[Data Liberation] Add WXR import CLI script

packages/playground/data-liberation/bin/import/blueprint-import-wxr.json

@@ -0,0 +1,18 @@
+{
+	"$schema": "../../../blueprints/public/blueprint-schema.json",
+	"constants": {
+		"WP_DEBUG": true,
+		"WP_DEBUG_LOG": true
+	},
+	"login": true,
+	"steps": [
+		{
+			"step": "activatePlugin",
+			"pluginPath": "data-liberation/plugin.php"
+		},
+		{
+			"step": "runPHP",
+			"code": "<?php require_once 'wordpress/wp-load.php';\n$upload_dir = wp_upload_dir();\nforeach ( wp_visit_file_tree( $upload_dir['basedir'] . '/import-wxr' ) as $event ) {\nforeach ( $event->files as $file ) {\nif ( $file->isFile() && pathinfo( $file->getPathname(), PATHINFO_EXTENSION ) === 'xml' ) {\ndata_liberation_import( $file->getPathname() );\n}\n}\n};"
+		}
+	]
+}

packages/playground/data-liberation/bin/import/import-wxr.sh

@@ -0,0 +1,48 @@
+#!/bin/bash
+#
+# A script that accepts a folder and imports all WXR files into a WordPress site
+#
+# Usage:
+#    ./import-wxr.sh <folder-name>
+#
+
+# Display help message
+show_help() {
+	echo "Usage: $0 [-h|--help] <folder-name>"
+	echo "Options:"
+	echo "  -h, --help Show this help message"
+}
+
+# Check if no arguments were provided. If so, display help message
+if [ $# -eq 0 ]; then
+	show_help
+	exit 1
+fi
+
+# Parse command line arguments. If an invalid argument is provided, display help message
+while [[ "$1" =~ ^- && ! "$1" == "--" ]]; do case $1 in
+  -h | --help )
+	show_help
+	exit 0
+	;;
+esac; shift; done
+if [[ "$1" == '--' ]]; then shift; fi
+
+# Check if filename is provided. If not, display error message.
+if [ -z "$1" ]; then
+	echo "Error: No folder provided"
+	show_help
+	exit 1
+fi
+
+# Check if the file exists
+if [ -d "$1" ]; then
+	bun ../../../cli/src/cli.ts \
+		server \
+		--mount=../../:/wordpress/wp-content/plugins/data-liberation \
+		--mount=$1:/wordpress/wp-content/uploads/import-wxr \
+		--blueprint=./blueprint-import-wxr.json
+else
+	echo "Error: Folder '$1' does not exist"
+	exit 1
+fi

packages/playground/data-liberation/bootstrap.php

@@ -52,6 +52,7 @@
 require_once __DIR__ . '/src/import/WP_File_Visitor_Event.php';
 require_once __DIR__ . '/src/import/WP_Imported_Entity.php';
 require_once __DIR__ . '/src/import/WP_Attachment_Downloader.php';
+require_once __DIR__ . '/src/import/WP_Attachment_Downloader_Event.php';
 require_once __DIR__ . '/src/import/WP_Stream_Importer.php';
 require_once __DIR__ . '/src/import/WP_Markdown_Importer.php';
 

packages/playground/data-liberation/phpunit.xml

@@ -10,6 +10,7 @@
       <file>tests/URLParserWHATWGComplianceTests.php</file>
       <file>tests/WPXMLProcessorTests.php</file>
       <file>tests/UrldecodeNTests.php</file>
+      <file>tests/WPStreamImporterTests.php</file>
     </testsuite>
   </testsuites>
 </phpunit>

packages/playground/data-liberation/plugin.php

@@ -25,39 +25,23 @@
     return [];
 });
 
