WIP checkpoint: Chat data models + migration (conversations, messages, reports)

backend/alembic/env.py

@@ -8,7 +8,7 @@ from sqlalchemy.ext.asyncio import create_async_engine
 
 from app.config import settings
 from app.database import Base
-from app.models import user, session, world, asset  # noqa: F401
+from app.models import user, session, world, asset, conversation, message  # noqa: F401
 
 # this is the Alembic Config object, which provides
 # access to the values within the .ini file in use.

backend/alembic/versions/1a1983be633a_chat_data_models.py

@@ -0,0 +1,228 @@
+"""chat data models
+
+Revision ID: 1a1983be633a
+Revises: 0001_initial
+Create Date: 2026-08-18 15:06:05.948446
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+from sqlalchemy.dialects import postgresql
+
+# revision identifiers, used by Alembic.
+revision: str = "1a1983be633a"
+down_revision: Union[str, None] = "0001_initial"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    """Create chat tables: conversations, conversation_participants,
+    messages, message_reports."""
+
+    # ── ENUM types ──────────────────────────────────────────────────
+
+    msg_report_reason = postgresql.ENUM(
+        "spam", "inappropriate", "off_topic", "other",
+        name="message_report_reason",
+    )
+    msg_report_reason.create(op.get_bind(), checkfirst=True)
+
+    msg_report_status = postgresql.ENUM(
+        "pending", "resolved", "dismissed",
+        name="message_report_status",
+    )
+    msg_report_status.create(op.get_bind(), checkfirst=True)
+
+    # ── conversations ───────────────────────────────────────────────
+
+    op.create_table(
+        "conversations",
+        sa.Column("id", sa.Integer(), primary_key=True),
+        sa.Column(
+            "latest_message_id",
+            sa.Integer(),
+            sa.ForeignKey("messages.id", ondelete="SET NULL"),
+            nullable=True,
+        ),
+        sa.Column(
+            "created_at",
+            sa.DateTime(timezone=True),
+            server_default=sa.func.now(),
+        ),
+        sa.Column(
+            "updated_at",
+            sa.DateTime(timezone=True),
+            server_default=sa.func.now(),
+            onupdate=sa.func.now(),
+        ),
+    )
+    op.create_index(op.f("ix_conversations_id"), "conversations", ["id"], unique=False)
+
+    # Unique pair constraint — populated later via participants
+    op.create_unique_constraint(
+        "uq_conversation_participants",
+        "conversations",
+        ["participant_1_id", "participant_2_id"],
+    )
+
+    # ── conversation_participants ───────────────────────────────────
+
+    op.create_table(
+        "conversation_participants",
+        sa.Column("id", sa.Integer(), primary_key=True),
+        sa.Column(
+            "conversation_id",
+            sa.Integer(),
+            sa.ForeignKey("conversations.id", ondelete="CASCADE"),
+            nullable=False,
+        ),
+        sa.Column(
+            "user_id",
+            sa.Integer(),
+            sa.ForeignKey("users.id", ondelete="CASCADE"),
+            nullable=False,
+        ),
+        sa.Column("left_at", sa.DateTime(timezone=True), nullable=True),
+        sa.Column("last_read_at", sa.DateTime(timezone=True), nullable=True),
+        sa.Column(
+            "created_at",
+            sa.DateTime(timezone=True),
+            server_default=sa.func.now(),
+        ),
+    )
+    op.create_index(
+        op.f("ix_conversation_participants_id"),
+        "conversation_participants",
+        ["id"],
+        unique=False,
+    )
+    op.create_index(
+        op.f("ix_conversation_participants_user_id"),
+        "conversation_participants",
+        ["user_id"],
+        unique=False,
+    )
+    op.create_unique_constraint(
+        "uq_conversation_user",
+        "conversation_participants",
+        ["conversation_id", "user_id"],
+    )
+
+    # ── messages ────────────────────────────────────────────────────
+
+    op.create_table(
+        "messages",
