Skip to content

V1.2.1

Choose a tag to compare

View all tags
@rozx rozx released this 07 Jun 05:19
· 79 commits to main since this release
v1.2.1
b0085cc
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Release Notes v1.2.1 / 发布说明 v1.2.1

A focused patch release fixing two data-loss bugs in cloaked-file detection and archive grouping, and expanding multi-part archive format detection to all archive types. No breaking changes.

一个聚焦的补丁版本，修复了伪装文件检测和档案分组中的两个数据丢失问题，并将多卷档案格式识别扩展到所有档案类型。无破坏性更改。

🐛 Bug Fixes / 错误修复

Don't rename ordinary files into bogus archive parts / 不再把普通文件改名成伪造的档案分卷

A plain non-archive file whose name ends in digits (e.g. 2382) could be renamed into a fake 7z multipart part (2.7z.382). It was then swept into a same-named real archive set and deleted at the end without ever being extracted — losing the user's data.

Root cause: _verify_with_signature() always returned True, so the weak extensionless/digit-suffix cloaking rules renamed any digit-suffixed file.

The signature gate now distinguishes deliberate cloaking from a guess:

  • If the name carries an explicit archive token (.7z / .rar / .zip, via ARCHIVE_TOKEN_RE), the rename is trusted — continuation parts legitimately have no signature (e.g. something.7z.002删除).
  • If the type was only guessed from a trailing number, a real magic-byte signature is now required before renaming.
  • On verification error the tool is now conservative and skips the rename instead of assuming the file is valid.

Genuine cloaked first parts and .7z.00N删 style parts still uncloak correctly.

名字以数字结尾的普通（非档案）文件（例如 2382）原先可能被改名成伪造的 7z 分卷（2.7z.382），随后被并入同名的真实档案集合，在结尾被删除却从未解压 —— 导致用户数据丢失。

根因：_verify_with_signature() 永远返回 True，导致弱规则（无扩展名 / 尾部数字）会改名任何带数字后缀的文件。

签名校验现在区分"有意伪装"与"猜测"：

  • 名字中带有显式档案标记（.7z / .rar / .zip，由 ARCHIVE_TOKEN_RE 识别）时，信任改名 —— 续卷本来就没有签名（例如 something.7z.002删除）。
  • 若类型只是从尾部数字猜测得到，改名前现在必须有真实的魔数签名。
  • 校验出错时改为保守处理，跳过改名，而不是假定文件有效。

真正的伪装首卷以及 .7z.00N删 之类的分卷仍能正确还原。

Don't merge different archive types that share a base name / 不再合并同名但类型不同的档案

The similarity check in _should_group_files ignored the archive extension, so a standalone .7z could be swept into a spanned .zip / .z01 group when they shared a base name and directory. The merged multipart group then deleted the .7z along with the zip parts (data loss) and mishandled the set during extraction.

Fix: the similarity path now requires ext1 == ext2, so only same-family archives group that way. .zip / .z01 still group via the exact (base, ext) check, and foo.7z stays its own single group alongside the foo.zip / foo.z01 set.

_should_group_files 的相似度判断忽略了档案扩展名，导致独立的 .7z 在与跨卷 .zip / .z01 同名且同目录时被并入同一组。合并后的多卷组随后把 .7z 连同 zip 分卷一起删除（数据丢失），解压时也处理错误。

修复：相似度路径现在要求 ext1 == ext2，只有同家族档案才走该路径分组。.zip / .z01 仍通过精确的 (base, ext) 判断分组，而 foo.7z 与 foo.zip / foo.z01 集合并存时保持为独立的单档案组。

✨ Improvements / 功能增强

Expanded multi-part archive format detection / 扩展多卷档案格式识别

Multi-part sets across all archive types are now recognized into a single group with the correct primary as the extraction entry point (previously only .7z.001, .tar.*.001, .zip/.z01, .rar/.r00/.partN.rar were handled correctly):

  • 7-Zip generic numbered split for any base extension — name.<ext>.001 / .002 … (zero-padded, 3+ digits), with .001 as the entry point. This covers the common name.zip.001 (produced by 7-Zip's "Split file") as well as name.rar.001, name.iso.001, and any other <ext>. Previously these were split into wrong groups or treated as single archives, leaving continuation parts behind.
  • WinZip ZIPX split — name.zipx + name.zx01 / .zx02 … (primary .zipx).
  • ARJ multi-volume — name.arj + name.a01 / .a02 … (primary .arj).
  • ACE multi-volume — name.ace + name.c00 / .c01 … (primary .ace). Note: ACE sets are now grouped/classified correctly, but the bundled 7-Zip cannot decode ACE, so extraction still fails and the source parts are retained (never deleted).

