wpp/weirsoe-party-protocol

Fork 0

Files

Asger Geel Weirsøe d15abf9d78

CI / test-and-quality (pull_request) Successful in 2m46s

Details

CI / test-and-quality (push) Successful in 2m50s

Details

docs: add fupogfakta game engine implementation plan

15 tasks across 8 batches covering:
- Celery infrastructure
- GameRun model + GameDriver interface
- FupOgFaktaConfig relational presets
- LieReaction model, reveal_order, ScoreEvent removal
- Full FupOgFaktaDriver with all phase transitions
- Platform play/pause/exit endpoints
- Fupogfakta lie/guess/react endpoints
- Angular frontend game screens rebuild
- Cleanup of obsolete manual-advance endpoints

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

2026-03-09 07:38:04 +01:00

72 KiB

Raw Permalink Blame History

Fup og Fakta — Game Engine Implementation Plan

For Claude: REQUIRED SUB-SKILL: Use superpowers:executing-plans to implement this plan task-by-task.

Goal: Implement a fully working Fup og Fakta game on top of a pluggable game-platform engine with Celery-driven timers, incremental reveal scoring, emoji reactions, and post-game awards.

Architecture: Platform layer (lobby/) provides GameRun, GameDriver interface, and Celery timer dispatch. Each game cartridge (starting with fupogfakta/) owns its models, config, and phase logic. The Angular frontend gets proper game screens replacing the current developer control panels.

Tech Stack: Django 6.0.2, Celery + Redis, Django Channels (WebSocket already built), Angular 19 standalone components, Vitest.

Reading Before You Start

docs/plans/2026-03-09-fupogfakta-game-engine-design.md — full design, read this first
lobby/views.py — existing REST endpoints (most will be replaced)
fupogfakta/models.py — existing models
realtime/broadcast.py — sync_broadcast_phase_event helper
partyhub/settings.py — env-driven config pattern used throughout

Run tests before starting: .venv/bin/python manage.py test lobby realtime → must be 78/78 green.

Batch 1 — Celery Infrastructure

No breaking changes. Additive only. Existing tests stay green throughout.

Task 1: Add Celery to dependencies

Files:

Modify: requirements.txt

Step 1: Add celery and redis to requirements

celery>=5.3,<6
redis>=5.0,<6

redis is the pure-Python Redis client Celery uses as a broker. (channels-redis uses it too but declares it as a dep — add explicitly for clarity.)

Step 2: Install

.venv/bin/pip install celery>=5.3,<6 redis>=5.0,<6

Expected: installs without errors.

Step 3: Commit

git add requirements.txt
git commit -m "chore: add celery + redis to requirements"

Task 2: Create Celery app

Files:

Create: partyhub/celery.py
Modify: partyhub/__init__.py

Step 1: Create partyhub/celery.py

import os
from celery import Celery

os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'partyhub.settings')

app = Celery('partyhub')
app.config_from_object('django.conf:settings', namespace='CELERY')
app.autodiscover_tasks()

Step 2: Read partyhub/__init__.py first, then update it

Add at the top (after any existing content):

from .celery import app as celery_app

__all__ = ('celery_app',)

Step 3: Add Celery config to partyhub/settings.py

Add after the CHANNEL_LAYERS block:

CELERY_BROKER_URL = f"redis://{env('CELERY_REDIS_HOST', '127.0.0.1')}:{env('CELERY_REDIS_PORT', '6379')}/1"
CELERY_RESULT_BACKEND = CELERY_BROKER_URL
CELERY_TASK_SERIALIZER = 'json'
CELERY_ACCEPT_CONTENT = ['json']
CELERY_TIMEZONE = TIME_ZONE
# In tests: use synchronous task execution (no broker needed)
CELERY_TASK_ALWAYS_EAGER = _testing
CELERY_TASK_EAGER_PROPAGATES = _testing

Step 4: Verify Django check still passes

.venv/bin/python manage.py check

Expected: System check identified no issues (0 silenced).

Step 5: Commit

git add partyhub/celery.py partyhub/__init__.py partyhub/settings.py
git commit -m "feat(platform): add celery app and broker config"

Task 3: Create the timer Celery task (skeleton)

Files:

Create: lobby/tasks.py

Step 1: Write failing test

In lobby/tests.py, add at the bottom:

class TimerTaskStaleGuardTest(TestCase):
    def test_stale_task_does_nothing_when_run_not_found(self):
        from lobby.tasks import handle_timer_expired
        # Should not raise — run_id that does not exist is silently ignored
        handle_timer_expired(run_id=99999, expected_state="LIE_PHASE")

Step 2: Run test to verify it fails

.venv/bin/python manage.py test lobby.tests.TimerTaskStaleGuardTest --verbosity=2

Expected: ImportError or ModuleNotFoundError for lobby.tasks.

Step 3: Create lobby/tasks.py

import logging
from celery import shared_task

logger = logging.getLogger(__name__)


@shared_task(bind=True, max_retries=0)
def handle_timer_expired(self, run_id: int, expected_state: str) -> None:
    """
    Fired by Celery when a phase timer expires.

    Guards against stale tasks: if the GameRun no longer exists or its
    current_state no longer matches expected_state (e.g. after pause/resume),
    the task silently exits.

    Full driver dispatch is wired in Batch 5 once GameRun model exists.
    """
    # Import here to avoid circular imports at module load time
    try:
        from lobby.models import GameRun  # will exist after Batch 2
    except ImportError:
        logger.warning("handle_timer_expired: GameRun model not yet available (pre-migration)")
        return

    try:
        run = GameRun.objects.select_related('session').get(pk=run_id)
    except GameRun.DoesNotExist:
        logger.info("handle_timer_expired: run %s not found — stale task, ignoring", run_id)
        return

    if run.current_state != expected_state:
        logger.info(
            "handle_timer_expired: run %s state is %r, expected %r — stale task, ignoring",
            run_id, run.current_state, expected_state,
        )
        return

    # Full transition logic wired in Batch 5
    logger.info("handle_timer_expired: run %s state=%r — TODO: dispatch driver", run_id, expected_state)

Step 4: Run test to verify it passes

.venv/bin/python manage.py test lobby.tests.TimerTaskStaleGuardTest --verbosity=2

Expected: PASS (ImportError path is handled gracefully).

Step 5: Commit

git add lobby/tasks.py lobby/tests.py
git commit -m "feat(platform): add handle_timer_expired celery task skeleton"

Batch 2 — GameRun Model + GameDriver Interface

Task 4: Add `game_type` field to GameSession and simplify status

Files:

Modify: fupogfakta/models.py
Create: fupogfakta/migrations/0005_gamesession_game_type_status_active.py (auto-generated)

Step 1: Write failing test

In lobby/tests.py, in an appropriate test class or new one:

class GameSessionGameTypeTest(TestCase):
    def setUp(self):
        self.user = User.objects.create_user(username='host2', password='pw')

    def test_game_session_has_game_type_field(self):
        session = GameSession.objects.create(host=self.user, code='GGTYPE')
        self.assertEqual(session.game_type, 'fupogfakta')

    def test_game_session_status_active_exists(self):
        session = GameSession.objects.create(host=self.user, code='ACTSTS')
        session.status = GameSession.Status.ACTIVE
        session.save()
        session.refresh_from_db()
        self.assertEqual(session.status, GameSession.Status.ACTIVE)

Step 2: Run test to verify it fails

.venv/bin/python manage.py test lobby.tests.GameSessionGameTypeTest --verbosity=2

Expected: AttributeError: type object 'GameSession' has no attribute 'game_type'

Step 3: Modify fupogfakta/models.py — add game_type and ACTIVE status

In the GameSession class, add game_type field and ACTIVE to the Status enum:

class Status(models.TextChoices):
    LOBBY = "lobby", "Lobby"
    ACTIVE = "active", "Aktiv"       # ADD THIS — game is running (state detail in GameRun)
    LIE = "lie", "Løgnfase"          # keep for now — removed in Batch 6
    GUESS = "guess", "Gættefase"     # keep for now
    REVEAL = "reveal", "Reveal"      # keep for now
    FINISHED = "finished", "Afsluttet"

And add the field (after current_round):

game_type = models.CharField(max_length=50, default='fupogfakta')

Step 4: Generate and apply migration

.venv/bin/python manage.py makemigrations fupogfakta --name gamesession_game_type_active_status
.venv/bin/python manage.py migrate

Step 5: Run test to verify it passes

.venv/bin/python manage.py test lobby.tests.GameSessionGameTypeTest --verbosity=2

Expected: PASS.

Step 6: Run full test suite — must stay green

.venv/bin/python manage.py test lobby realtime --verbosity=1

Expected: all passing.

