65 lines
2.2 KiB
Python
Generated
65 lines
2.2 KiB
Python
Generated
# 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: async."""
|
|
|
|
from pathlib import Path
|
|
import pytest # noqa: F401
|
|
from kreuzberg import extract_bytes, 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,
|
|
)
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
async def test_async_extract_bytes() -> None:
|
|
"""Async extract_bytes call on PDF document."""
|
|
content = Path("pdf/fake_memo.pdf").read_bytes()
|
|
mime_type = "application/pdf"
|
|
|
|
result = await extract_bytes(content, mime_type, None)
|
|
assert result.mime_type.strip() == "application/pdf" # noqa: S101
|
|
assert len(result.content) >= 50 # noqa: S101
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
async def test_async_extract_bytes_empty_mime() -> None:
|
|
"""extract_bytes empty MIME async."""
|
|
with pytest.raises(Exception): # noqa: B017
|
|
content = Path("text/plain.txt").read_bytes()
|
|
mime_type = ""
|
|
config = ExtractionConfig()
|
|
await extract_bytes(content, mime_type, config)
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
async def test_async_extract_bytes_invalid_mime() -> None:
|
|
"""extract_bytes unsupported MIME async."""
|
|
with pytest.raises(Exception): # noqa: B017
|
|
content = Path("text/plain.txt").read_bytes()
|
|
mime_type = "application/x-nonexistent"
|
|
config = ExtractionConfig()
|
|
await extract_bytes(content, mime_type, config)
|