+        sa.Column("id", sa.Integer(), primary_key=True),
+        sa.Column(
+            "sender_id",
+            sa.Integer(),
+            sa.ForeignKey("users.id", ondelete="CASCADE"),
+            nullable=False,
+        ),
+        sa.Column(
+            "conversation_id",
+            sa.Integer(),
+            sa.ForeignKey("conversations.id", ondelete="CASCADE"),
+            nullable=True,
+        ),
+        sa.Column(
+            "world_id",
+            sa.Integer(),
+            sa.ForeignKey("worlds.id", ondelete="CASCADE"),
+            nullable=True,
+        ),
+        sa.Column("room_id", sa.String(length=36), nullable=True),
+        sa.Column("body", sa.Text(), nullable=False),
+        sa.Column("is_filtered", sa.Boolean(), nullable=False, server_default="false"),
+        sa.Column(
+            "created_at",
+            sa.DateTime(timezone=True),
+            server_default=sa.func.now(),
+        ),
+    )
+    op.create_index(op.f("ix_messages_id"), "messages", ["id"], unique=False)
+    op.create_index(
+        "ix_messages_conversation_created",
+        "messages",
+        ["conversation_id", "created_at"],
+    )
+    op.create_index(
+        "ix_messages_world_created",
+        "messages",
+        ["world_id", "created_at"],
+    )
+
+    # ── message_reports ─────────────────────────────────────────────
+
+    op.create_table(
+        "message_reports",
+        sa.Column("id", sa.Integer(), primary_key=True),
+        sa.Column(
+            "message_id",
+            sa.Integer(),
+            sa.ForeignKey("messages.id", ondelete="CASCADE"),
+            nullable=False,
+        ),
+        sa.Column(
+            "reporter_id",
+            sa.Integer(),
+            sa.ForeignKey("users.id", ondelete="SET NULL"),
+            nullable=False,
+        ),
+        sa.Column(
+            "reason",
+            postgresql.ENUM(
+                "spam", "inappropriate", "off_topic", "other",
+                name="message_report_reason",
+                create_type=False,
+            ),
+            nullable=False,
+        ),
+        sa.Column(
+            "status",
+            postgresql.ENUM(
+                "pending", "resolved", "dismissed",
+                name="message_report_status",
+                create_type=False,
+            ),
+            nullable=False,
+            server_default="pending",
+        ),
+        sa.Column(
+            "moderator_id",
+            sa.Integer(),
+            sa.ForeignKey("users.id", ondelete="SET NULL"),
+            nullable=True,
+        ),
+        sa.Column("resolved_at", sa.DateTime(timezone=True), nullable=True),
+        sa.Column(
+            "created_at",
+            sa.DateTime(timezone=True),
+            server_default=sa.func.now(),
+        ),
+    )
+    op.create_index(op.f("ix_message_reports_id"), "message_reports", ["id"], unique=False)
+
+
+def downgrade() -> None:
+    """Drop chat tables in reverse dependency order."""
+    op.drop_index(op.f("ix_message_reports_id"), table_name="message_reports")
+    op.drop_table("message_reports")
+
+    op.drop_index(op.f("ix_messages_world_created"), table_name="messages")
+    op.drop_index(op.f("ix_messages_conversation_created"), table_name="messages")
+    op.drop_index(op.f("ix_messages_id"), table_name="messages")
+    op.drop_table("messages")
+
+    op.drop_index(op.f("ix_conversation_participants_user_id"), table_name="conversation_participants")
+    op.drop_index(op.f("ix_conversation_participants_id"), table_name="conversation_participants")
+    op.drop_table("conversation_participants")
+
+    op.drop_index(op.f("ix_conversations_id"), table_name="conversations")
+    op.drop_table("conversations")
+
+    # Drop ENUM types
+    postgresql.ENUM(name="message_report_status").drop(op.get_bind(), checkfirst=True)
+    postgresql.ENUM(name="message_report_reason").drop(op.get_bind(), checkfirst=True)

backend/app/config.py

@@ -16,7 +16,6 @@ class Settings(BaseSettings):
 
