3 small cleanup changes

darwin/cli_functions.py

@@ -1059,7 +1059,7 @@ def set_file_status(dataset_slug: str, status: str, files: List[str]) -> None:
     files: List[str]
         Names of the files we want to update.
     """
-    available_statuses = ["archived", "clear", "new", "restore-archived", "complete"]
+    available_statuses = ["archived", "new", "restore-archived", "complete"]
     if status not in available_statuses:
         _error(
             f"Invalid status '{status}', available statuses: {', '.join(available_statuses)}"
@@ -1075,8 +1075,6 @@ def set_file_status(dataset_slug: str, status: str, files: List[str]) -> None:
         )
         if status == "archived":
             dataset.archive(items)
-        elif status == "clear":
-            dataset.reset(items)
         elif status == "new":
             dataset.move_to_new(items)
         elif status == "restore-archived":

darwin/dataset/remote_dataset.py

@@ -467,68 +467,57 @@ def fetch_remote_files(
         """
 
     @abstractmethod
-    def archive(self, items: Iterator[DatasetItem]) -> None:
+    def archive(self, items: Iterable[DatasetItem]) -> None:
         """
         Archives (soft-deletion) the given ``DatasetItem``\\s belonging to this ``RemoteDataset``.
 
         Parameters
         ----------
-        items : Iterator[DatasetItem]
+        items : Iterable[DatasetItem]
             The ``DatasetItem``\\s to be archived.
         """
 
     @abstractmethod
-    def restore_archived(self, items: Iterator[DatasetItem]) -> None:
+    def restore_archived(self, items: Iterable[DatasetItem]) -> None:
         """
         Restores the archived ``DatasetItem``\\s that belong to this ``RemoteDataset``.
 
         Parameters
         ----------
-        items : Iterator[DatasetItem]
+        items : Iterable[DatasetItem]
             The ``DatasetItem``\\s to be restored.
         """
 
     @abstractmethod
-    def move_to_new(self, items: Iterator[DatasetItem]) -> None:
+    def move_to_new(self, items: Iterable[DatasetItem]) -> None:
         """
         Changes the given ``DatasetItem``\\s status to ``new``.
 
         Parameters
         ----------
-        items : Iterator[DatasetItem]
+        items : Iterable[DatasetItem]
             The ``DatasetItem``\\s whose status will change.
         """
 
     @abstractmethod
-    def reset(self, items: Iterator[DatasetItem]) -> None:
-        """
-        Resets the given ``DatasetItem``\\s.
-
-        Parameters
-        ----------
-        items : Iterator[DatasetItem]
-            The ``DatasetItem``\\s to be reset.
-        """
-
-    @abstractmethod
-    def complete(self, items: Iterator[DatasetItem]) -> None:
+    def complete(self, items: Iterable[DatasetItem]) -> None:
         """
         Completes the given ``DatasetItem``\\s.
 
         Parameters
         ----------
-        items : Iterator[DatasetItem]
+        items : Iterable[DatasetItem]
             The ``DatasetItem``\\s to be completed.
         """
 
     @abstractmethod
-    def delete_items(self, items: Iterator[DatasetItem]) -> None:
+    def delete_items(self, items: Iterable[DatasetItem]) -> None:
         """
         Deletes the given ``DatasetItem``\\s.
 
         Parameters
         ----------
-        items : Iterator[DatasetItem]
+        items : Iterable[DatasetItem]
             The ``DatasetItem``\\s to be deleted.
         """
 

darwin/dataset/remote_dataset_v2.py

@@ -10,6 +10,7 @@
     Sequence,
     Tuple,
     Union,
+    Iterable,
 )
 import numpy as np
 from pydantic import ValidationError
@@ -362,13 +363,13 @@ def fetch_remote_files(
             else:
                 return
 
-    def archive(self, items: Iterator[DatasetItem]) -> None:
+    def archive(self, items: Iterable[DatasetItem]) -> None:
         """
         Archives (soft-deletion) the given ``DatasetItem``\\s belonging to this ``RemoteDataset``.
 
         Parameters
         ----------
-        items : Iterator[DatasetItem]
+        items : Iterable[DatasetItem]
             The ``DatasetItem``\\s to be archived.
         """
         payload: Dict[str, Any] = {
@@ -379,13 +380,13 @@ def archive(self, items: Iterator[DatasetItem]) -> None:
         }
         self.client.api_v2.archive_items(payload, team_slug=self.team)
 
-    def restore_archived(self, items: Iterator[DatasetItem]) -> None:
+    def restore_archived(self, items: Iterable[DatasetItem]) -> None:
         """
         Restores the archived ``DatasetItem``\\s that belong to this ``RemoteDataset``.
 
