satorbis_kit.backends module

class satorbis_kit.backends.Backend

Bases: ABC

Abstract base class for compute backends (Dask, Sedona, etc.)

Follows SOLID principles: - Single Responsibility: Each backend handles one compute framework - Open/Closed: Open for extension via inheritance - Liskov Substitution: All backends are interchangeable - Interface Segregation: Clean, focused interface - Dependency Inversion: Depend on abstractions, not concretions

abstractmethod close()

Clean up resources.

abstractmethod compute(tasks: Iterable, show_progress: bool = True)

Execute delayed tasks and return results.

abstractmethod delayed(fn)

Mark a function as a delayed task for parallel execution.

class satorbis_kit.backends.BackendFactory

Bases: object

Factory for creating and managing compute backends.

Follows Factory Pattern + Singleton Pattern for managing backend registry.

create_dask_backend(client) → DaskBackend

Create a DaskBackend instance.

create_sedona_backend(spark_session) → SedonaBackend

Create a SedonaBackend instance.

get_backend_class(name: str) → type

Get a backend class by name.

Parameters:

name – Backend name

Returns:

Backend class

Raises:

ValueError – If backend name is unknown

classmethod get_instance() → BackendFactory

Get the singleton instance.

list_backends() → list[str]

List all registered backend names.

register_backend(name: str, backend_class: type) → None

Register a new backend type.

Parameters:

• name – Backend name

• backend_class – Backend class that subclasses Backend

Raises:

TypeError – If backend_class doesn’t subclass Backend

class satorbis_kit.backends.ComputeStrategy(*args, **kwargs)

Bases: Protocol

Protocol defining the compute strategy interface (Strategy Pattern)

close() → None

Clean up resources.

compute(tasks: Iterable, show_progress: bool = True) → Any

Execute delayed tasks and return results.

delayed(fn) → Any

Mark a function as a delayed task for parallel execution.

class satorbis_kit.backends.DaskBackend(client)

Bases: Backend

Dask-based backend for parallel computation.

close()

Close the Dask client.

compute(tasks: Iterable, show_progress: bool = True)

Execute Dask tasks with optional progress bar.

delayed(fn)

Create a Dask delayed task.

class satorbis_kit.backends.DaskConfig(scheduler_address: str | None = None, n_workers: int = 4, threads_per_worker: int = 1, processes: bool = True)

Bases: object

Configuration for Dask backend (Single Responsibility Principle).

n_workers: int = 4

processes: bool = True

scheduler_address: str | None = None

threads_per_worker: int = 1

class satorbis_kit.backends.SedonaBackend(spark_session)

Bases: Backend

Apache Sedona-based backend for geospatial processing.

close()

Close the Spark session.

compute(tasks: Iterable, show_progress: bool = True)

Execute Sedona tasks (to be implemented).

delayed(fn)

Create a Sedona task (to be implemented).

class satorbis_kit.backends.SedonaConfig(spark_master: str = 'local[*]', app_name: str = 'SatorbisKit')

Bases: object

Configuration for Sedona backend (Single Responsibility Principle).

app_name: str = 'SatorbisKit'

spark_master: str = 'local[*]'

satorbis_kit.backends.get_backend(name: str = 'dask', *, scheduler_address: str | None = None, n_workers: int = 4, threads_per_worker: int = 1, processes: bool = True, spark_master: str | None = None, **kwargs) → Iterator[Backend]

Factory/context manager for creating and managing compute backends.

Uses Strategy Pattern to select appropriate backend implementation. Uses Dependency Injection for configuration.

Parameters:

• name – Backend name (‘dask’ or ‘sedona’)

• scheduler_address – Dask scheduler address (for ‘dask’ backend)

• n_workers – Number of workers (for ‘dask’ backend)

• threads_per_worker – Threads per worker (for ‘dask’ backend)

• processes – Use processes instead of threads (for ‘dask’ backend)

• spark_master – Spark master URL (for ‘sedona’ backend)

• **kwargs – Additional backend-specific parameters

Yields:

Backend instance

Raises:

• ValueError – If unknown backend name

• NotImplementedError – If backend not yet implemented