-/**
- * Development debug code to run the import manually.
- * @TODO: Remove this in favor of a CLI command.
- */
 add_action('init', function() {
-    return;
-    $wxr_path = __DIR__ . '/tests/fixtures/wxr-simple.xml';
-    $importer = WP_Stream_Importer::create_for_wxr_file(
-        $wxr_path
-    );
-    while($importer->next_step()) {
-        // ...
+    if ( defined( 'WP_CLI' ) && WP_CLI ) {
+        /**
+         * Import a WXR file.
+         *
+         * <file>
+         * : The WXR file to import.
+         */
+        $command = function ( $args, $assoc_args ) {
+            $file = $args[0];
+            data_liberation_import( $file );
+        };
+
+        // Register the WP-CLI import command.
+		// Example usage: wp data-liberation /path/to/file.xml
+        WP_CLI::add_command( 'data-liberation', $command );
     }
-    return;
-    $importer->next_step();
-    $paused_importer_state = $importer->get_reentrancy_cursor();
-
-    echo "\n\n";
-    echo "moving to importer2\n";
-    echo "\n\n";
-
-    $importer2 = WP_Stream_Importer::create_for_wxr_file(
-        $wxr_path,
-        array(),
-        $paused_importer_state
-    );
-    $importer2->next_step();
-    $importer2->next_step();
-    $importer2->next_step();
-    // $importer2->next_step();
-    // var_dump($importer2);
-
-    die("YAY");
 });
 
 // Register admin menu

packages/playground/data-liberation/project.json

@@ -50,6 +50,16 @@
 				],
 				"parallel": false
 			}
+		},
+		"test:wp-phpunit": {
+			"executor": "nx:run-commands",
+			"options": {
+				"cwd": "packages/playground/data-liberation",
+				"commands": [
+					"bun ../cli/src/cli.ts run-blueprint --quiet --mount=./:/wordpress/wp-content/plugins/data-liberation --blueprint=./tests/import/blueprint-import.json"
+				],
+				"parallel": false
+			}
 		}
 	}
 }

packages/playground/data-liberation/src/functions.php

@@ -191,3 +191,37 @@ function wp_visit_file_tree( $dir ) {
 		new SplFileInfo( $dir )
 	);
 }
+
+/**
+ * Import a WXR file. Used by the CLI.
+ *
+ * @param string $path The path to the WXR file.
+ * @return void
+ */
+function data_liberation_import( $path ): bool {
+	$importer = WP_Stream_Importer::create_for_wxr_file( $path );
+
+	if ( ! $importer ) {
+		return false;
+	}
+
+	$is_wp_cli = defined( 'WP_CLI' ) && WP_CLI;
+
+	if ( $is_wp_cli ) {
+		WP_CLI::line( "Importing from {$path}" );
+	}
+
+	while ( $importer->next_step() ) {
+		// Output the current stage if running in WP-CLI.
+		if ( $is_wp_cli ) {
+			$current_stage = $importer->get_current_stage();
+			WP_CLI::line( "Import: stage {$current_stage}" );
+		}
+	}
+
+	if ( $is_wp_cli ) {
+		WP_CLI::success( 'Import ended' );
+	}
+
+	return true;
+}

packages/playground/data-liberation/src/import/WP_Attachment_Downloader.php

@@ -26,7 +26,7 @@ public function has_pending_requests() {
 	public function enqueue_if_not_exists( $url, $output_path ) {
 		$this->enqueued_resource_id = null;
 
-		$output_path = $this->output_root . '/' . ltrim( $output_path, '/' );
+		$output_path = ltrim( $output_path, '/' );
 		if ( file_exists( $output_path ) ) {
 			// @TODO: Reconsider the return value. The enqueuing operation failed,
 			//        but overall already having a file seems like a success.
@@ -99,13 +99,14 @@ public function poll() {
 		if ( ! $this->client->await_next_event() ) {
 			return false;
 		}
-		$event   = $this->client->get_event();
-		$request = $this->client->get_request();
-		// The request object we get from the client may be a redirect.
-		// Let's keep referring to the original request.
-		$original_request_id = $request->original_request()->id;
 