Safety is unchanged: source files are deleted only after a successful extraction, so an ordinary numbered file that is not actually an archive (e.g. report.2024.001) is retained when extraction fails.

现在所有档案类型的多卷集合都会被识别为单个分组，并以正确的主卷作为解压入口（此前仅 .7z.001、.tar.*.001、.zip/.z01、.rar/.r00/.partN.rar 能被正确处理）：

  • 任意扩展名的 7-Zip 通用编号分卷 —— name.<ext>.001 / .002 …（补零、3 位及以上），以 .001 为入口。覆盖常见的 name.zip.001（7-Zip"拆分文件"生成）以及 name.rar.001、name.iso.001 等任意 <ext>。此前这些会被分到错误的组或当成单个档案，导致续卷被遗留。
  • WinZip ZIPX 分卷 —— name.zipx + name.zx01 / .zx02 …（主卷 .zipx）。
  • ARJ 多卷 —— name.arj + name.a01 / .a02 …（主卷 .arj）。
  • ACE 多卷 —— name.ace + name.c00 / .c01 …（主卷 .ace）。注意： ACE 集合现在能被正确分组/识别，但捆绑的 7-Zip 无法解码 ACE，因此解压仍会失败，且源分卷会被保留（绝不删除）。

安全策略不变：仅在解压成功后才删除源文件，所以并非真正档案的编号文件（例如 report.2024.001）在解压失败时会被保留。

✅ Test Coverage / 测试覆盖

New/Updated Tests / 新增/更新测试

  • Cloaked false-positives (TestCloakedFalsePositives): a plain 2382 file is not detected/renamed, uncloak_file leaves a non-archive digit-suffixed file untouched, and a genuine cloaked 7z first part (real magic bytes) still uncloaks to secret.7z.001. Two existing _verify_with_signature tests were updated to pin the stricter behavior.
  • Grouping guard: unit (_should_group_files) regression that foo.7z vs foo.zip are not grouped, plus an integration (create_groups_by_name) regression asserting foo.7z stays a standalone group while foo.zip / .z01 / .z02 form one spanned set.
  • Multi-part format detection: a new test_regex.py for the pattern catalog, plus get_archive_base_name, _entry_point_priority, create_groups_by_name, nested-continuation, and ensure_contained_multipart_groups cases for the generic numbered split and ZIPX/ARJ/ACE formats.

Total: 199 tests pass (was 163 in v1.2.0, +36 new).

  • 伪装误报（TestCloakedFalsePositives）：普通的 2382 文件不会被检测/改名；uncloak_file 对非档案的数字后缀文件保持原样；而真正的伪装 7z 首卷（带真实魔数）仍会还原为 secret.7z.001。两个原有的 _verify_with_signature 测试更新为锁定更严格的行为。
  • 分组守卫：单元测试（_should_group_files）回归确认 foo.7z 与 foo.zip 不被分到一组；集成测试（create_groups_by_name）回归确认 foo.7z 保持独立组，而 foo.zip / .z01 / .z02 组成一个跨卷集合。
  • 多卷格式识别：新增 test_regex.py 覆盖模式表，以及 get_archive_base_name、_entry_point_priority、create_groups_by_name、嵌套续卷、ensure_contained_multipart_groups 针对通用编号分卷与 ZIPX/ARJ/ACE 格式的用例。

总计：199 个测试通过（v1.2.0 是 163 个，新增 36 个）。

🚀 Migration Notes / 迁移说明

For Users / 用户须知

  • No breaking changes.

  • Ordinary files whose names happen to end in digits (e.g. 2382) are no longer mistaken for archive parts, renamed, or deleted — they are left exactly as they are.

  • A standalone .7z that shares a base name with a spanned .zip set is now extracted on its own instead of being deleted with the zip parts.

  • 无破坏性更改。

  • 名字恰好以数字结尾的普通文件（例如 2382）不再被误认为档案分卷、被改名或被删除 —— 它们保持原样。

  • 与跨卷 .zip 集合同名的独立 .7z 现在会被单独解压，而不会随 zip 分卷一起被删除。

What's Changed

Full Changelog: v1.2.0...v1.2.1

Contributors

rozx
Assets 3
Loading