fix(tauri): deterministic CEF teardown on full app close (#1120)

[M3gA-Mind]

Summary

• Replace the fixed post-teardown sleep with a bounded poll so CEF child webviews (acct_*, prewarm) can finish closing before cef::shutdown().
• Stop CDP-backed provider scanners (Discord, Slack, Telegram, WhatsApp) and the iMessage scanner registry during the same early teardown path.
• Return the list of webviews requested closed from WebviewAccountsState::shutdown_all so shutdown ordering is observable and coordinated.
• Add structured log::info! diagnostics around webview drain timing and pending labels.
• Regenerate app/src-tauri/Cargo.lock so it matches the current 0.53.13 manifests and dependency graph (including removal of stale html2md entries).

Problem

Issue #1120: full app quit could race CEF teardown—child account webviews, CDP tasks, and fixed short sleeps did not reliably guarantee browsers were winding down before the runtime shut CEF down, risking helper processes and shutdown ordering issues.

Solution

Centralize early teardown: close prewarm + account webviews (collecting labels), stop long-lived scanner tasks, then poll AppHandle::get_webview for those labels until they disappear or a ~300ms budget expires, with a short initial yield for the UI thread. Unit tests in webview_accounts cover label bookkeeping for shutdown_all.

Submission Checklist

• Tests added or updated (happy path + at least one failure / edge case) per docs/TESTING-STRATEGY.md — webview_accounts tests extended for returned close labels / invariants.
• Diff coverage ≥ 80% — Local: pnpm test:coverage and pnpm test:rust completed successfully. CI Coverage Gate (diff-cover ≥ 80%) passed (merged Vitest lcov + cargo llvm-cov for core + Tauri per .github/workflows/coverage.yml).
• Coverage matrix updated — added/removed/renamed feature rows in docs/TEST-COVERAGE-MATRIX.md reflect this change — N/A: Tauri-shell shutdown behaviour only; no matrix row change.
• All affected feature IDs from the matrix are listed in the PR description under ## Related — N/A: see above.
• No new external network dependencies introduced (mock backend used per docs/TESTING-STRATEGY.md)
• Manual smoke checklist updated if this touches release-cut surfaces (docs/RELEASE-MANUAL-SMOKE.md) — N/A: no release doc change required.
• Linked issue closed via Closes #NNN in the ## Related section

Impact

• Desktop (Tauri / CEF): more deterministic shutdown on Cmd+Q, window close, tray quit, and restart paths that already call perform_early_teardown_*.
• Performance: bounded wait up to ~300ms on quit (replacing a fixed 50ms sleep) when child webviews exist; no change when none are open.

Related

• Closes Ensure CEF fully winds down on full app close #1120
• Follow-up PR(s)/TODOs: Broader E2E validation of zero lingering CEF helpers on macOS/Linux/Windows remains manual / environment-dependent per issue acceptance criteria.

Summary by CodeRabbit

• New Features

  • Consistent scanner shutdown API across platforms for reliable stop/restart behavior.
  • Shutdown now reports which account webviews were closed during teardown.

• Bug Fixes

  • Improved application shutdown by actively canceling background scanner tasks.
  • More reliable account removal that aborts associated scanner work instead of just bookkeeping.
  • Teardown waits for webviews to close instead of relying on fixed delays.

• Tests

  • Added tests validating scanner cancellation and repeatable shutdown behavior.

[coderabbitai]

📝 Walkthrough

Walkthrough

Scanner registries for multiple providers were refactored to store per-account tokio AbortHandle vectors and actively abort running scanner tasks on forget/forget_all. App early teardown was changed to close CEF webviews and poll until they drain (removing fixed sleeps) and to synchronously trigger scanner teardowns before CEF shutdown.

Changes

Scanner Lifecycle & Task Cancellation

Layer / File(s)	Summary
Imports & Mutex app/src-tauri/src/{discord,slack,telegram,whatsapp,imessage}_scanner/mod.rs	Swap tokio::sync::Mutex<HashSet<_>> for parking_lot::Mutex<HashMap<String, Vec<AbortHandle>>>; add tokio::task::AbortHandle usage.
Spawn API app/src-tauri/src/{discord,slack,telegram,whatsapp}_scanner/mod.rs	spawn_scanner now returns Vec<AbortHandle>; spawn_dom_poll returns an AbortHandle instead of detaching tasks.
Registry API & Behavior app/src-tauri/src/{discord,slack,telegram,whatsapp,imessage}_scanner/mod.rs	ensure_scanner becomes synchronous and records returned abort handles; forget(account_id) aborts stored handles; forget_all() aborts across accounts and returns aborted-handle count; macOS iMessage registry adds shutdown() (non-macOS stub present).
DOM Scan Targeting app/src-tauri/src/slack_scanner/mod.rs	dom_scan_once selects CDP targets using pinned/strict/relaxed rules (relaxed limited to /client), then connects/attaches/scan/detach explicitly.
Tests app/src-tauri/src/{discord,slack,telegram,whatsapp,imessage}_scanner/mod.rs	Unit tests updated/added to insert pending tasks, register their abort handles, and assert forget/forget_all/shutdown cause task cancellation and are repeatable.