Step 7: Commit

git add fupogfakta/models.py fupogfakta/migrations/ lobby/tests.py
git commit -m "feat(platform): add game_type field and ACTIVE status to GameSession"

Task 5: Create `GameDriver` interface

Files:

Create: lobby/driver.py

Step 1: Write failing test

# lobby/tests.py — add to bottom
class GameDriverInterfaceTest(TestCase):
    def test_cannot_instantiate_abstract_driver(self):
        from lobby.driver import GameDriver
        with self.assertRaises(TypeError):
            GameDriver()

    def test_phase_result_is_namedtuple(self):
        from lobby.driver import PhaseResult
        result = PhaseResult(next_state='SOME_STATE', duration_seconds=30, broadcast_payload={'type': 'x'})
        self.assertEqual(result.next_state, 'SOME_STATE')
        self.assertEqual(result.duration_seconds, 30)

Step 2: Run test to verify it fails

.venv/bin/python manage.py test lobby.tests.GameDriverInterfaceTest --verbosity=2

Expected: ModuleNotFoundError: No module named 'lobby.driver'

Step 3: Create lobby/driver.py

from abc import ABC, abstractmethod
from typing import NamedTuple


class PhaseResult(NamedTuple):
    """Return value from GameDriver phase methods."""
    next_state: str
    duration_seconds: int | None   # None = no timer, wait for manual trigger
    broadcast_payload: dict


class GameDriver(ABC):
    """
    Abstract base class for game cartridges.

    Each game app implements this and registers via GAME_DRIVERS in settings.
    The platform (lobby/) calls these methods; game logic never touches
    GameRun directly.
    """

    game_type: str  # must be set on concrete subclass, e.g. "fupogfakta"

    @abstractmethod
    def on_game_start(self, session, run, config: dict) -> PhaseResult:
        """Called when host presses Play from LOBBY. Returns initial phase."""

    @abstractmethod
    def on_timer_expired(self, session, run, config: dict) -> PhaseResult:
        """Called when Celery timer fires for run.current_state. Returns next phase."""

    @abstractmethod
    def on_exit(self, session, run) -> None:
        """
        Called when host presses Exit. Must delete all ephemeral game data
        (RoundQuestion, LieAnswer, Guess, LieReaction, Player, GameRun).
        GameSession row is kept.
        """

    def on_pause(self, session, run) -> None:
        """Default: no-op. Override if game needs to pause internal state."""

    def on_resume(self, session, run) -> None:
        """Default: no-op. Override if game needs to handle resume."""


def get_driver(game_type: str) -> GameDriver:
    """
    Look up and return a registered GameDriver instance by game_type.
    Raises KeyError if game_type is not registered.
    Import and call this from the platform task/views.
    """
    from django.conf import settings
    registry: dict[str, str] = getattr(settings, 'GAME_DRIVERS', {})
    if game_type not in registry:
        raise KeyError(f"No GameDriver registered for game_type={game_type!r}. "
                       f"Add it to GAME_DRIVERS in settings.py.")
    # Lazy import the driver class from dotted path
    module_path, class_name = registry[game_type].rsplit('.', 1)
    import importlib
    module = importlib.import_module(module_path)
    cls = getattr(module, class_name)
    return cls()

Step 4: Register driver path in partyhub/settings.py

Add after the CELERY block:

GAME_DRIVERS = {
    'fupogfakta': 'fupogfakta.driver.FupOgFaktaDriver',
}

Step 5: Run test to verify it passes

.venv/bin/python manage.py test lobby.tests.GameDriverInterfaceTest --verbosity=2

Expected: PASS.

Step 6: Commit

git add lobby/driver.py partyhub/settings.py lobby/tests.py
git commit -m "feat(platform): add GameDriver abstract interface and get_driver registry"

Task 6: Create `GameRun` model

Files:

Create: lobby/models.py (currently empty scaffold)
Create migration

Step 1: Write failing test

# lobby/tests.py
class GameRunModelTest(TestCase):
    def setUp(self):
        from django.contrib.auth import get_user_model
        User = get_user_model()
        self.user = User.objects.create_user(username='runhost', password='pw')
        self.session = GameSession.objects.create(host=self.user, code='RUNTEST')

    def test_create_game_run(self):
        from lobby.models import GameRun
        run = GameRun.objects.create(
            session=self.session,
            current_state='LIE_PHASE',
        )
        self.assertEqual(run.current_state, 'LIE_PHASE')
        self.assertFalse(run.is_paused)
        self.assertIsNone(run.phase_deadline)
        self.assertIsNone(run.celery_task_id)
        self.assertEqual(run.state_data, {})

    def test_game_run_is_one_to_one_with_session(self):
        from lobby.models import GameRun
        from django.db import IntegrityError
        GameRun.objects.create(session=self.session, current_state='LIE_PHASE')
        with self.assertRaises(IntegrityError):
            GameRun.objects.create(session=self.session, current_state='GUESS_PHASE')

Step 2: Run test to verify it fails

.venv/bin/python manage.py test lobby.tests.GameRunModelTest --verbosity=2

Expected: fails — GameRun does not exist.

Step 3: Read and update lobby/models.py

from django.db import models
from django.contrib.auth import get_user_model

User = get_user_model()


class GameRun(models.Model):
    """
    Ephemeral runtime state for an active game session.
    Created when host presses Play. Deleted when game ends (exit or finish).
    """
    session = models.OneToOneField(
        'fupogfakta.GameSession',
        on_delete=models.CASCADE,
        related_name='run',
    )
    current_state = models.CharField(max_length=64)
    phase_deadline = models.DateTimeField(null=True, blank=True)
    is_paused = models.BooleanField(default=False)
    paused_remaining_seconds = models.FloatField(null=True, blank=True)
    celery_task_id = models.CharField(max_length=255, null=True, blank=True)
    # Game-specific snapshot data (config, current question id, etc.)
    state_data = models.JSONField(default=dict, blank=True)
    created_at = models.DateTimeField(auto_now_add=True)

    def __str__(self):
        return f"GameRun({self.session.code}, state={self.current_state})"

Step 4: Generate and apply migration

.venv/bin/python manage.py makemigrations lobby --name add_gamerun
.venv/bin/python manage.py migrate

Step 5: Run test to verify it passes

.venv/bin/python manage.py test lobby.tests.GameRunModelTest --verbosity=2

Expected: PASS.

Step 6: Run full suite

.venv/bin/python manage.py test lobby realtime --verbosity=1

Expected: all green.

Step 7: Commit

git add lobby/models.py lobby/migrations/ lobby/tests.py
git commit -m "feat(platform): add GameRun model"

Batch 3 — Config System

Task 7: BaseGameConfig abstract model + FupOgFaktaConfig

Files:

Create: lobby/base_config.py
Modify: fupogfakta/models.py
Create migration

Step 1: Write failing test

# fupogfakta/tests.py (create if empty)
from django.test import TestCase
from django.contrib.auth import get_user_model

User = get_user_model()


class FupOgFaktaConfigTest(TestCase):
    def setUp(self):
        self.user = User.objects.create_user(username='cfghost', password='pw')

    def test_system_default_config_exists_after_migration(self):
        from fupogfakta.models import FupOgFaktaConfig
        default = FupOgFaktaConfig.objects.filter(user=None, is_default=True).first()
        # System default is created by a data migration (Task 8)
        self.assertIsNotNone(default)
        self.assertEqual(default.num_rounds, 3)
        self.assertEqual(default.lie_seconds, 45)

    def test_user_can_have_multiple_presets(self):
        from fupogfakta.models import FupOgFaktaConfig
        FupOgFaktaConfig.objects.create(user=self.user, name='Quick', is_default=True)
        FupOgFaktaConfig.objects.create(user=self.user, name='Long', is_default=False)
        self.assertEqual(FupOgFaktaConfig.objects.filter(user=self.user).count(), 2)

    def test_resolve_config_returns_user_default_if_exists(self):
        from fupogfakta.config import resolve_config
        user_cfg = FupOgFaktaConfig.objects.create(
            user=self.user, name='Mine', is_default=True, lie_seconds=20
        )
        result = resolve_config(self.user)
        self.assertEqual(result['lie_seconds'], 20)

    def test_resolve_config_falls_back_to_system_default(self):
        from fupogfakta.config import resolve_config
        # No user preset created
        result = resolve_config(self.user)
        self.assertEqual(result['lie_seconds'], 45)  # system default

Step 2: Run test to verify it fails

.venv/bin/python manage.py test fupogfakta --verbosity=2

Expected: module-level failures.

