qtalker---/backend/tests/test_ws_ritual_judgment.py

"""Workstream B WS integration tests: ritual_start/ritual_step/judgment
handlers in app.ws, plus the favor/essence/at_peace side effects and the
favor read-back bias on trait rolls at mint time."""

import uuid

import pytest
from sqlalchemy import select

import app.judgment as judgment_module
import app.ws
from app.entities import fallback_profile
from app.inventory import (
    CORRECT_JUDGMENT_ESSENCE,
    CROSS_OVER_ESSENCE,
    RITUAL_SUCCESS_ESSENCE,
)
from app.models.entity import Entity
from app.models.user import User
from app.rate_limit import RateLimiter


class FakeSpiritService:
    async def mint_profile(self, signature, channel, anomalies, language="en", entropy=None):
        return fallback_profile(signature)

    async def fragment(self, source, anomaly, language="en"):
        return "listen"

    async def wire_whisper(self, telemetry, language="en"):
        return "the wire hums"

    def chat_stream(self, entity, question, history, language="en"):
        async def gen():
            for token in ["I ", "am ", "here."]:
                yield token

        return gen()

    def ambient_ready(self):
        return False


async def _fake_synth(text, voice, profile, instability=0.0):
    return b"RIFFfake wav bytes"


@pytest.fixture(autouse=True)
def _fake_spirits(monkeypatch):
    monkeypatch.setattr(app.ws, "spirit_service", FakeSpiritService())
    monkeypatch.setattr(app.ws, "synthesize_spirit_voice", _fake_synth)
    # Generous, test-scoped limiters — the module-level ones are shared
    # singletons that accumulate real hit counts across the whole test
    # session (see backend/tests/test_ws_session.py's precedent), and this
    # file summons repeatedly.
    monkeypatch.setattr(app.ws, "summon_limiter", RateLimiter(max_requests=1000, window_seconds=60))
    monkeypatch.setattr(app.ws, "summon_ip_limiter", RateLimiter(max_requests=1000, window_seconds=60))
    monkeypatch.setattr(app.ws, "question_limiter", RateLimiter(max_requests=1000, window_seconds=60))
    monkeypatch.setattr(app.ws, "question_ip_limiter", RateLimiter(max_requests=1000, window_seconds=60))
    monkeypatch.setattr(app.ws, "fragment_limiter", RateLimiter(max_requests=1000, window_seconds=60))
    monkeypatch.setattr(app.ws, "fragment_ip_limiter", RateLimiter(max_requests=1000, window_seconds=60))
    monkeypatch.setattr(app.ws, "ritual_limiter", RateLimiter(max_requests=1000, window_seconds=60))
    monkeypatch.setattr(app.ws, "ritual_ip_limiter", RateLimiter(max_requests=1000, window_seconds=60))
    monkeypatch.setattr(app.ws, "judgment_limiter", RateLimiter(max_requests=1000, window_seconds=60))
    monkeypatch.setattr(app.ws, "judgment_ip_limiter", RateLimiter(max_requests=1000, window_seconds=60))


def _read_until(ws, msg_type, max_frames=60, **match):
    for _ in range(max_frames):
        frame = ws.receive_json()
        if frame.get("type") != msg_type:
            continue
        if all(frame.get(key) == value for key, value in match.items()):
            return frame
    raise AssertionError(f"never saw frame of type {msg_type!r} matching {match!r}")


def _login(sync_client, username):
    sync_client.post("/auth/register", json={"username": username, "password": "spookyspooky"})
    sync_client.post("/auth/login", json={"username": username, "password": "spookyspooky"})
    return sync_client.cookies.get("qm_session")


def _ws_connect(sync_client, token):
    return sync_client.websocket_connect(
        "/ws/session", headers={"cookie": f"qm_session={token}"}
    )


def _summon(ws):
    ws.send_json({"type": "summon"})
    entity_frame = _read_until(ws, "entity")
    _read_until(ws, "utterance", kind="greeting")
    return entity_frame


def _run_ritual(ws, steps=4):
    ws.send_json({"type": "ritual_start"})
    for i in range(1, steps + 1):
        ws.send_json({"type": "ritual_step", "step": i})
    result = _read_until(ws, "ritual_complete")
    # `_handle_ritual_step`'s reward call (essence credit / item roll) runs
    # *after* the `ritual_complete` frame is queued for send, so receiving
    # that frame doesn't by itself guarantee the reward has landed yet (the
    # sender task drains the queue concurrently with the handler's own
    # in-flight awaits). A trailing ping/pong forces a full round trip
    # through the single connection's sequential message loop, which can't
    # read the next message until the ritual_step handler (reward included)
    # has fully returned — so seeing the pong is a hard guarantee, not a
    # poll-and-hope.
    ws.send_json({"type": "ping"})
    _read_until(ws, "pong")
    return result


