sprint/WINTEST: functional Pester test suite for Windows engine#369
Closed
muunkky wants to merge 95 commits intoPeonPing:mainfrom
Closed
sprint/WINTEST: functional Pester test suite for Windows engine#369muunkky wants to merge 95 commits intoPeonPing:mainfrom
muunkky wants to merge 95 commits intoPeonPing:mainfrom
Conversation
…erride Completes the config key migration across all components: - Adapters (kilo, opencode): config templates use default_pack - TypeScript plugins: PeonConfig interface uses default_pack with active_pack as optional legacy fallback - install.ps1: all CLI commands and hook runtime read default_pack first with active_pack fallback, regex replacements write default_pack - install.sh: test sound lookup uses default_pack with fallback - hook-handle-use scripts: write session_override instead of agentskill - Skills docs: updated terminology throughout - Tests: updated assertions to match new key names, added legacy fallback test for TypeScript resolveActivePack
Reviewer approved commit 3f5a1f0. Routed executor close-out instructions and 1 planner card (DRY install.ps1 pack resolution) for sprint SMARTPACK.
All acceptance criteria verified as pre-existing: fnmatch-based path_rules matching in peon.sh, config.json template, override hierarchy, CLI commands (bind/unbind/bindings), and 9 BATS tests. peon.ps1 marked N/A (file does not exist in repo).
Verification-only card. All acceptance criteria confirmed as pre-existing in peon.sh, config.json, and BATS tests. No blockers found.
When a path_rule matches the current working directory, `peon status` now displays the matching rule (e.g., `path rule: */work/* -> glados`) in addition to the total count of configured rules.
Remove synchronous MediaPlayer/PresentationCore from peon.ps1 hook and win-play.ps1 to eliminate P0 deadlock caused by WPF dispatcher in headless PowerShell processes. - peon.ps1: replace inline audio block with Start-Process delegation to win-play.ps1 in a detached hidden window - peon.ps1: add 8-second System.Timers.Timer self-timeout before any I/O as safety net against unforeseen blocking - win-play.ps1: keep SoundPlayer for WAV, replace MediaPlayer with CLI player priority chain (ffplay -> mpv -> vlc) for non-WAV - install.ps1: print ffmpeg recommendation post-install if ffplay not found on PATH - Update Pester tests: assert zero MediaPlayer/PresentationCore refs, verify Start-Process delegation, Timer, and CLI player chain
…d5wz2f, review 1)
Review 1 approved at commit 57964e9. Routed executor close-out instructions and 1 backlog card (audio diagnostic logging) to planner.
Add write_state()/read_state() Python helpers (tempfile + os.replace) and Write-StateAtomic/Read-StateWithRetry PowerShell functions (PID-based temp + [System.IO.File]::Move). Replace all raw json.dump/Set-Content state writes and json.load/Get-Content state reads across peon.sh (main block + trainer blocks) and install.ps1 (embedded peon.ps1). Retry-on-read uses 50/100/200ms backoff with graceful fallback to empty defaults on corruption. BATS tests added for corrupted state recovery and concurrent Stop event safety.
Add Pack Selection Hierarchy table documenting the 5-layer override system (session_override > path_rules > pack_rotation > default_pack > hardcoded). Add Per-Project Pack Assignment section with bind/unbind CLI examples and manual config. Add bind/unbind/bindings CLI commands to Chinese README. Update llms.txt with hierarchy and bind/unbind context.
…tall.ps1 Consolidate the repeated default_pack -> active_pack -> "peon" fallback chain into a single Get-ActivePack helper function. Replaces ~10 inline expressions across both the installer script and the embedded peon.ps1 hook with calls to the helper. Also migrates the installer's initial config creation from active_pack to default_pack, aligning with the rename completed in peon.sh.
# Conflicts: # install.ps1 # tests/adapters-windows.Tests.ps1
…ack refactor Worktree merge for z0c9fd used --theirs which reverted HOOKBUG sprint changes (atomic state, audio delegation, safety timer). Restored install.ps1 from pre-merge state and manually applied Get-ActivePack helper extraction. 204/204 Pester tests pass.
APPROVAL routed to executor for close-out. Two BACKLOG items (Write-StateAtomic atomicity, ffplay install guidance) routed to planner for card creation.
Approved at commit 0a67a57. Executor gets close-out instructions. Planner gets 1 BACKLOG card (2 items: atomic state I/O hardening).