Step 3: Create lobby/base_config.py

from django.db import models
from django.contrib.auth import get_user_model

User = get_user_model()


class BaseGameConfig(models.Model):
    """
    Abstract base for per-user game configuration presets.
    Each game cartridge extends this with its own concrete fields.
    """
    name = models.CharField(max_length=100, default='Default')
    user = models.ForeignKey(
        User,
        on_delete=models.CASCADE,
        null=True,
        blank=True,
        help_text="null = system default (set in Django admin)",
    )
    is_default = models.BooleanField(
        default=False,
        help_text="If True and user=null, this is the system default for this game.",
    )
    created_at = models.DateTimeField(auto_now_add=True)

    class Meta:
        abstract = True

Step 4: Add FupOgFaktaConfig to fupogfakta/models.py

At the bottom of fupogfakta/models.py, add:

from lobby.base_config import BaseGameConfig


class FupOgFaktaConfig(BaseGameConfig):
    num_rounds = models.PositiveIntegerField(default=3)
    questions_per_round = models.PositiveIntegerField(default=3)
    min_players = models.PositiveIntegerField(default=2)
    max_players = models.PositiveIntegerField(default=8)
    lie_seconds = models.PositiveIntegerField(default=45)
    guess_seconds = models.PositiveIntegerField(default=30)
    reveal_seconds_per_lie = models.PositiveIntegerField(default=8)
    scoreboard_recap_seconds = models.PositiveIntegerField(default=10)
    awards_display_seconds = models.PositiveIntegerField(default=15)
    # Escalating scoring: index 0 = round 1, 1 = round 2, etc.
    points_correct = models.JSONField(default=list)
    points_bluff = models.JSONField(default=list)

    def points_correct_for_round(self, round_index: int) -> int:
        defaults = [1500, 3000, 4500]
        pts = self.points_correct or defaults
        return pts[min(round_index, len(pts) - 1)]

    def points_bluff_for_round(self, round_index: int) -> int:
        defaults = [500, 1000, 1500]
        pts = self.points_bluff or defaults
        return pts[min(round_index, len(pts) - 1)]

    class Meta:
        verbose_name = "Fup og Fakta config"
        ordering = ["name"]

    def __str__(self):
        owner = self.user.username if self.user else "SYSTEM"
        return f"{self.name} ({owner})"

Step 5: Create fupogfakta/config.py

from django.contrib.auth import get_user_model

User = get_user_model()

_CONFIG_FIELDS = [
    'num_rounds', 'questions_per_round', 'min_players', 'max_players',
    'lie_seconds', 'guess_seconds', 'reveal_seconds_per_lie',
    'scoreboard_recap_seconds', 'awards_display_seconds',
    'points_correct', 'points_bluff',
]


def resolve_config(user) -> dict:
    """
    Return config dict for a user starting a fupogfakta session.

    Resolution order:
    1. User's is_default=True preset (if any)
    2. System default (user=None, is_default=True)
    3. Model field defaults (hardcoded)
    """
    from fupogfakta.models import FupOgFaktaConfig

    cfg = (
        FupOgFaktaConfig.objects.filter(user=user, is_default=True).first()
        or FupOgFaktaConfig.objects.filter(user=None, is_default=True).first()
    )

    if cfg is None:
        # Fall back to model defaults
        return _model_defaults()

    return {field: getattr(cfg, field) for field in _CONFIG_FIELDS}


def _model_defaults() -> dict:
    from fupogfakta.models import FupOgFaktaConfig
    tmp = FupOgFaktaConfig()
    return {field: getattr(tmp, field) for field in _CONFIG_FIELDS}

Step 6: Generate migration and data migration for system default

.venv/bin/python manage.py makemigrations fupogfakta --name add_fupogfakta_config

Then create the data migration manually:

.venv/bin/python manage.py makemigrations fupogfakta --empty --name seed_system_default_config

Edit the generated file to add:

def seed_default(apps, schema_editor):
    FupOgFaktaConfig = apps.get_model('fupogfakta', 'FupOgFaktaConfig')
    FupOgFaktaConfig.objects.create(
        name='System Default',
        user=None,
        is_default=True,
        num_rounds=3,
        questions_per_round=3,
        min_players=2,
        max_players=8,
        lie_seconds=45,
        guess_seconds=30,
        reveal_seconds_per_lie=8,
        scoreboard_recap_seconds=10,
        awards_display_seconds=15,
        points_correct=[1500, 3000, 4500],
        points_bluff=[500, 1000, 1500],
    )

class Migration(migrations.Migration):
    dependencies = [('fupogfakta', '0005_...')]  # previous migration name
    operations = [migrations.RunPython(seed_default, migrations.RunPython.noop)]

.venv/bin/python manage.py migrate

Step 7: Register in admin — modify fupogfakta/admin.py

from django.contrib import admin
from .models import Category, Question, FupOgFaktaConfig

admin.site.register(Category)
admin.site.register(Question)

@admin.register(FupOgFaktaConfig)
class FupOgFaktaConfigAdmin(admin.ModelAdmin):
    list_display = ['name', 'user', 'is_default', 'num_rounds', 'lie_seconds', 'guess_seconds']
    list_filter = ['is_default', 'user']

Step 8: Run tests

.venv/bin/python manage.py test fupogfakta lobby realtime --verbosity=1

Expected: all green.

Step 9: Commit

git add lobby/base_config.py fupogfakta/models.py fupogfakta/migrations/ \
        fupogfakta/config.py fupogfakta/admin.py fupogfakta/tests.py
git commit -m "feat(config): add FupOgFaktaConfig model, resolve_config, and system default seed"

Batch 4 — FupOgFakta Game Models

Task 8: Add `LieReaction` model + `reveal_order` to RoundQuestion, remove ScoreEvent

Files:

Modify: fupogfakta/models.py
Create migrations

Step 1: Write failing tests

# fupogfakta/tests.py — add

class LieReactionModelTest(TestCase):
    def setUp(self):
        from django.contrib.auth import get_user_model
        User = get_user_model()
        host = User.objects.create_user(username='rhost', password='pw')
        from fupogfakta.models import GameSession, Player, Category, Question, RoundQuestion, LieAnswer
        self.session = GameSession.objects.create(host=host, code='REACT1')
        self.player = Player.objects.create(session=self.session, nickname='Alice')
        self.other = Player.objects.create(session=self.session, nickname='Bob')
        cat = Category.objects.create(name='Test', slug='test')
        q = Question.objects.create(category=cat, prompt='Q?', correct_answer='A')
        self.rq = RoundQuestion.objects.create(session=self.session, round_number=1, question=q, correct_answer='A')
        self.lie = LieAnswer.objects.create(round_question=self.rq, player=self.player, text='Fake')

    def test_player_can_react_to_lie(self):
        from fupogfakta.models import LieReaction
        r = LieReaction.objects.create(lie=self.lie, player=self.other, reaction='laugh')
        self.assertEqual(r.reaction, 'laugh')

    def test_duplicate_reaction_raises(self):
        from fupogfakta.models import LieReaction
        from django.db import IntegrityError
        LieReaction.objects.create(lie=self.lie, player=self.other, reaction='laugh')
        with self.assertRaises(IntegrityError):
            LieReaction.objects.create(lie=self.lie, player=self.other, reaction='laugh')

    def test_round_question_has_reveal_order_field(self):
        from fupogfakta.models import RoundQuestion
        self.assertIsNone(self.rq.reveal_order)

Step 2: Run test to verify it fails

.venv/bin/python manage.py test fupogfakta.tests.LieReactionModelTest --verbosity=2

Step 3: Update fupogfakta/models.py

Remove the entire ScoreEvent class.

Add reveal_order to RoundQuestion:

reveal_order = models.PositiveIntegerField(null=True, blank=True)

Add LieReaction at the bottom:

class LieReaction(models.Model):
    REACTION_CHOICES = [
        ('laugh', '😂 Laugh'),
        ('heart', '❤️ Heart'),
        ('fire', '🔥 Fire'),
        ('wow', '😮 Wow'),
    ]
    lie = models.ForeignKey(LieAnswer, on_delete=models.CASCADE, related_name='reactions')
    player = models.ForeignKey(Player, on_delete=models.CASCADE, related_name='lie_reactions')
    reaction = models.CharField(max_length=20, choices=REACTION_CHOICES)
    created_at = models.DateTimeField(auto_now_add=True)

    class Meta:
        unique_together = [('lie', 'player', 'reaction')]
        ordering = ['created_at']

Step 4: Generate migration

.venv/bin/python manage.py makemigrations fupogfakta --name add_liereaction_reveal_order_remove_scoreevent
.venv/bin/python manage.py migrate

Step 5: Run tests

.venv/bin/python manage.py test fupogfakta lobby realtime --verbosity=1

Expected: all green (ScoreEvent removal will break calculate_scores in lobby/views.py — fix in next step).

Step 6: Fix lobby/views.py — remove ScoreEvent references

In lobby/views.py, the calculate_scores view imports and uses ScoreEvent. Replace the score-saving logic with direct Player.score updates only (no audit model). Remove ScoreEvent from imports.

The score_events list building and ScoreEvent.objects.bulk_create(score_events) lines should be replaced with just the player.save() calls already present. Remove the score_events list and events_created from the response.

Step 7: Run full test suite

.venv/bin/python manage.py test fupogfakta lobby realtime --verbosity=1

Expected: all green.

Step 8: Commit

git add fupogfakta/models.py fupogfakta/migrations/ lobby/views.py fupogfakta/tests.py
git commit -m "feat(fupogfakta): add LieReaction + reveal_order, remove ScoreEvent"

Batch 5 — FupOgFakta Game Driver

This is the core game logic. Take it one state at a time.