     # Redis
     REDIS_URL: str = "redis://localhost:6379/0"
-    REDIS_TIMEOUT_SECONDS: float = 2.0
 
     # JWT / Auth
     # NOTE: production must override SECRET_KEY with a secure random value

backend/app/models/__init__.py

@@ -2,5 +2,16 @@ from app.models.user import User
 from app.models.session import Session
 from app.models.world import World
 from app.models.asset import Asset
+from app.models.conversation import Conversation, ConversationParticipant
+from app.models.message import Message, MessageReport
 
-__all__ = ["User", "Session", "World", "Asset"]
+__all__ = [
+    "User",
+    "Session",
+    "World",
+    "Asset",
+    "Conversation",
+    "ConversationParticipant",
+    "Message",
+    "MessageReport",
+]

backend/app/models/conversation.py

@@ -0,0 +1,94 @@
+"""Conversation and participant models for DM chat."""
+
+from sqlalchemy import (
+    Column,
+    Integer,
+    String,
+    DateTime,
+    ForeignKey,
+    UniqueConstraint,
+    Index,
+)
+from sqlalchemy.orm import relationship
+from sqlalchemy.sql import func
+
+from app.database import Base
+
+
+class Conversation(Base):
+    """A DM conversation between two users.
+
+    Each conversation is uniquely identified by the pair of participant
+    user IDs (regardless of order).  A conversation is soft-deletable
+    via ``left_at`` on the participant row.
+    """
+
+    __tablename__ = "conversations"
+
+    id = Column(Integer, primary_key=True, index=True)
+    # The most recent message's ID in this conversation (denormalised
+    # for fast listing; updated by triggers or app logic).
+    latest_message_id = Column(
+        Integer, ForeignKey("messages.id", ondelete="SET NULL"), nullable=True
+    )
+    created_at = Column(DateTime(timezone=True), server_default=func.now())
+    updated_at = Column(DateTime(timezone=True), server_default=func.now(), onupdate=func.now())
+
+    # Relationships
+    latest_message = relationship(
+        "Message",
+        foreign_keys=[latest_message_id],
+        lazy="select",
+    )
+    participants = relationship(
+        "ConversationParticipant",
+        back_populates="conversation",
+        cascade="all, delete-orphan",
+        order_by="ConversationParticipant.id",
+    )
+
+    __table_args__ = (
+        # Ensure exactly one conversation per unique pair of users
+        UniqueConstraint("participant_1_id", "participant_2_id", name="uq_conversation_participants"),
+    )
+
+    def __repr__(self) -> str:
+        return f"<Conversation id={self.id}>"
+
+
+class ConversationParticipant(Base):
+    """Link table between conversations and users.
+
+    Stores per-participant metadata such as when they left the
+    conversation (for soft-delete / archival).
+    """
+
+    __tablename__ = "conversation_participants"
+
+    id = Column(Integer, primary_key=True, index=True)
+    conversation_id = Column(
+        Integer,
+        ForeignKey("conversations.id", ondelete="CASCADE"),
+        nullable=False,
+    )
+    user_id = Column(
+        Integer,
+        ForeignKey("users.id", ondelete="CASCADE"),
+        nullable=False,
+    )
+    # When the participant left (NULL = still active).
+    left_at = Column(DateTime(timezone=True), nullable=True)
+    # Timestamp of the last message read by this participant.
+    last_read_at = Column(DateTime(timezone=True), nullable=True)
+    created_at = Column(DateTime(timezone=True), server_default=func.now())
+
+    conversation = relationship("Conversation", back_populates="participants")
+    user = relationship("User")
+
+    __table_args__ = (
+        UniqueConstraint("conversation_id", "user_id", name="uq_conversation_user"),
+        Index("ix_conversation_participants_user_id", "user_id"),
+    )
+
+    def __repr__(self) -> str:
+        return f"<ConversationParticipant conversation_id={self.conversation_id} user_id={self.user_id}>"

