|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Phpml\Dataset; |
|
6
|
|
|
|
|
7
|
|
|
use Phpml\Exception\DatasetException; |
|
8
|
|
|
use Phpml\Exception\FileException; |
|
9
|
|
|
|
|
10
|
|
|
class SvmDataset extends ArrayDataset |
|
11
|
|
|
{ |
|
12
|
|
|
public function __construct(string $filePath) |
|
13
|
|
|
{ |
|
14
|
|
|
[$samples, $targets] = self::readProblem($filePath); |
|
15
|
|
|
|
|
16
|
|
|
parent::__construct($samples, $targets); |
|
17
|
|
|
} |
|
18
|
|
|
|
|
19
|
|
|
private static function readProblem(string $filePath): array |
|
20
|
|
|
{ |
|
21
|
|
|
$handle = self::openFile($filePath); |
|
22
|
|
|
|
|
23
|
|
|
$samples = []; |
|
24
|
|
|
$targets = []; |
|
25
|
|
|
$maxIndex = 0; |
|
26
|
|
|
while (false !== $line = fgets($handle)) { |
|
27
|
|
|
[$sample, $target, $maxIndex] = self::processLine($line, $maxIndex); |
|
28
|
|
|
$samples[] = $sample; |
|
29
|
|
|
$targets[] = $target; |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
fclose($handle); |
|
33
|
|
|
|
|
34
|
|
|
foreach ($samples as &$sample) { |
|
35
|
|
|
$sample = array_pad($sample, $maxIndex + 1, 0); |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
return [$samples, $targets]; |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @return resource |
|
43
|
|
|
*/ |
|
44
|
|
|
private static function openFile(string $filePath) |
|
45
|
|
|
{ |
|
46
|
|
|
if (!file_exists($filePath)) { |
|
47
|
|
|
throw new FileException(sprintf('File "%s" missing.', basename($filePath))); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
$handle = fopen($filePath, 'rb'); |
|
51
|
|
|
if ($handle === false) { |
|
52
|
|
|
throw new FileException(sprintf('File "%s" can\'t be open.', basename($filePath))); |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
return $handle; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
private static function processLine(string $line, int $maxIndex): array |
|
59
|
|
|
{ |
|
60
|
|
|
$columns = self::parseLine($line); |
|
61
|
|
|
|
|
62
|
|
|
$target = self::parseTargetColumn($columns[0]); |
|
63
|
|
|
$sample = array_fill(0, $maxIndex + 1, 0); |
|
64
|
|
|
|
|
65
|
|
|
$n = count($columns); |
|
66
|
|
|
for ($i = 1; $i < $n; ++$i) { |
|
67
|
|
|
[$index, $value] = self::parseFeatureColumn($columns[$i]); |
|
68
|
|
|
if ($index > $maxIndex) { |
|
69
|
|
|
$maxIndex = $index; |
|
70
|
|
|
$sample = array_pad($sample, $maxIndex + 1, 0); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
$sample[$index] = $value; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
return [$sample, $target, $maxIndex]; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
private static function parseLine(string $line): array |
|
80
|
|
|
{ |
|
81
|
|
|
$line = explode('#', $line, 2)[0]; |
|
82
|
|
|
$line = rtrim($line); |
|
83
|
|
|
$line = str_replace("\t", ' ', $line); |
|
84
|
|
|
|
|
85
|
|
|
return explode(' ', $line); |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
private static function parseTargetColumn(string $column): float |
|
89
|
|
|
{ |
|
90
|
|
|
if (!is_numeric($column)) { |
|
91
|
|
|
throw new DatasetException(sprintf('Invalid target "%s".', $column)); |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
return (float) $column; |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
private static function parseFeatureColumn(string $column): array |
|
98
|
|
|
{ |
|
99
|
|
|
$feature = explode(':', $column, 2); |
|
100
|
|
|
if (count($feature) !== 2) { |
|
101
|
|
|
throw new DatasetException(sprintf('Invalid value "%s".', $column)); |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
$index = self::parseFeatureIndex($feature[0]); |
|
105
|
|
|
$value = self::parseFeatureValue($feature[1]); |
|
106
|
|
|
|
|
107
|
|
|
return [$index, $value]; |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
private static function parseFeatureIndex(string $index): int |
|
111
|
|
|
{ |
|
112
|
|
|
if (!is_numeric($index) || !ctype_digit($index)) { |
|
113
|
|
|
throw new DatasetException(sprintf('Invalid index "%s".', $index)); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
if ((int) $index < 1) { |
|
117
|
|
|
throw new DatasetException(sprintf('Invalid index "%s".', $index)); |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
return (int) $index - 1; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
private static function parseFeatureValue(string $value): float |
|
124
|
|
|
{ |
|
125
|
|
|
if (!is_numeric($value)) { |
|
126
|
|
|
throw new DatasetException(sprintf('Invalid value "%s".', $value)); |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
return (float) $value; |
|
130
|
|
|
} |
|
131
|
|
|
} |
|
132
|
|
|
|
php-ai /
php-ml