Task 9: Driver skeleton + LIE_PHASE entry

Files:

Create: fupogfakta/driver.py
Create: fupogfakta/phases.py (phase transition helpers)

Step 1: Write failing test

# fupogfakta/tests.py — add

class FupOgFaktaDriverStartTest(TestCase):
    def setUp(self):
        from django.contrib.auth import get_user_model
        User = get_user_model()
        self.host = User.objects.create_user(username='drvhost', password='pw')
        from fupogfakta.models import GameSession, Category, Question, FupOgFaktaConfig
        from lobby.models import GameRun
        self.session = GameSession.objects.create(host=self.host, code='DRVST1')
        cat = Category.objects.create(name='DrvCat', slug='drvcat')
        Question.objects.create(category=cat, prompt='Q1?', correct_answer='A1')
        Question.objects.create(category=cat, prompt='Q2?', correct_answer='A2')
        Question.objects.create(category=cat, prompt='Q3?', correct_answer='A3')
        self.run = GameRun.objects.create(session=self.session, current_state='LOBBY')
        self.config = {
            'num_rounds': 1, 'questions_per_round': 1,
            'lie_seconds': 30, 'guess_seconds': 20,
            'reveal_seconds_per_lie': 5, 'scoreboard_recap_seconds': 5,
            'awards_display_seconds': 5,
            'points_correct': [1500], 'points_bluff': [500],
            'min_players': 2, 'max_players': 8,
            'category_slug': 'drvcat',
        }

    def test_on_game_start_returns_lie_phase(self):
        from fupogfakta.driver import FupOgFaktaDriver
        driver = FupOgFaktaDriver()
        result = driver.on_game_start(self.session, self.run, self.config)
        self.assertEqual(result.next_state, 'LIE_PHASE')
        self.assertEqual(result.duration_seconds, 30)
        self.assertIn('prompt', result.broadcast_payload)

    def test_on_game_start_creates_round_question(self):
        from fupogfakta.driver import FupOgFaktaDriver
        from fupogfakta.models import RoundQuestion
        driver = FupOgFaktaDriver()
        driver.on_game_start(self.session, self.run, self.config)
        self.assertEqual(RoundQuestion.objects.filter(session=self.session).count(), 1)

Step 2: Run test to verify it fails

.venv/bin/python manage.py test fupogfakta.tests.FupOgFaktaDriverStartTest --verbosity=2

Step 3: Create fupogfakta/driver.py

import random
import logging

from lobby.driver import GameDriver, PhaseResult

logger = logging.getLogger(__name__)


class FupOgFaktaDriver(GameDriver):
    game_type = 'fupogfakta'

    def on_game_start(self, session, run, config: dict) -> PhaseResult:
        """Select first question and enter LIE_PHASE."""
        from fupogfakta.phases import enter_lie_phase
        return enter_lie_phase(session, run, config, round_number=1, question_number=1)

    def on_timer_expired(self, session, run, config: dict) -> PhaseResult:
        """Dispatch to the correct next-state handler based on current_state."""
        from fupogfakta import phases
        state = run.current_state
        state_data = run.state_data

        if state == 'LIE_PHASE':
            return phases.lie_phase_expired(session, run, config)
        if state == 'GUESS_PHASE':
            return phases.guess_phase_expired(session, run, config)
        if state.startswith('REVEAL_LIE_'):
            return phases.reveal_lie_expired(session, run, config)
        if state == 'REVEAL_TRUTH':
            return phases.reveal_truth_expired(session, run, config)
        if state == 'SCOREBOARD_RECAP':
            return phases.scoreboard_recap_expired(session, run, config)
        if state == 'POST_GAME_AWARDS':
            return phases.post_game_awards_expired(session, run, config)

        logger.error("FupOgFaktaDriver: unknown state %r for run %s", state, run.pk)
        return PhaseResult('FINISHED', None, {'type': 'phase.game_over'})

    def on_exit(self, session, run) -> None:
        """Delete all ephemeral game data. Keep GameSession row."""
        from fupogfakta.models import RoundQuestion, LieAnswer, Guess, LieReaction, Player
        from lobby.models import GameRun
        # Cascade deletes handle LieAnswer, Guess, LieReaction via RoundQuestion
        RoundQuestion.objects.filter(session=session).delete()
        Player.objects.filter(session=session).delete()
        GameRun.objects.filter(session=session).delete()
        session.status = session.Status.FINISHED
        session.save(update_fields=['status'])

Step 4: Create fupogfakta/phases.py

Start with just enter_lie_phase — other phases added in subsequent tasks:

import random
import logging

from lobby.driver import PhaseResult

logger = logging.getLogger(__name__)


def _pick_question(session, config: dict, round_number: int):
    """Pick a random unused question for this session from the configured category."""
    from fupogfakta.models import Category, Question, RoundQuestion
    category = Category.objects.get(slug=config['category_slug'])
    used_ids = RoundQuestion.objects.filter(session=session).values_list('question_id', flat=True)
    available = list(
        Question.objects.filter(category=category, is_active=True).exclude(pk__in=used_ids)
    )
    if not available:
        raise ValueError(f"No available questions for category {config['category_slug']!r}")
    return random.choice(available)


def enter_lie_phase(session, run, config: dict, round_number: int, question_number: int) -> PhaseResult:
    """Select a question, create RoundQuestion, return LIE_PHASE result."""
    from fupogfakta.models import RoundQuestion
    question = _pick_question(session, config, round_number)
    rq = RoundQuestion.objects.create(
        session=session,
        round_number=round_number,
        question=question,
        correct_answer=question.correct_answer,
    )
    run.state_data = {
        **run.state_data,
        'round_number': round_number,
        'question_number': question_number,
        'round_question_id': rq.id,
    }
    run.save(update_fields=['state_data'])
    return PhaseResult(
        next_state='LIE_PHASE',
        duration_seconds=config['lie_seconds'],
        broadcast_payload={
            'type': 'phase.lie_started',
            'round_question_id': rq.id,
            'prompt': question.prompt,
            'round_number': round_number,
            'question_number': question_number,
            'lie_seconds': config['lie_seconds'],
        },
    )


def lie_phase_expired(session, run, config: dict) -> PhaseResult:
    """Timer ran out in LIE_PHASE. Mix answers and enter GUESS_PHASE."""
    from fupogfakta.models import RoundQuestion, LieAnswer
    rq_id = run.state_data['round_question_id']
    rq = RoundQuestion.objects.get(pk=rq_id)

    # Mix answers (dedup, shuffle)
    lie_texts = list(LieAnswer.objects.filter(round_question=rq).values_list('text', flat=True))
    seen = set()
    answers = []
    for text in [rq.correct_answer, *lie_texts]:
        norm = text.strip().casefold()
        if norm and norm not in seen:
            seen.add(norm)
            answers.append(text.strip())
    random.shuffle(answers)
    rq.mixed_answers = answers
    rq.save(update_fields=['mixed_answers'])

    return PhaseResult(
        next_state='GUESS_PHASE',
        duration_seconds=config['guess_seconds'],
        broadcast_payload={
            'type': 'phase.guess_started',
            'round_question_id': rq.id,
            'answers': [{'text': t} for t in answers],
            'guess_seconds': config['guess_seconds'],
        },
    )


def guess_phase_expired(session, run, config: dict) -> PhaseResult:
    """Timer ran out in GUESS_PHASE. Compute reveal order, enter first REVEAL_LIE or REVEAL_TRUTH."""
    from fupogfakta.models import RoundQuestion, LieAnswer, Guess
    rq_id = run.state_data['round_question_id']
    rq = RoundQuestion.objects.prefetch_related('lies', 'guesses').get(pk=rq_id)

    # Determine which lies had at least one guesser
    guessed_lie_ids = set(
        Guess.objects.filter(round_question=rq, is_correct=False)
        .exclude(fooled_player=None)
        .values_list('fooled_player_id', flat=True)
    )
    # Map: liar player_id → lie
    lie_map = {lie.player_id: lie for lie in rq.lies.all()}
    reveal_lies = [lie for pid, lie in lie_map.items() if any(
        g.fooled_player_id == pid for g in rq.guesses.all()
    )]

    # Assign reveal_order to each lie being revealed
    for i, lie in enumerate(reveal_lies):
        lie.reveal_order = i
        lie.save(update_fields=['reveal_order'])

    run.state_data = {
        **run.state_data,
        'reveal_lies': [lie.id for lie in reveal_lies],
        'reveal_index': 0,
    }
    run.save(update_fields=['state_data'])

    if reveal_lies:
        return _reveal_lie_result(rq, reveal_lies[0], config, run.state_data)
    else:
        return _reveal_truth_result(rq, config)


def reveal_lie_expired(session, run, config: dict) -> PhaseResult:
    """A lie has been shown. Score the liar. Move to next lie or truth."""
    from fupogfakta.models import LieAnswer, Guess, Player, RoundQuestion
    rq_id = run.state_data['round_question_id']
    reveal_lies = run.state_data['reveal_lies']
    reveal_index = run.state_data['reveal_index']
    round_number = run.state_data['round_number']

