Module cevast.dataset.manager_factory

This module contains DatasetManager factory implementation.

"""This module contains DatasetManager factory implementation."""

from typing import Union, Type
from .dataset import DatasetSource, DatasetInvalidError
from . import managers


# TODO move to submodule managers, import * a check for subclass there
class DatasetManagerFactory:
    """
    Factory class providing the specific DatasetManager class based on DatasetSource.

    .. IMPORTANT:: DatasetManager classes are registered automatically. To add a new
    specialized DatasetManager class, it must inherit DatasetManager and be placed
    in "cevast.dataset.managers" package.
    """

    __classes = {}

    @classmethod
    def __load_classes(cls):
        """
        Automatically initialize lookup dictionary with subclasses of DatasetManager class
        that are visible to the Python interpreter (obtained from `type.__subclasses__()`).
        """
        for manager_class in managers.DatasetManager.__subclasses__():
            if hasattr(manager_class, 'dataset_source'):
                cls.__classes[str(manager_class.dataset_source)] = manager_class

    @classmethod
    def get_manager(cls, dataset_source: Union[DatasetSource, str]) -> Type[managers.DatasetManager]:
        """Return a corresponding DatasetManager class based on `dataset_source`."""
        if not cls.__classes:
            cls.__load_classes()

        # Validate and init dataset manager
        if not DatasetSource.validate(dataset_source):
            raise DatasetInvalidError("Dataset source %s is not valid." % dataset_source)
        if str(dataset_source) not in cls.__classes:
            raise DatasetInvalidError("Dataset source %s has no manager." % dataset_source)

        return cls.__classes[str(dataset_source)]

Classes

class DatasetManagerFactory

Factory class providing the specific DatasetManager class based on DatasetSource.

Important: DatasetManager classes are registered automatically. To add a new

specialized DatasetManager class, it must inherit DatasetManager and be placed in "cevast.dataset.managers" package.

Expand source code

class DatasetManagerFactory:
    """
    Factory class providing the specific DatasetManager class based on DatasetSource.

    .. IMPORTANT:: DatasetManager classes are registered automatically. To add a new
    specialized DatasetManager class, it must inherit DatasetManager and be placed
    in "cevast.dataset.managers" package.
    """

    __classes = {}

    @classmethod
    def __load_classes(cls):
        """
        Automatically initialize lookup dictionary with subclasses of DatasetManager class
        that are visible to the Python interpreter (obtained from `type.__subclasses__()`).
        """
        for manager_class in managers.DatasetManager.__subclasses__():
            if hasattr(manager_class, 'dataset_source'):
                cls.__classes[str(manager_class.dataset_source)] = manager_class

    @classmethod
    def get_manager(cls, dataset_source: Union[DatasetSource, str]) -> Type[managers.DatasetManager]:
        """Return a corresponding DatasetManager class based on `dataset_source`."""
        if not cls.__classes:
            cls.__load_classes()

        # Validate and init dataset manager
        if not DatasetSource.validate(dataset_source):
            raise DatasetInvalidError("Dataset source %s is not valid." % dataset_source)
        if str(dataset_source) not in cls.__classes:
            raise DatasetInvalidError("Dataset source %s has no manager." % dataset_source)

        return cls.__classes[str(dataset_source)]

Static methods

def get_manager(dataset_source: Union[DatasetSource, str]) ‑> Type[DatasetManager]

Return a corresponding DatasetManager class based on dataset_source.

Expand source code

@classmethod
def get_manager(cls, dataset_source: Union[DatasetSource, str]) -> Type[managers.DatasetManager]:
    """Return a corresponding DatasetManager class based on `dataset_source`."""
    if not cls.__classes:
        cls.__load_classes()

    # Validate and init dataset manager
    if not DatasetSource.validate(dataset_source):
        raise DatasetInvalidError("Dataset source %s is not valid." % dataset_source)
    if str(dataset_source) not in cls.__classes:
        raise DatasetInvalidError("Dataset source %s has no manager." % dataset_source)

    return cls.__classes[str(dataset_source)]