[Blueprints] Import WXRs via the DataLiberation importer (first stab)

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/Runner.php

@@ -618,10 +618,26 @@ private function createExecutionPlan(): array {
 		}
 
 		foreach ( $plan as $step ) {
-			// @TODO: Make sure this doesn't get included twice in the execution plan.
+			// @TODO: Make sure this doesn't get included twice in the execution plan,
+			//        e.g. if the Blueprint specified this step manually.
 			if ( $step instanceof ImportContentStep ) {
-				array_unshift( $plan, $this->createStepObject( 'installPlugin', [
-					'source' => $this->createDataReference( 'https://playground.wordpress.net/wordpress-importer.zip' ),
+				if($this->configuration->isRunningAsPhar()) {
+					throw new InvalidArgumentException( '@TODO: Importing content is not supported when running as phar.' );
+				} else {
+					$libraries_phar_path = __DIR__ . '/../../dist/php-toolkit.phar';
+					if(!file_exists($libraries_phar_path)) {
+						throw new InvalidArgumentException(
+							'In development, you must run `bash bin/build-libraries-phar.sh` to bundle importer libraries before importing content via a Blueprint. '.
+							'It generates a `dist/php-toolkit.phar` file bundling all the libraries required for importing content.'
+						);
+					}
+					$this->configuration->getLogger()->info( 'Loading importer libraries from ' . $libraries_phar_path );
+					$source = $this->createDataReference( new InlineFile( 'php-toolkit.phar', file_get_contents( $libraries_phar_path ) ) );
+				}
+				array_unshift( $plan, $this->createStepObject( 'writeFiles', [
+					'files' => [
+						'php-toolkit.phar' => $source,
+					],
 				] ) );
 				break;
 			}

components/Blueprints/RunnerConfiguration.php

@@ -235,4 +235,8 @@ public function isAllowedLocalFilesystemAccess(): bool {
 	public static function getPermissionCliFlag( string $permission ): string {
 		return $permission;
 	}
+
+	public function isRunningAsPhar(): bool {
+		return \Phar::running(false) !== '';
+	}
 }

components/Blueprints/Steps/ImportContentStep.php

@@ -44,7 +44,7 @@ public function run( Runtime $runtime, Tracker $progress ) {
 				$this->importWxr( $runtime, $content_definition );
 			} elseif ( $content_definition['type'] === 'posts' ) {
 				$progress[ $i ]->setCaption( 'Importing a post ' );
-				$this->importPosts( $runtime, $content_definition );
+				$this->importPosts( $runtime, $content_definition['source'] );
 			} else {
 				throw new RuntimeException( 'Unsupported content type: ' . $content_definition['type'] );
 			}
@@ -64,74 +64,75 @@ private function importWxr( Runtime $runtime, array $content_definition ): void
 			) );
 		}
 
+		// @TODO: Pass the data reference to the import script to enable streaming.
 		$wxrPath = $runtime->saveToTemporaryFile( $resolved );
+
+		// @TODO: Make it work when Blueprints are running as phar archive
+		$import_script_path = __DIR__ . '/scripts/import-content.php';
+		if ( ! file_exists( $import_script_path ) ) {
+			throw new BlueprintExecutionException( sprintf(
+				'Import script %s does not exist.',
+				$import_script_path
+			) );
+		}
+
+		$importer_script = file_get_contents( $import_script_path );
 		$runtime->evalPhpCodeInSubProcess(
 			<<<'PHP'
 <?php
-require_once getenv('DOCROOT') . '/wp-load.php';
-require_once getenv('DOCROOT') . '/wp-admin/includes/admin.php';
-
-kses_remove_filters();
-$admin_id = get_users(array('role' => 'Administrator') )[0]->ID;
-wp_set_current_user( $admin_id );
-
-wp_set_current_user( $admin_id );
-$importer = new WXR_Importer( array(
-'fetch_attachments' => true,
-// @TODO: Support custom author
-'default_author' => $admin_id
-) );
-$logger = new WP_Importer_Logger_CLI();
-$importer->set_logger( $logger );
-// Slashes from the imported content are lost if we don't call wp_slash here.
-add_action( 'wp_insert_post_data', function( $data ) {
-return wp_slash($data);
-});
-
-// Ensure that Site Editor templates are associated with the correct taxonomy.
-add_filter( 'wp_import_post_terms', function ( $terms, $post_id ) {
-foreach ( $terms as $post_term ) {
-if ( 'wp_theme' !== $term['taxonomy'] ) {continue;}
-$post_term = get_term_by('slug', $term['slug'], $term['taxonomy'] );
-if ( ! $post_term ) {
-$post_term = wp_insert_term(
-$term['slug'],
-$term['taxonomy']
-);
-$term_id = $post_term['term_id'];
-} else {
-$term_id = $post_term->term_id;
-}
-wp_set_object_terms( $post_id, $term_id, $term['taxonomy']) ;
-}
-return $terms;
-}, 10, 2 );
-$result = $importer->import( getenv('WXR_PATH') );
+// @TODO: Just call a function here, do not go through CLI arguments.
+// @TODO: Establish a communication channel between the main process and the subprocess
+//        to report progress and errors.
+// @TODO: Enforce chrooting of the imported static files.
+$_SERVER['argv'] = [
+	'import-wxr.php',
+	'wxr',
+	getenv('WXR_PATH'),
+	// @TODO: Support arbitrary media URLs to enable fetching assets during import.
+	// '--media-url',
+	// 'https://pd.w.org/'
+];
+?>
 PHP
+			. $importer_script
 			,
 			[
 				'WXR_PATH' => $wxrPath,
 			]
 		);
 	}
 
-	private function importPosts( Runtime $runtime, array $content_definition ): void {
-		$posts = $content_definition['source'];
-		if ( ! is_array( $posts ) ) {
-			throw new RuntimeException( 'Invalid posts data.' );
+	private function importPosts( Runtime $runtime, $post ): void {
+		// @TODO: Use the Data Liberation importer here.
+		$resolved = $runtime->resolve( $post );
+		if ( ! $resolved instanceof File ) {
+			throw new BlueprintExecutionException( sprintf(
+				'Imported content reference must be a file, but %s was a Directory.',
+				$post->get_human_readable_name()
+			) );
 		}
 
 		$runtime->evalPhpCodeInSubProcess(
 			<<<'PHP'
 <?php
 require_once getenv('DOCROOT') . '/wp-load.php';
 foreach (json_decode(getenv('POSTS'), true) as $post) {
-wp_insert_post(wp_slash($post));
+	$result = wp_insert_post(wp_slash($post));
+	if (is_wp_error($result)) {
+		throw new Exception( $result->get_error_message() );
+	}
 }
 PHP
 			,
 			[
-				'POSTS' => json_encode( $posts ),
+				'POSTS' => json_encode( [
+					[
+						'post_title'   => 'Test Post',
+						'post_content' => $resolved->getStream()->consume_all(),
+						'post_status'  => 'publish',
+						'post_type'    => 'post',
+					],
+				] ),
 			]
 		);
 	}