From e521fb8f803b1855f24336f95328e99b98c24eab Mon Sep 17 00:00:00 2001
From: Arkadiusz Kondas <arkadiusz.kondas@gmail.com>
Date: Wed, 6 Apr 2016 21:46:17 +0200
Subject: [PATCH] iris dataset loader

---
 src/Phpml/Dataset/Dataset.php            | 58 ++++++++++++++++++++++++
 src/Phpml/Dataset/Iris.php               | 11 +++++
 src/Phpml/Exception/DatasetException.php | 22 +++++++++
 tests/Phpml/Dataset/IrisTest.php         | 23 ++++++++++
 4 files changed, 114 insertions(+)
 create mode 100644 src/Phpml/Exception/DatasetException.php
 create mode 100644 tests/Phpml/Dataset/IrisTest.php

diff --git a/src/Phpml/Dataset/Dataset.php b/src/Phpml/Dataset/Dataset.php
index 316ee1d..cf1574e 100644
--- a/src/Phpml/Dataset/Dataset.php
+++ b/src/Phpml/Dataset/Dataset.php
@@ -1,9 +1,67 @@
 <?php
 
+declare (strict_types = 1);
+
+namespace Phpml\Dataset;
+
+use Phpml\Exception\DatasetException;
+
 abstract class Dataset
 {
     /**
      * @var string
      */
     protected $filepath;
+
+    /**
+     * @var array
+     */
+    private $samples = [];
+
+    /**
+     * @var array
+     */
+    private $lables = [];
+    
+    public function __construct()
+    {
+        $filepath =  dirname(__FILE__) . '/../../../data/' . $this->filepath;
+
+        if(!file_exists($filepath)) {
+            throw DatasetException::missingFile(basename($filepath));
+        }
+
+        $row = 0;
+        if (($handle = fopen($filepath, "r")) !== FALSE) {
+            while (($data = fgetcsv($handle, 1000, ",")) !== FALSE) {
+                $row++;
+                if($row==1) {
+                    continue;
+                }
+                $this->samples[] = array_slice($data, 0, 4);
+                $this->lables[] = $data[4];
+            }
+            fclose($handle);
+        } else {
+            throw DatasetException::cantOpenFile(basename($filepath));
+        }
+
+    }
+
+    /**
+     * @return array
+     */
+    public function getSamples()
+    {
+        return $this->samples;
+    }
+
+    /**
+     * @return array
+     */
+    public function getLabels()
+    {
+        return $this->lables;
+    }
+
 }
diff --git a/src/Phpml/Dataset/Iris.php b/src/Phpml/Dataset/Iris.php
index 0565558..b862a74 100644
--- a/src/Phpml/Dataset/Iris.php
+++ b/src/Phpml/Dataset/Iris.php
@@ -1,9 +1,20 @@
 <?php
 
+declare (strict_types = 1);
+
+namespace Phpml\Dataset;
+
+/**
+ * Classes: 3
+ * Samples per class: 50
+ * Samples total: 150
+ * Features per sample: 4
+ */
 class Iris extends Dataset
 {
     /**
      * @var string
      */
     protected $filepath = 'iris.csv';
+
 }
diff --git a/src/Phpml/Exception/DatasetException.php b/src/Phpml/Exception/DatasetException.php
new file mode 100644
index 0000000..f12c979
--- /dev/null
+++ b/src/Phpml/Exception/DatasetException.php
@@ -0,0 +1,22 @@
+<?php
+
+declare (strict_types = 1);
+
+namespace Phpml\Exception;
+
+class DatasetException extends \Exception
+{
+    /**
+     * @return DatasetException
+     */
+    public static function missingFile($filepath)
+    {
+        return new self(sprintf('Dataset file %s missing.', $filepath));
+    }
+
+    public static function cantOpenFile($filepath)
+    {
+        return new self(sprintf('Dataset file %s can\'t be open.', $filepath));
+    }
+
+}
diff --git a/tests/Phpml/Dataset/IrisTest.php b/tests/Phpml/Dataset/IrisTest.php
new file mode 100644
index 0000000..6bce532
--- /dev/null
+++ b/tests/Phpml/Dataset/IrisTest.php
@@ -0,0 +1,23 @@
+<?php
+declare(strict_types = 1);
+
+namespace tests\Phpml\Dataset;
+
+use Phpml\Dataset\Iris;
+
+class IrisTest extends \PHPUnit_Framework_TestCase
+{
+
+    public function testLoadingIrisDataset()
+    {
+        $iris = new Iris();
+
+        // whole dataset
+        $this->assertEquals(150, count($iris->getSamples()));
+        $this->assertEquals(150, count($iris->getLabels()));
+
+        // one sample features count
+        $this->assertEquals(4, count($iris->getSamples()[0]));
+    }
+    
+}
\ No newline at end of file