惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Microsoft Azure Blog
Microsoft Azure Blog
GbyAI
GbyAI
P
Proofpoint News Feed
Engineering at Meta
Engineering at Meta
Recent Announcements
Recent Announcements
L
LangChain Blog
B
Blog
阮一峰的网络日志
阮一峰的网络日志
Microsoft Security Blog
Microsoft Security Blog
博客园 - 【当耐特】
M
MIT News - Artificial intelligence
D
Docker
WordPress大学
WordPress大学
J
Java Code Geeks
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
The GitHub Blog
The GitHub Blog
博客园 - 叶小钗
Last Week in AI
Last Week in AI
Stack Overflow Blog
Stack Overflow Blog
有赞技术团队
有赞技术团队
MyScale Blog
MyScale Blog
H
Hackread – Cybersecurity News, Data Breaches, AI and More
MongoDB | Blog
MongoDB | Blog
博客园 - Franky

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
FastAPI doesn't speak your users' language. Here's how to...
Radomir Brkovic · 2026-06-25 · via DEV Community

Most FastAPI tutorials end at "it works in English." But the moment you ship to users in Germany, Brazil, or Japan, you realize FastAPI has no built-in answer for internationalization — not in models, not in validation errors, not in API responses.

This post shows exactly how FastKit Core solves i18n end-to-end: translatable database fields, translated validation errors, and a standardized response format that frontend teams actually love working with.


The problem in concrete terms

Let's say you're building a product catalog for a multinational SaaS. You need article titles in English, German, and French. Here's what you'd normally do in FastAPI:

# The "usual" approach — store as JSON manually
class Article(Base):
    __tablename__ = "articles"
    id: Mapped[int] = mapped_column(primary_key=True)
    title: "Mapped[dict] = mapped_column(JSON)  # {\"en\": \"...\", \"de\": \"...\", \"fr\": \"...\"}"

# Then manually handle locale everywhere
article = session.get(Article, 1)
locale = request.headers.get("Accept-Language", "en")[:2]
title = article.title.get(locale) or article.title.get("en")  # manual fallback

And then in every validation error response, you return Pydantic's default English-only messages. And then your frontend has to parse inconsistent JSON shapes across endpoints.

Three separate problems — all unsolved by FastAPI out of the box.


Solution 1: TranslatableMixin — i18n directly in your SQLAlchemy model

FastKit Core ships a TranslatableMixin that makes multi-language fields completely transparent. You declare which fields are translatable, and from then on they behave like normal strings — reading and writing the current locale automatically.

from sqlalchemy import JSON
from sqlalchemy.orm import Mapped, mapped_column
from fastkit_core.database import Base, BaseWithTimestamps, IntIdMixin, TranslatableMixin

class Article(Base, IntIdMixin, BaseWithTimestamps, TranslatableMixin):
    __translatable__ = ['title', 'content']
    __fallback_locale__ = 'en'

    title: Mapped[dict] = mapped_column(JSON)
    content: Mapped[dict] = mapped_column(JSON)
    author_id: Mapped[int]

That's the entire model. Now here's how you use it:

article = Article()

# Write English
article.title = "How to build scalable APIs"

# Switch to German and write
article.set_locale('de')
article.title = "Wie man skalierbare APIs entwickelt"

# Switch to French
article.set_locale('fr')
article.title = "Comment construire des APIs évolutives"

# Read back — always returns current locale
article.set_locale('en')
print(article.title)  # "How to build scalable APIs"

article.set_locale('de')
print(article.title)  # "Wie man skalierbare APIs entwickelt"

# Get all translations at once
print(article.get_translations('title'))
# {'en': 'How to build scalable APIs', 'de': 'Wie man skalierbare APIs entwickelt', 'fr': 'Comment construire des APIs évolutives'}

What's actually happening under the hood: the mixin overrides __getattribute__ and __setattr__ to intercept reads and writes on translatable fields, storing them in a {"locale": "value"} dict that gets serialized to a JSON column. SQLAlchemy event listeners handle serialization before insert/update and deserialization after load.

The result is transparent — you write article.title = "Hello" and the mixin takes care of the rest.