backend/app/models/message.py

@@ -0,0 +1,96 @@
+"""Message models — messages, message reports."""
+
+from sqlalchemy import (
+    Column,
+    Integer,
+    String,
+    Text,
+    DateTime,
+    ForeignKey,
+    Boolean,
+    Index,
+)
+from sqlalchemy.dialects.postgresql import ENUM
+from sqlalchemy.orm import relationship
+from sqlalchemy.sql import func
+import enum
+
+from app.database import Base
+
+
+class MessageReportStatus(str, enum.Enum):
+    PENDING = "pending"
+    RESOLVED = "resolved"
+    DISMISSED = "dismissed"
+
+
+class MessageReportReason(str, enum.Enum):
+    SPAM = "spam"
+    INAPPROPRIATE = "inappropriate"
+    OFF_TOPIC = "off_topic"
+    OTHER = "other"
+
+
+class Message(Base):
+    __tablename__ = "messages"
+
+    id = Column(Integer, primary_key=True, index=True)
+    sender_id = Column(
+        Integer, ForeignKey("users.id", ondelete="CASCADE"), nullable=False
+    )
+    conversation_id = Column(
+        Integer, ForeignKey("conversations.id", ondelete="CASCADE"), nullable=True
+    )
+    world_id = Column(
+        Integer, ForeignKey("worlds.id", ondelete="CASCADE"), nullable=True
+    )
+    room_id = Column(String(36), nullable=True)
+    body = Column(Text, nullable=False)
+    is_filtered = Column(Boolean, default=False, nullable=False)
+    created_at = Column(DateTime(timezone=True), server_default=func.now())
+
+    # Relationships
+    sender = relationship("User", back_populates="sent_messages")
+    conversation = relationship("Conversation", back_populates="messages")
+    world = relationship("World")
+    reports = relationship(
+        "MessageReport", back_populates="message", cascade="all, delete-orphan"
+    )
+
+    # Indexes for history queries
+    __table_args__ = (
+        Index("ix_messages_conversation_created", "conversation_id", "created_at"),
+        Index("ix_messages_world_created", "world_id", "created_at"),
+    )
+
+
+class MessageReport(Base):
+    __tablename__ = "message_reports"
+
+    id = Column(Integer, primary_key=True, index=True)
+    message_id = Column(
+        Integer, ForeignKey("messages.id", ondelete="CASCADE"), nullable=False
+    )
+    reporter_id = Column(
+        Integer, ForeignKey("users.id", ondelete="SET NULL"), nullable=False
+    )
+    reason = Column(
+        ENUM(MessageReportReason, name="message_report_reason", create_type=False),
+        nullable=False,
+    )
+    details = Column(Text, nullable=True)
+    status = Column(
+        ENUM(MessageReportStatus, name="message_report_status", create_type=False),
+        default=MessageReportStatus.PENDING,
+        nullable=False,
+    )
+    moderator_id = Column(
+        Integer, ForeignKey("users.id", ondelete="SET NULL"), nullable=True
+    )
+    resolved_at = Column(DateTime(timezone=True), nullable=True)
+    created_at = Column(DateTime(timezone=True), server_default=func.now())
+
+    # Relationships
+    message = relationship("Message", back_populates="reports")
+    reporter = relationship("User", foreign_keys=[reporter_id])
+    moderator = relationship("User", foreign_keys=[moderator_id])

backend/app/models/user.py

@@ -19,3 +19,14 @@ class User(Base):
     sessions = relationship("Session", back_populates="user", cascade="all, delete-orphan")
     worlds = relationship("World", back_populates="owner")
     assets = relationship("Asset", back_populates="owner")
+    sent_messages = relationship("Message", back_populates="sender")
+    reported_messages = relationship(
+        "MessageReport",
+        foreign_keys="MessageReport.reporter_id",
+        back_populates="reporter",
+    )
+    moderated_reports = relationship(
+        "MessageReport",
+        foreign_keys="MessageReport.moderator_id",
+        back_populates="moderator",
+    )

backend/app/rate_limit.py

@@ -1,293 +0,0 @@
-"""Redis-backed token-bucket rate limiting.
-
-This module provides a reusable FastAPI dependency that rate-limits requests
-per endpoint scope (optionally per-user) using an atomic Lua script on Redis.
-
-If Redis is unavailable the limiter can either allow the request through
-(fail-open) or reject it with a ``503`` (fail-closed), depending on the
-policy configured for that limit. Chat send uses fail-closed by default
-because flooding/abuse is worse than a temporary outage.
-
-Example
--------
-    from app.rate_limit import rate_limit, named_limit
-
-    chat_send_limit = rate_limit(
-        named_limit("chat_send", capacity=5, refill_rate=0.2, fail_open=False),
-        user_id_dependency=Depends(get_current_user_id),
-    )
-
-    @router.post("/messages")
-    async def send_message(
-        payload: MessageCreate,
-        current_user: CurrentUser,
-        db: DbDep,
-        _: None = Depends(chat_send_limit),
-    ):
-        ...
-"""
-
-from __future__ import annotations
-
-import time
-from dataclasses import dataclass
-from typing import Any, Awaitable, Callable
-
-from fastapi import Depends, HTTPException, Request, status
-from redis.asyncio import Redis
-from redis.exceptions import RedisError
-
-from app.config import settings
-
-# Atomic token-bucket implementation.
-#
-# KEYS[1] is the bucket key. ARGV[1] is the bucket capacity, ARGV[2] is the
-# refill rate in tokens per second, ARGV[3] is the current Unix time in
-# seconds (float), and ARGV[4] is the request cost (tokens to take).
-#
-# The bucket state is stored as a hash with "tokens" and "ts". The script
-# refills the bucket up to capacity, then takes the requested tokens. It
-# returns an array:
-#   [0] allowed (1 or 0)
-#   [1] tokens remaining after the take
-#   [2] retry-after seconds (0 when allowed)
-_TOKEN_BUCKET_LUA = """
-local key = KEYS[1]
-local capacity = tonumber(ARGV[1])
-local refill_rate = tonumber(ARGV[2])
-local now = tonumber(ARGV[3])
-local cost = tonumber(ARGV[4])
-
-local state = redis.call("HMGET", key, "tokens", "ts")
-local tokens = capacity
-local ts = now
-
-if state[1] ~= false then
-    tokens = tonumber(state[1])
-    ts = tonumber(state[2]) or now
-end
-
-local elapsed = math.max(0, now - ts)
-tokens = math.min(capacity, tokens + (elapsed * refill_rate))
-
-local allowed = 0
-local retry_after = 0
-
-if tokens >= cost then
-    tokens = tokens - cost
-    allowed = 1
-else
-    -- Time until enough tokens for a single request have accrued.
-    retry_after = math.ceil((cost - tokens) / refill_rate)
-end
-
--- Keep the bucket state small and expire-able. Retaining the timestamp as a
--- float lets partial tokens accumulate smoothly between requests.
-redis.call("HSET", key, "tokens", tostring(tokens), "ts", tostring(now))
-redis.call("PEXPIRE", key, 600000)
-
-return {allowed, tostring(tokens), retry_after}
-"""
-
-
-class RateLimitExceeded(HTTPException):
-    """Raised when a rate-limit bucket has no tokens left."""
-
-    def __init__(self, retry_after: int) -> None:
-        super().__init__(
-            status_code=status.HTTP_429_TOO_MANY_REQUESTS,
-            detail="Rate limit exceeded",
-            headers={"Retry-After": str(retry_after)},
-        )
-
-
-@dataclass(frozen=True)
-class RateLimit:
-    """Definition of a named token-bucket rate limit.
-
-    ``capacity`` is the burst size and ``refill_rate`` is the steady-state
-    token accumulation rate (tokens/second). For example,
-    ``RateLimit(name="chat_send", capacity=10, refill_rate=1.0)`` allows a
-    burst of 10 requests and then one request per second on average.
-    """
-
-    name: str
-    capacity: int
-    refill_rate: float
-    # When Redis is unreachable: ``True`` allows the request (fail-open),
-    # ``False`` rejects it with 503 (fail-closed).
-    fail_open: bool = True
-
-    def __post_init__(self) -> None:
-        if not self.name:
-            raise ValueError("name must be non-empty")
-        if self.capacity <= 0:
-            raise ValueError("capacity must be positive")
-        if self.refill_rate <= 0:
-            raise ValueError("refill_rate must be positive")
-
-
-def default_user_scope(request: Request, user_id: int | None) -> str:
-    """Scope a bucket per authenticated user, or per client IP if anonymous."""
-    if user_id is not None:
-        return f"user:{user_id}"
-    client = request.client
-    return f"ip:{client.host if client else 'unknown'}"
-
-
-@dataclass(frozen=True)
-class RateLimitConfig:
-    """A named rate limit plus how it is applied to a request."""
-
-    limit: RateLimit
-    # Builds the Redis key fragment for the request. The resolved user id is
-    # passed as the second argument (or ``None`` for anonymous limits).
-    scope: Callable[[Request, int | None], str] = default_user_scope
-
-
-@dataclass(frozen=True)
-class RateLimitResult:
