Implemented automated architecture validation to enforce design decisions: Architecture Validation System: - Created .architecture-rules.yaml with comprehensive rule definitions - Implemented validate_architecture.py script with AST-based validation - Added pre-commit hook configuration for automatic validation - Comprehensive documentation in docs/architecture/architecture-patterns.md Key Design Rules Enforced: - API-001 to API-004: API endpoint patterns (Pydantic models, no business logic, exception handling, auth) - SVC-001 to SVC-004: Service layer patterns (domain exceptions, db session params, no HTTP concerns) - MDL-001 to MDL-002: Model separation (SQLAlchemy vs Pydantic) - EXC-001 to EXC-002: Exception handling (custom exceptions, no bare except) - JS-001 to JS-003: JavaScript patterns (apiClient, logger, Alpine components) - TPL-001: Template patterns (extend base.html) Features: - Validates separation of concerns (routes vs services vs models) - Enforces proper exception handling (domain exceptions in services, HTTP in routes) - Checks database session patterns and Pydantic model usage - JavaScript and template validation - Detailed error reporting with suggestions - Integration with pre-commit hooks and CI/CD UI Fix: - Fixed icon names in content-pages.html (pencil→edit, trash→delete) Documentation: - Added architecture patterns guide with examples - Created scripts/README.md for validator usage - Updated mkdocs.yml with architecture documentation - Built and verified documentation successfully Usage: python scripts/validate_architecture.py # Validate all python scripts/validate_architecture.py --verbose # With details python scripts/validate_architecture.py --errors-only # Errors only 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
18 KiB
Architecture Patterns & Design Decisions
This document describes the architectural patterns and design decisions that must be followed throughout the codebase.
Note: These patterns are enforced automatically by
scripts/validate_architecture.py. Run the validator before committing code.
Table of Contents
- Core Principles
- Layered Architecture
- API Endpoint Patterns
- Service Layer Patterns
- Model Patterns
- Exception Handling
- JavaScript Patterns
- Validation
Core Principles
1. Separation of Concerns
Each layer has specific responsibilities:
- Routes/Endpoints: HTTP handling, validation, authentication, response formatting
- Services: Business logic, data processing, orchestration
- Models: Data structure and persistence
❌ Bad Example - Business logic in endpoint:
@router.post("/vendors")
async def create_vendor(vendor: VendorCreate, db: Session = Depends(get_db)):
# ❌ BAD: Business logic in endpoint
if db.query(Vendor).filter(Vendor.subdomain == vendor.subdomain).first():
raise HTTPException(status_code=409, detail="Vendor exists")
db_vendor = Vendor(**vendor.dict())
db.add(db_vendor)
db.commit()
db.refresh(db_vendor)
return db_vendor
✅ Good Example - Delegated to service:
@router.post("/vendors", response_model=VendorResponse)
async def create_vendor(
vendor: VendorCreate,
current_user: User = Depends(get_current_admin),
db: Session = Depends(get_db)
):
try:
# ✅ GOOD: Delegate to service
result = vendor_service.create_vendor(db, vendor)
return result
except VendorAlreadyExistsError as e:
raise HTTPException(status_code=409, detail=str(e))
except Exception as e:
logger.error(f"Failed to create vendor: {e}")
raise HTTPException(status_code=500, detail="Internal server error")
2. Type Safety
Use Pydantic for API validation, SQLAlchemy for database models.
3. Proper Exception Handling
Services throw domain exceptions, routes convert to HTTP responses.
