Project started 🍾

app/__init__.py

@@ -0,0 +1,14 @@
+"""mcma-backend — self-hosted, offline-first music service.
+
+Hexagonal (ports & adapters) architecture:
+
+* ``app.domain``         — pure business core: entities, value objects, errors,
+                           and *ports* (Protocols). No framework imports.
+* ``app.application``    — use cases / services. Orchestrate domain via ports.
+* ``app.infrastructure`` — driven adapters: ORM models, repositories, db, redis,
+                           source backends, ML/HTTP clients.
+* ``app.api``            — driving adapter: FastAPI routers & schemas.
+* ``app.core``           — cross-cutting concerns: config, logging, security.
+"""
+
+__version__ = "0.1.0"

app/api/__init__.py

@@ -0,0 +1 @@
+"""Driving adapter — FastAPI routers, schemas, dependency wiring."""

app/api/deps.py

@@ -0,0 +1,30 @@
+"""Shared FastAPI dependencies — the composition root for request-scoped wiring.
+
+Concrete adapters are bound to ports here so routers and services stay
+decoupled from infrastructure. Repository/service providers are added in
+later steps as the domain grows.
+"""
+
+from collections.abc import AsyncIterator
+from typing import Annotated
+
+from fastapi import Depends
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from app.infrastructure.db import get_sessionmaker
+
+
+async def get_session() -> AsyncIterator[AsyncSession]:
+    """Request-scoped DB session. Commits on success, rolls back on exception."""
+    session = get_sessionmaker()()
+    try:
+        yield session
+        await session.commit()
+    except Exception:
+        await session.rollback()
+        raise
+    finally:
+        await session.close()
+
+
+SessionDep = Annotated[AsyncSession, Depends(get_session)]

app/api/errors.py

@@ -0,0 +1,50 @@
+"""Maps domain exceptions to HTTP responses. The only place that knows both."""
+
+from fastapi import FastAPI, Request, status
+from fastapi.responses import JSONResponse
+
+from app.core.logging import get_logger
+from app.domain.errors import (
+    AlreadyExistsError,
+    AuthenticationError,
+    ConflictError,
+    DependencyUnavailableError,
+    DomainError,
+    NotFoundError,
+    PermissionDeniedError,
+    ValidationError,
+)
+
+log = get_logger(__name__)
+
+_STATUS_BY_ERROR: dict[type[DomainError], int] = {
+    NotFoundError: status.HTTP_404_NOT_FOUND,
+    AlreadyExistsError: status.HTTP_409_CONFLICT,
+    ConflictError: status.HTTP_409_CONFLICT,
+    ValidationError: status.HTTP_422_UNPROCESSABLE_CONTENT,
+    AuthenticationError: status.HTTP_401_UNAUTHORIZED,
+    PermissionDeniedError: status.HTTP_403_FORBIDDEN,
+    DependencyUnavailableError: status.HTTP_503_SERVICE_UNAVAILABLE,
+}
+
+
+def _error_body(code: str, message: str) -> dict[str, dict[str, str]]:
+    return {"error": {"code": code, "message": message}}
+
+
+def register_exception_handlers(app: FastAPI) -> None:
+    @app.exception_handler(DomainError)
+    async def _handle_domain_error(_request: Request, exc: DomainError) -> JSONResponse:
+        http_status = _STATUS_BY_ERROR.get(type(exc), status.HTTP_400_BAD_REQUEST)
+        return JSONResponse(
+            status_code=http_status,
+            content=_error_body(exc.code, exc.message),
+        )
+
+    @app.exception_handler(Exception)
+    async def _handle_unexpected(_request: Request, exc: Exception) -> JSONResponse:
+        log.error("unhandled_exception", exc_info=exc)
+        return JSONResponse(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            content=_error_body("internal_error", "An unexpected error occurred."),
+        )

app/api/health.py

@@ -0,0 +1,83 @@
+"""Health & readiness endpoints — used by compose healthchecks and the admin UI.
+
+* ``/health``  — liveness: the process is up. Always 200 if serving.
+* ``/health/ready`` — readiness: checks DB, Redis, and (optionally) ML.
+  Returns 503 if a *required* dependency is down. ML is optional — its absence
+  degrades, never fails, readiness (graceful degradation, see plan §6.5).
+"""
+
+import asyncio
+from typing import Literal
+
+from fastapi import APIRouter, Response, status
+from pydantic import BaseModel
+from sqlalchemy import text
+
+from app.core.config import get_settings
+from app.core.logging import get_logger
+from app.infrastructure.cache import get_redis
+from app.infrastructure.db import get_sessionmaker
+
+log = get_logger(__name__)
+router = APIRouter(tags=["health"])
+
+CheckStatus = Literal["ok", "down", "skipped"]
+
+# A readiness probe must answer fast and never hang — bound every dependency
+# check. A check that exceeds this is reported "down".
+CHECK_TIMEOUT_SECONDS = 2.0
+
+
+class HealthResponse(BaseModel):
+    status: Literal["ok"] = "ok"
+
+
+class ReadinessResponse(BaseModel):
+    status: Literal["ready", "degraded"]
+    checks: dict[str, CheckStatus]
+
+
+@router.get("/health", response_model=HealthResponse)
+async def health() -> HealthResponse:
+    return HealthResponse()
+
+
+async def _check_db() -> CheckStatus:
+    try:
+        async with asyncio.timeout(CHECK_TIMEOUT_SECONDS):
+            async with get_sessionmaker()() as session:
+                await session.execute(text("SELECT 1"))
+        return "ok"
+    except Exception as exc:
+        log.warning("healthcheck_db_down", error=str(exc))
+        return "down"
+
+
+async def _check_redis() -> CheckStatus:
+    try:
+        async with asyncio.timeout(CHECK_TIMEOUT_SECONDS):
+            await get_redis().ping()
+        return "ok"
+    except Exception as exc:
+        log.warning("healthcheck_redis_down", error=str(exc))
+        return "down"
+
+
+async def _check_ml() -> CheckStatus:
+    # Optional dependency. A real client lands in step 12; absence is fine.
+    return "skipped" if get_settings().ml_service_url is None else "ok"
+
+
+@router.get("/health/ready", response_model=ReadinessResponse)
+async def readiness(response: Response) -> ReadinessResponse:
+    db, redis, ml = await asyncio.gather(_check_db(), _check_redis(), _check_ml())
+    checks: dict[str, CheckStatus] = {"database": db, "redis": redis, "ml": ml}
+
+    required_down = db == "down" or redis == "down"
+    if required_down:
+        response.status_code = status.HTTP_503_SERVICE_UNAVAILABLE
+
+    return ReadinessResponse(
+        status="degraded" if required_down else "ready",
+        checks=checks,
+    )

app/api/middleware.py

@@ -0,0 +1,52 @@
+"""HTTP middleware: bind a correlation id and log each request."""
+
+import time
+import uuid
+
+from starlette.types import ASGIApp, Message, Receive, Scope, Send
+
+from app.core.logging import correlation_id, get_logger
+
+log = get_logger("http")
+
+_HEADER = "x-correlation-id"
+
+
+class CorrelationIdMiddleware:
+    """Pure-ASGI middleware: reuse inbound ``X-Correlation-Id`` or mint one,
+    bind it for downstream logs, echo it back, and log request completion."""
+
+    def __init__(self, app: ASGIApp) -> None:
+        self.app = app
+
+    async def __call__(self, scope: Scope, receive: Receive, send: Send) -> None:
+        if scope["type"] != "http":
+            await self.app(scope, receive, send)
+            return
+
+        headers = dict(scope["headers"])
+        inbound = headers.get(_HEADER.encode())
+        cid = inbound.decode() if inbound else uuid.uuid4().hex
+        token = correlation_id.set(cid)
+        started = time.perf_counter()
+        status_code = 0
+
+        async def send_wrapper(message: Message) -> None:
+            nonlocal status_code
+            if message["type"] == "http.response.start":
+                status_code = message["status"]
+                message.setdefault("headers", [])
+                message["headers"].append((_HEADER.encode(), cid.encode()))
+            await send(message)
+
+        try:
+            await self.app(scope, receive, send_wrapper)
+        finally:
+            log.info(
+                "request",
+                method=scope["method"],
+                path=scope["path"],
+                status=status_code,
+                duration_ms=round((time.perf_counter() - started) * 1000, 1),
+            )
+            correlation_id.reset(token)

app/application/__init__.py

@@ -0,0 +1,6 @@
+"""Application layer — use cases / services.
+
+Services orchestrate the domain through *ports* (Protocols defined in
+``app.domain``). They never import concrete adapters directly; adapters are
+injected at the composition root (``app.main`` / ``app.api.deps``).
+"""

app/cli.py

@@ -0,0 +1,25 @@
+"""Management CLI (``mcma``). Admin/seed commands land here in later steps.
+
+For now it exposes ``mcma version``. ``mcma create-admin`` arrives with auth
+(plan §11 step 3).
+"""
+
+import argparse
+
+from app import __version__
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(prog="mcma", description="mcma-backend management CLI")
+    sub = parser.add_subparsers(dest="command")
+    sub.add_parser("version", help="Print the backend version")
+
+    args = parser.parse_args()
+    if args.command == "version":
+        print(__version__)
+    else:
+        parser.print_help()
+
+
+if __name__ == "__main__":
+    main()

app/core/__init__.py

@@ -0,0 +1 @@
+"""Cross-cutting concerns: configuration, logging, security."""

app/core/config.py

@@ -0,0 +1,74 @@
+"""Application settings — single source of truth, sourced from environment.
+
+Nothing is hardcoded. All knobs come from env vars (or a local ``.env`` during
+development). Access the cached singleton via :func:`get_settings`.
+"""
+
+from functools import lru_cache
+from pathlib import Path
+from typing import Literal
+
+from pydantic import Field, SecretStr, field_validator
+from pydantic_settings import BaseSettings, SettingsConfigDict
+
+
+class Settings(BaseSettings):
+    model_config = SettingsConfigDict(
+        env_file=".env",
+        env_file_encoding="utf-8",
+        extra="ignore",
+        case_sensitive=False,
+    )
+
+    # -- runtime ----------------------------------------------------------
+    environment: Literal["dev", "test", "prod"] = "dev"
+    log_level: str = "INFO"
+    log_json: bool = Field(
+        default=False,
+        description="Structured JSON logs (enable in prod).",
+    )
+
+    # -- database ---------------------------------------------------------
+    # Async driver required (asyncpg). Example:
+    # postgresql+asyncpg://mcma:mcma@db:5432/mcma
+    database_url: str = "postgresql+asyncpg://mcma:mcma@localhost:5432/mcma"
+    db_echo: bool = False
+    db_pool_size: int = 5
+    db_max_overflow: int = 10
+
+    # -- redis (cache + arq broker) --------------------------------------
+    redis_url: str = "redis://localhost:6379/0"
+
+    # -- auth -------------------------------------------------------------
+    jwt_secret: SecretStr = SecretStr("change-me-in-prod")
+    jwt_algorithm: str = "HS256"
+    access_token_ttl_seconds: int = 60 * 15  # 15 min
+    refresh_token_ttl_seconds: int = 60 * 60 * 24 * 30  # 30 days (offline-first)
+
+    # -- media / storage --------------------------------------------------
+    media_path: Path = Path("/data/media")
+    transcode_cache_path: Path = Path("/data/transcode-cache")
+    max_parallel_downloads: int = 2
+
+    # -- external services (all optional; graceful degradation) ----------
+    ml_service_url: str | None = None
+    acoustid_api_key: SecretStr | None = None
+    musicbrainz_user_agent: str = "mcma-backend/0.1.0 ( https://github.com/your/repo )"
+    youtube_cookies_path: Path | None = None
+
+    @field_validator("database_url")
+    @classmethod
+    def _require_async_driver(cls, v: str) -> str:
+        if "+asyncpg" not in v:
+            raise ValueError("database_url must use the asyncpg driver: postgresql+asyncpg://")
+        return v
+
+    @property
+    def is_prod(self) -> bool:
+        return self.environment == "prod"
+
+
+@lru_cache
+def get_settings() -> Settings:
+    """Cached settings singleton. Patch the cache in tests via ``get_settings.cache_clear()``."""
+    return Settings()

app/core/logging.py

@@ -0,0 +1,52 @@
+"""Structured logging via structlog.
+
+Emits key=value (dev) or JSON (prod) with a per-request/task ``correlation_id``
+bound through a contextvar. Call :func:`configure_logging` once at startup.
+"""
+
+import logging
+from contextvars import ContextVar
+from typing import cast
+
+import structlog
+from structlog.typing import EventDict, FilteringBoundLogger, Processor, WrappedLogger
+
+# Bound by middleware (HTTP) and worker entrypoints (tasks) so every log line
+# downstream carries the same id without explicit passing.
+correlation_id: ContextVar[str | None] = ContextVar("correlation_id", default=None)
+
+
+def _add_correlation_id(
+    _logger: WrappedLogger, _method: str, event_dict: EventDict
+) -> EventDict:
+    cid = correlation_id.get()
+    if cid is not None:
+        event_dict["correlation_id"] = cid
+    return event_dict
+
+
+def configure_logging(*, level: str = "INFO", json: bool = False) -> None:
+    shared: list[Processor] = [
+        structlog.contextvars.merge_contextvars,
+        _add_correlation_id,
+        structlog.processors.add_log_level,
+        structlog.processors.TimeStamper(fmt="iso", utc=True),
+        structlog.processors.StackInfoRenderer(),
+        structlog.processors.format_exc_info,
+    ]
+    renderer = (
+        structlog.processors.JSONRenderer() if json else structlog.dev.ConsoleRenderer(colors=True)
+    )
+
+    structlog.configure(
+        processors=[*shared, renderer],
+        wrapper_class=structlog.make_filtering_bound_logger(logging.getLevelName(level.upper())),
+        logger_factory=structlog.PrintLoggerFactory(),
+        cache_logger_on_first_use=True,
+    )
+    # Route stdlib logging (uvicorn, sqlalchemy) through structlog formatting.
+    logging.basicConfig(level=level.upper(), format="%(message)s")
+
+
+def get_logger(name: str | None = None) -> FilteringBoundLogger:
+    return cast(FilteringBoundLogger, structlog.get_logger(name))

app/domain/__init__.py

@@ -0,0 +1,5 @@
+"""Pure business core — entities, value objects, errors, and ports (Protocols).
+
+This package MUST NOT import frameworks (FastAPI, SQLAlchemy, redis, …).
+Adapters depend on the domain; the domain depends on nothing but stdlib.
+"""

app/domain/errors.py

@@ -0,0 +1,63 @@
+"""Domain exception hierarchy — framework-agnostic.
+
+Services raise these; the API layer maps them to HTTP responses
+(see ``app.api.errors``). The domain never references HTTP status codes.
+"""
+
+
+class DomainError(Exception):
+    """Base for all expected, business-meaningful failures.
+
+    ``code`` is a stable, machine-readable identifier returned to clients.
+    """
+
+    code: str = "domain_error"
+
+    def __init__(self, message: str | None = None) -> None:
+        super().__init__(message or self.__class__.__doc__ or self.code)
+        self.message = str(self)
+
+
+class NotFoundError(DomainError):
+    """Requested resource does not exist."""
+
+    code = "not_found"
+
+
+class AlreadyExistsError(DomainError):
+    """Resource conflicts with an existing one (e.g. duplicate)."""
+
+    code = "already_exists"
+
+
+class ConflictError(DomainError):
+    """Operation conflicts with current state (e.g. stale version on write)."""
+
+    code = "conflict"
+
+
+class ValidationError(DomainError):
+    """Input is well-formed but violates a business rule."""
+
+    code = "validation_error"
+
+
+class AuthenticationError(DomainError):
+    """Caller could not be authenticated."""
+
+    code = "authentication_error"
+
+
+class PermissionDeniedError(DomainError):
+    """Caller authenticated but not authorized for this action."""
+
+    code = "permission_denied"
+
+
+class DependencyUnavailableError(DomainError):
+    """An external dependency (source, ML, MusicBrainz) is unavailable.
+
+    Callers should degrade gracefully rather than propagate as a hard 500.
+    """
+
+    code = "dependency_unavailable"