# --- entity frame never leaks traits ---------------------------------------


@pytest.mark.asyncio
async def test_entity_frame_never_includes_traits(sync_client):
    _login(sync_client, "no-leak")
    with _ws_connect(sync_client, sync_client.cookies.get("qm_session")) as ws:
        _read_until(ws, "session")
        entity_frame = _summon(ws)
        assert "traits" not in entity_frame["entity"]


# --- ritual ------------------------------------------------------------


@pytest.mark.asyncio
async def test_ritual_success_reveals_true_traits_and_credits_essence(
    sync_client, db_session, monkeypatch
):
    monkeypatch.setattr(app.ws.judgment, "roll_ritual_success", lambda traits, rng=None: True)

    token = _login(sync_client, "ritual-winner")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )

    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        _summon(ws)
        result = _run_ritual(ws)

    assert result["success"] is True
    assert result["revealed"] is not None
    assert set(result["revealed"].keys()) == {
        "alignment",
        "power",
        "volatility",
        "deceptiveness",
    }
    for v in result["revealed"].values():
        assert 0.0 <= v <= 1.0

    # trickle (summon) + ritual success milestone. `_run_ritual`'s trailing
    # ping/pong (see its docstring comment) guarantees the reward has fully
    # landed before we get here.
    from app.inventory import SUMMON_ESSENCE_TRICKLE

    db_session.expire_all()
    user = await db_session.get(User, user_id)
    assert user.essence == SUMMON_ESSENCE_TRICKLE + RITUAL_SUCCESS_ESSENCE


@pytest.mark.asyncio
async def test_ritual_failure_reveals_nothing_and_grants_no_essence(
    sync_client, db_session, monkeypatch
):
    monkeypatch.setattr(app.ws.judgment, "roll_ritual_success", lambda traits, rng=None: False)

    token = _login(sync_client, "ritual-loser")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )

    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        _summon(ws)
        result = _run_ritual(ws)

    assert result["success"] is False
    assert result["revealed"] is None

    db_session.expire_all()
    user = await db_session.get(User, user_id)
    from app.inventory import SUMMON_ESSENCE_TRICKLE

    assert user.essence == SUMMON_ESSENCE_TRICKLE


@pytest.mark.asyncio
async def test_ritual_complete_only_fires_after_all_steps(sync_client, monkeypatch):
    monkeypatch.setattr(app.ws.judgment, "roll_ritual_success", lambda traits, rng=None: True)
    _login(sync_client, "ritual-partial")

    with _ws_connect(sync_client, sync_client.cookies.get("qm_session")) as ws:
        _read_until(ws, "session")
        _summon(ws)
        ws.send_json({"type": "ritual_start"})
        ws.send_json({"type": "ritual_step", "step": 1})
        ws.send_json({"type": "ritual_step", "step": 2})
        ws.send_json({"type": "ping"})
        pong = _read_until(ws, "pong")
        assert pong == {"type": "pong"}
        # No ritual_complete should have arrived yet (only 2/4 steps done) —
        # if it had, it'd have been consumed as the "pong" read above
        # skipped it via _read_until's scan, so assert explicitly by
        # finishing the remaining steps and confirming exactly one
        # ritual_complete follows.
        ws.send_json({"type": "ritual_step", "step": 3})
        ws.send_json({"type": "ritual_step", "step": 4})
        result = _read_until(ws, "ritual_complete")
        assert result["success"] is True