Layered Architecture
┌─────────────────────────────────────────────────────────┐
│ API Layer │
│ (app/api/v1/**/*.py, app/routes/**/*.py) │
│ │
│ Responsibilities: │
│ - HTTP request/response handling │
│ - Authentication/Authorization │
│ - Input validation (Pydantic models) │
│ - Exception handling (domain → HTTP) │
│ │
│ ❌ Should NOT: │
│ - Contain business logic │
│ - Directly access database (except via services) │
│ - Raise domain exceptions │
└─────────────────────────────────────────────────────────┘
↓
┌─────────────────────────────────────────────────────────┐
│ Service Layer │
│ (app/services/**/*.py) │
│ │
│ Responsibilities: │
│ - Business logic │
│ - Data validation (business rules) │
│ - Database operations │
│ - Orchestration of multiple operations │
│ │
│ ❌ Should NOT: │
│ - Know about HTTP (no HTTPException) │
│ - Create database sessions (accept as parameter) │
│ - Handle HTTP-specific concerns │
└─────────────────────────────────────────────────────────┘
↓
┌─────────────────────────────────────────────────────────┐
│ Data Layer │
│ (app/models/**/*.py) │
│ │
│ Responsibilities: │
│ - Database schema (SQLAlchemy models) │
│ - API schemas (Pydantic models) │
│ - Data structure definitions │
│ │
│ ❌ Should NOT: │
│ - Mix SQLAlchemy and Pydantic in same class │
│ - Contain business logic │
└─────────────────────────────────────────────────────────┘
API Endpoint Patterns
Rule API-001: Use Pydantic Models
All endpoints MUST use Pydantic models for request/response.
# ✅ GOOD: Pydantic models for type safety
class VendorCreate(BaseModel):
name: str = Field(..., max_length=200)
subdomain: str = Field(..., max_length=100)
is_active: bool = True
class VendorResponse(BaseModel):
id: int
name: str
subdomain: str
created_at: datetime
class Config:
from_attributes = True # For SQLAlchemy compatibility
@router.post("/vendors", response_model=VendorResponse)
async def create_vendor(vendor: VendorCreate, db: Session = Depends(get_db)):
result = vendor_service.create_vendor(db, vendor)
return result
# ❌ BAD: Raw dict, no validation
@router.post("/vendors")
async def create_vendor(data: dict):
return {"name": data["name"]} # No type safety!
Rule API-002: No Business Logic in Endpoints
Endpoints should only handle HTTP concerns.
# ✅ GOOD: Delegate to service
@router.post("/vendors")
async def create_vendor(vendor: VendorCreate, db: Session = Depends(get_db)):
result = vendor_service.create_vendor(db, vendor)
return result
# ❌ BAD: Business logic in endpoint
@router.post("/vendors")
async def create_vendor(vendor: VendorCreate, db: Session = Depends(get_db)):
# ❌ Database operations belong in service!
db_vendor = Vendor(**vendor.dict())
db.add(db_vendor)
db.commit()
return db_vendor
Rule API-003: Proper Exception Handling
Catch service exceptions and convert to HTTPException.
# ✅ GOOD: Proper exception handling
@router.post("/vendors", response_model=VendorResponse)
async def create_vendor(vendor: VendorCreate, db: Session = Depends(get_db)):
try:
result = vendor_service.create_vendor(db, vendor)
return result
except VendorAlreadyExistsError as e:
raise HTTPException(status_code=409, detail=str(e))
except ValueError as e:
raise HTTPException(status_code=400, detail=str(e))
except Exception as e:
logger.error(f"Unexpected error creating vendor: {e}")
raise HTTPException(status_code=500, detail="Internal server error")
Rule API-004: Authentication
Protected endpoints must use dependency injection for auth.
# ✅ GOOD: Use Depends for auth
@router.post("/vendors")
async def create_vendor(
vendor: VendorCreate,
current_user: User = Depends(get_current_admin), # ✅ Auth required
db: Session = Depends(get_db)
):
result = vendor_service.create_vendor(db, vendor)
return result
Service Layer Patterns
Rule SVC-001: No HTTPException in Services
Services should NOT know about HTTP. Raise domain exceptions instead.