CEF Shutdown Orchestration

Layer / File(s)	Summary
Teardown Constants & Helpers app/src-tauri/src/lib.rs	Add CEF close/timeout constants and helpers to close prewarm and account webviews and to shut down the iMessage scanner.
Polling/Drain Helpers app/src-tauri/src/lib.rs	Add pending_cef_webview_labels() and sync/async wait_for_cef_webviews_to_close_*() to poll app.get_webview() until labels drain or budget elapses.
Early Teardown Flow app/src-tauri/src/lib.rs	perform_early_teardown_sync() and perform_early_teardown_async() refactored to use close+wait sequence and to remove prior unconditional 50ms sleeps.
App-Level Scanner Teardown app/src-tauri/src/webview_accounts/mod.rs	Add teardown_all_account_scanners() calling each provider registry's forget_all(); teardown_account_scanners() now calls registry.inner().forget(account_id) synchronously.
Shutdown Return & Wiring app/src-tauri/src/webview_accounts/mod.rs	shutdown_all() now returns Vec<String> of closed labels and logs per-label outcomes; webview_account_open calls provider ensure_scanner directly (no tokio::spawn).
Shutdown Tests app/src-tauri/src/webview_accounts/mod.rs	Teardown tests updated to use abort_handle(), yield_now().await, and assert aborted tasks are finished; assertions verify returned closed label list.

Sequence Diagrams

sequenceDiagram
    participant Client as Caller (forget/forget_all)
    participant Reg as ScannerRegistry
    participant Handle as AbortHandle(s)
    participant Task as Scanner Task
    participant Tokio as Tokio Executor

    Client->>Reg: forget(account_id)
    Reg->>Handle: lookup Vec<AbortHandle>
    Reg->>Handle: call abort() on each
    Handle->>Task: cancellation signal
    Tokio->>Task: task polled -> stops
    Task-->>Reg: stopped (implicit)

Loading

sequenceDiagram
    participant User as User quits
    participant App as App (lib.rs)
    participant WA as WebviewAccountsState
    participant Reg as ScannerRegistry(s)
    participant Web as CEF Webviews
    participant CEF as CEF Runtime

    User->>App: ExitRequested
    App->>WA: close_early_cef_webviews()
    WA->>Reg: teardown_all_account_scanners() -> forget_all()
    Reg->>Reg: abort stored scanner handles
    App->>Web: pending_cef_webview_labels()
    loop poll until budget or empty
        App->>Web: get_webview(label)
        alt exists
            App->>App: sleep interval, retry
        else closed
            App->>App: proceed
        end
    end
    App->>CEF: cef::shutdown()

Loading

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~60 minutes

Possibly related PRs

• #1128 — Centralizes early teardown logic in lib.rs and adjusts CEF/webview shutdown sequencing (directly related to teardown changes).
• #1028 — Changes Slack scanner target selection and scan flow (overlaps Slack dom_scan_once adjustments).
• #751 — Refactors per-provider scanners to return/track AbortHandle(s) and registry abort semantics (same scanner/registry surface).

Suggested reviewers

• senamakel

Poem

🐰 I held the abort and let it go,

Tasks once running, now gentle and slow;
Webviews close as the timers unwind,
Clean shutdowns — peace for my mind. ✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main change: fixing deterministic CEF teardown on app close, matching the core objective of the PR.
Linked Issues check	✅ Passed	All linked issue objectives are addressed: child webviews closed before CEF shutdown [#1120], long-lived scanner tasks stopped [#1120], fixed sleeps replaced with bounded polling [#1120], teardown ordering explicit and centralized [#1120], and improved logging/diagnostics [#1120].
Out of Scope Changes check	✅ Passed	All changes directly support the linked objective to achieve deterministic CEF teardown: scanner registry refactoring enables task cancellation, teardown logic polling replaces sleeps, webview shutdown returns labels for observability, and iMessage scanner gets shutdown hooks.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

Tip

💬 Introducing Slack Agent: The best way for teams to turn conversations into code.

Slack Agent is built on CodeRabbit's deep understanding of your code, so your team can collaborate across the entire SDLC without losing context.

• Generate code and open pull requests
• Plan features and break down work
• Investigate incidents and troubleshoot customer tickets together
• Automate recurring tasks and respond to alerts with triggers
• Summarize progress and report instantly

Built for teams:

• Shared memory across your entire org—no repeating context
• Per-thread sandboxes to safely plan and execute work
• Governance built-in—scoped access, auditability, and budget controls

One agent for your entire SDLC. Right inside Slack.

