Skip to content

feat: safely import PDF and DOCX packages - #8

Open
ZZDR1023 wants to merge 3 commits into
mainfrom
ZZDR1023/issue-4-safe-package-input
Open

feat: safely import PDF and DOCX packages#8
ZZDR1023 wants to merge 3 commits into
mainfrom
ZZDR1023/issue-4-safe-package-input

Conversation

@ZZDR1023

@ZZDR1023 ZZDR1023 commented Aug 12, 2026

Copy link
Copy Markdown
Owner

Refs #4

Hardens local package ingestion at the PDF, DOCX, ZIP, filesystem, and OOXML metadata trust boundaries. It enforces exact normalized PDF text limits, classifies CFB/OLE DOCX containers as encrypted, rejects normalized external/escaping relationships and VBA/macro/ActiveX indicators, and bounds all traversed package entries and depth. Validation includes focused and full tests, real LibreOffice-generated PDF/DOCX imports, generated encrypted/macro/adversarial DOCX rejection, build, smoke, and audit.

@ZZDR1023

Copy link
Copy Markdown
Owner Author

主控边界复核退回:PDF 最终输出字符数可被分隔符绕过;加密 DOCX 分类不符契约;TargetMode 空白可绕过 external 拒绝;改名 VBA relationship/content type 被接受;非支持文件可绕过 package entry limit。完整复现和必须补的 Red tests 见 Issue #4 最新评论。请通过新的 Orca regression Dispatch 更新本 PR。

@ZZDR1023

Copy link
Copy Markdown
Owner Author

当前 head 的主控复核通过

39cf5fc3155a853ac292f0389971572bab3b517f 已完成 PDF/DOCX parser boundary recovery。主控在干净 archive 独立通过 8 个 Vitest files / 31 tests、7 个 Python tests、build、smoke、audit、真实 LibreOffice PDF/DOCX import,以及 CFB/external relation/VBA/OOXML ambiguity/entry-flood/text-limit 对抗 fixture;CI verify 31692543694 通过。

完整复核和人验收步骤在 Issue #4。代码与自动化验证通过,待人验收;请勿在此前合并

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant