Files
nisse/nisse-dev/tests/test_control.py
T
dongarandClaude Fable 5 a251104a44
CI / checks (push) Successful in 9s
CI / checks (pull_request) Successful in 9s
Add control loop, gating primitives, and transport interface (A4/B1)
Step-2 PR A: the approval loop end to end with a stub agent, no LLM
and no network.

- transport.py: Transport interface per A1 (send_message,
  send_approval_request, ask_question, on_message) with an
  InMemoryTransport; no platform SDK outside an implementation
- gates.py: PlanRevision with monotonic rev-N tags; Gate.evaluate
  enforces revision binding (stale tags refused with a notice),
  operator allowlisting, and approve-word-only semantics; any other
  reply is rejection feedback. Channel seam ready for Gitea/Vikunja
  approval sources
- control.py: B1 state machine subset with structural gating — the
  agent's only actions are SubmitPlan (ends its turn) and AskHuman;
  no proceed tool exists. Gate state, revision counters, task_id,
  and pending questions checkpoint through StateStore, and boot
  resumes an open gate or gates to AWAIT_INPUT, never blind-resumes
- gate.open/approve/reject A9 events complete the log taxonomy
- CLAUDE.md: record the plan-first working convention
- Tests: 15 unit (gates + orchestrator flows), 2 integration
  (mid-gate restart with revisions intact; unresolved intents gate
  the whole boot)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-09-05 21:28:07 +00:00

156 lines
4.9 KiB
Python

import io
import json
from pathlib import Path
import pytest
from nisse_dev.control import (
AWAIT_ARCH_SIGNOFF,
AWAIT_INPUT,
AWAIT_MILESTONE_SIGNOFF,
EXECUTING,
IDLE,
AskHuman,
Orchestrator,
StubAgent,
SubmitPlan,
)
from nisse_dev.oplog import OpLog
from nisse_dev.state import StateStore
from nisse_dev.transport import InMemoryTransport
class Harness:
def __init__(self, tmp_path: Path, agent: object | None = None) -> None:
self.stream = io.StringIO()
self.store = StateStore(
tmp_path / "state.db", oplog=OpLog("nisse-dev", self.stream)
)
self.transport = InMemoryTransport()
self.orch = Orchestrator(
self.store, self.transport, agent or StubAgent(), operators={"don"}
)
self.orch.start()
@property
def state(self) -> str:
state = self.store.get_state()
assert state is not None
return state.state
def events(self, kind: str) -> list[dict]:
return [
e
for e in map(json.loads, self.stream.getvalue().splitlines())
if e["event"] == kind
]
def outbound_texts(self) -> list[str]:
return [m.text for m in self.transport.outbound]
@pytest.fixture
def h(tmp_path: Path) -> Harness:
return Harness(tmp_path)
def test_boots_idle_and_greets(h: Harness) -> None:
assert h.state == IDLE
assert "ready" in h.outbound_texts()[0]
def test_use_project_submits_arch_plan_and_opens_gate(h: Harness) -> None:
h.transport.receive("don", "use demo")
assert h.state == AWAIT_ARCH_SIGNOFF
request = h.transport.outbound[-1]
assert request.kind == "approval_request"
assert "rev-1" in request.text
[opened] = h.events("gate.open")
assert opened["gate"] == "arch"
assert opened["rev"] == "rev-1"
assert opened["task_id"] is not None
def test_full_happy_path_to_executing(h: Harness) -> None:
h.transport.receive("don", "use demo")
h.transport.receive("don", "approve")
assert h.state == AWAIT_MILESTONE_SIGNOFF
h.transport.receive("don", "approve")
assert h.state == AWAIT_MILESTONE_SIGNOFF # approved, awaiting start
h.transport.receive("don", "start milestone 1")
assert h.state == EXECUTING
approvals = h.events("gate.approve")
assert [e["gate"] for e in approvals] == ["arch", "milestone-1"]
def test_reject_replans_with_feedback_and_bumps_revision(h: Harness) -> None:
h.transport.receive("don", "use demo")
h.transport.receive("don", "too vague, add tests")
assert h.state == AWAIT_ARCH_SIGNOFF
request = h.transport.outbound[-1]
assert "rev-2" in request.text
assert "too vague, add tests" in request.text
[rejected] = h.events("gate.reject")
assert rejected["rev"] == "rev-1"
def test_stale_approval_is_refused_then_current_works(h: Harness) -> None:
h.transport.receive("don", "use demo")
h.transport.receive("don", "needs work") # now at rev-2
h.transport.receive("don", "approve rev-1")
assert h.state == AWAIT_ARCH_SIGNOFF
assert "not applied" in h.outbound_texts()[-1]
assert h.events("gate.approve") == []
h.transport.receive("don", "approve rev-2")
assert h.state == AWAIT_MILESTONE_SIGNOFF
def test_non_operator_messages_are_ignored(h: Harness) -> None:
h.transport.receive("don", "use demo")
before = len(h.transport.outbound)
h.transport.receive("mallory", "approve")
assert h.state == AWAIT_ARCH_SIGNOFF
assert len(h.transport.outbound) == before
def test_start_milestone_requires_approval_first(h: Harness) -> None:
h.transport.receive("don", "use demo")
h.transport.receive("don", "approve")
h.transport.receive("don", "start milestone 1")
assert h.state == AWAIT_MILESTONE_SIGNOFF # it was rejection feedback
h.transport.receive("don", "approve rev-2")
h.transport.receive("don", "start milestone 1")
assert h.state == EXECUTING
def test_status_reports_state_and_gate(h: Harness) -> None:
h.transport.receive("don", "use demo")
h.transport.receive("don", "status")
status = h.outbound_texts()[-1]
assert "AWAIT_ARCH_SIGNOFF" in status
assert "demo" in status
assert "rev-1" in status
class QuestionThenPlan:
def plan(
self,
gate: str,
project: str,
feedback: str | None = None,
answer: str | None = None,
) -> SubmitPlan | AskHuman:
if answer is None:
return AskHuman("monorepo or polyrepo?")
return SubmitPlan(f"[stub] {gate} plan using {answer}")
def test_ask_human_pauses_and_answer_resumes(tmp_path: Path) -> None:
h = Harness(tmp_path, agent=QuestionThenPlan())
h.transport.receive("don", "use demo")
assert h.state == AWAIT_INPUT
assert h.transport.outbound[-1].kind == "question"
h.transport.receive("don", "monorepo")
assert h.state == AWAIT_ARCH_SIGNOFF
assert "monorepo" in h.transport.outbound[-1].text