👉 Get started

---

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

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@app/src-tauri/src/lib.rs`:
- Around line 1034-1048: The sync waiter wait_for_cef_webviews_to_close_sync
blocks the Tauri main event-loop with std::thread::sleep and must not be used
from RunEvent::ExitRequested or the tray quit handler; replace its blocking
polling by either calling the existing async variant
wait_for_cef_webviews_to_close_async (via perform_early_teardown_async) so
sleeps use tokio::time::sleep and don't block the event loop, or remove the
synchronous polling entirely and rely on the event-loop sweep; update
invocations in the ExitRequested and tray quit handler to use the async teardown
path (or spawn a dedicated background thread to run the current polling logic
off the main thread) and ensure any shared helpers (pending_cef_webview_labels)
remain accessible to the chosen path.

In `@app/src-tauri/src/webview_accounts/mod.rs`:
- Around line 635-642: The shutdown code currently only pushes label into
closed_labels when wv.close() succeeds, which means failed closes are not
tracked by the shutdown poll; change the logic so you record the label as soon
as app.get_webview(&label) returns Some (i.e. immediately push label into
closed_labels or a "tracked" set after get_webview succeeds), then attempt
wv.close() and log any Err from wv.close() without removing the label, ensuring
get_webview, wv.close, and closed_labels are the referenced symbols to modify so
failed closes remain visible to the teardown logic.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: c3a4d123-775f-421b-bee7-344fca04bdd4

📥 Commits

Reviewing files that changed from the base of the PR and between 995e5cc and 33bcd22.

⛔ Files ignored due to path filters (1)

• app/src-tauri/Cargo.lock is excluded by !**/*.lock

📒 Files selected for processing (7)

• app/src-tauri/src/discord_scanner/mod.rs
• app/src-tauri/src/imessage_scanner/mod.rs
• app/src-tauri/src/lib.rs
• app/src-tauri/src/slack_scanner/mod.rs
• app/src-tauri/src/telegram_scanner/mod.rs
• app/src-tauri/src/webview_accounts/mod.rs
• app/src-tauri/src/whatsapp_scanner/mod.rs

[coderabbitai]

♻️ Duplicate comments (1)

app/src-tauri/src/lib.rs (1)

1004-1012: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Track the prewarm label even when close() fails.

This only adds cef-prewarm to closed_labels on Ok(()). If the webview still exists but wv.close() errors, the async drain poll loses visibility into that live CEF child and shutdown can still race on it. This is the same failure mode you just fixed for account webviews in app/src-tauri/src/webview_accounts/mod.rs.

Proposed fix

 fn close_early_cef_webviews<R: tauri::Runtime>(app: &AppHandle<R>) -> Vec<String> {
     let mut closed_labels = Vec::new();
-    if teardown_cef_prewarm(app).is_ok() {
+    if app.get_webview(CEF_PREWARM_LABEL).is_some() {
         closed_labels.push(CEF_PREWARM_LABEL.to_string());
+        if let Err(err) = teardown_cef_prewarm(app) {
+            log::warn!("[cef-prewarm] early teardown close failed: {err}");
+        }
     }
     if let Some(state) = app.try_state::<webview_accounts::WebviewAccountsState>() {
         closed_labels.extend(state.shutdown_all(app));
     }
     closed_labels

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@app/src-tauri/src/lib.rs` around lines 1004 - 1012, The function
close_early_cef_webviews currently only pushes CEF_PREWARM_LABEL when
teardown_cef_prewarm(app) returns Ok, which loses visibility if the prewarm
webview exists but teardown errors; change the logic so CEF_PREWARM_LABEL is
appended to closed_labels regardless of teardown_cef_prewarm's Result (push
before or in both branches), and preserve/report the error from
teardown_cef_prewarm (e.g., log or ignore after recording the label) so the
prewarm child remains tracked even when teardown fails; refer to
close_early_cef_webviews, teardown_cef_prewarm, and CEF_PREWARM_LABEL to locate
and update the code.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Duplicate comments:
In `@app/src-tauri/src/lib.rs`:
- Around line 1004-1012: The function close_early_cef_webviews currently only
pushes CEF_PREWARM_LABEL when teardown_cef_prewarm(app) returns Ok, which loses
visibility if the prewarm webview exists but teardown errors; change the logic
so CEF_PREWARM_LABEL is appended to closed_labels regardless of
teardown_cef_prewarm's Result (push before or in both branches), and
preserve/report the error from teardown_cef_prewarm (e.g., log or ignore after
recording the label) so the prewarm child remains tracked even when teardown
fails; refer to close_early_cef_webviews, teardown_cef_prewarm, and
CEF_PREWARM_LABEL to locate and update the code.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 28faa4d4-ca90-4bfa-a3fe-6a01eefe6008

📥 Commits

Reviewing files that changed from the base of the PR and between 33bcd22 and ea9c9a6.

📒 Files selected for processing (2)

• app/src-tauri/src/lib.rs
• app/src-tauri/src/webview_accounts/mod.rs