[Blueprints] Import WXRs via the DataLiberation importer (#127)

Author: adamziel

.gitignore

@@ -19,6 +19,7 @@ new-wp
 test.txt
 secrets.php
 testing-grounds.php
+test-script.php
 plugins/git-repo/git-test-server-data
 *.secret.json
 plugins/static-files-editor/build
@@ -30,4 +31,4 @@ components/DataLiberation/Tests/test-output-html/
 components/DataLiberation/Tests/test-output-md
 blueprint-dev.json
 examples/create-wp-site/data-liberation.zip
-untracked
+untracked

bin/build-libraries-phar.sh

@@ -20,7 +20,7 @@ cd $PROJECT_DIR
 mkdir -p $BUILD_DIR
 rm $DIST_DIR/wordpress-libraries.* > /dev/null 2>&1 || true
 export BOX_BASE_PATH=$(type -a box | grep -v 'alias' | awk '{print $3}')
-php $BUILD_DIR/box.php compile -d $PROJECT_DIR -c $PROJECT_DIR/phar-box.json
+php $BUILD_DIR/box.php compile -d $PROJECT_DIR -c $PROJECT_DIR/phar-libraries.json
 php -d 'phar.readonly=0' $BUILD_DIR/truncate-composer-checks.php $DIST_DIR/wordpress-libraries.phar
 cd $DIST_DIR
 php $BUILD_DIR/smoke-test.php

bin/build-phar/smoke-test.php

@@ -1,6 +1,6 @@
 <?php
 
-require_once __DIR__ . '/../../dist/wordpress-libraries.phar';
+require_once __DIR__ . '/../../dist/php-toolkit.phar';
 
 /**
  * None of this will actually try to parse a file or import
@@ -9,7 +9,9 @@
  */
 $c = WordPress\DataLiberation\Importer\StreamImporter::create_for_wxr_file(__DIR__ . '/nosuchfile.xml', [
     'uploads_path' => __DIR__ . '/uploads',
-    'new_site_url' => 'https://smoke-test.org'
+    'new_site_url' => 'https://smoke-test.org',
+	'new_site_content_root_url' => 'https://smoke-test.org',
+	'new_media_root_url' => 'https://smoke-test.org',
 ]);
 
 WordPress\DataLiberation\URL\WPURL::parse('https://example.com');

components/Blueprints/DataReference/DataReference.php

@@ -11,13 +11,20 @@ class DataReference {
 	 * @var int
 	 */
 	public $id;
+
+	/**
+	 * @var array
+	 */
+	public $original_definition;
+
 	/**
 	 * @var int
 	 */
 	private static $instanceCounter = 0;
 
-	public function __construct() {
+	public function __construct($original_definition = null) {
 		$this->id = self::$instanceCounter ++;
+		$this->original_definition = $original_definition;
 	}
 
 	static public function create( $reference, array $additional_reference_classes = [] ) {
@@ -29,11 +36,7 @@ static public function create( $reference, array $additional_reference_classes =
 		], $additional_reference_classes );
 		foreach ( $classes as $class ) {
 			if ( $class::is_valid( $reference ) ) {
-				if ( method_exists( $class, 'from_blueprint_data' ) ) {
-					return $class::from_blueprint_data( $reference );
-				} else {
-					return new $class( $reference );
-				}
+				return new $class( $reference );
 			}
 		}
 		throw new InvalidArgumentException(

components/Blueprints/DataReference/DataReferenceResolver.php

@@ -74,10 +74,17 @@ public function startEagerResolution( array $dataReferences, Tracker $dataResolu
 	 */
 	public function resolve( DataReference $reference ) {
 		// TODO: Comment this. Make semantics clearer.
-		if ( isset( $this->resolvedDataReferences[ $reference->id ] ) ) {
-			return $this->resolvedDataReferences[ $reference->id ];
+		if ( ! isset( $this->resolvedDataReferences[ $reference->id ] ) ) {
+			$this->resolvedDataReferences[ $reference->id ] = $this->resolve_uncached( $reference );
 		}
+		return $this->resolvedDataReferences[ $reference->id ];
+	}
 
+	// @TODO: Clean up the semantics of this class. Resolve() and separate resolve_uncached() seem confusing. There's
+	//        a bunch of implicit behaviors related to caching. Ideally we would either have a self-contained resolution
+	//        method, or co-locate the resolution logic with the data reference classes and only use this class for
+	//        caching.
+	public function resolve_uncached( DataReference $reference ) {
 		$progress_tracker = $this->subTrackers[ $reference->id ] ?? new Tracker();
 
 		if ( $reference instanceof WordPressOrgPlugin ) {

components/Blueprints/DataReference/ExecutionContextPath.php

@@ -20,7 +20,7 @@ class ExecutionContextPath extends DataReference {
 	 */
 	public function __construct( string $path ) {
 		$this->path = $path;
-		parent::__construct();
+		parent::__construct($path);
 	}
 