         Parameters
         ----------
-        items : Iterator[DatasetItem]
+        items : Iterable[DatasetItem]
             The ``DatasetItem``\\s to be restored.
         """
         payload: Dict[str, Any] = {
@@ -396,13 +397,13 @@ def restore_archived(self, items: Iterator[DatasetItem]) -> None:
         }
         self.client.api_v2.restore_archived_items(payload, team_slug=self.team)
 
-    def move_to_new(self, items: Iterator[DatasetItem]) -> None:
+    def move_to_new(self, items: Iterable[DatasetItem]) -> None:
         """
         Changes the given ``DatasetItem``\\s status to ``new``.
 
         Parameters
         ----------
-        items : Iterator[DatasetItem]
+        items : Iterable[DatasetItem]
             The ``DatasetItem``\\s whose status will change.
         """
 
@@ -417,25 +418,13 @@ def move_to_new(self, items: Iterator[DatasetItem]) -> None:
             team_slug=self.team,
         )
 
-    def reset(self, items: Iterator[DatasetItem]) -> None:
-        """
-        Deprecated
-        Resets the  given ``DatasetItem``\\s.
-
-        Parameters
-        ----------
-        items : Iterator[DatasetItem]
-            The ``DatasetItem``\\s to be resetted.
-        """
-        raise ValueError("Reset is deprecated for version 2 datasets")
-
-    def complete(self, items: Iterator[DatasetItem]) -> None:
+    def complete(self, items: Iterable[DatasetItem]) -> None:
         """
         Completes the given ``DatasetItem``\\s.
 
         Parameters
         ----------
-        items : Iterator[DatasetItem]
+        items : Iterable[DatasetItem]
             The ``DatasetItem``\\s to be completed.
         """
         (workflow_id, stages) = self._fetch_stages("complete")
@@ -449,13 +438,13 @@ def complete(self, items: Iterator[DatasetItem]) -> None:
             team_slug=self.team,
         )
 
-    def delete_items(self, items: Iterator[DatasetItem]) -> None:
+    def delete_items(self, items: Iterable[DatasetItem]) -> None:
         """
         Deletes the given ``DatasetItem``\\s.
 
         Parameters
         ----------
-        items : Iterator[DatasetItem]
+        items : Iterable[DatasetItem]
             The ``DatasetItem``\\s to be deleted.
         """
         self.client.api_v2.delete_items(

docs/DEV.md

@@ -51,6 +51,40 @@ To run end-to-end tests locally, copy `e2e_tests/.env.example` to `.env` and pop
 pytest e2e_tests
 ```
 
+### Code Formatting and Linting
+The project uses two main tools for code quality:
+
+1. **Black** - The uncompromising code formatter
+   - Automatically formats Python code to a consistent style
+   - Run locally before committing:
+   ```
+   black .
+   ```
+   - CI will check formatting with `black --check`
+
+2. **Ruff** - An extremely fast Python linter
+   - Enforces code style and catches potential errors
+   - Run locally:
+   ```
+   ruff check .
+   ```
+
+Both tools are automatically run in CI/CD pipelines for all Python files changed in pull requests. The workflow will:
+- Check code formatting with Black
+- Run Ruff linting checks
+- Fail the build if any issues are found
+
+To ensure your code passes CI checks, you can run these tools locally before pushing:
+```bash
+# Format code
+black .
+
+# Run linter
+ruff check .
+```
+
+For VS Code users, it's recommended to enable format-on-save with Black and install the Ruff extension for real-time linting feedback.
+
 ## Useful Aliases
 Aliases can be helpful for testing and development. Add them to your shell configuration file .bashrc .zshrc etc for ease of use and development
 ```

tests/darwin/cli_functions_test.py

@@ -289,25 +289,6 @@ def test_calls_dataset_archive(
                     )
                     mock.assert_called_once_with(fetch_remote_files_mock.return_value)
 
-    def test_calls_dataset_clear(
-        self, dataset_identifier: str, remote_dataset: RemoteDataset
-    ):
-        with patch.object(
-            Client, "get_remote_dataset", return_value=remote_dataset
-        ) as get_remote_dataset_mock:
-            with patch.object(
-                RemoteDatasetV2, "fetch_remote_files"
-            ) as fetch_remote_files_mock:
-                with patch.object(RemoteDatasetV2, "reset") as mock:
-                    set_file_status(dataset_identifier, "clear", ["one.jpg", "two.jpg"])
-                    get_remote_dataset_mock.assert_called_once_with(
-                        dataset_identifier=dataset_identifier
-                    )
-                    fetch_remote_files_mock.assert_called_once_with(
-                        {"item_names": "one.jpg,two.jpg"}
-                    )
-                    mock.assert_called_once_with(fetch_remote_files_mock.return_value)
-
     def test_calls_dataset_new(
         self, dataset_identifier: str, remote_dataset: RemoteDataset
     ):