feat: Improve auto-updater coverage and fix CI coverage command

[cuteolaf]

• add test-only pull hook, shutdown delay toggle, and new unit tests to updater.rs, plus extra coverage for version.rs and watcher.rs
• update the CI workflow to use distinct output flags for json vs html in the cargo llvm-cov nextest step, resolving the duplicate --output-path error

Summary by CodeRabbit

• Tests
  • Expanded test coverage for the auto-updater: pull process, restart signaling, mandatory update handling, update orchestration, version parsing defaults, and watcher behavior.
• Chores
  • CI workflow updated to change where HTML coverage reports are written.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

📝 Walkthrough

Walkthrough

Adds test-only hook support to the auto-updater pull flow, expands unit tests across updater, version, and watcher modules, and updates the CI workflow to change the cargo-llvm-cov HTML output option.

Changes

Cohort / File(s)	Summary
CI Workflow Configuration .github/workflows/ci.yml	Replaced cargo-llvm-cov --output-path coverage.html with --output-dir coverage-html, changing where HTML coverage is produced.
Auto-Updater Test Infrastructure & Tests crates/auto-updater/src/updater.rs	Introduced a test-only PullHook type alias and pull_hook: Option<PullHook> field on AutoUpdater (behind #[cfg(test)]); added test-only SHUTDOWN_DELAY_SECS variant and conditional shutdown behavior; modified pull_image to invoke the hook when present; expanded test suite covering pull hook behavior, mandatory update handling, restart signaling, image formatting, and end-to-end update/pull interactions.
Version Unit Tests crates/auto-updater/src/version.rs	Added unit tests: default equals current version and invalid parse inputs ("1.2", "1.2.3.4", "a.b.c"). No runtime API changes.
Watcher Unit Tests crates/auto-updater/src/watcher.rs	Added tests for VersionWatcher accessor, builder default/custom intervals, and status logic when required version equals current version. No runtime API changes.

Sequence Diagram(s)

(omitted — changes are test hooks and unit tests without a new multi-component runtime control flow that warrants a sequence diagram)

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

• feat: CI publish HTML test coverage report to GitHub Pages #8: Also updates .github/workflows/ci.yml to change cargo-llvm-cov HTML output handling (switch to --output-dir coverage-html).

Poem

🐰 In tests I hop where hooks abide,

I mimic pulls and skip the tide.
Coverage finds its tidy nest,
Versions checked, the code at rest.
A cheerful nibble, then a jest. 🥕

Pre-merge checks and finishing touches

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the two main changes: improving auto-updater test coverage and fixing the CI coverage command flag.
Docstring Coverage	✅ Passed	Docstring coverage is 84.38% which is sufficient. The required threshold is 80.00%.

✨ Finishing touches

• 📝 Generate docstrings

---

📜 Recent review details

Configuration used: defaults

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 19a6e0d and f572dd8.

📒 Files selected for processing (1)

• crates/auto-updater/src/updater.rs

🚧 Files skipped from review as they are similar to previous changes (1)

• crates/auto-updater/src/updater.rs

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)

• GitHub Check: Docker

---

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.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[coderabbitai]

Actionable comments posted: 1

🤖 Fix all issues with AI Agents

In @crates/auto-updater/src/updater.rs:
- Around line 280-286: The test currently calls
updater.check_and_update().await.unwrap() and then discards the boolean returned
by matches!(result, UpdateResult::NoUpdateAvailable |
UpdateResult::AlreadyUpToDate), so the assertion has no effect; replace the bare
matches! usage with an actual assertion such as assert!(matches!(result,
UpdateResult::NoUpdateAvailable | UpdateResult::AlreadyUpToDate)) (or an
appropriate assert_eq!/matches! variant) so the test fails when
updater.check_and_update().await.unwrap() does not return one of the expected
UpdateResult variants.

🧹 Nitpick comments (1)

crates/auto-updater/src/updater.rs (1)

311-322: Edge case: future_version() returns current version when all components are u32::MAX.

While extremely unlikely in practice, if current has all components at u32::MAX, this returns the same version rather than a "future" one. The test using this helper (make_future_requirement) would then test the "already up to date" path instead of the intended "needs update" path.

This is a minor robustness concern for test reliability. Consider adding a panic! or using a sentinel version for clarity.

🔎 Optional: fail explicitly if no future version is possible

         } else if current.major < u32::MAX {
             Version::new(current.major + 1, 0, 0)
         } else {
-            current
+            panic!("Cannot create future version: all components are u32::MAX")
         }

📜 Review details

Configuration used: defaults

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 299d25a and 19a6e0d.

📒 Files selected for processing (4)

• .github/workflows/ci.yml
• crates/auto-updater/src/updater.rs
• crates/auto-updater/src/version.rs
• crates/auto-updater/src/watcher.rs

🧰 Additional context used

🧬 Code graph analysis (1)

crates/auto-updater/src/updater.rs (2)

crates/auto-updater/src/version.rs (2)

• new (16-22)
• current (25-31)

crates/auto-updater/src/watcher.rs (2)

• new (20-29)
• new (113-117)

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)

• GitHub Check: Clippy

🔇 Additional comments (8)

.github/workflows/ci.yml (1)

74-74: LGTM!

The separation of --output-path for JSON and --output-dir for HTML correctly resolves the duplicate flag conflict. The downstream references to coverage-html/ in the artifact upload and GitHub Pages deployment are consistent with this change.

crates/auto-updater/src/version.rs (1)

138-148: LGTM!

The new tests provide good coverage for:

• Verifying Default trait consistency with current()
• Validating parse() correctly rejects malformed version strings (too few/many parts, non-numeric)

crates/auto-updater/src/watcher.rs (1)

158-191: LGTM!

The new tests provide comprehensive coverage for:

• Current version accessor returning expected value
• Builder pattern with custom and default intervals
• Status correctly reports UpToDate when requirement matches current version

Accessing check_interval directly is acceptable since tests are in the same module.

crates/auto-updater/src/updater.rs (5)

9-30: LGTM!

The test-only infrastructure is well-designed:

• PullHook type alias enables mocking Docker pull operations
• SHUTDOWN_DELAY_SECS eliminates test delays while preserving production graceful shutdown timing
• Conditional compilation keeps production binaries clean

---

110-118: LGTM!

The conditional handling for mandatory updates is correct. In tests, returning Ok(()) allows the test to continue and verify behavior, while production code exits to trigger orchestrator restart.

---

126-129: LGTM!

Clean hook injection pattern that short-circuits real Docker calls during tests while preserving production behavior.

---

336-449: LGTM!

Excellent test coverage for the update flow:

• test_handle_update_triggers_restart_signal verifies pull hook invocation and restart signal emission
• test_check_and_update_reports_updated validates the full update path with watcher integration
• test_handle_update_mandatory_triggers_shutdown confirms mandatory flag handling

The hook-based approach cleanly isolates tests from Docker dependencies.

---

451-469: LGTM!

Good negative test case verifying that pull errors propagate correctly through handle_update.