- opencode/kilo adapter tests: active_pack -> default_pack - hook-handle-use test: agentskill -> session_override
Add tests/windows-setup.ps1 with reusable helper functions: - Extract-PeonHookScript: extracts peon.ps1 from install.ps1 here-string - New-PeonTestEnvironment: creates isolated temp dir with config, state, mock packs, and mock win-play.ps1 audio logger - Invoke-PeonHook: pipes CESP JSON to peon.ps1 via Process API - New-CespJson, Get-PeonState, Get-PeonConfig, Get-AudioLog helpers Add tests/peon-engine.Tests.ps1 with 25 smoke tests validating the harness infrastructure and core peon.ps1 functional behavior: - Extraction produces valid PS syntax - Test env creates all required files and accepts overrides - SessionStart/Stop events play correct sounds - Disabled config skips audio - Mock win-play.ps1 logs calls without playing real audio Update CI workflow to run both Pester test files.
Add clarifying comment in tests/windows-setup.ps1 explaining that CLAUDE_PEON_DIR and PEON_TEST env vars exist for structural parity with the BATS harness and are not consumed by peon.ps1.
Add tests/windows-setup.ps1 with reusable helper functions: - Extract-PeonHookScript: extracts peon.ps1 from install.ps1 here-string - New-PeonTestEnvironment: creates isolated temp dir with config, state, mock packs, and mock win-play.ps1 audio logger - Invoke-PeonHook: pipes CESP JSON to peon.ps1 via Process API - New-CespJson, Get-PeonState, Get-PeonConfig, Get-AudioLog helpers Add tests/peon-engine.Tests.ps1 with 25 smoke tests validating the harness infrastructure and core peon.ps1 functional behavior: - Extraction produces valid PS syntax - Test env creates all required files and accepts overrides - SessionStart/Stop events play correct sounds - Disabled config skips audio - Mock win-play.ps1 logs calls without playing real audio Update CI workflow to run both Pester test files.
16 integration tests covering: - Pack name input validation (path traversal, shell injection, charset) - Session ID sanitization (malicious IDs fallback to "default") - Config/state mutation correctness (agentskill mode, pack_rotation) - Hook mode vs CLI mode behavior (stdin JSON vs arg) - win-play.ps1 WAV/MP3 branching (SoundPlayer vs CLI players) - Volume clamping at boundaries (0.0, 1.0) - Player priority chain (ffplay -> mpv -> vlc -> silent exit)
15 Pester tests covering the full pack selection override hierarchy: - Default pack fallback (active_pack, empty fallback to "peon") - Session override mode (per-session pack from state, agentskill alias) - Session override fallback (unmatched session, missing pack cleanup) - Default key for Cursor users without conversation_id - Pack rotation (random selection from array, single-pack array) - Edge cases (empty rotation, missing mode key, legacy string format) path_rules tests are deferred as the feature is not yet implemented in peon.ps1 (Windows) -- only exists in peon.sh (Unix).
New test file tests/peon-adapters.Tests.ps1 with 48 tests that actually execute adapter scripts with controlled input and verify JSON output shape. Category A (simple translators): codex, gemini, copilot, windsurf, kiro, openclaw, deepagents -- event mapping verified via mock peon.ps1 stdin capture. Category B (filesystem watchers): amp, antigravity, kimi -- pure functions (Emit-Event, Process-WireLine) extracted and tested in isolation. Category C (structural): deepagents.ps1 added to syntax validation and ExecutionPolicy Bypass checks in adapters-windows.Tests.ps1. Also adds edge case tests (missing peon.ps1, unknown events, no stdin) and CESP JSON shape validation across all Category A adapters.
Implements all test scenarios from card 1dnbzv covering: - Event routing: SessionStart, Stop, PermissionRequest, PostToolUseFailure, SubagentStart, Notification suppression, Cursor camelCase remap - Config behavior: enabled:false, category toggles, volume passthrough, missing config resilience - State management: Stop debounce, no-repeat sound selection, session TTL expiry, corrupted state recovery, empty stdin handling Scenario 14 (spam detection) is skipped due to a production bug in ConvertTo-Hashtable that corrupts prompt_timestamps arrays when reading state back from JSON -- single-element arrays become hashtables, preventing accumulation across invocations.
# Conflicts: # tests/peon-engine.Tests.ps1 # tests/windows-setup.ps1
Adds the new functional adapter test file to the Pester Run.Path array in test.yml so all 48 tests execute in CI on windows-latest.
Scenarios 1 and 7 asserted "agentskill" but hook-handle-use.ps1 sets "session_override". Updated assertions and scenario 7 description to match the actual source behavior. All 16 tests pass.
