feat(dataset): default dataset add action in configuration (#3398)

renku/core/dataset/dataset_add.py

@@ -371,7 +371,7 @@ def copy_file(file: DatasetAddMetadata, dataset: Dataset, storage: Optional[ISto
     if not file.has_action:
         return []
 
-    # NOTE: If file is in a sub-directory of a dataset's remote storage URI, only update the metadata
+    # NOTE: If file is in a subdirectory of a dataset's remote storage URI, only update the metadata
     if file.from_cloud_storage:
         if dataset.storage and is_uri_subfolder(resolve_uri(dataset.storage), file.url):
             file.action = DatasetAddAction.METADATA_ONLY

renku/core/dataset/providers/local.py

@@ -22,6 +22,7 @@
 from typing import TYPE_CHECKING, List, Optional
 
 from renku.core import errors
+from renku.core.config import get_value
 from renku.core.dataset.providers.api import (
     AddProviderInterface,
     ExporterApi,
@@ -114,14 +115,32 @@ def get_metadata(
         if flags > 1:
             raise errors.ParameterError("--move, --copy and --link are mutually exclusive.")
 
-        prompt_action = True if flags == 0 else False
+        prompt_action = False
 
         if move:
             default_action = DatasetAddAction.MOVE
         elif link:
             default_action = DatasetAddAction.SYMLINK
-        else:
+        elif copy:
             default_action = DatasetAddAction.COPY
+        else:
+            prompt_action = True
+            action = get_value("renku", "default_dataset_add_action")
+            if action:
+                prompt_action = False
+                if action.lower() == "copy":
+                    default_action = DatasetAddAction.COPY
+                elif action.lower() == "move":
+                    default_action = DatasetAddAction.MOVE
+                elif action.lower() == "link":
+                    default_action = DatasetAddAction.SYMLINK
+                else:
+                    raise errors.ParameterError(
+                        f"Invalid default action for adding to datasets in Renku config: '{action}'. "
+                        "Valid values are 'copy', 'link', and 'move'."
+                    )
+            else:
+                default_action = DatasetAddAction.COPY
 
         ends_with_slash = False
         u = urllib.parse.urlparse(uri)
@@ -180,6 +199,14 @@ def get_file_metadata(src: Path) -> DatasetAddMetadata:
 
         destination_root = get_destination_root()
 
+        if not is_subpath(source_root, project_context.path):
+            if link:
+                raise errors.ParameterError(f"Cannot use '--link' for files outside of project: '{uri}'")
+            if default_action == DatasetAddAction.SYMLINK:
+                # NOTE: A default action of 'link' cannot be used for external files
+                action = DatasetAddAction.COPY
+                prompt_action = True
+
         results = []
         if source_root.is_dir():
             for file in source_root.rglob("*"):
@@ -195,8 +222,9 @@ def get_file_metadata(src: Path) -> DatasetAddMetadata:
 
         if not force and prompt_action:
             communication.confirm(
-                f"The following files will be copied to {destination.relative_to(project_context.path)} "
-                "(use '--move' or '--link' to move or symlink them instead, '--copy' to not show this warning):\n\t"
+                f"The following files will be copied to {destination.relative_to(project_context.path)}:\n\t"
+                "(use '--move' or '--link' to move or symlink them instead, '--copy' to not show this warning).\n\t"
+                "(run 'renku config set renku.default_dataset_add_action copy' to make copy the default action).\n\t"
                 + "\n\t".join(str(e.source) for e in results)
                 + "\nProceed?",
                 abort=True,

renku/ui/cli/config.py

@@ -93,6 +93,10 @@
 | ``dataverse.server_url``       | URL for the Dataverse API server    | ``None``  |
 |                                | to use                              |           |
 +--------------------------------+-------------------------------------+-----------+
+| ``default_dataset_add_action`` | Default action when adding files to | ``None``  |
+|                                | datasets. Can be either ``copy`` or |           |
+|                                | ``move``.                           |           |
++--------------------------------+-------------------------------------+-----------+
 | ``lfs_threshold``              | Threshold file size below which     | ``100kb`` |
 |                                | files are not added to git LFS      |           |
 +--------------------------------+-------------------------------------+-----------+

tests/cli/test_datasets.py

@@ -669,6 +669,71 @@ def test_add_local_actions(runner, project, action, existing_paths, missing_path
         assert path.is_symlink()
 
 
+@pytest.mark.parametrize("action, source_exists_after", [("--copy", True), ("--move", False)])
+def test_add_non_local_actions(runner, project, directory_tree, action, source_exists_after):
+    """Test adding data outside the project with different actions."""
+    path = directory_tree / "file1"
+
+    result = runner.invoke(cli, ["dataset", "add", action, "--create", "local", path])
+
+    assert 0 == result.exit_code, format_result_exception(result)
+    assert source_exists_after == path.exists()
+    assert (project.path / "data" / "local" / "file1").exists()
+
+
+def test_add_non_local_link_action(runner, project, directory_tree):
+    """Test cannot add and link data outside the project."""
+    path = directory_tree / "file1"
+
+    result = runner.invoke(cli, ["dataset", "add", "--link", "--create", "local", path])
+
+    assert 2 == result.exit_code, format_result_exception(result)
+    assert "Cannot use '--link' for files outside of project:" in result.output
+
+
+@pytest.mark.parametrize("action, source_exists_after", [("copy", True), ("move", False)])
+@pytest.mark.serial
+def test_add_default_configured_actions(runner, project, directory_tree, action, source_exists_after):
+    """Test adding data with different actions set in Renku configuration file."""
+    path = directory_tree / "file1"
+    set_value("renku", "default_dataset_add_action", action, global_only=True)
+
+    result = runner.invoke(cli, ["dataset", "add", "--create", "local", path])
+
+    assert 0 == result.exit_code, format_result_exception(result)
+    assert "The following files will be copied to" not in result.output
+    assert path.exists() is source_exists_after
+    assert (project.path / "data" / "local" / "file1").exists()
+
+
+@pytest.mark.serial
+def test_add_default_configured_link(runner, project, directory_tree):
+    """Test adding data with default ``link`` action should prompt the user."""
+    path = directory_tree / "file1"
+    set_value("renku", "default_dataset_add_action", "link", global_only=True)
+
+    result = runner.invoke(cli, ["dataset", "add", "--create", "local", path], input="y\n")
+
+    assert 0 == result.exit_code, format_result_exception(result)
+    assert "The following files will be copied to" in result.output
+    assert path.exists()
+    assert (project.path / "data" / "local" / "file1").exists()
+    assert not (project.path / "data" / "local" / "file1").is_symlink()
+
+
+@pytest.mark.serial
+def test_add_default_configured_invalid_action(runner, project, directory_tree):
+    """Test adding data with an invalid actions set in Renku configuration file."""
+    path = directory_tree / "file1"
+    set_value("renku", "default_dataset_add_action", "invalid", global_only=True)
+
+    result = runner.invoke(cli, ["dataset", "add", "--create", "local", path])
+
+    assert 2 == result.exit_code, format_result_exception(result)
+    assert "Invalid default action for adding to datasets in Renku config: 'invalid'." in result.output
+    assert "Valid values are 'copy', 'link', and 'move'." in result.output
+
+
 def test_add_an_empty_directory(runner, project, directory_tree):
     """Test adding an empty directory to a dataset."""
     path = directory_tree / "empty-directory"