 	/**

components/Blueprints/DataReference/GitPath.php

@@ -25,20 +25,17 @@ class GitPath extends DataReference {
 
 	/**
 	 * Constructor.
-	 *
-	 * @param  string  $git_repository  The git repository URL.
-	 * @param  string|null  $ref  The git reference.
-	 * @param  string|null  $path  The path within the repository.
 	 */
 	public function __construct(
-		string $git_repository,
-		?string $ref = null,
-		?string $path = null
+		$definition
 	) {
-		$this->git_repository = $git_repository;
-		$this->ref            = $ref;
-		$this->path           = $path;
-		parent::__construct();
+		if ( ! isset( $definition['gitRepository'] ) ) {
+			throw new InvalidArgumentException( 'Invalid git path data' );
+		}
+		$this->git_repository = $definition['gitRepository'];
+		$this->ref            = $definition['ref'] ?? null;
+		$this->path           = $definition['path'] ?? null;
+		parent::__construct($definition);
 	}
 
 	/**
@@ -72,25 +69,6 @@ public function get_filename(): string {
 		return basename( $this->path );
 	}
 
-	/**
-	 * Create an instance from an array.
-	 *
-	 * @param  array  $data  The array data.
-	 *
-	 * @return self The created instance.
-	 */
-	public static function from_blueprint_data( array $data ): self {
-		if ( ! isset( $data['gitRepository'] ) ) {
-			throw new InvalidArgumentException( 'Invalid git path data' );
-		}
-
-		return new self(
-			$data['gitRepository'],
-			$data['ref'] ?? null,
-			$data['path'] ?? null
-		);
-	}
-
 	/**
 	 * Check if an array represents a valid git path.
 	 *

components/Blueprints/DataReference/InlineDirectory.php

@@ -25,13 +25,31 @@ class InlineDirectory extends DataReference {
 	/**
 	 * Constructor.
 	 *
-	 * @param  string  $name  The directory name.
-	 * @param  array  $children  The directory children.
+	 * @param  array  $data  The blueprint data array.
 	 */
-	public function __construct( string $name, array $children ) {
-		$this->name     = $name;
+	public function __construct( array $data ) {
+		if ( ! isset( $data['directoryName'] ) || ! isset( $data['files'] ) || ! is_array( $data['files'] ) ) {
+			throw new InvalidArgumentException( 'Invalid inline directory data' );
+		}
+
+		$this->name = $data['directoryName'];
+		
+		$children = [];
+		foreach ( $data['files'] as $fileName => $child ) {
+			if ( is_string( $child ) ) {
+				$children[$fileName] = new InlineFile( [
+					'filename' => $fileName,
+					'content' => $child
+				] );
+			} elseif ( self::is_valid( $child ) ) {
+				$children[$fileName] = new self( $child );
+			} else {
+				throw new InvalidArgumentException( 'Invalid inline directory child' );
+			}
+		}
+		
 		$this->children = $children;
-		parent::__construct();
+		parent::__construct( $data );
 	}
 
 	/**
@@ -81,31 +99,6 @@ public function as_directory(): Directory {
 		return new Directory( $this->as_filesystem(), $this->get_name() );
 	}
 
-	/**
-	 * Create an instance from an array.
-	 *
-	 * @param  array  $data  The array data.
-	 *
-	 * @return self The created instance.
-	 */
-	public static function from_blueprint_data( array $data ): self {
-		if ( ! isset( $data['directoryName'] ) || ! isset( $data['files'] ) || ! is_array( $data['files'] ) ) {
-			throw new InvalidArgumentException( 'Invalid inline directory data' );
-		}
-
-		$children = [];
-		foreach ( $data['files'] as $fileName => $child ) {
-			if ( is_string( $child ) ) {
-				$children[$fileName] = new InlineFile( $fileName, $child );
-			} elseif ( self::is_valid( $child ) ) {
-				$children[$fileName] = self::from_blueprint_data( $child );
-			} else {
-				throw new InvalidArgumentException( 'Invalid inline directory child' );
-			}
-		}
-
-		return new self( $data['directoryName'], $children );
-	}
 
 	/**
 	 * Check if an array represents a valid inline directory.

components/Blueprints/DataReference/InlineFile.php

@@ -24,10 +24,14 @@ class InlineFile extends DataReference {
 	 * @param  string  $filename  The filename.
 	 * @param  string  $content  The content.
 	 */
-	public function __construct( string $filename, string $content ) {
-		$this->filename = $filename;
-		$this->content  = $content;
-		parent::__construct();
+	public function __construct( array $definition ) {
+		if ( ! isset( $definition['filename'] ) || ! isset( $definition['content'] ) ) {
+			throw new InvalidArgumentException( 'Invalid inline file definition' );
+		}
+
+		$this->filename = $definition['filename'];
+		$this->content  = $definition['content'];
+		parent::__construct($definition);
 	}
 
 	/**
@@ -48,21 +52,6 @@ public function get_content(): string {
 		return $this->content;
 	}
 
-	/**
-	 * Create an instance from an array.
-	 *
-	 * @param  array  $data  The array data.
-	 *
-	 * @return self The created instance.
-	 */
-	public static function from_blueprint_data( array $data ): self {
-		if ( ! isset( $data['filename'] ) || ! isset( $data['content'] ) ) {
-			throw new InvalidArgumentException( 'Invalid inline file data' );
-		}
-
-		return new self( $data['filename'], $data['content'] );
-	}
-
 	/**
 	 * Check if an array represents a valid inline file.
 	 *

components/Blueprints/DataReference/URLReference.php

@@ -20,7 +20,7 @@ class URLReference extends DataReference {
 	 */
 	public function __construct( string $url ) {
 		$this->url = $url;
-		parent::__construct();
+		parent::__construct($url);
 	}
 
 	/**