| """STRATA data acquisition subpackage. | |
| The registry and manifest data models depend only on the standard library plus | |
| PyYAML. The network download dependencies (``huggingface_hub``, ``requests``) are | |
| imported lazily inside :mod:`strata.data.acquire`, so registry/manifest logic can | |
| be exercised (and unit-tested) without network libraries installed, and so | |
| ``HF_HOME`` can be configured before ``huggingface_hub`` initializes. | |
| """ | |
| from __future__ import annotations | |
| from strata.data.registry import DatasetSpec, Registry, load_registry | |
| from strata.data.languages import CORE_GRAPH_SOURCES, CORE_LANGUAGE_CODES, CORE_LANGUAGES | |
| from strata.data.wikipedia import discover_wikipedia_shards, iter_wikipedia_documents | |
| __all__ = [ | |
| "CORE_GRAPH_SOURCES", | |
| "CORE_LANGUAGES", | |
| "CORE_LANGUAGE_CODES", | |
| "DatasetSpec", | |
| "Registry", | |
| "discover_wikipedia_shards", | |
| "iter_wikipedia_documents", | |
| "load_registry", | |
| ] | |