    # Score the liar for this lie
    lie = LieAnswer.objects.get(pk=reveal_lies[reveal_index])
    guessers = list(Guess.objects.filter(round_question_id=rq_id, fooled_player=lie.player).select_related('player'))
    points_bluff = _points_bluff(config, round_number)
    if guessers:
        delta = points_bluff * len(guessers)
        lie.player.score += delta
        lie.player.save(update_fields=['score'])

    # Next reveal
    next_index = reveal_index + 1
    run.state_data = {**run.state_data, 'reveal_index': next_index}
    run.save(update_fields=['state_data'])

    rq = RoundQuestion.objects.get(pk=rq_id)
    if next_index < len(reveal_lies):
        next_lie = LieAnswer.objects.get(pk=reveal_lies[next_index])
        return _reveal_lie_result(rq, next_lie, config, run.state_data)
    else:
        return _reveal_truth_result(rq, config)


def reveal_truth_expired(session, run, config: dict) -> PhaseResult:
    """Truth revealed. Score correct guessers. Enter SCOREBOARD_RECAP."""
    from fupogfakta.models import Guess, RoundQuestion
    rq_id = run.state_data['round_question_id']
    round_number = run.state_data['round_number']
    rq = RoundQuestion.objects.get(pk=rq_id)

    correct_guessers = list(Guess.objects.filter(round_question=rq, is_correct=True).select_related('player'))
    points_correct = _points_correct(config, round_number)
    for guess in correct_guessers:
        guess.player.score += points_correct
        guess.player.save(update_fields=['score'])

    leaderboard = _leaderboard(session)
    return PhaseResult(
        next_state='SCOREBOARD_RECAP',
        duration_seconds=config['scoreboard_recap_seconds'],
        broadcast_payload={
            'type': 'phase.scoreboard',
            'leaderboard': leaderboard,
            'round_number': round_number,
        },
    )


def scoreboard_recap_expired(session, run, config: dict) -> PhaseResult:
    """Scoreboard shown. Advance to next question, next round, or post-game awards."""
    round_number = run.state_data['round_number']
    question_number = run.state_data['question_number']
    num_rounds = config['num_rounds']
    questions_per_round = config['questions_per_round']

    if question_number < questions_per_round:
        # Next question in same round
        return enter_lie_phase(session, run, config, round_number, question_number + 1)
    elif round_number < num_rounds:
        # Next round
        return enter_lie_phase(session, run, config, round_number + 1, 1)
    else:
        # All rounds done → post-game awards
        return _post_game_awards_result(session, config)


def post_game_awards_expired(session, run, config: dict) -> PhaseResult:
    """Awards shown. Game over."""
    return PhaseResult(
        next_state='FINISHED',
        duration_seconds=None,
        broadcast_payload={
            'type': 'phase.game_over',
            'leaderboard': _leaderboard(session),
        },
    )


# ── helpers ──────────────────────────────────────────────────────────────────

def _points_correct(config: dict, round_number: int) -> int:
    pts = config.get('points_correct') or [1500, 3000, 4500]
    return pts[min(round_number - 1, len(pts) - 1)]


def _points_bluff(config: dict, round_number: int) -> int:
    pts = config.get('points_bluff') or [500, 1000, 1500]
    return pts[min(round_number - 1, len(pts) - 1)]


def _leaderboard(session) -> list:
    from fupogfakta.models import Player
    return list(
        Player.objects.filter(session=session)
        .order_by('-score', 'nickname')
        .values('id', 'nickname', 'score')
    )


def _reveal_lie_result(rq, lie, config: dict, state_data: dict) -> PhaseResult:
    from fupogfakta.models import Guess
    guessers = list(
        Guess.objects.filter(round_question=rq, fooled_player=lie.player)
        .select_related('player')
        .values('player__nickname')
    )
    return PhaseResult(
        next_state=f'REVEAL_LIE_{state_data["reveal_index"]}',
        duration_seconds=config['reveal_seconds_per_lie'],
        broadcast_payload={
            'type': 'phase.reveal_lie',
            'lie_text': lie.text,
            'author': lie.player.nickname,
            'guessers': [g['player__nickname'] for g in guessers],
            'guesser_count': len(guessers),
        },
    )


def _reveal_truth_result(rq, config: dict) -> PhaseResult:
    from fupogfakta.models import Guess
    correct_guessers = list(
        Guess.objects.filter(round_question=rq, is_correct=True)
        .select_related('player')
        .values('player__nickname')
    )
    return PhaseResult(
        next_state='REVEAL_TRUTH',
        duration_seconds=config['reveal_seconds_per_lie'],
        broadcast_payload={
            'type': 'phase.reveal_truth',
            'correct_answer': rq.correct_answer,
            'correct_guessers': [g['player__nickname'] for g in correct_guessers],
        },
    )


def _post_game_awards_result(session, config: dict) -> PhaseResult:
    from fupogfakta.models import LieReaction, LieAnswer
    from django.db.models import Count

    # Most laughs overall per player
    top_laugh = (
        LieReaction.objects.filter(lie__round_question__session=session, reaction='laugh')
        .values('lie__player__nickname')
        .annotate(total=Count('id'))
        .order_by('-total')
        .first()
    )
    # Most hearts on a single lie
    top_heart = (
        LieReaction.objects.filter(lie__round_question__session=session, reaction='heart')
        .values('lie__player__nickname', 'lie__text')
        .annotate(total=Count('id'))
        .order_by('-total')
        .first()
    )

    awards = []
    if top_laugh:
        awards.append({
            'award': 'most_hilarious',
            'label': 'Most Hilarious Liar 😂',
            'winner': top_laugh['lie__player__nickname'],
            'count': top_laugh['total'],
        })
    if top_heart:
        awards.append({
            'award': 'most_beloved',
            'label': 'Most Beloved Lie ❤️',
            'winner': top_heart['lie__player__nickname'],
            'lie': top_heart['lie__text'],
            'count': top_heart['total'],
        })

    return PhaseResult(
        next_state='POST_GAME_AWARDS',
        duration_seconds=config['awards_display_seconds'],
        broadcast_payload={
            'type': 'phase.awards',
            'awards': awards,
            'leaderboard': _leaderboard(session),
        },
    )

Step 5: Run tests

.venv/bin/python manage.py test fupogfakta.tests.FupOgFaktaDriverStartTest --verbosity=2

Expected: PASS.

Step 6: Run full suite

.venv/bin/python manage.py test fupogfakta lobby realtime --verbosity=1

Expected: all green.

Step 7: Commit

git add fupogfakta/driver.py fupogfakta/phases.py fupogfakta/tests.py
git commit -m "feat(fupogfakta): implement game driver and all phase transitions"

Batch 6 — Platform REST Endpoints + Celery Wiring

Task 10: Wire `handle_timer_expired` to the driver

Files:

Modify: lobby/tasks.py

Step 1: Write failing test

# lobby/tests.py — add

class TimerTaskDispatchTest(TestCase):
    def setUp(self):
        from django.contrib.auth import get_user_model
        User = get_user_model()
        self.host = User.objects.create_user(username='taskhost', password='pw')
        from fupogfakta.models import GameSession, Category, Question, Player
        from lobby.models import GameRun
        self.session = GameSession.objects.create(host=self.host, code='TASKTS')
        self.cat = Category.objects.create(name='TskCat', slug='tskcat')
        for i in range(3):
            Question.objects.create(category=self.cat, prompt=f'Q{i}', correct_answer=f'A{i}')
        Player.objects.create(session=self.session, nickname='P1')
        Player.objects.create(session=self.session, nickname='P2')
        self.run = GameRun.objects.create(
            session=self.session,
            current_state='LIE_PHASE',
            state_data={
                'round_number': 1, 'question_number': 1,
                'round_question_id': None,  # set below
                'config': {
                    'num_rounds': 1, 'questions_per_round': 1,
                    'lie_seconds': 30, 'guess_seconds': 20,
                    'reveal_seconds_per_lie': 5, 'scoreboard_recap_seconds': 5,
                    'awards_display_seconds': 5,
                    'points_correct': [1500], 'points_bluff': [500],
                    'min_players': 2, 'max_players': 8,
                    'category_slug': 'tskcat',
                },
            },
        )
        from fupogfakta.models import RoundQuestion, Question as Q
        q = Q.objects.filter(category=self.cat).first()
        rq = RoundQuestion.objects.create(session=self.session, round_number=1, question=q, correct_answer=q.correct_answer)
        self.run.state_data['round_question_id'] = rq.id
        self.run.save(update_fields=['state_data'])
        self.rq = rq

    def test_task_transitions_lie_to_guess(self):
        from lobby.tasks import handle_timer_expired
        from lobby.models import GameRun
        handle_timer_expired(run_id=self.run.pk, expected_state='LIE_PHASE')
        self.run.refresh_from_db()
        self.assertEqual(self.run.current_state, 'GUESS_PHASE')

    def test_stale_task_does_not_transition(self):
        from lobby.tasks import handle_timer_expired
        from lobby.models import GameRun
        handle_timer_expired(run_id=self.run.pk, expected_state='WRONG_STATE')
        self.run.refresh_from_db()
        self.assertEqual(self.run.current_state, 'LIE_PHASE')  # unchanged

