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
fix: never fetch remote HTML images and keep external links
myhloli2d016d626e84
diff --git a/docs/en/reference/output_files.md b/docs/en/reference/output_files.mdindex cf1366b3..6ff61e7a 100644--- a/docs/en/reference/output_files.md+++ b/docs/en/reference/output_files.md@@ -76,7 +76,7 @@ and report `pdf_title_geometry_conflict`; unavailable expansion clearance report`pdf_title_clearance_unavailable`. Only rendering context changes, with no input,asset, protocol or public-option changes.-Requires `docvortex>=0.4.7,<1` (the current minimum dependency declared by MinerU 4.0). DocVortex clears native TXT PDF display equation+Requires `docvortex>=0.4.12,<1` (the current minimum dependency declared by MinerU 4.0). DocVortex clears native TXT PDF display equation`content` once when producing model output; MinerU Flash TXT uses that output.Flash OCR already leaves display equation content empty. MinerU does not clearequation content again. Both Flash paths retain the bbox, orientation, image anddiff --git a/docs/zh/reference/output_files.md b/docs/zh/reference/output_files.mdindex ec77c92c..f4dd021c 100644--- a/docs/zh/reference/output_files.md+++ b/docs/zh/reference/output_files.md@@ -47,7 +47,7 @@ pdf_bytes = render(result.middle_json, RenderFormat.PDF,`pdf_title_layout_expanded` 记录扩展;`pdf_layout_font_exception` 记录局部正文较大或空间不足的调整原因、参考/目标/最终字号、原框和绘制框。原始几何重叠时不扩大占用并报告 `pdf_title_geometry_conflict`;原始间距过紧、没有安全扩展区域时报告 `pdf_title_clearance_unavailable`。字号和绘制区域仅存在于渲染上下文,不修改 MiddleJson 或素材,不新增接口参数。-使用 `docvortex>=0.4.7,<1`(MinerU 4.0 当前声明的最低依赖)。DocVortex 在生产原生 TXT PDF 模型输出时将行间公式的 `content` 清空一次,MinerU Flash TXT 直接使用该输出;Flash OCR 原本不填充行间公式内容,MinerU 不再重复清空。两条 Flash 路径保留 bbox、方向、图片及检测到的编号区域,PDF、Markdown、HTML、DOCX、EPUB、LaTeX 沿用图片回退。行内公式、非 Flash tier 的公式文本不变。旧缓存不会自动改写,重新解析才获得新几何和空内容公式。+使用 `docvortex>=0.4.12,<1`(MinerU 4.0 当前声明的最低依赖)。DocVortex 在生产原生 TXT PDF 模型输出时将行间公式的 `content` 清空一次,MinerU Flash TXT 直接使用该输出;Flash OCR 原本不填充行间公式内容,MinerU 不再重复清空。两条 Flash 路径保留 bbox、方向、图片及检测到的编号区域,PDF、Markdown、HTML、DOCX、EPUB、LaTeX 沿用图片回退。行内公式、非 Flash tier 的公式文本不变。旧缓存不会自动改写,重新解析才获得新几何和空内容公式。```pythonfrom pathlib import Pathdiff --git a/mineru/parser/api_server.py b/mineru/parser/api_server.pyindex 8f387064..6960193b 100644--- a/mineru/parser/api_server.py+++ b/mineru/parser/api_server.py@@ -1480,7 +1480,6 @@ async def _run_job(source_uri=extracted.source_uri,local_resource_root=extracted.local_resource_root,transport_encoding=extracted.transport_encoding,- fetch_remote_images=True,)result = await parse_async(str(tmp_path),diff --git a/mineru/parser/mineru_parser.py b/mineru/parser/mineru_parser.pyindex bceaec71..50857123 100644--- a/mineru/parser/mineru_parser.py+++ b/mineru/parser/mineru_parser.py@@ -193,7 +193,6 @@ class MinerUParser(DocumentParser):resolved_source_context = HtmlSourceContext(source_uri=resolved_path.as_uri(),local_resource_root=resolved_path.parent,- fetch_remote_images=True,)source_properties = (DocumentProperties()diff --git a/pyproject.toml b/pyproject.tomlindex 6c8ddda2..378f28eb 100644--- a/pyproject.toml+++ b/pyproject.toml@@ -39,7 +39,7 @@ classifiers = ["Programming Language :: Python :: 3.14",]dependencies = [- "docvortex>=0.4.11,<1",+ "docvortex>=0.4.12,<1","gradio>=6.8,<7","mineru[torch] ; sys_platform == 'darwin' and platform_machine == 'arm64'","mineru-vl-utils>=2.0.4,<3",diff --git a/tests/unittest/test_flash_html.py b/tests/unittest/test_flash_html.pyindex 16d4b86d..06d827eb 100644--- a/tests/unittest/test_flash_html.py+++ b/tests/unittest/test_flash_html.py@@ -386,6 +386,44 @@ def test_html_parse_server_local_source_keeps_relative_assets(tmp_path: Path) ->assert image_body["image_base64"].startswith("data:image/png;base64,")+def test_html_remote_image_never_downloaded_and_zip_keeps_external_link(tmp_path: Path) -> None:+ """远程图片永不下载:SDK 与 parse-server 均保留外链,自包含 zip 正常生成且 markdown 保留链接。"""+ # .invalid 顶级域保证不可解析;若回归为服务端下载,本测试会因取图失败或超时显式暴露。+ url = "https://example.invalid/chart.png"+ source = tmp_path / "remote-image.html"+ source.write_text(f'<html><body><h1>Remote</h1><img src="{url}" alt="chart"></body></html>', encoding="utf-8")++ result = parse(str(source))+ body = _image_body(result.middle_json)+ assert body.image_url == url and not body.image_path and not body.image_base64++ archive_bytes = api_server._build_self_contained_zip_output(result)+ with ZipFile(BytesIO(archive_bytes)) as archive:+ names = archive.namelist()+ assert "markdown.md" in names and url in archive.read("markdown.md").decode()+ assert not [name for name in names if name.startswith("images/")]++ file_store = FileStore(tmp_path / "api-files")+ request = CreateJobRequest.model_validate(+ {+ "files": [{"source": {"type": "local", "path": str(source)}}],+ "tier": "standard",+ "output_formats": ["markdown", "zip"],+ }+ )+ record = api_server.JobStore().create(request, file_store)++ asyncio.run(api_server._run_job(record, request, file_store, image_analysis=True, allow_local_source=True))++ parsed_file = record.files[0]+ assert parsed_file.status == "completed"+ assert parsed_file.output_files is not None and parsed_file.output_files.zip is not None+ zip_record = file_store.get_file(parsed_file.output_files.zip.file_id)+ assert zip_record.sha256sum is not None+ with ZipFile(BytesIO(file_store.read_blob(zip_record.sha256sum))) as archive:+ assert url in archive.read("markdown.md").decode()++def test_html_doclib_local_bridge_uses_flash_parser(tmp_path: Path) -> None:"""验证 Doclib 本地 Flash 桥接把 HTML 文件交给统一 MinerUParser。"""source = tmp_path / "doclib.html"