DatasetLoader

atmosphere.DatasetLoader(client)

Loads dataset records from ATProto.

This class fetches dataset index records and can create Dataset objects from them. Note that loading a dataset requires having the corresponding Python class for the sample type.

Examples

>>> client = AtmosphereClient()
>>> loader = DatasetLoader(client)
>>>
>>> # List available datasets
>>> datasets = loader.list()
>>> for ds in datasets:
...     print(ds["name"], ds["schemaRef"])
>>>
>>> # Get a specific dataset record
>>> record = loader.get("at://did:plc:abc/ac.foundation.dataset.record/xyz")

Methods

Name	Description
get	Fetch a dataset record by AT URI.
get_blob_urls	Get fetchable URLs for blob-stored dataset shards.
get_blobs	Get the blob references from a dataset record.
get_metadata	Get the metadata from a dataset record.
get_storage_type	Get the storage type of a dataset record.
get_urls	Get the WebDataset URLs from a dataset record.
list_all	List dataset records from a repository.
to_dataset	Create a Dataset object from an ATProto record.

get

atmosphere.DatasetLoader.get(uri)

Fetch a dataset record by AT URI.

Parameters

Name	Type	Description	Default
uri	str | AtUri	The AT URI of the dataset record.	required

Returns

Name	Type	Description
	dict	The dataset record as a dictionary.

Raises

Name	Type	Description
	ValueError	If the record is not a dataset record.

get_blob_urls

atmosphere.DatasetLoader.get_blob_urls(uri)

Get fetchable URLs for blob-stored dataset shards.

This resolves the PDS endpoint and constructs URLs that can be used to fetch the blob data directly.

Parameters

Name	Type	Description	Default
uri	str | AtUri	The AT URI of the dataset record.	required

Returns

Name	Type	Description
	list[str]	List of URLs for fetching the blob data.

Raises

Name	Type	Description
	ValueError	If storage type is not blobs or PDS cannot be resolved.

get_blobs

atmosphere.DatasetLoader.get_blobs(uri)

Get the blob references from a dataset record.

Parameters

Name	Type	Description	Default
uri	str | AtUri	The AT URI of the dataset record.	required

Returns

Name	Type	Description
	list[dict]	List of blob reference dicts with keys: $type, ref, mimeType, size.

Raises

Name	Type	Description
	ValueError	If the storage type is not blobs.

get_metadata

atmosphere.DatasetLoader.get_metadata(uri)

Get the metadata from a dataset record.

Parameters

Name	Type	Description	Default
uri	str | AtUri	The AT URI of the dataset record.	required

Returns

Name	Type	Description
	Optional[dict]	The metadata dictionary, or None if no metadata.

get_storage_type

atmosphere.DatasetLoader.get_storage_type(uri)

Get the storage type of a dataset record.

Parameters

Name	Type	Description	Default
uri	str | AtUri	The AT URI of the dataset record.	required

Returns

Name	Type	Description
	str	Either “external” or “blobs”.

Raises

Name	Type	Description
	ValueError	If storage type is unknown.

get_urls

atmosphere.DatasetLoader.get_urls(uri)

Get the WebDataset URLs from a dataset record.

Parameters

Name	Type	Description	Default
uri	str | AtUri	The AT URI of the dataset record.	required

Returns

Name	Type	Description
	list[str]	List of WebDataset URLs.

Raises

Name	Type	Description
	ValueError	If the storage type is not external URLs.

list_all

atmosphere.DatasetLoader.list_all(repo=None, limit=100)

List dataset records from a repository.

Parameters

Name	Type	Description	Default
repo	Optional[str]	The DID of the repository. Defaults to authenticated user.	None
limit	int	Maximum number of records to return.	100

Returns

Name	Type	Description
	list[dict]	List of dataset records.

to_dataset

atmosphere.DatasetLoader.to_dataset(uri, sample_type)

Create a Dataset object from an ATProto record.

This method creates a Dataset instance from a published record. You must provide the sample type class, which should match the schema referenced by the record.

Supports both external URL storage and ATProto blob storage.

Parameters

Name	Type	Description	Default
uri	str | AtUri	The AT URI of the dataset record.	required
sample_type	Type[ST]	The Python class for the sample type.	required

Returns

Name	Type	Description
	Dataset[ST]	A Dataset instance configured from the record.

Raises

Name	Type	Description
	ValueError	If no storage URLs can be resolved.

Examples

>>> loader = DatasetLoader(client)
>>> dataset = loader.to_dataset(uri, MySampleType)
>>> for batch in dataset.shuffled(batch_size=32):
...     process(batch)