@pytest.mark.asyncio
async def test_fresh_summon_resets_ritual_progress(sync_client, monkeypatch):
    monkeypatch.setattr(app.ws.judgment, "roll_ritual_success", lambda traits, rng=None: True)
    _login(sync_client, "ritual-reset")

    with _ws_connect(sync_client, sync_client.cookies.get("qm_session")) as ws:
        _read_until(ws, "session")
        _summon(ws)
        ws.send_json({"type": "ritual_start"})
        ws.send_json({"type": "ritual_step", "step": 1})
        ws.send_json({"type": "ritual_step", "step": 2})
        # A brand-new summon should discard the in-progress ritual — the
        # next 4 steps on the new entity shouldn't complete after only 2
        # more (i.e. carry over the old count).
        _summon(ws)
        ws.send_json({"type": "ritual_start"})
        ws.send_json({"type": "ritual_step", "step": 1})
        ws.send_json({"type": "ritual_step", "step": 2})
        ws.send_json({"type": "ping"})
        pong = _read_until(ws, "pong")
        assert pong == {"type": "pong"}


# --- judgment: trust / banish -----------------------------------------------


@pytest.mark.asyncio
async def test_judgment_trust_correct_on_benevolent_entity(sync_client, db_session, monkeypatch):
    monkeypatch.setattr(
        app.ws.judgment,
        "judge_verdict",
        lambda verdict, traits, **kw: judgment_module.JudgmentOutcome(
            True, 0.05, CORRECT_JUDGMENT_ESSENCE, False, "reward"
        ),
    )
    token = _login(sync_client, "truster")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )

    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        _summon(ws)
        ws.send_json({"type": "judgment", "verdict": "trust"})
        result = _read_until(ws, "judgment_result")

    assert result == {
        "type": "judgment_result",
        "correct": True,
        "favor_delta": 0.05,
        "essence_delta": CORRECT_JUDGMENT_ESSENCE,
        "at_peace": False,
        "consequence": "reward",
    }

    db_session.expire_all()
    user = await db_session.get(User, user_id)
    assert user.favor == pytest.approx(0.05)


@pytest.mark.asyncio
async def test_judgment_wrong_trust_emits_escalation_consequence(sync_client, db_session, monkeypatch):
    monkeypatch.setattr(
        app.ws.judgment,
        "judge_verdict",
        lambda verdict, traits, **kw: judgment_module.JudgmentOutcome(
            False, -0.10, 0, False, "escalation"
        ),
    )
    token = _login(sync_client, "wrong-truster")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )

    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        _summon(ws)
        ws.send_json({"type": "judgment", "verdict": "trust"})
        result = _read_until(ws, "judgment_result")

    assert result["consequence"] == "escalation"
    assert result["correct"] is False
    assert result["favor_delta"] == -0.10

    db_session.expire_all()
    user = await db_session.get(User, user_id)
    assert user.favor == pytest.approx(-0.10)


@pytest.mark.asyncio
async def test_judgment_favor_clamped_at_negative_one(sync_client, db_session, monkeypatch):
    monkeypatch.setattr(
        app.ws.judgment,
        "judge_verdict",
        lambda verdict, traits, **kw: judgment_module.JudgmentOutcome(
            False, -0.10, 0, False, "escalation"
        ),
    )
    token = _login(sync_client, "favor-floor")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )

    async with app.ws.session_maker() as db:
        user = await db.get(User, user_id)
        user.favor = -0.95
        await db.commit()

    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        _summon(ws)
        ws.send_json({"type": "judgment", "verdict": "trust"})
        result = _read_until(ws, "judgment_result")

    assert result["favor_delta"] == -0.10  # the raw per-event delta, unclamped
    db_session.expire_all()
    user = await db_session.get(User, user_id)
    assert user.favor == pytest.approx(-1.0)  # but the stored balance is clamped


@pytest.mark.asyncio
async def test_judgment_favor_clamped_at_positive_one(sync_client, db_session, monkeypatch):
    monkeypatch.setattr(
        app.ws.judgment,
        "judge_verdict",
        lambda verdict, traits, **kw: judgment_module.JudgmentOutcome(
            True, 0.08, CROSS_OVER_ESSENCE, True, "crossed_over"
        ),
    )
    token = _login(sync_client, "favor-ceiling")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )

    async with app.ws.session_maker() as db:
        user = await db.get(User, user_id)
        user.favor = 0.97
        await db.commit()