Locale from request — automatic with LocaleMiddleware

FastKit Core ships a ready-made LocaleMiddleware that you register once and forget. Add it to your app and every request automatically sets the correct locale for both model reads and translation lookups:

from fastapi import FastAPI
from fastkit_core.http import LocaleMiddleware

app = FastAPI()
app.add_middleware(LocaleMiddleware)

That's it. No custom middleware to write, no set_locale() calls scattered through your handlers.

The middleware resolves locale with the following priority:

  1. Accept-Language request header (e.g. de, fr, en)
  2. ?lang= query parameter (e.g. /articles?lang=de)
  3. locale cookie
  4. Default: en This means a German user whose browser sends Accept-Language: de gets German responses automatically. A user who explicitly visits /articles?lang=fr gets French. A returning user with a locale cookie set during login gets their saved preference. All without touching a single route handler.

After this middleware is registered, every TranslatableMixin model in every request automatically reads and writes the correct locale. No passing locale around. No article.title.get(locale) everywhere.

Fallback behavior

If a translation doesn't exist for the requested locale, it falls back to __fallback_locale__ automatically:

article.set_locale('es')  # Spanish not set
print(article.title)  # "How to build scalable APIs" — English fallback


Solution 2: Translated validation errors with _()

FastAPI + Pydantic gives you validation out of the box, but errors always come out in English, with Pydantic's internal message format. If you're serving a German user, they get:

{
  "detail": [
    {
      "loc": ["body", "email"],
      "msg": "value is not a valid email address",
      "type": "value_error.email"
    }
  ]
}

FastKit Core replaces this with a Laravel-style _() helper backed by JSON translation files, and a BaseSchema that formats errors into a clean, consistent structure.

Setting up translation files

translations/
├── en.json
├── de.json
└── fr.json

// translations/en.json
{
  "validation": {
    "required": "The {field} field is required.",
    "string_too_short": "The {field} field must be at least {min_length} characters.",
    "string_too_long": "The {field} field must not exceed {max_length} characters.",
    "email": "The {field} field must be a valid email address.",
    "value_error": "{field} is invalid."
  },
  "messages": {
    "welcome": "Welcome, {name}!",
    "article_created": "Article created successfully."
  }
}

// translations/de.json
{
  "validation": {
    "required": "Das Feld {field} ist erforderlich.",
    "string_too_short": "Das Feld {field} muss mindestens {min_length} Zeichen lang sein.",
    "string_too_long": "Das Feld {field} darf maximal {max_length} Zeichen haben.",
    "email": "Das Feld {field} muss eine gültige E-Mail-Adresse sein.",
    "value_error": "{field} ist ungültig."
  },
  "messages": {
    "welcome": "Willkommen, {name}!",
    "article_created": "Artikel erfolgreich erstellt."
  }
}

Using _() in your code

from fastkit_core.i18n import _, set_locale

set_locale('en')
print(_('messages.welcome', name='Maria'))
# "Welcome, Maria!"

set_locale('de')
print(_('messages.welcome', name='Maria'))
# "Willkommen, Maria!"

The locale context is shared between _() and TranslatableMixin — setting it once via middleware affects both model reads and translation lookups.

BaseSchema — structured, translated validation errors

BaseSchema extends Pydantic's BaseModel with a format_errors() classmethod that maps Pydantic's internal error types to your translation keys and returns a clean dict:

from pydantic import EmailStr
from fastkit_core.validation import BaseSchema

class ArticleCreate(BaseSchema):
    title: str
    email: EmailStr
    content: str

When validation fails, instead of calling exc.errors() and getting Pydantic's raw output, you call format_errors():

from pydantic import ValidationError
from fastkit_core.validation import BaseSchema
from fastkit_core.i18n import set_locale

set_locale('de')

try:
    ArticleCreate(title="", email="not-an-email", content="Hello")
except ValidationError as exc:
    errors = ArticleCreate.format_errors(exc)
    print(errors)

Output (in German):

{
  "email": ["Das Feld email muss eine gültige E-Mail-Adresse sein."],
  "title": ["Das Feld title ist erforderlich."]
}

Same code, different locale set in middleware → different language in errors. No if/else, no per-field translations.


