php-dataset/src/DatasetManager.php

<?php

namespace NoccyLabs\Dataset;

use Iterator;

/**
 * DatasetManager is the central class of noccylabs/dataset.
 * 
 * @author Christopher Vagnetoft <cvagnetoft@gmail.com>
 * @copyright (c) 2022, NoccyLabs
 * @package noccylabs/dataset
 */
class DatasetManager
{
    private static array $packageVersions = [];

    private static array $datasets = [];

    public function __construct()
    {
        if (count(self::$datasets) == 0) {
            $this->scanForDatasets();
        }
    }

    /**
     * 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 Iterator 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): Iterator
    {
        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
     * 
     * 
     */
    private function scanForDatasets()
    {
        $root = $this->determineVendorPath();
        if (!$root) {
            // Skip loading if we couldn't determine the root
            return;
        }

        $glob = glob($root."/*/*/dataset.json");

        self::$packageVersions = [];
        $fn = realpath($root."/composer/installed.php");
        if (file_exists($fn)) {
            $versions = include $fn;
            foreach ($versions['versions'] as $name=>$version) {
                self::$packageVersions[$name] = $version['version'];
            }
        }

        foreach ($glob as $match) {
            $path = dirname($match);
            $package = basename(dirname($path))."/".basename($path);
            $this->scanPackageDatasets($package, $path);
        }
    }

    /**
     * 
     * 
     * 
     * @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)
    {
        $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);

    }

    /**
     * 
     * 
     * 
     * 
     */
    private function loadDatasets(array $datasets, ?string $prefix, string $package, string $path)
    {
        foreach ($datasets as $name=>$options) {
            if (!array_key_exists('filename', $options)) {
                $this->loadDatasets($options, ltrim($prefix . "." . $name, "."), $package, $path);
                return;
            }
            $options['filename'] = $path . "/" . $options['filename'];
            $pn = sprintf("%s#%s", $package, ltrim($prefix.".".$name,"."));
            $pv = self::$packageVersions[$package]??null;
            $ds = new Dataset($pn, $options, $pv);
            $this->registerDataset($ds);
        }
    }

    /**
     * 
     * 
     * 
     */
    private function determineVendorPath(): ?string
    {
        $d = defined("NOCCYLABS_DATASET_TEST") ? (dirname(__DIR__)."/tests") : __DIR__;
        while ($d != dirname($d)) {
            if (file_exists($d."/autoload.php")) break;
            $d = dirname($d);
        }

        if (file_exists($d."/autoload.php")) {
            return $d;
        }
        return null;
    }

    /**
     * Register a dataset
     * 
     * 
     */
    public function registerDataset(Dataset $dataset)
    {
        $id = $dataset->getIdentifier();
        if (array_key_exists($id, self::$datasets)) {
            // Don't overwrite previously registered datasets. Investigate how
            // this can be handled better in the future.
            return;
        }

        self::$datasets[$id] = $dataset;
    }

}