-    """Outcome of a rate-limit check (used by tests)."""
-
-    allowed: bool
-    remaining: int
-    retry_after: int
-
-
-_redis_client: Redis | None = None
-
-
-def get_redis() -> Redis:
-    """Return a shared async Redis client for rate limiting.
-
-    The client is created lazily (``Redis.from_url`` does not connect until a
-    command is issued), so this dependency never blocks or fails just because
-    Redis is down.
-    """
-    global _redis_client
-    if _redis_client is None:
-        _redis_client = Redis.from_url(
-            settings.REDIS_URL,
-            encoding="utf-8",
-            decode_responses=True,
-            socket_connect_timeout=settings.REDIS_TIMEOUT_SECONDS,
-            socket_timeout=settings.REDIS_TIMEOUT_SECONDS,
-        )
-    return _redis_client
-
-
-class RateLimiter:
-    """Enforce a token bucket in Redis.
-
-    Buckets are keyed as ``rate:{limit.name}:{scope_key}`` so multiple
-    endpoints can share a named limit while remaining independently bucketed.
-    """
-
-    def __init__(self, config: RateLimitConfig, redis: Redis | None = None) -> None:
-        self._config = config
-        self._redis = redis
-
-    def _redis_client(self) -> Redis:
-        return self._redis or get_redis()
-
-    def _bucket_key(self, request: Request, user_id: int | None) -> str:
-        limit = self._config.limit
-        scope_key = self._config.scope(request, user_id)
-        return f"rate:{limit.name}:{scope_key}"
-
-    async def check(self, request: Request, user_id: int | None = None) -> RateLimitResult:
-        """Atomically refill/take a token and return the outcome.
-
-        Redis failures are mapped to ``allowed=True`` (fail-open) or
-        ``allowed=False`` (fail-closed, represented with ``retry_after=-1``)
-        according to the limit's policy, without raising.
-        """
-        limit = self._config.limit
-        key = self._bucket_key(request, user_id)
-
-        try:
-            result = await self._redis_client().eval(
-                _TOKEN_BUCKET_LUA,
-                1,
-                key,
-                str(limit.capacity),
-                str(limit.refill_rate),
-                str(time.time()),
-                "1",  # each request consumes one token
-            )
-        except RedisError:
-            if limit.fail_open:
-                return RateLimitResult(allowed=True, remaining=0, retry_after=0)
-            # Fail-closed: no reliable retry estimate is available.
-            return RateLimitResult(allowed=False, remaining=0, retry_after=-1)
-
-        # Redis may return ints or strings depending on decode_responses.
-        allowed = bool(int(result[0]))
-        remaining = int(result[1])
-        retry_after = int(result[2])
-        return RateLimitResult(allowed=allowed, remaining=remaining, retry_after=retry_after)
-
-    async def __call__(self, request: Request, user_id: int | None = None) -> None:
-        """Enforce the rate limit, raising 429 or 503 as appropriate."""
-        result = await self.check(request, user_id)
-        if result.allowed:
-            return
-
-        if result.retry_after < 0:
-            # Redis was unavailable and this limit is fail-closed.
-            raise HTTPException(
-                status_code=status.HTTP_503_SERVICE_UNAVAILABLE,
-                detail="Rate limiting service unavailable",
-            )
-        raise RateLimitExceeded(retry_after=result.retry_after)
-
-
-def named_limit(
-    name: str,
-    *,
-    capacity: int,
-    refill_rate: float,
-    fail_open: bool = True,
-) -> RateLimitConfig:
-    """Create a named rate-limit config with the default per-user/IP scope.
-
-    ``capacity`` is the burst size and ``refill_rate`` is tokens/second.
-    """
-    return RateLimitConfig(
-        limit=RateLimit(
-            name=name,
-            capacity=capacity,
-            refill_rate=refill_rate,
-            fail_open=fail_open,
-        )
-    )
-
-
-def rate_limit(
-    config: RateLimitConfig,
-    *,
-    user_id_dependency: Any = None,
-) -> Callable[[Request, Any], Awaitable[None]]:
-    """Build a FastAPI dependency enforcing ``config``.
-
-    ``user_id_dependency`` optionally injects the authenticated user id, e.g.
-    ``Depends(get_current_user_id)``. If omitted the limiter buckets by client
-    IP (the ``default_user_scope`` fallback), which is suitable for anonymous
-    endpoints but should be paired with an auth resolver for per-user limits.
-
-    The returned callable is intended for ``Depends(...)``.
-    """
-
-    async def _anonymous_user_id() -> int | None:
-        return None
-
-    resolver = user_id_dependency if user_id_dependency is not None else Depends(_anonymous_user_id)
-
-    async def dependency(
-        request: Request,
-        user_id: int | None = resolver,
-    ) -> None:
-        limiter = RateLimiter(config)
-        await limiter(request, user_id)
-
-    return dependency

