Initial unit tests, code cleanup

src/Dataset.php

@@ -7,6 +7,10 @@ use NoccyLabs\Dataset\Readers\JsonReader;
 
 class Dataset
 {
+    protected string $packageName;
+
+    protected string $datasetName;
+
     protected string $identifier;
     
     protected array $options;
@@ -18,6 +22,8 @@ class Dataset
         $this->identifier = $identifier;
         $this->options = $options;
         $this->version = $version;
+
+        [$this->packageName, $this->datasetName] = explode("#", $identifier, 2);
     }
 
     public function getIdentifier(): string
@@ -25,11 +31,31 @@ class Dataset
         return $this->identifier;
     }
 
+    public function getPackageName(): string
+    {
+        return $this->packageName;
+    }
+
+    public function getDatasetName(): string
+    {
+        return $this->datasetName;
+    }
+
     public function getVersion(): ?string
     {
         return $this->version;
     }
 
+    public function getComment(): ?string
+    {
+        return array_key_exists('comment', $this->options) ? $this->options['comment'] : null;
+    }
+
+    public function getLicense(): ?string
+    {
+        return array_key_exists('license', $this->options) ? $this->options['license'] : null;
+    }
+
     public function open(): ReaderInterface
     {
         $filename = $this->options['filename'];

src/DatasetManager.php

@@ -22,16 +22,46 @@ class DatasetManager
         }
     }
 
+    /**
+     * Return all the available datasets
+     * 
+     * @return Array<Dataset> The available datasets
+     */
     public function getAvailableDatasets(): array
     {
         return self::$datasets;
     }
 
+    /**
+     * Directly return a reader for a specific dataset.
+     * 
+     * @param string $identifier The dataset identifier
+     * @return ReaderInterface A reader for the data
+     * @throws InvalidDatasetException if the dataset can not be opened
+     * @throws UnknownDatasetException if the dataset does not exist
+     */
     public function openDataset(string $identifier): ReaderInterface
     {
         return $this->getDataset($identifier)->open();
     }
 
+
+    /**
+     * Return a Dataset object containing metadata and methods to retrieve
+     * a reader for the data in the set.
+     * 
+     * @param string $identifier The dataset identifier
+     * @throws UnknownDatasetException if the dataset does not exist
+     */
+    public function getDataset(string $identifier): Dataset
+    {
+        if (!array_key_exists($identifier, self::$datasets)) {
+            throw UnknownDatasetException::DatasetNotFound();
+        }
+
+        return self::$datasets[$identifier];
+    }
+
     /**
      * Find the vendor directory and try to locate all bundled datasets
      * 
@@ -67,8 +97,8 @@ class DatasetManager
      * 
      * 
      * 
-     * @param string The package name (org/package)
-     * @param string The full path to the package (..../vendor/org/package)
+     * @param string $package The package name (org/package)
+     * @param string $path The full path to the package (..../vendor/org/package)
      */
     private function scanPackageDatasets(string $package, string $path)
     {
@@ -84,10 +114,15 @@ class DatasetManager
         }
 
         $this->loadDatasets($json['datasets'], null, $package, $path);
-        //printf("found %d sets in %s\n", count($json['datasets']), $package);
 
     }
 
+    /**
+     * 
+     * 
+     * 
+     * 
+     */
     private function loadDatasets(array $datasets, ?string $prefix, string $package, string $path)
     {
         foreach ($datasets as $name=>$options) {
@@ -103,14 +138,21 @@ class DatasetManager
         }
     }
 
+    /**
+     * 
+     * 
+     * 
+     */
     private function determineVendorPath(): ?string
     {
-        if (defined("COMPOSER_VENDOR_PATH")) {
-            return COMPOSER_VENDOR_PATH;
+        $d = defined("NOCCYLABS_DATASET_TEST") ? (dirname(__DIR__)."/tests") : __DIR__;
+        while ($d != dirname($d)) {
+            if (file_exists($d."/autoload.php")) break;
+            $d = dirname($d);
         }
-        if (file_exists(__DIR__."/../../../autoload.php")) {
-            // we are installed as a composer package
-            return dirname(__DIR__, 3);
+
+        if (file_exists($d."/autoload.php")) {
+            return $d;
         }
         return null;
     }
@@ -127,17 +169,4 @@ class DatasetManager
         self::$datasets[$id] = $dataset;
     }
 
-
-
-    /**
-     * 
-     * 
-     * @throws InvalidDatasetException if the dataset can not be opened
-     * @throws UnknownDatasetExcception if the dataset does not exist
-     */
-    public function getDataset(string $identifier): Dataset
-    {
-
-        return self::$datasets[$identifier];
-    }
 }

src/InvalidDatasetException.php

@@ -0,0 +1,8 @@
+<?php
+
+namespace NoccyLabs\Dataset;
+
+use RuntimeException;
+
+class InvalidDatasetException extends RuntimeException
+{}

src/Readers/CsvReader.php

@@ -8,6 +8,8 @@ class CsvReader implements ReaderInterface
 {
     private array $files = [];
 
+    private array $options = [];
+
     private int $currentFile = 0;
 
     private ?int $loadedFile = null;
@@ -21,6 +23,7 @@ class CsvReader implements ReaderInterface
     public function __construct(string $filename, array $options)
     {
         $this->files = glob($filename);
+        $this->options = $options;
     }
 
     private function checkLoadedSlice()

src/Readers/JsonReader.php

@@ -8,6 +8,8 @@ class JsonReader implements ReaderInterface
 {
     private array $files = [];
 
+    private array $options = [];
+
     private int $currentFile = 0;
 
     private ?int $loadedFile = null;
@@ -21,6 +23,7 @@ class JsonReader implements ReaderInterface
     public function __construct(string $filename, array $options)
     {
         $this->files = glob($filename);
+        $this->options = $options;
     }
 
     private function checkLoadedSlice()

src/UnknownDatasetExeption.php

@@ -0,0 +1,13 @@
+<?php
+
+namespace NoccyLabs\Dataset;
+
+use RuntimeException;
+
+class UnknownDatasetException extends RuntimeException
+{
+    public static function DatasetNotFound(): self
+    {
+        return new UnknownDatasetException("The requested dataset is not available");
+    }
+}