🐛 Isolate Playwright E2E DB and add live-target guardrails

[ebigunso]

Summary

• add Playwright global setup and teardown to launch a harness-managed local API for E2E
• enforce local-only API targets by default and fail fast on unsafe targets
• provision disposable isolated SQLite DB artifacts under .playwright-cli/e2e-db
• add backend API bind env support via API_BIND_ADDR for harness orchestration
• document safe E2E workflow and validation mapping updates
• fix auth bootstrap timeout by setting login form fallback to method=post action=/api/login

Validation

• npm run check
• npm run test:unit
• cargo test -p sleep-api
• negative guard check with unsafe E2E_API_BASE_URL fails fast as expected
• npm run test:e2e:auth (passes)

[Copilot]

Pull request overview

This PR adds comprehensive E2E test isolation infrastructure to prevent accidental data corruption during Playwright test runs. The implementation introduces a harness-managed local API with disposable SQLite databases, enforces localhost-only API targets by default, and adds explicit safety guardrails that fail fast if unsafe targets are detected.

Changes:

• Added Playwright global setup/teardown to spawn an isolated API process with a unique disposable DB for each E2E run
• Implemented safety guardrails that block non-localhost API targets unless explicitly overridden with ALLOW_NON_ISOLATED_E2E=1
• Added backend API_BIND_ADDR configuration support to enable E2E harness orchestration
• Refactored E2E tests to use stable selectors and wait for API responses to improve reliability
• Updated documentation to describe safe E2E workflow and validation requirements

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 9 comments.

Show a summary per file

File	Description
sleep-ui/tests/global.setup.ts	New global setup that spawns isolated API with disposable DB, validates safety preconditions, and waits for API readiness
sleep-ui/tests/global.teardown.ts	New global teardown that terminates spawned API process and cleans up DB artifacts
sleep-ui/tests/e2e.spec.ts	Refactored tests to use helper functions for stable selectors, wait for API responses, and handle conditional fields
sleep-ui/tests/auth.setup.ts	Added safety assertions to prevent running auth bootstrap against unsafe targets
sleep-ui/playwright.config.ts	Integrated global setup/teardown, added URL validation, configured webServer to use isolated API
sleep-ui/package.json	Added test:e2e:safe script alias (redundant with test:e2e)
sleep-ui/README.md	Documented safe E2E workflow and explicit bypass mechanism
sleep-api/src/main.rs	Updated to use configurable bind address from API_BIND_ADDR env var
sleep-api/src/config.rs	Added api_bind_addr() function with default 0.0.0.0:8080
docs/coding-agent/references/validation.md	Updated with E2E isolation validation requirements and canonical commands
docs/coding-agent/references/how-to-run.md	Documented API/UI runtime requirements, bind address config, and E2E workflows
docs/coding-agent/plans/active/sleep-ui-e2e-db-isolation-guardrails-plan.md	Added comprehensive plan document describing isolation approach and validation strategy

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[ebigunso]

Addressed review follow-ups from PR53:

• Must fixes:

  • Hardened teardown for cross-platform process termination + DB cleanup retries.
  • Removed redundant test:e2e:safe script alias.
  • Documented strict-port behavior for Playwright E2E startup.
  • Reduced runtime-state collision risk by scoping state path to PLAYWRIGHT_E2E_RUN_ID.

• Should fixes:

  • Made E2E date/offset generation deterministic (removed Date.now()-based randomness in key paths).
  • Replaced URL regex fallback checks with pathname-based checks.

Validation rerun:

• npm run check
• npm run test:e2e:auth

[Copilot]

Pull request overview

Copilot reviewed 13 out of 13 changed files in this pull request and generated 3 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[ebigunso]

Correction to previous note (plain text):

Addressed newest PR53 feedback and main conflict.

1. Must-fix bypass runtime-state path in global setup:

• Runtime directory is now created before writing runtime-state in bypass mode.

2. Must-fix detached teardown reliability on Unix:

• Teardown now signals Unix process groups using negative pid, with fallback to pid-level signals, plus DB cleanup retries.

3. Should-fix readiness strictness:

• Readiness check now accepts expected /api/session states 200 or 401, instead of broad less-than-500.

4. Main conflict:

• Merged origin/main and resolved login form conflict while keeping both method=post action=/api/login and data-testid=login-form-action.

Validation evidence:

• cd sleep-ui && npm run check
• cd sleep-ui && npm run test:unit
• cd sleep-ui && npm run test:e2e:auth
• git diff --name-only --diff-filter=U returned no output.

[Copilot]

Pull request overview

Copilot reviewed 15 out of 15 changed files in this pull request and generated 10 comments.