    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        _summon(ws)
        ws.send_json({"type": "judgment", "verdict": "cross_over"})
        _read_until(ws, "judgment_result")

    db_session.expire_all()
    user = await db_session.get(User, user_id)
    assert user.favor == pytest.approx(1.0)


# --- judgment: cross_over / at_peace ----------------------------------------


@pytest.mark.asyncio
async def test_judgment_cross_over_correct_sets_at_peace_and_pays_most_essence(
    sync_client, db_session, monkeypatch
):
    monkeypatch.setattr(
        app.ws.judgment,
        "judge_verdict",
        lambda verdict, traits, **kw: judgment_module.JudgmentOutcome(
            True, 0.08, CROSS_OVER_ESSENCE, True, "crossed_over"
        ),
    )
    token = _login(sync_client, "crosser")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )

    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        entity_frame = _summon(ws)
        entity_id = uuid.UUID(entity_frame["entity"]["id"])
        ws.send_json({"type": "judgment", "verdict": "cross_over"})
        result = _read_until(ws, "judgment_result")

    assert result["consequence"] == "crossed_over"
    assert result["at_peace"] is True
    assert result["essence_delta"] == CROSS_OVER_ESSENCE

    db_session.expire_all()
    entity = await db_session.get(Entity, entity_id)
    assert entity.at_peace is True

    from app.inventory import SUMMON_ESSENCE_TRICKLE

    user = await db_session.get(User, user_id)
    assert user.essence == SUMMON_ESSENCE_TRICKLE + CROSS_OVER_ESSENCE


@pytest.mark.asyncio
async def test_judgment_cross_over_resisted_on_demon_has_no_effect(sync_client, db_session, monkeypatch):
    monkeypatch.setattr(
        app.ws.judgment,
        "judge_verdict",
        lambda verdict, traits, **kw: judgment_module.JudgmentOutcome(
            False, 0.0, 0, False, "resisted"
        ),
    )
    token = _login(sync_client, "resisted-demon")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )

    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        entity_frame = _summon(ws)
        entity_id = uuid.UUID(entity_frame["entity"]["id"])
        ws.send_json({"type": "judgment", "verdict": "cross_over"})
        result = _read_until(ws, "judgment_result")

    assert result == {
        "type": "judgment_result",
        "correct": False,
        "favor_delta": 0.0,
        "essence_delta": 0,
        "at_peace": False,
        "consequence": "resisted",
    }

    db_session.expire_all()
    entity = await db_session.get(Entity, entity_id)
    assert entity.at_peace is False
    from app.inventory import SUMMON_ESSENCE_TRICKLE

    user = await db_session.get(User, user_id)
    assert user.essence == SUMMON_ESSENCE_TRICKLE
    assert user.favor == 0.0


# --- judgment: test ----------------------------------------------------


@pytest.mark.asyncio
async def test_judgment_test_verdict_is_always_allowed_and_neutral(sync_client, db_session):
    token = _login(sync_client, "tester")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )

    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        _summon(ws)
        # No ritual attempted at all — should not crash, just no effect.
        ws.send_json({"type": "judgment", "verdict": "test"})
        result = _read_until(ws, "judgment_result")

    assert result == {
        "type": "judgment_result",
        "correct": False,
        "favor_delta": 0,
        "essence_delta": 0,
        "at_peace": False,
        "consequence": "neutral",
    }

    db_session.expire_all()
    user = await db_session.get(User, user_id)
    assert user.favor == 0.0


@pytest.mark.asyncio
async def test_judgment_test_verdict_correct_after_successful_ritual(sync_client, monkeypatch):
    monkeypatch.setattr(app.ws.judgment, "roll_ritual_success", lambda traits, rng=None: True)
    _login(sync_client, "tester-after-ritual")

    with _ws_connect(sync_client, sync_client.cookies.get("qm_session")) as ws:
        _read_until(ws, "session")
        _summon(ws)
        _run_ritual(ws)
        ws.send_json({"type": "judgment", "verdict": "test"})
        result = _read_until(ws, "judgment_result")

    assert result["correct"] is True
    assert result["consequence"] == "neutral"
    assert result["favor_delta"] == 0
    assert result["essence_delta"] == 0