Step 2: Update lobby/tasks.py — full implementation

import logging
from celery import shared_task
from asgiref.sync import async_to_sync
from channels.layers import get_channel_layer

logger = logging.getLogger(__name__)


def _schedule_next(run, duration_seconds: int) -> str:
    """Schedule handle_timer_expired for run after duration_seconds. Returns task id."""
    from datetime import timedelta
    from django.utils import timezone
    eta = timezone.now() + timedelta(seconds=duration_seconds)
    result = handle_timer_expired.apply_async(
        kwargs={'run_id': run.pk, 'expected_state': run.current_state},
        eta=eta,
    )
    return result.id


def _broadcast(session_code: str, payload: dict) -> None:
    channel_layer = get_channel_layer()
    group_name = f"game_{session_code.upper()}"
    async_to_sync(channel_layer.group_send)(
        group_name,
        {'type': 'phase_event', 'payload': payload},
    )


@shared_task(bind=True, max_retries=0)
def handle_timer_expired(self, run_id: int, expected_state: str) -> None:
    from lobby.models import GameRun
    from lobby.driver import get_driver
    from django.utils import timezone

    try:
        run = GameRun.objects.select_related('session').get(pk=run_id)
    except GameRun.DoesNotExist:
        logger.info("handle_timer_expired: run %s not found — stale task", run_id)
        return

    if run.current_state != expected_state:
        logger.info("handle_timer_expired: run %s state mismatch (%r != %r) — stale", run_id, run.current_state, expected_state)
        return

    if run.is_paused:
        logger.info("handle_timer_expired: run %s is paused — ignoring", run_id)
        return

    config = run.state_data.get('config', {})
    driver = get_driver(run.session.game_type)

    if run.current_state == 'FINISHED':
        # Clean up
        driver.on_exit(run.session, run)
        return

    result = driver.on_timer_expired(run.session, run, config)

    if result.next_state == 'FINISHED':
        _broadcast(run.session.code, result.broadcast_payload)
        driver.on_exit(run.session, run)
        return

    run.current_state = result.next_state
    run.phase_deadline = timezone.now() + __import__('datetime').timedelta(seconds=result.duration_seconds) if result.duration_seconds else None
    run.celery_task_id = None
    run.save(update_fields=['current_state', 'phase_deadline', 'celery_task_id', 'state_data'])

    if result.duration_seconds:
        task_id = _schedule_next(run, result.duration_seconds)
        run.celery_task_id = task_id
        run.save(update_fields=['celery_task_id'])

    _broadcast(run.session.code, result.broadcast_payload)

Step 3: Run tests

.venv/bin/python manage.py test lobby.tests.TimerTaskDispatchTest --verbosity=2

Expected: PASS (CELERY_TASK_ALWAYS_EAGER=True in test mode).

Step 4: Commit

git add lobby/tasks.py lobby/tests.py
git commit -m "feat(platform): wire handle_timer_expired to GameDriver dispatch"

Task 11: Platform play/pause/exit endpoints

Files:

Modify: lobby/views.py — add play, pause, exit views
Modify: lobby/urls.py — register routes

Step 1: Write failing tests

# lobby/tests.py — add

class PlayPauseExitTest(TestCase):
    def setUp(self):
        from django.contrib.auth import get_user_model
        User = get_user_model()
        self.host = User.objects.create_user(username='ppe_host', password='pw')
        self.client.force_login(self.host)
        from fupogfakta.models import GameSession, Category, Question, Player
        from fupogfakta.models import FupOgFaktaConfig
        self.session = GameSession.objects.create(host=self.host, code='PPETEST')
        cat = Category.objects.create(name='PPECat', slug='ppecat')
        for i in range(3):
            Question.objects.create(category=cat, prompt=f'Q{i}', correct_answer=f'A{i}')
        Player.objects.create(session=self.session, nickname='P1')
        Player.objects.create(session=self.session, nickname='P2')

    def test_play_creates_game_run_and_transitions_to_active(self):
        from lobby.models import GameRun
        resp = self.client.post(f'/lobby/sessions/PPETEST/play',
                                content_type='application/json',
                                data={'category_slug': 'ppecat'})
        self.assertEqual(resp.status_code, 201)
        self.assertTrue(GameRun.objects.filter(session__code='PPETEST').exists())

    def test_pause_sets_is_paused(self):
        from lobby.models import GameRun
        self.client.post(f'/lobby/sessions/PPETEST/play',
                         content_type='application/json',
                         data={'category_slug': 'ppecat'})
        resp = self.client.post('/lobby/sessions/PPETEST/pause')
        self.assertEqual(resp.status_code, 200)
        run = GameRun.objects.get(session__code='PPETEST')
        self.assertTrue(run.is_paused)

    def test_exit_deletes_game_run(self):
        from lobby.models import GameRun
        self.client.post('/lobby/sessions/PPETEST/play',
                         content_type='application/json',
                         data={'category_slug': 'ppecat'})
        resp = self.client.post('/lobby/sessions/PPETEST/exit')
        self.assertEqual(resp.status_code, 200)
        self.assertFalse(GameRun.objects.filter(session__code='PPETEST').exists())

Step 2: Add views to lobby/views.py

Add these three views at the bottom:

@require_POST
@login_required
def play_session(request: HttpRequest, code: str) -> JsonResponse:
    """Start game (or resume if paused)."""
    from lobby.models import GameRun
    from lobby.driver import get_driver
    from lobby.tasks import _schedule_next, _broadcast
    from fupogfakta.config import resolve_config
    from django.utils import timezone

    session_code = _normalize_session_code(code)
    try:
        session = GameSession.objects.get(code=session_code)
    except GameSession.DoesNotExist:
        return JsonResponse({'error': 'Session not found'}, status=404)

    if session.host_id != request.user.id:
        return JsonResponse({'error': 'Only host can start game'}, status=403)

    # Resume from pause
    run = GameRun.objects.filter(session=session).first()
    if run and run.is_paused:
        remaining = run.paused_remaining_seconds or 0
        run.phase_deadline = timezone.now() + __import__('datetime').timedelta(seconds=remaining)
        run.is_paused = False
        run.paused_remaining_seconds = None
        run.save(update_fields=['phase_deadline', 'is_paused', 'paused_remaining_seconds'])
        task_id = _schedule_next(run, int(remaining))
        run.celery_task_id = task_id
        run.save(update_fields=['celery_task_id'])
        _broadcast(session.code, {'type': 'phase.resumed', 'phase_deadline': run.phase_deadline.isoformat()})
        return JsonResponse({'status': 'resumed'})

    if run:
        return JsonResponse({'error': 'Game already running'}, status=409)

    # Fresh start
    payload = _json_body(request)
    category_slug = str(payload.get('category_slug', '')).strip()
    if not category_slug:
        return JsonResponse({'error': 'category_slug required'}, status=400)

    config = resolve_config(request.user)
    config['category_slug'] = category_slug

    run = GameRun.objects.create(
        session=session,
        current_state='LOBBY',
        state_data={'config': config},
    )

    driver = get_driver(session.game_type)
    result = driver.on_game_start(session, run, config)

    run.current_state = result.next_state
    run.state_data['config'] = config
    run.save(update_fields=['current_state', 'state_data'])

    if result.duration_seconds:
        run.phase_deadline = timezone.now() + __import__('datetime').timedelta(seconds=result.duration_seconds)
        task_id = _schedule_next(run, result.duration_seconds)
        run.celery_task_id = task_id
        run.save(update_fields=['phase_deadline', 'celery_task_id'])

    session.status = GameSession.Status.ACTIVE
    session.save(update_fields=['status'])

    _broadcast(session.code, result.broadcast_payload)

    return JsonResponse({'status': 'started', 'state': run.current_state}, status=201)


@require_POST
@login_required
def pause_session(request: HttpRequest, code: str) -> JsonResponse:
    from lobby.models import GameRun
    from lobby.tasks import _broadcast
    from django.utils import timezone
    from celery.app.control import Control

    session_code = _normalize_session_code(code)
    try:
        session = GameSession.objects.get(code=session_code)
    except GameSession.DoesNotExist:
        return JsonResponse({'error': 'Session not found'}, status=404)

    if session.host_id != request.user.id:
        return JsonResponse({'error': 'Only host can pause'}, status=403)

    try:
        run = GameRun.objects.get(session=session)
    except GameRun.DoesNotExist:
        return JsonResponse({'error': 'Game not running'}, status=400)

    if run.is_paused:
        return JsonResponse({'error': 'Already paused'}, status=400)

    # Revoke Celery task
    if run.celery_task_id:
        from partyhub.celery import app as celery_app
        celery_app.control.revoke(run.celery_task_id, terminate=True)

    remaining = (run.phase_deadline - timezone.now()).total_seconds() if run.phase_deadline else 0
    run.is_paused = True
    run.paused_remaining_seconds = max(remaining, 0)
    run.celery_task_id = None
    run.save(update_fields=['is_paused', 'paused_remaining_seconds', 'celery_task_id'])