Comments suppressed due to low confidence (1)

docs/coding-agent/plans/active/pr53-unresolved-comments-and-main-conflict-resolution-plan.md:120

• This plan document describes future work to resolve PR #53 review comments and merge conflicts. It's unusual to include a plan for future PR revisions within the PR itself, as this creates a circular reference (the PR contains a plan to fix the PR).

Consider keeping such plans external to the PR (in a separate branch or working directory) until the work is complete, or mark them more clearly as "meta-documentation" that tracks the evolution of the PR. The current inclusion isn't harmful but may be confusing to reviewers who expect plan documents to describe completed or in-progress work, not future iterations.

# PR53 Unresolved Comments and Main Conflict Resolution Plan

- status: draft
- owner: orchestrator
- work_type: mixed
- created: 2026-02-22

## Goal
Resolve new unresolved PR #53 review comments (must + should as requested) and cleanly merge `origin/main` into `feat/e2e-db-isolation-guardrails` without regressions.

## Definition of Done
- New unresolved **must** comments are fixed:
  - bypass runtime-state write path creates parent runtime dir.
  - teardown termination is reliable for Unix detached/process-group behavior.
- New unresolved **should** comment is addressed:
  - readiness predicate in setup is tightened from broad `<500` behavior.
- Branch is rebased/merged with `origin/main` and has no unresolved conflicts.
- Required validations pass and PR #53 is updated with response notes.

## Scope
- `sleep-ui/tests/global.setup.ts`
- `sleep-ui/tests/global.teardown.ts`
- `sleep-ui/src/routes/login/+page.svelte` (merge conflict)
- PR #53 thread responses and validation evidence

## Non-goals
- No broad E2E harness redesign.
- No unrelated UI/test refactors.
- No backend behavior changes beyond what is needed for comment/conflict resolution.

## Task Waves
- Wave 1 (sequential triage): `Task_1`
- Wave 2 (parallel implementation): `Task_2`, `Task_3`, `Task_4`
- Wave 3 (sequential verification): `Task_5`

## Tasks

### Task_1
- type: research
- title: Confirm unresolved comments and merge conflict targets
- owns:
  - PR #53 review thread metadata
  - merge-base status with `origin/main`
- depends_on: []
- acceptance:
  - unresolved items are classified into must/should.
  - exact conflicted file list is confirmed.
- validation:
  - required: true
    owner: orchestrator
    kind: review
    detail: Classification accepted before edits.

### Task_2
- type: impl
- title: Fix runtime-state bypass path and readiness strictness
- owns:
  - `sleep-ui/tests/global.setup.ts`
- depends_on: [Task_1]
- acceptance:
  - `ALLOW_NON_ISOLATED_E2E=1` path cannot fail from missing runtime dir.
  - readiness probe checks expected successful/known states (not generic `<500`).
- validation:
  - required: true
    owner: worker
    kind: command
    detail: targeted auth smoke passes and unsafe/bypass paths behave as expected.

### Task_3
- type: impl
- title: Harden teardown for detached process cleanup on Unix
- owns:
  - `sleep-ui/tests/global.teardown.ts`
- depends_on: [Task_1]
- acceptance:
  - teardown targets detached process group on Unix where needed.
  - DB cleanup runs with retry only after termination attempts.
- validation:
  - required: true
    owner: worker
    kind: command
    detail: `npm run test:e2e:auth` finishes without leaked API process symptoms.

### Task_4
- type: impl
- title: Resolve merge conflict with origin/main in login page
- owns:
  - `sleep-ui/src/routes/login/+page.svelte`
- depends_on: [Task_1]
- acceptance:
  - preserve both `method="post" action="/api/login"` and incoming `data-testid` form hook.
  - no conflict markers remain.
- validation:
  - required: true
    owner: worker
    kind: command
    detail: `git diff --name-only --diff-filter=U` is empty after merge.

### Task_5
- type: validation
- title: Run required validations and post PR53 updates
- owns:
  - `sleep-ui/**`
- depends_on: [Task_2, Task_3, Task_4]
- acceptance:
  - `npm run check` and `npm run test:unit` pass.
  - `npm run test:e2e:auth` pass.
  - PR #53 comment with resolution mapping is posted.
- validation:
  - required: true
    owner: worker
    kind: command
    detail: `cd sleep-ui && npm run check && npm run test:unit && npm run test:e2e:auth`

## Progress Log
- 2026-02-22: Drafted from Researcher triage of newest unresolved threads and merge conflict.

## Decision Log
- 2026-02-22: Treat two teardown/setup threads as must-fix and readiness strictness as should-fix per user request to address both categories.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.