first

services/asf-core/app/db/__init__.py

@@ -0,0 +1,15 @@
+from .mongo import (
+    get_mongo_db,
+    get_mongo_client,
+    MongoCollections,
+    create_indexes,
+    get_collection,
+)
+
+__all__ = [
+    "get_mongo_db",
+    "get_mongo_client",
+    "MongoCollections",
+    "create_indexes",
+    "get_collection",
+]

services/asf-core/app/db/mongo.py

@@ -0,0 +1,167 @@
+"""MongoDB client + collection provider.
+
+Every query is (a) tenant-scoped and (b) backed by explicit indexes.
+"""
+
+from __future__ import annotations
+
+from contextlib import asynccontextmanager
+from typing import AsyncIterator
+
+from motor.motor_asyncio import (
+    AsyncIOMotorClient,
+    AsyncIOMotorCollection,
+    AsyncIOMotorDatabase,
+)
+from pymongo import IndexModel
+import structlog
+
+from ..config import get_settings
+
+log = structlog.get_logger(__name__)
+
+
+_client: AsyncIOMotorClient | None = None
+_db: AsyncIOMotorDatabase | None = None
+
+
+class MongoCollections(str):
+    ORGANIZATIONS = "organizations"
+    USERS = "users"
+    TEAMS = "teams"
+    PROJECTS = "projects"
+    ARTIFACTS = "artifacts"
+    ARTIFACT_RELATIONSHIPS = "artifact_relationships"
+    INTEGRATIONS = "integrations"
+    JOBS = "jobs"
+    AI_EXECUTIONS = "ai_executions"
+    AUDIT_EVENTS = "audit_events"
+
+
+def get_mongo_client() -> AsyncIOMotorClient:
+    global _client
+    if _client is None:
+        s = get_settings()
+        _client = AsyncIOMotorClient(
+            s.mongo_uri,
+            minPoolSize=s.mongo_min_pool_size,
+            maxPoolSize=s.mongo_max_pool_size,
+            uuidRepresentation="standard",
+            tz_aware=True,
+        )
+    return _client
+
+
+def get_mongo_db() -> AsyncIOMotorDatabase:
+    global _db
+    if _db is None:
+        s = get_settings()
+        _db = get_mongo_client()[s.mongo_db]
+    return _db
+
+
+def get_collection(name: MongoCollections | str) -> AsyncIOMotorCollection:
+    return get_mongo_db()[name]
+
+
+async def create_indexes() -> None:
+    """Create (idempotent) indexes. Called once on service startup.
+
+    Rule of thumb for tenanted data: compound index with ``tenant_id`` first,
+    then the most common filter / sort key.
+    """
+
+    db = get_mongo_db()
+
+    # --- organizations ---
+    await db.organizations.create_indexes(
+        [
+            IndexModel("slug", unique=True, name="orgs_slug_unique"),
+        ]
+    )
+
+    # --- users ---
+    await db.users.create_indexes(
+        [
+            IndexModel([("tenant_id", 1), ("email", 1)], unique=True, name="users_tenant_email_unique"),
+        ]
+    )
+
+    # --- projects ---
+    await db.projects.create_indexes(
+        [
+            IndexModel([("tenant_id", 1), ("key", 1)], unique=True, name="projects_tenant_key_unique"),
+            IndexModel([("tenant_id", 1), ("is_active", 1)], name="projects_tenant_active"),
+        ]
+    )
+
+    # --- artifacts ---
+    await db.artifacts.create_indexes(
+        [
+            IndexModel([("tenant_id", 1), ("project_id", 1)], name="arts_tenant_project"),
+            IndexModel([("tenant_id", 1), ("project_id", 1), ("type", 1)], name="arts_tenant_project_type"),
+            IndexModel([("tenant_id", 1), ("external_id", 1)], name="arts_tenant_external_id", sparse=True),
+            IndexModel([("tenant_id", 1), ("created_at", -1)], name="arts_tenant_created_desc"),
+        ]
+    )
+
+    # --- artifact_relationships ---
+    # Two indexes: one for "where does this artifact point TO", one for "what points TO this artifact"
+    await db.artifact_relationships.create_indexes(
+        [
+            IndexModel([("tenant_id", 1), ("source_artifact_id", 1), ("relationship_type", 1)],
+                     name="rels_tenant_source_type"),
+            IndexModel([("tenant_id", 1), ("target_artifact_id", 1), ("relationship_type", 1)],
+                     name="rels_tenant_target_type"),
+            IndexModel([("source_artifact_id", 1), ("target_artifact_id", 1), ("relationship_type", 1)],
+                     unique=True, name="rels_unique_edge"),
+        ]
+    )
+
+    # --- jobs ---
+    await db.jobs.create_indexes(
+        [
+            IndexModel([("tenant_id", 1), ("status", 1), ("priority", -1), ("created_at", 1)],
+                     name="jobs_tenant_status_priority"),
+            IndexModel([("tenant_id", 1), ("project_id", 1), ("created_at", -1)],
+                     name="jobs_tenant_project_created_desc"),
+            IndexModel([("correlation_id", 1)], name="jobs_correlation_id", sparse=True),
+        ]
+    )
+
+    # --- audit_events (append-only, oldest-to-newest scans normal) ---
+    await db.audit_events.create_indexes(
+        [
+            IndexModel([("tenant_id", 1), ("created_at", -1)], name="audit_tenant_created_desc"),
+            IndexModel([("tenant_id", 1), ("actor_id", 1), ("created_at", -1)], name="audit_tenant_actor_desc"),
+            IndexModel([("correlation_id", 1)], name="audit_correlation_id", sparse=True),
+        ]
+    )
+
+    # --- ai_executions ---
+    await db.ai_executions.create_indexes(
+        [
+            IndexModel([("tenant_id", 1), ("created_at", -1)], name="aie_tenant_created_desc"),
+            IndexModel([("tenant_id", 1), ("job_id", 1)], name="aie_tenant_job", sparse=True),
+        ]
+    )
+
+    await log.adebug("mongo_indexes_ready")
+
+
+async def close_mongo() -> None:
+    global _client, _db
+    if _client is not None:
+        _client.close()
+    _client = None
+    _db = None
+
+
+@asynccontextmanager
+async def mongo_lifespan() -> AsyncIterator[None]:
+    """Async lifespan helper for FastAPI startup / shutdown."""
+    await create_indexes()
+    try:
+        yield
+    finally:
+        await close_mongo()