firestore-service
Guide for creating Firestore services with async operations, transactions, and proper error handling following this project's patterns.
When & Why to Use This Skill
This Claude skill provides a comprehensive framework for developing robust, production-ready Firestore services in Python. It specializes in implementing asynchronous database operations, ensuring data integrity through atomic transactions, and enforcing standardized CRUD patterns. By following these architectural guidelines, developers can build scalable, maintainable, and high-performance backend systems integrated with Google Cloud Firestore.
Use Cases
- Architecting scalable backend services that require non-blocking, asynchronous Firestore interactions for high concurrency.
- Implementing complex business logic requiring atomic multi-document updates and data consistency via async transactions.
- Standardizing the service layer across a Python or FastAPI project with consistent error handling, dependency injection, and directory structures.
- Integrating automated security audit logging for sensitive database operations (Create, Update, Delete) to meet compliance requirements.
- Streamlining the development of database-centric features with pre-defined patterns for testing using Firebase emulators and unit test mocking.
| name | firestore-service |
|---|---|
| description | Guide for creating Firestore services with async operations, transactions, and proper error handling following this project's patterns. |
Firestore Service Creation
Use this skill when creating services that interact with Firestore using async operations.
For comprehensive coding guidelines, see AGENTS.md in the repository root.
Service Structure
Create services in app/services/ with the following structure:
"""
Resource service with async Firestore operations.
"""
from datetime import UTC, datetime
from typing import TYPE_CHECKING
from google.cloud import firestore
from app.core.firebase import get_async_firestore_client
from app.exceptions import ResourceAlreadyExistsError, ResourceNotFoundError
from app.middleware import log_audit_event
from app.models.resource import RESOURCE_COLLECTION, Resource, ResourceCreate, ResourceUpdate
# Note: Models are organized in subdirectories:
# - app/models/resource/requests.py (ResourceCreate, ResourceUpdate)
# - app/models/resource/responses.py (Resource, RESOURCE_COLLECTION)
if TYPE_CHECKING:
from google.cloud.firestore import AsyncClient, AsyncDocumentReference, AsyncTransaction
class ResourceService:
"""
Service for resource CRUD operations using async Firestore.
"""
def __init__(self) -> None:
self.collection_name = RESOURCE_COLLECTION
def _get_client(self) -> AsyncClient:
return get_async_firestore_client()
Transactional Operations
Use @firestore.async_transactional for atomic operations. Define transaction methods as static:
@staticmethod
@firestore.async_transactional
async def _create_in_transaction( # pragma: no cover
transaction: AsyncTransaction,
doc_ref: AsyncDocumentReference,
data: dict,
) -> None:
# Tested via E2E tests with Firebase emulators; unit tests mock this method
snapshot = await doc_ref.get(transaction=transaction)
if snapshot.exists:
raise ResourceAlreadyExistsError("Resource already exists")
transaction.set(doc_ref, data)
CRUD Operations
Create
async def create_resource(self, user_id: str, resource_data: ResourceCreate) -> Resource:
"""
Create a new resource for the given user.
"""
client = self._get_client()
doc_ref = client.collection(self.collection_name).document(user_id)
now = datetime.now(UTC)
resource_dict = {
"id": user_id,
**resource_data.model_dump(),
"created_at": now,
"updated_at": now,
}
transaction = client.transaction()
await self._create_in_transaction(transaction, doc_ref, resource_dict)
log_audit_event("create", user_id, "resource", user_id, "success")
return Resource(**resource_dict)
Read
async def get_resource(self, user_id: str) -> Resource:
"""
Get resource by user ID.
Raises:
ResourceNotFoundError: If resource does not exist.