    _broadcast(session.code, {'type': 'phase.paused', 'remaining_seconds': run.paused_remaining_seconds})
    return JsonResponse({'status': 'paused', 'remaining_seconds': run.paused_remaining_seconds})


@require_POST
@login_required
def exit_session(request: HttpRequest, code: str) -> JsonResponse:
    from lobby.models import GameRun
    from lobby.driver import get_driver
    from celery.app.control import Control

    session_code = _normalize_session_code(code)
    try:
        session = GameSession.objects.get(code=session_code)
    except GameSession.DoesNotExist:
        return JsonResponse({'error': 'Session not found'}, status=404)

    if session.host_id != request.user.id:
        return JsonResponse({'error': 'Only host can exit'}, status=403)

    run = GameRun.objects.filter(session=session).first()
    if run:
        if run.celery_task_id:
            from partyhub.celery import app as celery_app
            celery_app.control.revoke(run.celery_task_id, terminate=True)
        driver = get_driver(session.game_type)
        driver.on_exit(session, run)

    return JsonResponse({'status': 'exited'})

Step 3: Add routes to lobby/urls.py

path("sessions/<str:code>/play", views.play_session, name="play_session"),
path("sessions/<str:code>/pause", views.pause_session, name="pause_session"),
path("sessions/<str:code>/exit", views.exit_session, name="exit_session"),

Step 4: Run tests

.venv/bin/python manage.py test lobby.tests.PlayPauseExitTest --verbosity=2

Expected: PASS.

Step 5: Run full suite

.venv/bin/python manage.py test fupogfakta lobby realtime --verbosity=1

Step 6: Commit

git add lobby/views.py lobby/urls.py lobby/tests.py
git commit -m "feat(platform): add play/pause/exit session endpoints"

Task 12: FupOgFakta game-specific endpoints (lie, guess, react)

Files:

Create: fupogfakta/views.py
Create: fupogfakta/urls.py
Modify: partyhub/urls.py

Step 1: Write failing tests

# fupogfakta/tests.py — add

class SubmitLieViewTest(TestCase):
    def setUp(self):
        # Create session in LIE_PHASE with a RoundQuestion active
        from django.contrib.auth import get_user_model
        User = get_user_model()
        self.host = User.objects.create_user(username='liehost', password='pw')
        from fupogfakta.models import GameSession, Player, Category, Question, RoundQuestion
        from lobby.models import GameRun
        self.session = GameSession.objects.create(host=self.host, code='LIETEST', status='active')
        cat = Category.objects.create(name='LieCat', slug='liecat')
        q = Question.objects.create(category=cat, prompt='What is?', correct_answer='Real Answer')
        self.rq = RoundQuestion.objects.create(session=self.session, round_number=1, question=q, correct_answer='Real Answer')
        self.player = Player.objects.create(session=self.session, nickname='Liar')
        self.run = GameRun.objects.create(
            session=self.session, current_state='LIE_PHASE',
            state_data={'round_question_id': self.rq.id, 'config': {'lie_seconds': 45}}
        )

    def test_submit_lie_succeeds(self):
        resp = self.client.post(
            '/fupogfakta/LIETEST/lie',
            content_type='application/json',
            data={'session_token': self.player.session_token, 'text': 'My fake answer'},
        )
        self.assertEqual(resp.status_code, 201)

    def test_submit_correct_answer_as_lie_rejected(self):
        resp = self.client.post(
            '/fupogfakta/LIETEST/lie',
            content_type='application/json',
            data={'session_token': self.player.session_token, 'text': 'real answer'},
        )
        self.assertEqual(resp.status_code, 422)
        self.assertIn('lie_matches_correct_answer', resp.json().get('error_code', ''))

Step 2: Create fupogfakta/views.py

import json
from django.http import HttpRequest, JsonResponse
from django.views.decorators.http import require_POST

from fupogfakta.models import GameSession, Player, RoundQuestion, LieAnswer, Guess, LieReaction
from lobby.models import GameRun


def _json_body(request: HttpRequest) -> dict:
    try:
        return json.loads(request.body) if request.body else {}
    except json.JSONDecodeError:
        return {}


def _get_active_run_and_player(code: str, session_token: str):
    """Returns (session, run, player) or raises."""
    try:
        session = GameSession.objects.get(code=code.upper())
    except GameSession.DoesNotExist:
        return None, None, None
    try:
        run = GameRun.objects.get(session=session)
    except GameRun.DoesNotExist:
        return session, None, None
    try:
        player = Player.objects.get(session=session, session_token=session_token)
    except Player.DoesNotExist:
        return session, run, None
    return session, run, player


@require_POST
def submit_lie(request: HttpRequest, code: str) -> JsonResponse:
    payload = _json_body(request)
    session_token = str(payload.get('session_token', '')).strip()
    lie_text = str(payload.get('text', '')).strip()

    if not session_token:
        return JsonResponse({'error': 'session_token required'}, status=400)
    if not lie_text or len(lie_text) > 255:
        return JsonResponse({'error': 'text must be 1-255 characters'}, status=400)

    session, run, player = _get_active_run_and_player(code, session_token)
    if not session:
        return JsonResponse({'error': 'Session not found'}, status=404)
    if not run:
        return JsonResponse({'error': 'Game not running'}, status=400)
    if not player:
        return JsonResponse({'error': 'Invalid session_token'}, status=403)
    if run.current_state != 'LIE_PHASE':
        return JsonResponse({'error': 'Not in lie phase'}, status=400)

    rq_id = run.state_data.get('round_question_id')
    try:
        rq = RoundQuestion.objects.get(pk=rq_id, session=session)
    except RoundQuestion.DoesNotExist:
        return JsonResponse({'error': 'No active question'}, status=400)

    # Reject if lie matches correct answer (case-insensitive)
    if lie_text.strip().casefold() == rq.correct_answer.strip().casefold():
        return JsonResponse({
            'error': 'Your lie matches the correct answer — try something else!',
            'error_code': 'lie_matches_correct_answer',
        }, status=422)

    from django.db import IntegrityError
    try:
        lie = LieAnswer.objects.create(round_question=rq, player=player, text=lie_text)
    except IntegrityError:
        return JsonResponse({'error': 'Lie already submitted'}, status=409)

    # Broadcast progress (anonymous count only)
    from realtime.broadcast import sync_broadcast_phase_event
    total_players = Player.objects.filter(session=session).count()
    submitted = LieAnswer.objects.filter(round_question=rq).count()
    sync_broadcast_phase_event(session.code, 'phase.lie_progress', {
        'submitted': submitted,
        'total': total_players,
    })

    return JsonResponse({'lie_id': lie.id}, status=201)


@require_POST
def submit_guess(request: HttpRequest, code: str) -> JsonResponse:
    payload = _json_body(request)
    session_token = str(payload.get('session_token', '')).strip()
    selected_text = str(payload.get('selected_text', '')).strip()

    if not session_token:
        return JsonResponse({'error': 'session_token required'}, status=400)
    if not selected_text:
        return JsonResponse({'error': 'selected_text required'}, status=400)

    session, run, player = _get_active_run_and_player(code, session_token)
    if not session:
        return JsonResponse({'error': 'Session not found'}, status=404)
    if not run:
        return JsonResponse({'error': 'Game not running'}, status=400)
    if not player:
        return JsonResponse({'error': 'Invalid session_token'}, status=403)
    if run.current_state != 'GUESS_PHASE':
        return JsonResponse({'error': 'Not in guess phase'}, status=400)

    rq_id = run.state_data.get('round_question_id')
    rq = RoundQuestion.objects.get(pk=rq_id, session=session)

    allowed = {t.strip().casefold() for t in rq.mixed_answers}
    if selected_text.casefold() not in allowed:
        return JsonResponse({'error': 'Answer not in this round'}, status=400)

    correct_normalized = rq.correct_answer.strip().casefold()
    is_correct = selected_text.casefold() == correct_normalized

    fooled_player_id = None
    if not is_correct:
        fooled_player_id = (
            LieAnswer.objects.filter(round_question=rq, text__iexact=selected_text)
            .values_list('player_id', flat=True)
            .first()
        )

    from django.db import IntegrityError
    try:
        guess = Guess.objects.create(
            round_question=rq,
            player=player,
            selected_text=selected_text,
            is_correct=is_correct,
            fooled_player_id=fooled_player_id,
        )
    except IntegrityError:
        return JsonResponse({'error': 'Guess already submitted'}, status=409)

    return JsonResponse({'guess_id': guess.id, 'is_correct': is_correct}, status=201)


@require_POST
def submit_reaction(request: HttpRequest, code: str) -> JsonResponse:
    payload = _json_body(request)
    session_token = str(payload.get('session_token', '')).strip()
    lie_id = payload.get('lie_id')
    reaction = str(payload.get('reaction', '')).strip()