-		while ( true ) {
+		do {
+			$event   = $this->client->get_event();
+			$request = $this->client->get_request();
+			// The request object we get from the client may be a redirect.
+			// Let's keep referring to the original request.
+			$original_request_id = $this->client->get_request()->original_request()->id;
+
 			switch ( $event ) {
 				case Client::EVENT_GOT_HEADERS:
 					if ( ! $request->is_redirected() ) {
@@ -129,7 +130,7 @@ public function poll() {
 						fclose( $this->fps[ $original_request_id ] );
 					}
 					if ( isset( $this->output_paths[ $original_request_id ] ) ) {
-						$partial_file = $this->output_root . '/' . $this->output_paths[ $original_request_id ] . '.partial';
+						$partial_file = $this->output_paths[ $original_request_id ] . '.partial';
 						if ( file_exists( $partial_file ) ) {
 							unlink( $partial_file );
 						}
@@ -162,7 +163,7 @@ public function poll() {
 					}
 					break;
 			}
-		}
+		} while ( $this->client->await_next_event() );
 
 		return true;
 	}

packages/playground/data-liberation/src/import/WP_Stream_Importer.php

@@ -185,6 +185,15 @@ public function next_step() {
 		}
 	}
 
+	/**
+	 * Get the current stage.
+	 *
+	 * @return string
+	 */
+	public function get_current_stage() {
+		return $this->stage;
+	}
+
 	/**
 	 * Advance the cursor to the oldest finished download. For example:
 	 *
@@ -400,7 +409,7 @@ private function enqueue_attachment_download( string $raw_url, $context_path = n
 
 		$enqueued = $this->downloader->enqueue_if_not_exists( $url, $output_path );
 		if ( $enqueued ) {
-			$resource_id   = $this->downloader->get_last_enqueued_resource_id();
+			$resource_id   = $this->downloader->get_enqueued_resource_id();
 			$entity_cursor = $this->entity_iterator->get_reentrancy_cursor();
 			$this->active_downloads[ $entity_cursor ][ $resource_id ] = true;
 		}

packages/playground/data-liberation/tests/WPStreamImporterTests.php

@@ -0,0 +1,23 @@
+<?php
+
+use PHPUnit\Framework\TestCase;
+
+/**
+ * Tests for the WPStreamImporter class.
+ */
+class WPStreamImporterTests extends TestCase {
+
+	protected function setUp(): void {
+		parent::setUp();
+
+		if ( ! isset( $_SERVER['SERVER_SOFTWARE'] ) || $_SERVER['SERVER_SOFTWARE'] !== 'PHP.wasm' ) {
+			$this->markTestSkipped( 'Test only runs in Playground' );
+		}
+	}
+
+	public function test_import_simple_wxr() {
+		$import = data_liberation_import( __DIR__ . '/wxr/small-export.xml' );
+
+		$this->assertTrue( $import );
+	}
+}

packages/playground/data-liberation/tests/import/blueprint-import.json

@@ -1,15 +1,14 @@
 {
 	"$schema": "../../../blueprints/public/blueprint-schema.json",
-	"constants": {
-		"WP_DEBUG": true,
-		"WP_DEBUG_DISPLAY": true,
-		"WP_DEBUG_LOG": true
-	},
 	"login": true,
 	"steps": [
 		{
 			"step": "activatePlugin",
 			"pluginPath": "data-liberation/plugin.php"
+		},
+		{
+			"step": "runPHP",
+			"code": "<?php require_once 'wordpress/wp-load.php'; $base = '/wordpress/wp-content/plugins/data-liberation/';\nrequire $base . 'vendor/autoload.php';\ntry {\n$arguments = [\n'--stderr',\n'--configuration', $base . 'phpunit.xml'\n];\n$res = (new PHPUnit\\TextUI\\Application())->run($arguments);\nif ( $res !== 0 ) {\ntrigger_error('PHPUnit failed', E_USER_ERROR);\n}\n} catch (Throwable $e) {\ntrigger_error('PHPUnit failed: ' . $e->getMessage(), E_USER_ERROR);\n};"
 		}
 	]
 }

packages/playground/data-liberation/tests/import/run.sh

@@ -4,4 +4,4 @@ bun ../../../cli/src/cli.ts \
     server \
     --mount=../../:/wordpress/wp-content/plugins/data-liberation \
     --mount=../../../../docs:/wordpress/wp-content/docs \
-    --blueprint=/Users/cloudnik/www/Automattic/core/plugins/playground/packages/playground/data-liberation/tests/import/blueprint-import.json
+    --blueprint=./blueprint-import.json