app/infrastructure/__init__.py

@@ -0,0 +1,6 @@
+"""Driven adapters — concrete implementations of domain ports.
+
+ORM models, repositories, the async DB engine, the Redis client, source
+backends, and external HTTP clients live here. Everything framework- and
+vendor-specific is confined to this package.
+"""

app/infrastructure/cache/__init__.py

@@ -0,0 +1,5 @@
+"""Redis adapter: shared connection pool for cache and arq broker."""
+
+from app.infrastructure.cache.redis import close_redis, get_redis
+
+__all__ = ["close_redis", "get_redis"]

app/infrastructure/cache/redis.py

@@ -0,0 +1,26 @@
+"""Redis client provider — a single shared async connection pool."""
+
+from redis.asyncio import Redis
+
+from app.core.config import get_settings
+
+_client: Redis | None = None
+
+
+def get_redis() -> Redis:
+    """Return the process-wide Redis client (created on first use)."""
+    global _client
+    if _client is None:
+        _client = Redis.from_url(
+            str(get_settings().redis_url),
+            encoding="utf-8",
+            decode_responses=True,
+        )
+    return _client
+
+
+async def close_redis() -> None:
+    global _client
+    if _client is not None:
+        await _client.aclose()
+        _client = None

app/infrastructure/db/__init__.py

@@ -0,0 +1,17 @@
+"""Database adapter: declarative base, async engine, session factory."""
+
+from app.infrastructure.db.base import Base
+from app.infrastructure.db.engine import (
+    dispose_engine,
+    get_engine,
+    get_sessionmaker,
+    session_scope,
+)
+
+__all__ = [
+    "Base",
+    "dispose_engine",
+    "get_engine",
+    "get_sessionmaker",
+    "session_scope",
+]

app/infrastructure/db/base.py

@@ -0,0 +1,22 @@
+"""Declarative base with a fixed naming convention.
+
+The naming convention makes Alembic autogenerate deterministic, named
+constraints — essential for clean, reversible migrations.
+"""
+
+from sqlalchemy import MetaData
+from sqlalchemy.orm import DeclarativeBase
+
+NAMING_CONVENTION = {
+    "ix": "ix_%(column_0_label)s",
+    "uq": "uq_%(table_name)s_%(column_0_name)s",
+    "ck": "ck_%(table_name)s_%(constraint_name)s",
+    "fk": "fk_%(table_name)s_%(column_0_name)s_%(referred_table_name)s",
+    "pk": "pk_%(table_name)s",
+}
+
+
+class Base(DeclarativeBase):
+    """Base for all ORM models. Import models so Alembic sees their metadata."""
+
+    metadata = MetaData(naming_convention=NAMING_CONVENTION)

