From 0542c7e844c46c733f3e50ac30141db043fbaa95 Mon Sep 17 00:00:00 2001
From: Christopher Vagnetoft <cvagnetoft@gmail.com>
Date: Sat, 3 Sep 2022 14:11:20 +0200
Subject: [PATCH] Dataset scanning

---
 src/Dataset.php        |  7 ++++--
 src/DatasetManager.php | 52 ++++++++++++++++++++++++++++++++++++++++--
 2 files changed, 55 insertions(+), 4 deletions(-)

diff --git a/src/Dataset.php b/src/Dataset.php
index 315d63c..e740795 100644
--- a/src/Dataset.php
+++ b/src/Dataset.php
@@ -5,11 +5,14 @@ namespace NoccyLabs\Dataset;
 
 class Dataset
 {
-    protected string $identfier;
+    protected string $identifier;
+    
+    protected array $options;
 
-    public function __construct(string $identifier, string $filename, array $metadata)
+    public function __construct(string $identifier, array $options)
     {
         $this->identifier = $identifier;
+        $this->options = $options;
     }
 
     public function getIdentifier(): string
diff --git a/src/DatasetManager.php b/src/DatasetManager.php
index 0af5b62..f6de99d 100644
--- a/src/DatasetManager.php
+++ b/src/DatasetManager.php
@@ -22,6 +22,16 @@ class DatasetManager
         }
     }
 
+    public function getAvailableDatasets(): array
+    {
+        return self::$datasets;
+    }
+
+    /**
+     * Find the vendor directory and try to locate all bundled datasets
+     * 
+     * 
+     */
     private function scanForDatasets()
     {
         $root = $this->determineVendorPath();
@@ -33,9 +43,47 @@ class DatasetManager
         $glob = glob($root."/*/*/dataset.json");
 
         foreach ($glob as $match) {
-            $path = dirname($glob);
+            $path = dirname($match);
             $package = basename(dirname($path))."/".basename($path);
-            printf("found: %s\n", $package);
+            $this->scanPackageDatasets($package, $path);
+        }
+    }
+
+    /**
+     * 
+     * 
+     * 
+     * @param string The package name (org/package)
+     * @param string The full path to the package (..../vendor/org/package)
+     */
+    private function scanPackageDatasets(string $package, string $path)
+    {
+        $file = $path."/dataset.json";
+        $json = @json_decode(
+            @file_get_contents($file),
+            true
+        );
+
+        if (!$json) {
+            trigger_error("Unable to parse dataset.json in {$package}");
+            return;
+        }
+
+        $this->loadDatasets($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 $name=>$info) {
+            if (!array_key_exists('filename', $info)) {
+                $this->loadDatasets($info, ltrim($prefix . "." . $name, "."), $package, $path);
+                return;
+            }
+            $pn = sprintf("%s#%s", $package, ltrim($prefix.".".$name,"."));
+            $ds = new Dataset($pn, $info);
+            $this->registerDataset($ds);
         }
     }