# Conflicts: # .github/workflows/test.yml
Item A: Replace brittle regex (?<=\d),(?=\d) with InvariantCulture enforcement before ConvertTo-Json. The regex corrupted integer arrays like [1,2,3] -> [1.2.3] on non-English locales. Now we save/restore CurrentCulture around the serialization call. Item B: Anchor here-string extraction on the unique marker comment "# peon-ping hook for Claude Code" inside install.ps1. Previously the regex hookScript = @'(.+?)'@ assumed exactly one here-string, which would silently misextract if a second were added. Card: WINTEST-xk4ymm
# Conflicts: # tests/windows-setup.ps1
Change $config.Run.Path from an explicit array of 5 test files to "tests/" so Pester auto-discovers all *.Tests.ps1 files. This ensures new test files are picked up without CI workflow edits. windows-setup.ps1 and hookbug-integration.ps1 are not *.Tests.ps1 files, so Pester correctly ignores them.
- Step 2A-2D: 4 parallel test cards (event routing, adapters, security, packs) - Step 2.5: harness hardening (locale serialization, extraction regex) - Step 3: CI auto-discovery for all Pester test files - Archived superseded card gtb6dm - Umbrella card j30alo checkboxes updated
|
@muunkky is attempting to deploy a commit to the Gary Sheng's projects Team on Vercel. A member of the Team first needs to authorize it. |
Contributor
Author
|
Superseded by a clean PR that excludes .gitban/ project management content. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Motivation
The Windows hook engine (
peon.ps1embedded ininstall.ps1) had zero functional tests. Structural syntax checks existed inadapters-windows.Tests.ps1, but nothing verified that events actually route to the correct CESP categories, that config toggles suppress sounds, that state management (debounce, no-repeat, spam detection, TTL expiry) works, or that security boundaries hold. The SMARTPACKDEBT fixes were written and reviewed without any way to run them on Windows. This sprint builds the test infrastructure that proves the engine works.What changed
Shared test harness (
tests/windows-setup.ps1)New-PeonTestEnvironmentextracts the embedded peon.ps1 from install.ps1 using AST parsing, creates an isolated temp directory with mockwin-play.ps1(logs calls instead of playing audio), mock pack manifests with real CESP category structure, default config, and empty state. SupportsConfigOverridesandStateOverridesfor per-test customization.Invoke-PeonHookpipes CESP JSON to the extracted script and captures exit code + audio log.New-CespJsongenerates well-formed event payloads.Engine tests (
tests/peon-engine.Tests.ps1— 17 scenarios)Event routing (SessionStart, Stop, PermissionRequest, PostToolUseFailure, SubagentStart), notification suppression (permission_prompt, idle_prompt), Cursor camelCase remapping, config behavior (enabled toggle, category toggles, volume passthrough, missing config resilience), and state management (Stop debounce, no-repeat logic, spam detection threshold, session TTL expiry, corrupted state recovery, empty stdin).
Adapter tests (
tests/peon-adapters.Tests.ps1— 12 scenarios)Functional tests for all 12 Windows PowerShell adapters: each adapter is invoked with a mock event and verified to produce correct CESP JSON output with proper
hook_event_namemapping. Also validates daemon mode flags, FileSystemWatcher usage patterns, and absence ofExecutionPolicy Bypass.Security tests (
tests/peon-security.Tests.ps1— 15 scenarios)hook-handle-use.ps1: pack name injection, path traversal, nonexistent pack handling, CLI vs hook mode behavior.win-play.ps1: path traversal in file argument, volume clamping (negative, >1.0, non-numeric), missing file handling, backend-specific argument validation (ffplay, mpv, vlc, pwsh MediaPlayer).Pack selection tests (
tests/peon-packs.Tests.ps1— 7 scenarios)Pack selection hierarchy: default_pack config, default_pack with active_pack legacy fallback, session_override via state, round-robin rotation, random rotation, session_override priority over rotation. Scenarios 4-7 (path_rules) deferred pending engine port.
CI integration
.github/workflows/test.yml: switched from explicit file list toGet-ChildItem -Filter *.Tests.ps1auto-discovery so new test files are picked up automatically.Fixes found during testing
install.ps1: removed stray backticks from merge artifacts causing parse errorsinstall.ps1: hardened config serialization extraction regex in the test harnessinstall.ps1: fixed staleactive_packtest assertions from SMARTPACK renamesinstall.ps1: fixedsession_overriderotation mode handlingVerification
Real Windows 10 run (PowerShell 5.1):
The 1 skipped test is Scenario 14 (spam detection after 3 rapid UserPromptSubmit events) — blocked by a known
ConvertTo-Hashtablebug where PS 5.1 pipeline semantics unwrap single-element arrays. Tracked as card 8ny6qr.Risks and limitations
Extract-PeonHookScriptwill need updating.Deferred work