67 lines
2.1 KiB
Python
67 lines
2.1 KiB
Python
|
# This file is auto-generated by alef — DO NOT EDIT.
|
||
|
|
# alef:hash:4e15143f4af1ae8bafbdb1506ef057da924484c66a19483966333558ad437e75
|
||
|
|
# To regenerate: alef generate
|
||
|
|
# To verify freshness: alef verify --exit-code
|
||
|
|
# Issues & docs: https://github.com/kreuzberg-dev/alef
|
||
|
|
|
||
|
|
"""E2e tests for category: embeddings."""
|
||
|
|
|
||
|
|
import pytest # noqa: F401
|
||
|
|
from kreuzberg import embed_texts, get_embedding_preset, list_embedding_presets, ExtractionConfig
|
||
|
|
|
||
|
|
|
||
|
|
def _alef_e2e_text(value: object) -> str:
|
||
|
|
return "" if value is None else str(value)
|
||
|
|
|
||
|
|
|
||
|
|
def _alef_e2e_item_texts(item: object) -> tuple[str, ...]:
|
||
|
|
raw_items = getattr(item, "items", None)
|
||
|
|
items_text = " ".join(str(value) for value in raw_items) if isinstance(raw_items, list) else ""
|
||
|
|
return (
|
||
|
|
_alef_e2e_text(item),
|
||
|
|
_alef_e2e_text(getattr(item, "kind", None)),
|
||
|
|
_alef_e2e_text(getattr(item, "name", None)),
|
||
|
|
_alef_e2e_text(getattr(item, "source", None)),
|
||
|
|
_alef_e2e_text(getattr(item, "alias", None)),
|
||
|
|
_alef_e2e_text(getattr(item, "text", None)),
|
||
|
|
_alef_e2e_text(getattr(item, "signature", None)),
|
||
|
|
items_text,
|
||
|
|
)
|
||
|
|
|
||
|
|
|
||
|
|
def test_embed_texts_different_preset() -> None:
|
||
|
|
"""embed_texts: multilingual preset."""
|
||
|
|
texts = ["Hello world", "Test"]
|
||
|
|
config = ExtractionConfig(model={"name": "multilingual", "type": "preset"})
|
||
|
|
|
||
|
|
result = embed_texts(texts, config)
|
||
|
|
assert len(result) >= 2 # noqa: S101
|
||
|
|
|
||
|
|
|
||
|
|
def test_get_embedding_preset_known() -> None:
|
||
|
|
"""get_embedding_preset: known preset."""
|
||
|
|
name = "balanced"
|
||
|
|
|
||
|
|
_ = get_embedding_preset(name)
|
||
|
|
|
||
|
|
|
||
|
|
def test_get_embedding_preset_nominal() -> None:
|
||
|
|
"""get_embedding_preset: nominal case."""
|
||
|
|
name = "balanced"
|
||
|
|
|
||
|
|
_ = get_embedding_preset(name)
|
||
|
|
|
||
|
|
|
||
|
|
def test_get_embedding_preset_unknown() -> None:
|
||
|
|
"""get_embedding_preset: unknown preset fails."""
|
||
|
|
name = "nonexistent-xyz"
|
||
|
|
|
||
|
|
result = get_embedding_preset(name)
|
||
|
|
assert not result # noqa: S101
|
||
|
|
|
||
|
|
|
||
|
|
def test_list_embedding_presets_sanity() -> None:
|
||
|
|
"""list_embedding_presets: returns at least one."""
|
||
|
|
result = list_embedding_presets()
|
||
|
|
assert result # noqa: S101
|