app/infrastructure/db/engine.py

@@ -0,0 +1,61 @@
+"""Async engine + session factory, created lazily from settings.
+
+The engine is process-global and cached. The API binds a session per request
+(see ``app.api.deps``); workers and scripts use :func:`session_scope`.
+"""
+
+from collections.abc import AsyncIterator
+from contextlib import asynccontextmanager
+from functools import lru_cache
+
+from sqlalchemy.ext.asyncio import (
+    AsyncEngine,
+    AsyncSession,
+    async_sessionmaker,
+    create_async_engine,
+)
+
+from app.core.config import get_settings
+
+
+@lru_cache
+def get_engine() -> AsyncEngine:
+    settings = get_settings()
+    return create_async_engine(
+        settings.database_url,
+        echo=settings.db_echo,
+        pool_size=settings.db_pool_size,
+        max_overflow=settings.db_max_overflow,
+        pool_pre_ping=True,  # survive Postgres restarts / idle drops
+    )
+
+
+@lru_cache
+def get_sessionmaker() -> async_sessionmaker[AsyncSession]:
+    return async_sessionmaker(
+        bind=get_engine(),
+        expire_on_commit=False,
+        autoflush=False,
+    )
+
+
+@asynccontextmanager
+async def session_scope() -> AsyncIterator[AsyncSession]:
+    """Transactional session for workers/scripts: commit on success, rollback on error."""
+    session = get_sessionmaker()()
+    try:
+        yield session
+        await session.commit()
+    except Exception:
+        await session.rollback()
+        raise
+    finally:
+        await session.close()
+
+
+async def dispose_engine() -> None:
+    """Dispose the pooled engine on shutdown. Safe to call if never initialized."""
+    if get_engine.cache_info().currsize:
+        await get_engine().dispose()
+        get_engine.cache_clear()
+        get_sessionmaker.cache_clear()

app/main.py

@@ -0,0 +1,51 @@
+"""FastAPI composition root: wiring, lifespan, middleware, routers."""
+
+from collections.abc import AsyncIterator
+from contextlib import asynccontextmanager
+
+from fastapi import FastAPI
+
+from app.api.errors import register_exception_handlers
+from app.api.health import router as health_router
+from app.api.middleware import CorrelationIdMiddleware
+from app.core.config import get_settings
+from app.core.logging import configure_logging, get_logger
+from app.infrastructure.cache import close_redis
+from app.infrastructure.db import dispose_engine
+
+log = get_logger(__name__)
+
+
+@asynccontextmanager
+async def lifespan(_app: FastAPI) -> AsyncIterator[None]:
+    settings = get_settings()
+    log.info("startup", environment=settings.environment)
+    yield
+    log.info("shutdown")
+    await dispose_engine()
+    await close_redis()
+
+
+def create_app() -> FastAPI:
+    settings = get_settings()
+    configure_logging(level=settings.log_level, json=settings.log_json)
+
+    app = FastAPI(
+        title="mcma-backend",
+        version="0.1.0",
+        summary="Self-hosted, offline-first music service.",
+        lifespan=lifespan,
+    )
+
+    app.add_middleware(CorrelationIdMiddleware)
+    register_exception_handlers(app)
+
+    app.include_router(health_router)
+    # Versioned API routers (auth, library, …) are mounted in later steps:
+    #   app.include_router(api_v1_router, prefix="/api/v1")
+    #   app.include_router(subsonic_router, prefix="/rest")
+
+    return app
+
+
+app = create_app()

app/workers/__init__.py

@@ -0,0 +1,4 @@
+"""arq worker — background tasks (downloads, enrichment, transcoding).
+
+CPU/IO-heavy work runs here, never in the request cycle (plan §2.6).
+"""

app/workers/arq_worker.py

@@ -0,0 +1,32 @@
+"""arq worker settings — the queue runtime. Task functions register here.
+
+Run with: ``arq app.workers.arq_worker.WorkerSettings``.
+Tasks (download, enrich, transcode) are appended to ``functions`` in later steps.
+"""
+
+from typing import Any, ClassVar
+
+from arq.connections import RedisSettings
+
+from app.core.config import get_settings
+from app.core.logging import configure_logging, get_logger
+
+log = get_logger("worker")
+
+
+async def startup(_ctx: dict[str, Any]) -> None:
+    settings = get_settings()
+    configure_logging(level=settings.log_level, json=settings.log_json)
+    log.info("worker_startup", environment=settings.environment)
+
+
+async def shutdown(_ctx: dict[str, Any]) -> None:
+    log.info("worker_shutdown")
+
+
+class WorkerSettings:
+    functions: ClassVar[list[Any]] = []  # populated as tasks are implemented
+    on_startup = startup
+    on_shutdown = shutdown
+    max_jobs = get_settings().max_parallel_downloads
+    redis_settings = RedisSettings.from_dsn(str(get_settings().redis_url))