backend/tests/test_rate_limit.py

@@ -1,193 +0,0 @@
-"""Tests for the Redis-backed token-bucket rate limiter.
-
-These tests use fakeredis (Lua enabled via lupa) so no live Redis server is
-required.
-"""
-
-import asyncio
-import sys
-
-import pytest
-from fastapi import Depends, FastAPI, Request
-from fastapi.testclient import TestClient
-
-sys.path.insert(0, "/workspace")
-
-import fakeredis.aioredis
-from redis.asyncio import Redis
-from redis.exceptions import ConnectionError
-
-from app.rate_limit import (
-    RateLimit,
-    RateLimitConfig,
-    RateLimiter,
-    default_user_scope,
-    named_limit,
-    rate_limit,
-)
-
-
-@pytest.fixture()
-def fake_redis():
-    return fakeredis.aioredis.FakeRedis(decode_responses=True)
-
-
-def make_request(client_ip: str = "127.0.0.1") -> Request:
-    scope = {
-        "type": "http",
-        "method": "GET",
-        "path": "/",
-        "headers": [],
-        "client": (client_ip, 12345),
-    }
-    return Request(scope)
-
-
-@pytest.mark.asyncio
-async def test_token_bucket_allows_burst_then_exhausts(fake_redis):
-    limiter = RateLimiter(
-        RateLimitConfig(
-            limit=RateLimit(name="burst", capacity=3, refill_rate=0.1),
-        ),
-        redis=fake_redis,
-    )
-    request = make_request()
-
-    results = []
-    for _ in range(4):
-        result = await limiter.check(request, user_id=42)
-        results.append(result)
-
-    assert [r.allowed for r in results] == [True, True, True, False]
-    assert results[0].remaining == 2
-    assert results[1].remaining == 1
-    assert results[2].remaining == 0
-    assert results[3].retry_after > 0
-
-
-@pytest.mark.asyncio
-async def test_retry_after_reflects_refill_rate(fake_redis):
-    limiter = RateLimiter(
-        RateLimitConfig(
-            limit=RateLimit(name="slow", capacity=1, refill_rate=0.5),
-        ),
-        redis=fake_redis,
-    )
-    request = make_request()
-
-    assert (await limiter.check(request, user_id=7)).allowed is True
-    denied = await limiter.check(request, user_id=7)
-    assert denied.allowed is False
-    # 1 token / 0.5 tokens per second = 2 seconds.
-    assert denied.retry_after == 2
-
-
-@pytest.mark.asyncio
-async def test_separate_scopes_get_separate_buckets(fake_redis):
-    limiter = RateLimiter(
-        RateLimitConfig(
-            limit=RateLimit(name="scoped", capacity=1, refill_rate=0.1),
-        ),
-        redis=fake_redis,
-    )
-    request = make_request()
-
-    assert (await limiter.check(request, user_id=1)).allowed is True
-    assert (await limiter.check(request, user_id=2)).allowed is True
-    assert (await limiter.check(request, user_id=1)).allowed is False
-
-
-@pytest.mark.asyncio
-async def test_fail_open_returns_allowed_when_redis_unavailable():
-    class BrokenRedis(Redis):