    VALID_REACTIONS = {'laugh', 'heart', 'fire', 'wow'}
    if not session_token:
        return JsonResponse({'error': 'session_token required'}, status=400)
    if reaction not in VALID_REACTIONS:
        return JsonResponse({'error': f'reaction must be one of {sorted(VALID_REACTIONS)}'}, status=400)

    session, run, player = _get_active_run_and_player(code, session_token)
    if not session:
        return JsonResponse({'error': 'Session not found'}, status=404)
    if not run or run.current_state != 'GUESS_PHASE':
        return JsonResponse({'error': 'Reactions only allowed during guess phase'}, status=400)
    if not player:
        return JsonResponse({'error': 'Invalid session_token'}, status=403)

    # Cannot react to own lie
    try:
        lie = LieAnswer.objects.get(pk=lie_id, round_question__session=session)
    except LieAnswer.DoesNotExist:
        return JsonResponse({'error': 'Lie not found'}, status=404)
    if lie.player_id == player.id:
        return JsonResponse({'error': 'Cannot react to your own lie'}, status=400)

    from django.db import IntegrityError
    try:
        LieReaction.objects.create(lie=lie, player=player, reaction=reaction)
    except IntegrityError:
        # Already reacted with this emoji — idempotent, return ok
        pass

    return JsonResponse({'status': 'ok'})

Step 3: Create fupogfakta/urls.py

from django.urls import path
from . import views

app_name = 'fupogfakta'

urlpatterns = [
    path('<str:code>/lie', views.submit_lie, name='submit_lie'),
    path('<str:code>/guess', views.submit_guess, name='submit_guess'),
    path('<str:code>/react', views.submit_reaction, name='submit_reaction'),
]

Step 4: Register in partyhub/urls.py

Read the file first, then add:

path('fupogfakta/', include('fupogfakta.urls')),

Step 5: Run tests

.venv/bin/python manage.py test fupogfakta.tests.SubmitLieViewTest --verbosity=2

Expected: PASS.

Step 6: Commit

git add fupogfakta/views.py fupogfakta/urls.py partyhub/urls.py fupogfakta/tests.py
git commit -m "feat(fupogfakta): add lie/guess/react endpoints"

Batch 7 — Frontend Game Screens

The Angular host-shell and player-shell currently work as developer control panels. Replace them with proper game screens that react to WebSocket events.

Task 13: WebSocket state machine in Angular

Files:

Create: frontend/angular/src/app/game-state.service.ts

The service opens a WebSocket to ws/game/{sessionCode}/ and maintains reactive state from phase.* events. Both host and player shells subscribe to this service.

// game-state.service.ts
import { Injectable, signal } from '@angular/core';

export type GamePhase =
  | 'LOBBY' | 'LIE_PHASE' | 'GUESS_PHASE'
  | `REVEAL_LIE_${number}` | 'REVEAL_TRUTH'
  | 'SCOREBOARD_RECAP' | 'POST_GAME_AWARDS' | 'FINISHED';

export interface GameState {
  phase: GamePhase;
  sessionCode: string;
  payload: Record<string, unknown>;
}

@Injectable({ providedIn: 'root' })
export class GameStateService {
  readonly state = signal<GameState>({ phase: 'LOBBY', sessionCode: '', payload: {} });
  private ws: WebSocket | null = null;

  connect(sessionCode: string, token: string, role: 'player' | 'host') {
    const query = role === 'host' ? 'role=host' : `session_token=${token}`;
    this.ws = new WebSocket(`ws://${location.host}/ws/game/${sessionCode}/?${query}`);
    this.ws.onmessage = (ev) => {
      const msg = JSON.parse(ev.data);
      this.state.set({ phase: msg.type.replace('phase.', '').toUpperCase() as GamePhase, sessionCode, payload: msg });
    };
  }

  disconnect() { this.ws?.close(); }

  ping() { this.ws?.send(JSON.stringify({ type: 'ping' })); }
}

Step 1: Write Vitest test

// frontend/angular/src/app/game-state.service.spec.ts
import { TestBed } from '@angular/core/testing';
import { GameStateService } from './game-state.service';

describe('GameStateService', () => {
  it('initialises in LOBBY phase', () => {
    const svc = TestBed.inject(GameStateService);
    expect(svc.state().phase).toBe('LOBBY');
  });
});

Step 2: Run test

cd frontend/angular && npm test -- --run 2>&1 | grep -E 'PASS|FAIL|GameState'

Step 3: Rebuild host-shell for real gameplay

Replace host-shell.component.ts content. The host screen shows:

LOBBY: session code large, player list, "Play" button with category select
LIE_PHASE: question prompt, lie-submission progress bar (N/total), countdown timer, Pause button
GUESS_PHASE: answers displayed (no correct answer highlighted), countdown timer, Pause button
REVEAL_LIE_*: lie text, author revealed, guessers listed, score delta animation
REVEAL_TRUTH: correct answer, who guessed right, score deltas
SCOREBOARD_RECAP: full leaderboard
POST_GAME_AWARDS: award cards
FINISHED: final leaderboard, "New Game" button

Host makes REST calls only for: POST /lobby/sessions/{code}/play, POST pause, POST exit.

Step 4: Rebuild player-shell for real gameplay

Player screen shows:

LOBBY: "Waiting for host to start…" with players list
LIE_PHASE: text input for lie, submit button, countdown — replaced by "Submitted ✓" once sent
GUESS_PHASE: answer buttons, once selected show emoji reaction buttons for other lies (cannot react to own lie), countdown
REVEAL_* / SCOREBOARD / AWARDS: display-only, shows own score delta highlighted
FINISHED: final leaderboard

Player makes REST calls only for: POST /fupogfakta/{code}/lie, POST /fupogfakta/{code}/guess, POST /fupogfakta/{code}/react.

Step 5: Run Angular tests

cd frontend/angular && npm test -- --run 2>&1 | tail -10

Step 6: Commit

git add frontend/angular/src/
git commit -m "feat(frontend): rebuild host and player screens as real game UI"

Batch 8 — Cleanup

Task 14: Remove obsolete lobby/views.py endpoints

The old manual-advance endpoints (start_round, show_question, submit_lie, mix_answers, submit_guess, calculate_scores, reveal_scoreboard, finish_game, start_next_round) are now replaced by the driver + platform endpoints. Remove them from lobby/views.py and lobby/urls.py, then remove or update the tests that covered them.

Run full suite after:

.venv/bin/python manage.py test fupogfakta lobby realtime --verbosity=1

Commit:

git commit -m "chore: remove obsolete manual-advance lobby endpoints"

Task 15: Update TODO.md

Mark completed items, add new backlog items discovered during implementation:

Rate limiting on fupogfakta/lie and /guess endpoints
Session-code brute-force protection on /join
TTS / read-aloud integration (Fase 4)

git add TODO.md && git commit -m "docs: update TODO after game engine implementation"

Running Order Summary

Batch 1 — Celery infra        (Tasks 1-3)   no breaking changes
Batch 2 — GameRun + Driver    (Tasks 4-6)   additive
Batch 3 — Config system       (Task 7)      additive
Batch 4 — Game models         (Task 8)      remove ScoreEvent
Batch 5 — FupOgFakta driver   (Task 9)      new game logic
Batch 6 — Platform endpoints  (Tasks 10-12) new REST API
Batch 7 — Frontend            (Tasks 13)    replace UI
Batch 8 — Cleanup             (Tasks 14-15) remove old code

Each batch is independently mergeable. Run .venv/bin/python manage.py test fupogfakta lobby realtime before every commit.

72 KiB Raw Permalink Blame History

Fup og Fakta — Game Engine Implementation Plan

Reading Before You Start

Batch 1 — Celery Infrastructure

Task 1: Add Celery to dependencies

Task 2: Create Celery app

Task 3: Create the timer Celery task (skeleton)

Batch 2 — GameRun Model + GameDriver Interface

Task 4: Add game_type field to GameSession and simplify status

Task 5: Create GameDriver interface

Task 6: Create GameRun model

Batch 3 — Config System

Task 7: BaseGameConfig abstract model + FupOgFaktaConfig

Batch 4 — FupOgFakta Game Models

Task 8: Add LieReaction model + reveal_order to RoundQuestion, remove ScoreEvent

Batch 5 — FupOgFakta Game Driver

Task 9: Driver skeleton + LIE_PHASE entry

Batch 6 — Platform REST Endpoints + Celery Wiring

Task 10: Wire handle_timer_expired to the driver

Task 11: Platform play/pause/exit endpoints

Task 12: FupOgFakta game-specific endpoints (lie, guess, react)

Batch 7 — Frontend Game Screens

Task 13: WebSocket state machine in Angular

Batch 8 — Cleanup

Task 14: Remove obsolete lobby/views.py endpoints

Task 15: Update TODO.md

Running Order Summary

72 KiB

Raw Permalink Blame History

Task 4: Add `game_type` field to GameSession and simplify status

Task 5: Create `GameDriver` interface

Task 6: Create `GameRun` model

Task 8: Add `LieReaction` model + `reveal_order` to RoundQuestion, remove ScoreEvent

Task 10: Wire `handle_timer_expired` to the driver