"""
client = self._get_client()
doc_ref = client.collection(self.collection_name).document(user_id)
snapshot = await doc_ref.get()
if not snapshot.exists:
raise ResourceNotFoundError("Resource not found")
data = snapshot.to_dict()
if not data:
raise ResourceNotFoundError("Resource not found")
return Resource(**data)
Update
Use transactions to ensure atomicity and return merged data:
@staticmethod
@firestore.async_transactional
async def _update_in_transaction( # pragma: no cover
transaction: AsyncTransaction,
doc_ref: AsyncDocumentReference,
updates: dict,
) -> dict | None:
# Tested via E2E tests with Firebase emulators; unit tests mock this method
snapshot = await doc_ref.get(transaction=transaction)
if not snapshot.exists:
return None
existing_data = snapshot.to_dict() or {}
transaction.update(doc_ref, updates)
return {**existing_data, **updates}
async def update_resource(self, user_id: str, resource_data: ResourceUpdate) -> Resource:
"""
Update an existing resource.
Raises:
ResourceNotFoundError: If resource does not exist.
"""
client = self._get_client()
doc_ref = client.collection(self.collection_name).document(user_id)
update_dict = {k: v for k, v in resource_data.model_dump(exclude_unset=True).items() if v is not None}
if not update_dict:
return await self.get_resource(user_id)
update_dict["updated_at"] = datetime.now(UTC)
transaction = client.transaction()
merged_data = await self._update_in_transaction(transaction, doc_ref, update_dict)
if merged_data is None:
raise ResourceNotFoundError("Resource not found")
log_audit_event("update", user_id, "resource", user_id, "success")
return Resource(**merged_data)
Delete
@staticmethod
@firestore.async_transactional
async def _delete_in_transaction( # pragma: no cover
transaction: AsyncTransaction,
doc_ref: AsyncDocumentReference,
) -> dict | None:
# Tested via E2E tests with Firebase emulators; unit tests mock this method
snapshot = await doc_ref.get(transaction=transaction)
if not snapshot.exists:
return None
data = snapshot.to_dict()
transaction.delete(doc_ref)
return data
async def delete_resource(self, user_id: str) -> Resource:
"""
Delete a resource by user ID.
Raises:
ResourceNotFoundError: If resource does not exist.
"""
client = self._get_client()
doc_ref = client.collection(self.collection_name).document(user_id)
transaction = client.transaction()
deleted_data = await self._delete_in_transaction(transaction, doc_ref)
if deleted_data is None:
raise ResourceNotFoundError("Resource not found")
log_audit_event("delete", user_id, "resource", user_id, "success")
return Resource(**deleted_data)
Audit Logging
Use log_audit_event() from app.middleware for security-relevant operations:
from app.middleware import log_audit_event
# After successful operation
log_audit_event("create", user_id, "resource", resource_id, "success")
log_audit_event("update", user_id, "resource", resource_id, "success")
log_audit_event("delete", user_id, "resource", resource_id, "success", details={"reason": "user_request"})
Dependency Registration
Register the service in app/dependencies.py:
from typing import Annotated
from fastapi import Depends
from app.services.resource import ResourceService
def get_resource_service() -> ResourceService:
"""
Dependency provider for ResourceService.
"""
return ResourceService()
ResourceServiceDep = Annotated[ResourceService, Depends(get_resource_service)]
Collection Constants
Define collection names in the response model file:
# app/models/resource/responses.py
RESOURCE_COLLECTION = "resources"
Import in service:
from app.models.resource import RESOURCE_COLLECTION
Type Hints
Use TYPE_CHECKING for Firestore types to avoid import issues:
from typing import TYPE_CHECKING
if TYPE_CHECKING:
from google.cloud.firestore import AsyncClient, AsyncDocumentReference, AsyncTransaction
Testing
Transaction methods are tested via E2E tests with Firebase emulators. Add # pragma: no cover comment and explanatory note:
@staticmethod
@firestore.async_transactional
async def _create_in_transaction( # pragma: no cover
transaction: AsyncTransaction,
doc_ref: AsyncDocumentReference,
data: dict,
) -> None:
# Tested via E2E tests with Firebase emulators; unit tests mock this method
...
Unit tests mock the transactional methods or the service itself:
from unittest.mock import AsyncMock
mock_service = AsyncMock(spec=ResourceService)
mock_service.create_resource.return_value = make_resource()