sc2_datasets.torch.datasets.sc2_dataset
=======================================

.. py:module:: sc2_datasets.torch.datasets.sc2_dataset


Classes
-------

.. autoapisummary::

   sc2_datasets.torch.datasets.sc2_dataset.SC2Dataset


Module Contents
---------------

.. py:class:: SC2Dataset(names_urls: list[sc2_datasets.available_replaypacks.DatasetProperties], unpack_dir: pathlib.Path | str = Path('./data/unpack').resolve(), download_dir: pathlib.Path | str = Path('./data/download').resolve(), download: bool = True, unpack_n_workers: int = 16, transform: Callable | None = None, validator: Callable | None = None)

   Bases: :py:obj:`torch.utils.data.Dataset`


   Inherits from PyTorch Dataset and ensures that the dataset for SC2EGSet is downloaded.

   :param names_urls: Specifies the URL of the dataset which will be used to download the files.
   :type names_urls: list[DatasetProperties]
   :param unpack_dir: Specifies the path of a directory where the dataset files will be unpacked.
   :type unpack_dir: Path | str
   :param download_dir: Specifies the path of a directory where the dataset files will be downloaded.
   :type download_dir: Path | str
   :param unpack_n_workers: Specifies the number of workers that will be used for unpacking the archive, defaults to 16.
   :type unpack_n_workers: int, optional
   :param transform: PyTorch transform function that takes SC2ReplayData and returns something.
   :type transform: Func[SC2ReplayData, T]
   :param validator: Specifies the validation option for fetched data, defaults to None.
   :type validator: Callable | None, optional


   .. py:attribute:: transform
      :value: None



   .. py:attribute:: download_dir


   .. py:attribute:: unpack_dir


   .. py:attribute:: names_urls


   .. py:attribute:: download
      :value: True



   .. py:attribute:: unpack_n_workers
      :value: 16



   .. py:attribute:: validator
      :value: None



   .. py:attribute:: skip_files
      :type:  dict[str, set[str]]


   .. py:attribute:: len
      :value: 0



   .. py:method:: ensure_downloaded()

      Ensures that the dataset was downloaded before accessing the __len__ or __getitem__ methods.



   .. py:method:: __len__() -> int

      Returns the number of items that are within the dataset



   .. py:method:: __getitem__(index: Any) -> tuple[Any, Any] | sc2_datasets.replay_data.sc2_replay_data.SC2ReplayData

      Exposes logic of getting a single parsed item by using dataset[index].

      :param index: Specifies the index of an item that should be retrieved.
      :type index: Any

      :raises IndexError: To support negative indexing, if the index is less than zero twice,            IndexError is raised.
      :raises IndexError: If the index is greater than length of the dataset, IndexError is raised.

      :returns: Returns a parsed SC2ReplayData from an underlying SC2ReplaypackDataset,
                or a result of a transform that was passed to the dataset.
      :rtype: tuple[Any, Any] | SC2ReplayData