Phpml\Dataset\SvmDataset

Complexity

Total Complexity

21

Size/Duplication

Total Lines	121
Duplicated Lines	0 %

Coupling/Cohesion

Components	1
Dependencies	1

Importance

Changes

0

9 Methods

Rating	Name	Duplication	Size	Complexity
A	__construct()	0	6	1
A	readProblem()	0	21	3
A	openFile()	0	13	3
A	processLine()	0	20	3
A	parseLine()	0	10	1
A	parseTargetColumn()	0	8	2
A	parseFeatureColumn()	0	12	2
A	parseFeatureIndex()	0	12	4
A	parseFeatureValue()	0	8	2

<?php

declare(strict_types=1);

namespace Phpml\Dataset;

use Phpml\Exception\DatasetException;
use Phpml\Exception\FileException;

class SvmDataset extends ArrayDataset
{
    public function __construct(string $filePath)
    {
        [$samples, $targets] = self::readProblem($filePath);

The variable $samples does not exist. Did you forget to declare it?

The variable $targets does not exist. Did you forget to declare it?

        parent::__construct($samples, $targets);
    }

    private static function readProblem(string $filePath): array
    {
        $handle = self::openFile($filePath);

        $samples = [];
        $targets = [];
        $maxIndex = 0;
        while (($line = fgets($handle)) !== false) {
            [$sample, $target, $maxIndex] = self::processLine($line, $maxIndex);

The variable $sample seems only to be defined at a later point. Did you maybe move this code here without moving the variable definition?

The variable $target does not exist. Did you mean $targets?

            $samples[] = $sample;

The variable $sample seems only to be defined at a later point. Did you maybe move this code here without moving the variable definition?

            $targets[] = $target;

The variable $target does not exist. Did you mean $targets?

        }

        fclose($handle);

        foreach ($samples as &$sample) {
            $sample = array_pad($sample, $maxIndex + 1, 0);
        }

        return [$samples, $targets];
    }

    private static function openFile(string $filePath)
    {
        if (!file_exists($filePath)) {
            throw FileException::missingFile(basename($filePath));
        }

        $handle = fopen($filePath, 'rb');
        if ($handle === false) {
            throw FileException::cantOpenFile(basename($filePath));
        }

        return $handle;
    }

    private static function processLine(string $line, int $maxIndex): array
    {
        $columns = self::parseLine($line);

        $target = self::parseTargetColumn($columns[0]);
        $sample = array_fill(0, $maxIndex + 1, 0);

        $n = count($columns);
        for ($i = 1; $i < $n; ++$i) {
            [$index, $value] = self::parseFeatureColumn($columns[$i]);

The variable $index does not exist. Did you forget to declare it?

The variable $value does not exist. Did you forget to declare it?

            if ($index > $maxIndex) {
                $maxIndex = $index;
                $sample = array_pad($sample, $maxIndex + 1, 0);
            }

            $sample[$index] = $value;
        }

        return [$sample, $target, $maxIndex];
    }

    private static function parseLine(string $line): array
    {
        $line = explode('#', $line, 2)[0];
        $line = rtrim($line);
        $line = str_replace("\t", ' ', $line);

        $columns = explode(' ', $line);

        return $columns;
    }

    private static function parseTargetColumn(string $column): float
    {
        if (!is_numeric($column)) {
            throw DatasetException::invalidTarget($column);
        }

        return (float) $column;
    }

    private static function parseFeatureColumn(string $column): array
    {
        $feature = explode(':', $column, 2);
        if (count($feature) != 2) {
            throw DatasetException::invalidValue($column);
        }

        $index = self::parseFeatureIndex($feature[0]);
        $value = self::parseFeatureValue($feature[1]);

        return [$index, $value];
    }

    private static function parseFeatureIndex(string $index): int
    {
        if (!is_numeric($index) || !ctype_digit($index)) {
            throw DatasetException::invalidIndex($index);
        }

        if ((int) $index < 1) {
            throw DatasetException::invalidIndex($index);
        }

        return (int) $index - 1;
    }

    private static function parseFeatureValue(string $value): float
    {
        if (!is_numeric($value)) {
            throw DatasetException::invalidValue($value);
        }

        return (float) $value;
    }
}