Add connection class and address other issues (5d2b22ac) · Commits · NDIP / Nova Packages / nova-galaxy

src/nova/galaxy/data_store.py

+7 −3

Original line number	Diff line number	Diff line
		@@ -3,7 +3,7 @@
		from typing import TYPE_CHECKING

		if TYPE_CHECKING:
		from .nova import Nova # Only imports for type checking
		from .nova import NovaConnection # Only imports for type checking


		class Datastore:
		@@ -12,7 +12,11 @@ class Datastore:
		The constructor is not intended for external use. Use nova.galaxy.Nova.create_data_store() instead.
		"""

		def __init__(self, name: str, nova_instance: "Nova", history_id: str):
		def __init__(self, name: str, nova_connection: "NovaConnection", history_id: str) -> None:
		self.name = name
		self.nova = nova_instance
		self.nova_connection = nova_connection
		self.history_id = history_id
		self.persist_store = False

		def persist(self) -> None:
		self.persist_store = True

src/nova/galaxy/dataset.py

+31 −7

Original line number	Diff line number	Diff line
		@@ -53,7 +53,11 @@ class AbstractData(ABC):
		raise NotImplementedError()

		@abstractmethod
		def download(self, local_path: str) -> None:
		def download(self, local_path: str) -> Any:
		raise NotImplementedError()

		@abstractmethod
		def get_content(self) -> Any:
		raise NotImplementedError()

		def cancel_upload(self) -> None:
		@@ -70,7 +74,7 @@ class Dataset(AbstractData):
		self.store: Datastore

		def upload(self, store: Datastore) -> None:
		galaxy_instance = store.nova.galaxy_instance
		galaxy_instance = store.nova_connection.galaxy_instance
		dataset_client = DatasetClient(galaxy_instance)
		history_id = galaxy_instance.histories.get_histories(name=store.name)[0]["id"]
		dataset_id = galaxy_instance.tools.upload_file(path=self.path, history_id=history_id)
		@@ -78,11 +82,19 @@ class Dataset(AbstractData):
		self.store = store
		dataset_client.wait_for_dataset(self.id)

		def download(self, local_path: str) -> None:
		def download(self, local_path: str) -> AbstractData:
		"""Downloads this dataset to the local path given."""
		if self.store and self.id:
		dataset_client = DatasetClient(self.store.nova.galaxy_instance)
		dataset_client = DatasetClient(self.store.nova_connection.galaxy_instance)
		dataset_client.download_dataset(self.id, use_default_filename=False, file_path=local_path)
		return self
		else:
		raise Exception("Dataset is not present in Galaxy.")

		def get_content(self) -> Any:
		if self.store and self.id:
		dataset_client = DatasetClient(self.store.nova_connection.galaxy_instance)
		return dataset_client.download_dataset(self.id, use_default_filename=False, file_path=None).decode("utf-8")
		else:
		raise Exception("Dataset is not present in Galaxy.")

		@@ -100,18 +112,30 @@ class DatasetCollection(AbstractData):
		"""Will need to handle this differently than single datasets."""
		raise NotImplementedError

		def download(self, local_path: str) -> None:
		def download(self, local_path: str) -> AbstractData:
		"""Downloads this dataset collection to the local path given."""
		if self.store and self.id:
		dataset_client = DatasetCollectionClient(self.store.nova.galaxy_instance)
		dataset_client = DatasetCollectionClient(self.store.nova_connection.galaxy_instance)
		dataset_client.download_dataset_collection(self.id, file_path=local_path)
		return self
		else:
		raise Exception("Dataset collection is not present in Galaxy.")

		def get_content(self) -> Any:
		if self.store and self.id:
		dataset_client = DatasetCollectionClient(self.store.nova_connection.galaxy_instance)
		info = dataset_client.show_dataset_collection(self.id)
		output = ""
		for element in info["elements"]:
		output += f"{element['element_identifier']}\n"
		return output
		else:
		raise Exception("Dataset collection is not present in Galaxy.")


		def upload_datasets(store: Datastore, datasets: Dict[str, AbstractData]) -> Dict[str, str]:
		"""Helper method to upload multiple datasets or collections in parallel."""
		galaxy_instance = store.nova.galaxy_instance
		galaxy_instance = store.nova_connection.galaxy_instance
		dataset_client = DatasetClient(galaxy_instance)
		history_id = galaxy_instance.histories.get_histories(name=store.name)[0]["id"]
		dataset_ids = {}

src/nova/galaxy/nova.py

+39 −15

Original line number	Diff line number	Diff line
		"""The NOVA class is responsible for managing interactions with a Galaxy server instance."""

		from typing import Optional
		from contextlib import contextmanager
		from typing import Generator, List, Optional

		from bioblend import galaxy

		@@ -20,6 +21,35 @@ class GalaxyConnectionError(Exception):
		super().__init__(self.message)


		class NovaConnection:
		"""Manages datastore for current connection.