# ✅ GOOD: Domain exception
class VendorAlreadyExistsError(Exception):
"""Raised when vendor with same subdomain already exists"""
pass
class VendorService:
def create_vendor(self, db: Session, vendor_data: VendorCreate):
if self._vendor_exists(db, vendor_data.subdomain):
raise VendorAlreadyExistsError(
f"Vendor with subdomain '{vendor_data.subdomain}' already exists"
)
# Business logic...
vendor = Vendor(**vendor_data.dict())
db.add(vendor)
db.commit()
return vendor
# ❌ BAD: HTTPException in service
class VendorService:
def create_vendor(self, db: Session, vendor_data: VendorCreate):
if self._vendor_exists(db, vendor_data.subdomain):
# ❌ Service shouldn't know about HTTP!
raise HTTPException(status_code=409, detail="Vendor exists")
Rule SVC-002: Create Custom Exception Classes
Don't use generic Exception. Create specific domain exceptions.
# ✅ GOOD: Specific exceptions
class VendorError(Exception):
"""Base exception for vendor-related errors"""
pass
class VendorNotFoundError(VendorError):
"""Raised when vendor is not found"""
pass
class VendorAlreadyExistsError(VendorError):
"""Raised when vendor already exists"""
pass
class VendorService:
def get_vendor(self, db: Session, vendor_code: str):
vendor = db.query(Vendor).filter(Vendor.vendor_code == vendor_code).first()
if not vendor:
raise VendorNotFoundError(f"Vendor '{vendor_code}' not found")
return vendor
# ❌ BAD: Generic Exception
class VendorService:
def get_vendor(self, db: Session, vendor_code: str):
vendor = db.query(Vendor).filter(Vendor.vendor_code == vendor_code).first()
if not vendor:
raise Exception("Vendor not found") # ❌ Too generic!
return vendor
Rule SVC-003: Database Session as Parameter
Services should receive database session as parameter, not create it internally.
# ✅ GOOD: db session as parameter
class VendorService:
def create_vendor(self, db: Session, vendor_data: VendorCreate):
vendor = Vendor(**vendor_data.dict())
db.add(vendor)
db.commit()
db.refresh(vendor)
return vendor
def _vendor_exists(self, db: Session, subdomain: str) -> bool:
return db.query(Vendor).filter(Vendor.subdomain == subdomain).first() is not None
# ❌ BAD: Creating session internally
class VendorService:
def create_vendor(self, vendor_data: VendorCreate):
# ❌ Don't create session here - makes testing hard
db = SessionLocal()
vendor = Vendor(**vendor_data.dict())
db.add(vendor)
db.commit()
return vendor
Benefits:
- Testability (can inject mock session)
- Transaction control (caller controls commit/rollback)
- Resource management (caller handles session lifecycle)
Rule SVC-004: Use Pydantic for Input Validation
Service methods should accept Pydantic models for complex inputs.
# ✅ GOOD: Pydantic model ensures validation
class VendorService:
def create_vendor(self, db: Session, vendor_data: VendorCreate):
# vendor_data is already validated by Pydantic
vendor = Vendor(**vendor_data.dict())
db.add(vendor)
db.commit()
return vendor
Model Patterns
Rule MDL-001: SQLAlchemy for Database Models
# ✅ GOOD: SQLAlchemy model
from sqlalchemy import Column, Integer, String, Boolean
from app.database import Base
class Vendor(Base):
__tablename__ = "vendors"
id = Column(Integer, primary_key=True, index=True)
name = Column(String(200), nullable=False)
subdomain = Column(String(100), unique=True, nullable=False)
is_active = Column(Boolean, default=True)
Rule MDL-002: Separate Pydantic from SQLAlchemy
NEVER mix SQLAlchemy and Pydantic in the same class.
# ❌ BAD: Mixing SQLAlchemy and Pydantic
class Vendor(Base, BaseModel): # ❌ Don't do this!
