django-patterns

Django & DRF Professional Patterns

Core Expertise

• Modern Django: 5.x features, async views/middleware, ASGI deployment.
• Background & Real-time: Celery integration, Django Channels.
• ORM Optimization: select_related, prefetch_related, custom managers.
• Security: JWT auth, OAuth2, RBAC, protection against SQLi/XSS/CSRF.

Critical rules (non-obvious)

• N+1 queries: always use select_related (FK) / prefetch_related (M2M) — never iterate and query inside loops
• get_or_create race condition: wrap in transaction.atomic() in concurrent environments
• Never call save() inside pre_save signal — causes infinite recursion; use update_fields
• bulk_create skips signals and save() — don't use when signal logic is required
• Migrations on large tables: use RunSQL with CONCURRENTLY index creation to avoid locks

ORM: select_related vs prefetch_related

# FK / OneToOne → select_related (JOIN)
books = Book.objects.select_related("author", "author__publisher").all()

# M2M / reverse FK → prefetch_related (separate query)
authors = Author.objects.prefetch_related("books", "books__tags").all()

# Custom prefetch with queryset
from django.db.models import Prefetch
Author.objects.prefetch_related(
    Prefetch("books", queryset=Book.objects.filter(published=True), to_attr="active_books")
)

ORM: annotations and aggregations

from django.db.models import Count, Avg, Q, F, Value
from django.db.models.functions import Coalesce

Author.objects.annotate(
    book_count=Count("books"),
    avg_rating=Coalesce(Avg("books__rating"), Value(0.0)),
    high_rated=Count("books", filter=Q(books__rating__gte=4)),
).filter(book_count__gt=0).order_by("-book_count")

ORM: F expressions (avoid race conditions)

# BAD — race condition
product = Product.objects.get(pk=pk)
product.stock -= quantity
product.save()

# GOOD — atomic at DB level
Product.objects.filter(pk=pk).update(stock=F("stock") - quantity)