		Should not be instantiated manually. Use Nova.connect() instead. Any stores created using the connection will
		be automatically purged after connection is closed, unless Datastore.persist() is called for that store.
		"""

		def __init__(self, galaxy_instance: galaxy.GalaxyInstance):
		self.galaxy_instance = galaxy_instance
		self.datastores: List[Datastore] = []

		def create_data_store(self, name: str) -> Datastore:
		"""Creates a datastore with the given name."""
		histories = self.galaxy_instance.histories.get_histories(name=name)
		if len(histories) > 0:
		store = Datastore(name, self, histories[0]["id"])
		self.datastores.append(store)
		return store
		history_id = self.galaxy_instance.histories.create_history(name=name)["id"]
		store = Datastore(name, self, history_id)
		self.datastores.append(store)
		return store

		def remove_data_store(self, store: Datastore) -> None:
		"""Permanently deletes the data store with the given name."""
		history = self.galaxy_instance.histories.get_histories(name=store.name)[0]["id"]
		self.galaxy_instance.histories.delete_history(history_id=history, purge=True)


		class Nova:
		"""
		Class to manage a Galaxy connection.
		@@ -46,7 +76,8 @@ class Nova:
		self.galaxy_api_key = galaxy_key
		self.galaxy_instance: galaxy.GalaxyInstance

		def connect(self) -> None:
		@contextmanager
		def connect(self) -> Generator:
		"""
		Connects to the Galaxy instance using the provided URL and API key.

		@@ -62,16 +93,9 @@ class Nova:
		raise ValueError("Galaxy URL must be a string")
		self.galaxy_instance = galaxy.GalaxyInstance(url=self.galaxy_url, key=self.galaxy_api_key)
		self.galaxy_instance.config.get_version()

		def create_data_store(self, name: str) -> Datastore:
		"""Creates a datastore with the given name."""
		histories = self.galaxy_instance.histories.get_histories(name=name)
		if len(histories) > 0:
		return Datastore(name, self, histories[0]["id"])
		history_id = self.galaxy_instance.histories.create_history(name=name)["id"]
		return Datastore(name, self, history_id)

		def remove_data_store(self, name: str) -> None:
		"""Permanently deletes the data store with the given name."""
		history = self.galaxy_instance.histories.get_histories(name=name)[0]["id"]
		self.galaxy_instance.histories.delete_history(history_id=history, purge=True)
		conn = NovaConnection(self.galaxy_instance)
		yield conn
		# Remove all data stores after execution
		for store in conn.datastores:
		if not store.persist_store:
		conn.remove_data_store(store)

src/nova/galaxy/outputs.py

+8 −2

Original line number	Diff line number	Diff line
		@@ -28,7 +28,13 @@ class Outputs:
		self.data.append(data)

		def get_dataset(self, name: str) -> AbstractData:
		try:
		return next(filter(lambda x: isinstance(x, Dataset) and x.name == name, self.data))
		except StopIteration as e:
		raise Exception(f"There is no dataset: {name}") from e

		def get_collection(self, name: str) -> AbstractData:
		try:
		return next(filter(lambda x: isinstance(x, DatasetCollection) and x.name == name, self.data))
		except StopIteration as e:
		raise Exception(f"There is no dataset collection: {name}") from e

src/nova/galaxy/tool.py

+1 −1

Original line number	Diff line number	Diff line
		@@ -35,7 +35,7 @@ class Tool(AbstractWork):
		def run(self, data_store: Datastore, params: Parameters) -> Outputs:
		"""Runs this tool in a blocking manner and returns a map of the output datasets and collections."""
		outputs = Outputs()
		galaxy_instance = data_store.nova.galaxy_instance
		galaxy_instance = data_store.nova_connection.galaxy_instance
		datasets_to_upload = {}

		# Set Tool Inputs