across the dataset. |
26
|
|
|
* |
27
|
|
|
* @param string $data The data to analyze |
28
|
|
|
* |
29
|
|
|
* @return string[] |
30
|
|
|
*/ |
31
|
1 |
|
public function sniff($data) |
32
|
|
|
{ |
33
|
1 |
|
$lineTerminator = $this->getOption('lineTerminator') ?: "\n"; |
34
|
1 |
|
$delimiters = $this->getOption('delimiters'); |
35
|
1 |
|
$lines = collect(explode($lineTerminator, $this->removeQuotedStrings($data))); |
36
|
|
|
return collect($delimiters)->flip()->map(function($x, $char) use ($lines) { |
|
|
|
|
37
|
|
|
|
38
|
|
|
// standard deviation |
39
|
|
|
$sd = $lines->map(function($line, $line_no) use ($char) { |
|
|
|
|
40
|
1 |
|
$delimited = collect(s($line)->split($char)) |
41
|
|
|
->map(function($str) { |
42
|
1 |
|
return $str->length(); |
43
|
1 |
|
}); |
44
|
|
|
// standard deviation |
45
|
1 |
|
$avg = $delimited->average(); |
46
|
1 |
|
return sqrt($delimited->recollect(function(Collection $d, $len) use ($avg) { |
47
|
1 |
|
return $d->add(pow($len - $avg, 2)); |
48
|
1 |
|
}) |
49
|
1 |
|
->sum() / $delimited->count()); |
50
|
1 |
|
}); |
51
|
1 |
|
return $sd->average(); |
52
|
|
|
|
53
|
1 |
|
}) |
54
|
1 |
|
->sort() |
55
|
1 |
|
->getKeyAt(1); |
56
|
|
|
} |
57
|
|
|
} |