Transforms complex documents like PDFs and Office docs into LLM-ready markdown/JSON for your Agentic workflows.
Python ai4sciencedocument-analysisdocxextract-datalayout-analysisocrparserpdfpdf-converterpdf-extractor-llmpdf-extractor-pretrainpdf-extractor-ragpdf-parserpptxpythonxlsx
"""验证 ZIP 导出的临时目录隔离、用户文件保护与异常清理。"""
from __future__ import annotations
import base64
import zipfile
from io import BytesIO
from pathlib import Path
from typing import Literal
import pytest
from PIL import Image
from mineru.kit.common import save_parse_result
from mineru.parser.base import ParseResult
from mineru.parser.writer import DataWriter
from mineru.types import ImageBlock, ImageBodyBlock, MiddleJson, PageInfo
@pytest.fixture
def image_bytes() -> bytes:
"""生成可逐字节核对的最小 PNG 素材。"""
output = BytesIO()
Image.new("RGB", (2, 2), "red").save(output, format="PNG")
return output.getvalue()
@pytest.fixture
def parse_result(image_bytes: bytes) -> ParseResult:
"""构造包含内嵌图片的解析结果,覆盖素材落盘与引用。"""
return ParseResult(
MiddleJson(
pages=[
PageInfo(
page_idx=0,
blocks=[
ImageBlock(
type="image",
index=0,
content=[
ImageBodyBlock(
type="image_body",
index=0,
content="",
image_base64="data:image/png;base64," + base64.b64encode(image_bytes).decode(),
)
],
)
],
)
],
is_full_document=True,
metadata={"file_suffix": "docx", "producer": {"name": "mineru", "version": "test"}},
)
)
@pytest.mark.parametrize("sibling_kind", ["file", "directory"])
def test_zip_export_preserves_existing_sibling(
tmp_path: Path,
parse_result: ParseResult,
image_bytes: bytes,
sibling_kind: Literal["file", "directory"],
) -> None:
"""与输出同名的隐藏文件或目录不得混入 ZIP,也不得被清理。"""
dest = tmp_path / "report.zip"
sibling = tmp_path / ".report"
if sibling_kind == "directory":
sibling.mkdir()
sentinel = sibling / "private.txt"
else:
sentinel = sibling
sentinel.write_bytes(b"unrelated user data")
save_parse_result(parse_result, dest, "zip")
assert sentinel.read_bytes() == b"unrelated user data"
assert set(tmp_path.iterdir()) == {sibling, dest}
with zipfile.ZipFile(dest) as archive:
names = set(archive.namelist())
images = [name for name in names if name.startswith("images/")]
assert len(images) == 1
assert names == {"markdown.md", "middle_json.json", "structured_content.json", *images}
assert archive.read(images[0]) == image_bytes
assert images[0] in archive.read("markdown.md").decode()
def test_zip_export_cleans_up_after_save_failure(
tmp_path: Path, parse_result: ParseResult, monkeypatch: pytest.MonkeyPatch
) -> None:
"""结果保存失败时清理部分素材,同时保留已有用户目录和内容。"""
dest = tmp_path / "report.zip"
sibling = tmp_path / ".report"
sibling.mkdir()
sentinel = sibling / "private.txt"
sentinel.write_bytes(b"unrelated user data")
def fail_save(self: ParseResult, writer: DataWriter) -> None:
"""模拟写出部分图片后序列化失败。"""
writer.write("images/partial.png", b"partial image")
raise RuntimeError("save failed")
monkeypatch.setattr(ParseResult, "save", fail_save)
with pytest.raises(RuntimeError, match="save failed"):
save_parse_result(parse_result, dest, "zip")
assert sentinel.read_bytes() == b"unrelated user data"
assert set(tmp_path.iterdir()) == {sibling}
assert set(sibling.iterdir()) == {sentinel}
def test_zip_export_cleans_up_after_archive_failure(
tmp_path: Path, parse_result: ParseResult, monkeypatch: pytest.MonkeyPatch
) -> None:
"""ZIP 写入失败时清理暂存文档和图片,同时保留已有用户目录和内容。"""
dest = tmp_path / "report.zip"
sibling = tmp_path / ".report"
sibling.mkdir()
sentinel = sibling / "private.txt"
sentinel.write_bytes(b"unrelated user data")
def fail_write(self: zipfile.ZipFile, filename: Path, arcname: str) -> None:
"""模拟归档写入过程中发生 I/O 错误。"""
raise OSError("archive write failed")
monkeypatch.setattr(zipfile.ZipFile, "write", fail_write)
with pytest.raises(OSError, match="archive write failed"):
save_parse_result(parse_result, dest, "zip")
assert sentinel.read_bytes() == b"unrelated user data"
assert set(tmp_path.iterdir()) == {sibling, dest}
assert set(sibling.iterdir()) == {sentinel}