__tablename__ = "vendors"
name: str = Column(String(200))
# ✅ GOOD: Separate models
# Database model (app/models/vendor.py)
class Vendor(Base):
__tablename__ = "vendors"
id = Column(Integer, primary_key=True)
name = Column(String(200), nullable=False)
# API model (app/api/v1/admin/vendors.py)
class VendorCreate(BaseModel):
name: str = Field(..., max_length=200)
class VendorResponse(BaseModel):
id: int
name: str
class Config:
from_attributes = True
Exception Handling
Rule EXC-001: Domain-Specific Exceptions
Create exception hierarchy in app/exceptions/
# app/exceptions/vendor_exceptions.py
class VendorError(Exception):
"""Base exception for vendor domain"""
pass
class VendorNotFoundError(VendorError):
"""Vendor does not exist"""
pass
class VendorAlreadyExistsError(VendorError):
"""Vendor already exists"""
pass
class VendorValidationError(VendorError):
"""Vendor data validation failed"""
pass
Rule EXC-002: Never Use Bare Except
# ❌ BAD: Bare except
try:
result = do_something()
except: # ❌ Catches EVERYTHING including KeyboardInterrupt!
pass
# ✅ GOOD: Specific exceptions
try:
result = do_something()
except ValueError as e:
logger.error(f"Validation error: {e}")
except DatabaseError as e:
logger.error(f"Database error: {e}")
except Exception as e:
logger.error(f"Unexpected error: {e}")
raise
JavaScript Patterns
Rule JS-001: Use apiClient Directly
// ✅ GOOD
const vendors = await apiClient.get('/api/v1/vendors');
// ❌ BAD
const vendors = await window.apiClient.get('/api/v1/vendors');
Rule JS-002: Use Centralized Logger
// ✅ GOOD: Centralized logger
const vendorLog = window.LogConfig.createLogger('vendors');
vendorLog.info('Loading vendors...');
vendorLog.error('Failed to load vendors:', error);
// ❌ BAD: console
console.log('Loading vendors...'); // ❌ Use logger instead
Rule JS-003: Alpine Components Pattern
// ✅ GOOD: Proper Alpine component
function vendorsManager() {
return {
// ✅ Inherit base layout functionality
...data(),
// ✅ Set page identifier for sidebar
currentPage: 'vendors',
// Component state
vendors: [],
loading: false,
// ✅ Init with guard
async init() {
if (window._vendorsInitialized) {
return;
}
window._vendorsInitialized = true;
await this.loadVendors();
}
};
}
Validation
Running the Validator
# Validate entire codebase
python scripts/validate_architecture.py
# Validate specific directory
python scripts/validate_architecture.py app/api/
# Verbose output with context
python scripts/validate_architecture.py --verbose
# Errors only (suppress warnings)
python scripts/validate_architecture.py --errors-only
Pre-commit Hook
Install pre-commit to validate automatically before commits:
# Install pre-commit
pip install pre-commit
# Setup hooks
pre-commit install
# Run manually
pre-commit run --all-files
CI/CD Integration
Add to your CI pipeline:
# .github/workflows/ci.yml
- name: Validate Architecture
run: |
python scripts/validate_architecture.py
Quick Reference
| Layer | Responsibility | Can Use | Cannot Use |
|---|---|---|---|
| API Endpoints | HTTP handling, auth, validation | Pydantic, HTTPException, Depends | Direct DB access, business logic |
| Services | Business logic, orchestration | DB session, domain exceptions | HTTPException, HTTP concepts |
| Models | Data structure | SQLAlchemy OR Pydantic | Mixing both in same class |
Common Violations and Fixes
Violation: Business logic in endpoint
# Before
@router.post("/vendors")
async def create_vendor(vendor: VendorCreate, db: Session = Depends(get_db)):
db_vendor = Vendor(**vendor.dict())
db.add(db_vendor)
db.commit()
return db_vendor
# After ✅
@router.post("/vendors")
async def create_vendor(vendor: VendorCreate, db: Session = Depends(get_db)):
try:
return vendor_service.create_vendor(db, vendor)
except VendorAlreadyExistsError as e:
raise HTTPException(status_code=409, detail=str(e))
Violation: HTTPException in service
# Before
class VendorService:
def create_vendor(self, db: Session, vendor_data):
if exists:
raise HTTPException(status_code=409, detail="Exists")
# After ✅
class VendorService:
def create_vendor(self, db: Session, vendor_data):
if exists:
raise VendorAlreadyExistsError("Vendor already exists")
Remember: These patterns are enforced automatically. Run python scripts/validate_architecture.py before committing!