# --- at_peace + re-contact ----------------------------------------------


@pytest.mark.asyncio
async def test_at_peace_entity_is_not_recontacted_a_fresh_one_mints_instead(
    sync_client, db_session
):
    anomalies = [
        {"type": "anomaly", "source": "radio", "frequency": 101.0 + i, "magnitude": 5.0 + i}
        for i in range(4)
    ]

    token = _login(sync_client, "peace-seeker")
    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        for anomaly in anomalies:
            ws.send_json(anomaly)
        first_frame = _read_until(ws, "entity")
        first_id = uuid.UUID(first_frame["entity"]["id"])
        first_signature = None  # not exposed to the client; fetched below

    # Manually mark the entity at_peace, as a completed cross_over would.
    async with app.ws.session_maker() as db:
        entity = await db.get(Entity, first_id)
        entity.at_peace = True
        first_signature = entity.signature
        await db.commit()

    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        for anomaly in anomalies:
            ws.send_json(anomaly)
        second_frame = _read_until(ws, "entity")

    assert second_frame["is_new"] is True
    second_id = uuid.UUID(second_frame["entity"]["id"])
    assert second_id != first_id

    db_session.expire_all()
    second_entity = await db_session.get(Entity, second_id)
    assert second_entity.at_peace is False
    # Salted variant of the same base signature, not a raw collision.
    assert second_entity.signature != first_signature
    assert second_entity.signature.startswith(first_signature + ":")

    first_entity = await db_session.get(Entity, first_id)
    assert first_entity is not None  # memorialized, never deleted
    assert first_entity.at_peace is True


# --- favor read-back bias on trait rolls at mint time -----------------------


@pytest.mark.asyncio
async def test_high_favor_user_mints_less_volatile_deceptive_entities(sync_client, db_session):
    token = _login(sync_client, "high-favor-summoner")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )
    async with app.ws.session_maker() as db:
        user = await db.get(User, user_id)
        user.favor = 1.0
        await db.commit()

    anomalies = [
        {"type": "anomaly", "source": "radio", "frequency": 201.0 + i, "magnitude": 5.0 + i}
        for i in range(4)
    ]
    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        for anomaly in anomalies:
            ws.send_json(anomaly)
        entity_frame = _read_until(ws, "entity")

    entity_id = uuid.UUID(entity_frame["entity"]["id"])
    db_session.expire_all()
    entity = await db_session.get(Entity, entity_id)

    # Recompute what the unbiased roll would have been for this signature
    # and confirm the stored traits were nudged toward more legible
    # (lower volatility/deceptiveness), never the other direction.
    from app.entities import roll_traits

    unbiased = roll_traits(entity.signature)
    assert entity.traits["volatility"] <= unbiased["volatility"]
    assert entity.traits["deceptiveness"] <= unbiased["deceptiveness"]
    # alignment/power are untouched by the bias.
    assert entity.traits["alignment"] == pytest.approx(unbiased["alignment"])
    assert entity.traits["power"] == pytest.approx(unbiased["power"])


@pytest.mark.asyncio
async def test_low_favor_user_mints_more_volatile_deceptive_entities(sync_client, db_session):
    token = _login(sync_client, "low-favor-summoner")
    user_id = uuid.UUID(
        sync_client.get("/auth/me", headers={"cookie": f"qm_session={token}"}).json()["id"]
    )
    async with app.ws.session_maker() as db:
        user = await db.get(User, user_id)
        user.favor = -1.0
        await db.commit()

    anomalies = [
        {"type": "anomaly", "source": "radio", "frequency": 301.0 + i, "magnitude": 5.0 + i}
        for i in range(4)
    ]
    with _ws_connect(sync_client, token) as ws:
        _read_until(ws, "session")
        for anomaly in anomalies:
            ws.send_json(anomaly)
        entity_frame = _read_until(ws, "entity")

    entity_id = uuid.UUID(entity_frame["entity"]["id"])
    db_session.expire_all()
    entity = await db_session.get(Entity, entity_id)

    from app.entities import roll_traits

    unbiased = roll_traits(entity.signature)
    assert entity.traits["volatility"] >= unbiased["volatility"]
    assert entity.traits["deceptiveness"] >= unbiased["deceptiveness"]