add list option

Author: johnnygreco

src/data_designer/cli/commands/download.py

@@ -24,13 +24,18 @@ def personas_command(
         "--dry-run",
         help="Show what would be downloaded without actually downloading",
     ),
+    list_available: bool = typer.Option(
+        False,
+        "--list",
+        help="List available persona datasets and their sizes",
+    ),
 ) -> None:
-    """Download persona datasets for synthetic data generation.
-
-    Persona datasets contain diverse character profiles that can be used
-    to generate contextually rich synthetic data.
+    """Download Nemotron-Personas datasets for synthetic data generation.
 
     Examples:
+        # List available datasets
+        data-designer download personas --list
+
         # Interactive selection
         data-designer download personas
 
@@ -44,4 +49,8 @@ def personas_command(
         data-designer download personas --all --dry-run
     """
     controller = DownloadController(DATA_DESIGNER_HOME)
-    controller.run_personas(locales=locales, all_locales=all_locales, dry_run=dry_run)
+
+    if list_available:
+        controller.list_personas()
+    else:
+        controller.run_personas(locales=locales, all_locales=all_locales, dry_run=dry_run)

src/data_designer/cli/controllers/download_controller.py

@@ -28,6 +28,25 @@ def __init__(self, config_dir: Path):
         self.config_dir = config_dir
         self.service = DownloadService(config_dir)
 
+    def list_personas(self) -> None:
+        """List available persona datasets and their sizes."""
+        print_header("Available Nemotron-Persona Datasets")
+        console.print()
+
+        available_locales = self.service.get_available_locales()
+
+        print_text("📦 Available locales:")
+        console.print()
+
+        for locale in available_locales.keys():
+            size = DATASET_SIZES[locale]
+            already_downloaded = self.service.is_locale_downloaded(locale)
+            status = " (downloaded)" if already_downloaded else ""
+            print_text(f"  • {locale}: {size}{status}")
+
+        console.print()
+        print_info(f"Total: {len(available_locales)} datasets available")
+
     def run_personas(self, locales: list[str] | None, all_locales: bool, dry_run: bool = False) -> None:
         """Main entry point for persona dataset downloads.
 

tests/cli/commands/test_download_command.py

@@ -14,7 +14,7 @@ def test_personas_command_interactive_mode(mock_download_controller: MagicMock)
     mock_controller_instance = MagicMock(spec=DownloadController)
     mock_download_controller.return_value = mock_controller_instance
 
-    personas_command(locales=None, all_locales=False, dry_run=False)
+    personas_command(locales=None, all_locales=False, dry_run=False, list_available=False)
 
     mock_download_controller.assert_called_once_with(DATA_DESIGNER_HOME)
     mock_controller_instance.run_personas.assert_called_once_with(locales=None, all_locales=False, dry_run=False)
@@ -26,7 +26,7 @@ def test_personas_command_with_specific_locales(mock_download_controller: MagicM
     mock_controller_instance = MagicMock(spec=DownloadController)
     mock_download_controller.return_value = mock_controller_instance
 
-    personas_command(locales=["en_US", "ja_JP"], all_locales=False, dry_run=False)
+    personas_command(locales=["en_US", "ja_JP"], all_locales=False, dry_run=False, list_available=False)
 
     mock_download_controller.assert_called_once_with(DATA_DESIGNER_HOME)
     mock_controller_instance.run_personas.assert_called_once_with(
@@ -40,7 +40,7 @@ def test_personas_command_with_all_flag(mock_download_controller: MagicMock) ->
     mock_controller_instance = MagicMock(spec=DownloadController)
     mock_download_controller.return_value = mock_controller_instance
 
-    personas_command(locales=None, all_locales=True, dry_run=False)
+    personas_command(locales=None, all_locales=True, dry_run=False, list_available=False)
 
     mock_download_controller.assert_called_once_with(DATA_DESIGNER_HOME)
     mock_controller_instance.run_personas.assert_called_once_with(locales=None, all_locales=True, dry_run=False)
@@ -52,7 +52,20 @@ def test_personas_command_with_dry_run_flag(mock_download_controller: MagicMock)
     mock_controller_instance = MagicMock(spec=DownloadController)
     mock_download_controller.return_value = mock_controller_instance
 
-    personas_command(locales=["en_US"], all_locales=False, dry_run=True)
+    personas_command(locales=["en_US"], all_locales=False, dry_run=True, list_available=False)
 
     mock_download_controller.assert_called_once_with(DATA_DESIGNER_HOME)
     mock_controller_instance.run_personas.assert_called_once_with(locales=["en_US"], all_locales=False, dry_run=True)
+
+
+@patch("data_designer.cli.commands.download.DownloadController")
+def test_personas_command_with_list_flag(mock_download_controller: MagicMock) -> None:
+    """Test personas_command with --list flag."""
+    mock_controller_instance = MagicMock(spec=DownloadController)
+    mock_download_controller.return_value = mock_controller_instance
+
+    personas_command(locales=None, all_locales=False, dry_run=False, list_available=True)
+
+    mock_download_controller.assert_called_once_with(DATA_DESIGNER_HOME)
+    mock_controller_instance.list_personas.assert_called_once()
+    mock_controller_instance.run_personas.assert_not_called()

tests/cli/controllers/test_download_controller.py

@@ -34,6 +34,18 @@ def test_init(tmp_path: Path) -> None:
     assert controller.service.config_dir == tmp_path
 
 
+def test_list_personas(controller: DownloadController) -> None:
+    """Test list_personas displays all available datasets."""
+    controller.list_personas()
+    # Method should complete without errors and print to console
+
+
+def test_list_personas_with_downloaded_datasets(controller_with_datasets: DownloadController) -> None:
+    """Test list_personas shows (downloaded) status for existing datasets."""
+    controller_with_datasets.list_personas()
+    # Method should complete without errors and show downloaded status
+
+
 @patch("data_designer.cli.controllers.download_controller.confirm_action", return_value=False)
 @patch("data_designer.cli.controllers.download_controller.select_multiple_with_arrows", return_value=["en_US"])
 @patch("data_designer.cli.controllers.download_controller.check_ngc_cli_with_instructions", return_value=True)