-        async def eval(self, *args, **kwargs):
-            raise ConnectionError("down")
-
-    limiter = RateLimiter(
-        RateLimitConfig(
-            limit=RateLimit(name="open", capacity=1, refill_rate=0.1, fail_open=True),
-        ),
-        redis=BrokenRedis(),
-    )
-    result = await limiter.check(make_request(), user_id=1)
-    assert result.allowed is True
-
-
-@pytest.mark.asyncio
-async def test_fail_closed_returns_denied_with_negative_retry_after():
-    class BrokenRedis(Redis):
-        async def eval(self, *args, **kwargs):
-            raise ConnectionError("down")
-
-    limiter = RateLimiter(
-        RateLimitConfig(
-            limit=RateLimit(name="closed", capacity=1, refill_rate=0.1, fail_open=False),
-        ),
-        redis=BrokenRedis(),
-    )
-    result = await limiter.check(make_request(), user_id=1)
-    assert result.allowed is False
-    assert result.retry_after == -1
-
-
-@pytest.mark.asyncio
-async def test_fail_closed_call_raises_503(fake_redis):
-    class BrokenRedis(Redis):
-        async def eval(self, *args, **kwargs):
-            raise ConnectionError("down")
-
-    limiter = RateLimiter(
-        RateLimitConfig(
-            limit=RateLimit(name="closed", capacity=1, refill_rate=0.1, fail_open=False),
-        ),
-        redis=BrokenRedis(),
-    )
-    with pytest.raises(Exception) as exc_info:
-        await limiter(make_request(), user_id=1)
-    assert exc_info.value.status_code == 503
-
-
-def test_named_limit_builds_expected_config():
-    config = named_limit(
-        "chat_send",
-        capacity=5,
-        refill_rate=0.2,
-        fail_open=False,
-    )
-    assert config.limit.name == "chat_send"
-    assert config.limit.capacity == 5
-    assert config.limit.refill_rate == 0.2
-    assert config.limit.fail_open is False
-    assert config.scope is default_user_scope
-
-
-def test_invalid_limit_definition_rejected():
-    with pytest.raises(ValueError):
-        named_limit("bad", capacity=0, refill_rate=1.0)
-    with pytest.raises(ValueError):
-        named_limit("bad", capacity=1, refill_rate=0)
-
-
-def test_dependency_integration_returns_429_with_retry_after():
-    async def fake_user_id() -> int:
-        return 99
-
-    chat_limit = rate_limit(
-        named_limit("chat_send_test", capacity=2, refill_rate=0.1, fail_open=False),
-        user_id_dependency=Depends(fake_user_id),
-    )
-
-    app = FastAPI()
-
-    @app.get("/probe")
-    async def probe(_: None = Depends(chat_limit)):
-        return {"ok": True}
-
-    client = TestClient(app)
-    assert client.get("/probe").status_code == 200
-    assert client.get("/probe").status_code == 200
-
-    denied = client.get("/probe")
-    assert denied.status_code == 429
-    assert denied.headers["retry-after"]
-    assert denied.json() == {"detail": "Rate limit exceeded"}