fix(policy): fail closed on invalid .html4ignore states - #697

👋 Jules, reporting for duty! I'm here to lend a hand with this pull request.

When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down.

I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job!

For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with @jules. You can find this option in the Pull Request section of your global Jules UI settings. You can always switch back!

New to Jules? Learn more at jules.google/docs.

For security, I will only act on instructions from the user who triggered this task.

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

_{Comment @coderabbitai help to get the list of available commands.}

Current-head repair finding for 868be4cba8fdfe77fafb5bbf30420c0f34f681b1:

1. ignore_file.exists() follows symlinks. A dangling .html4ignore symlink therefore reports false and the function takes the same path as “no policy file”, so the directory is processed instead of failing closed. Add a deterministic regression with .html4ignore -> missing-target; it must raise IgnoreFileReadException / skip directory processing. The minimum existence check must use Files.exists(path, LinkOption.NOFOLLOW_LINKS) so the directory entry itself is observed.

2. The current implementation does not eliminate TOCTOU. It checks isFile, isSymbolicLink, canRead, and length() and then opens the path later with useLines; every path check can become stale before the open. Oracle’s Files.exists contract explicitly warns that its result is immediately outdated in security-sensitive use, and SecureDirectoryStream exists for race-resistant relative file operations. Official references: https://docs.oracle.com/en/java/javase/25/docs/api/java.base/java/nio/file/Files.html and https://docs.oracle.com/en/java/javase/25/docs/api/java.base/java/nio/file/SecureDirectoryStream.html

GREEN acceptance: (a) dangling-symlink RED → GREEN, (b) policy entry is opened without following links through a race-resistant directory-relative handle where supported, with explicit fail-closed behavior when that guarantee cannot be provided, (c) size/line/pattern limits are enforced on the opened object rather than a pre-open path snapshot, and (d) PR/security notes stop claiming TOCTOU is fixed until those invariants are demonstrated. No gate weakening or exception swallowing.

@jules 현재 head 868be4cba8fdfe77fafb5bbf30420c0f34f681b1을 다음 acceptance로 수리해 주세요.

• 먼저 .html4ignore -> missing-target dangling symlink fixture를 추가해 현재 구현의 fail-open을 RED로 재현합니다.
• File.exists() 대신 directory entry 자체를 NOFOLLOW_LINKS로 관찰해 dangling symlink도 invalid policy로 fail closed해야 합니다.
• 다만 그 변경만으로 TOCTOU fixed라고 쓰면 안 됩니다. path 검사 후 useLines()로 재-open하는 현재 구조의 race를 제거하도록, 지원되는 provider에서는 SecureDirectoryStream의 directory-relative newByteChannel(..., NOFOLLOW_LINKS) 또는 동등한 race-resistant open을 사용하고, 그런 보장을 제공할 수 없는 provider에서는 명시적으로 fail closed하는 경계를 설계하세요.
• size/line/pattern limits는 열린 policy object에 적용하고, symlink/path-swap으로 외부 target을 읽지 않는 회귀를 추가하세요.
• .jules/sentinel.md와 PR 문구의 HIGH/TOCTOU 해결 주장은 exact-head regression과 구현이 증명하는 범위로만 낮추세요.
• 기존 valid semantic delta/tests를 보존하고 force-push/rebase/gate weakening 없이 ordinary descendant commit으로 올려 주세요.