76 lines
2.7 KiB
Python
Generated
76 lines
2.7 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: format_specific."""
|
|
|
|
from pathlib import Path
|
|
import pytest # noqa: F401
|
|
from kreuzberg import extract_bytes_sync, extract_file_sync
|
|
|
|
|
|
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_format_docx_standalone() -> None:
|
|
"""Standalone DOCX extraction using extract_bytes_sync."""
|
|
content = Path("docx/fake.docx").read_bytes()
|
|
mime_type = "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
|
|
|
|
result = extract_bytes_sync(content, mime_type, None)
|
|
assert len(result.content) >= 20 # noqa: S101
|
|
|
|
|
|
def test_format_hwpx_standalone() -> None:
|
|
"""Standalone HWPX extraction using extract_bytes_sync."""
|
|
content = Path("hwpx/simple.hwpx").read_bytes()
|
|
mime_type = "application/haansofthwpx"
|
|
|
|
result = extract_bytes_sync(content, mime_type, None)
|
|
assert len(result.content) >= 20 # noqa: S101
|
|
assert "Hello from HWPX" in result.content # noqa: S101
|
|
|
|
|
|
def test_format_pdf_text() -> None:
|
|
"""Standalone PDF text extraction using extract_bytes_sync."""
|
|
content = Path("pdf/fake_memo.pdf").read_bytes()
|
|
mime_type = "application/pdf"
|
|
|
|
result = extract_bytes_sync(content, mime_type, None)
|
|
assert len(result.content) >= 50 # noqa: S101
|
|
assert any(v in result.content for v in ["Mallori", "May"]) # noqa: S101
|
|
|
|
|
|
def test_format_pptx() -> None:
|
|
"""PPTX presentation extraction using extract_file_sync."""
|
|
path = "pptx/simple.pptx"
|
|
mime_type = "application/vnd.openxmlformats-officedocument.presentationml.presentation"
|
|
|
|
_ = extract_file_sync(path, mime_type, None)
|
|
|
|
|
|
def test_format_xlsx() -> None:
|
|
"""XLSX spreadsheet extraction using extract_file_sync."""
|
|
path = "xlsx/stanley_cups.xlsx"
|
|
mime_type = "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
|
|
|
|
_ = extract_file_sync(path, mime_type, None)
|