Solution 3: Standardized response format

The third piece of the puzzle is consistent API responses. FastKit Core provides four response helpers that enforce the same JSON envelope across every endpoint:

from fastkit_core.http import success_response, error_response, paginated_response

success_response

@router.get("/{id}")
async def show(id: int, service: ArticleService = Depends(get_service)):
    article = await service.get_or_404(id)
    return success_response(data=article)

{
  "success": true,
  "data": {
    "id": 1,
    "title": "How to build scalable APIs",
    "content": "...",
    "author_id": 42
  }
}

paginated_response

@router.get("")
async def index(page: int = 1, service: ArticleService = Depends(get_service)):
    items, meta = await service.paginate(page=page, per_page=20)
    return paginated_response(items=items, pagination=meta)

{
  "success": true,
  "data": [...],
  "pagination": {
    "page": 1,
    "per_page": 20,
    "total": 143,
    "total_pages": 8,
    "has_next": true,
    "has_prev": false
  }
}

error_response — with translated validation errors

from fastkit_core.http import error_response
from fastkit_core.i18n import _

@router.post("")
async def store(data: ArticleCreate, service: ArticleService = Depends(get_service)):
    try:
        article = await service.create(data)
        return success_response(data=article, status_code=201)
    except ValidationError as exc:
        errors = ArticleCreate.format_errors(exc)
        return error_response(
            message=_('validation.failed'),
            errors=errors,
            status_code=422
        )

{
  "success": false,
  "message": "Validierung fehlgeschlagen.",
  "errors": {
    "title": ["Das Feld title ist erforderlich."],
    "email": ["Das Feld email muss eine gültige E-Mail-Adresse sein."]
  }
}

The frontend team gets success: true/false to branch on, data always at the same key, errors always as { field: [messages] }, and pagination always in the same shape. No defensive parsing, no "what does this endpoint return?" questions.


Putting it all together — a full multilingual endpoint

First, register the middleware once in your main.py:

from fastapi import FastAPI
from fastkit_core.http import LocaleMiddleware

app = FastAPI()
app.add_middleware(LocaleMiddleware)

That's the entire setup. From this point, every request automatically carries the correct locale through your entire stack — models, translations, validation errors, and responses.

The router itself stays completely clean:

from fastapi import APIRouter, Depends
from fastkit_core.database import get_async_db
from fastkit_core.http import success_response, paginated_response
from sqlalchemy.ext.asyncio import AsyncSession
from .service import ArticleService
from .schemas import ArticleCreate, ArticleResponse

router = APIRouter(prefix='/articles', tags=['Articles'])

def get_service(session: AsyncSession = Depends(get_async_db)) -> ArticleService:
    return ArticleService(session)

@router.get("")
async def index(
    page: int = 1,
    service: ArticleService = Depends(get_service)
):
    items, meta = await service.paginate(page=page, per_page=20)
    return paginated_response(items=items, pagination=meta)

@router.get("/{id}")
async def show(id: int, service: ArticleService = Depends(get_service)):
    article = await service.get_or_404(id)
    return success_response(data=article)

@router.post("", status_code=201)
async def store(data: ArticleCreate, service: ArticleService = Depends(get_service)):
    article = await service.create(data)
    return success_response(data=article, status_code=201)

A German user hitting GET /articles with Accept-Language: de gets article titles in German. An English user gets them in English. Validation errors come back in the right language automatically. The response shape is identical either way.


Why this matters beyond convenience

In a multinational SaaS, i18n is not a feature you add later — it's infrastructure. Retrofitting it after the fact means touching every model, every validation schema, every response, and every frontend integration.

FastKit Core makes it a day-one decision with almost no ceremony:

  • One mixin on your model declares which fields are translatable
  • One middleware sets the locale per request
  • One base schema formats errors in the right language
  • Four response helpers standardize every endpoint output The patterns were built from real production experience migrating Laravel applications to FastAPI — where Laravel's i18n and response conventions were things we sorely missed.

Getting started

pip install fastkit-core

Full documentation at fastkit.org/docs — including the TranslationManager API, locale middleware configuration, and how TranslatableMixin works with async sessions.

If this solved a problem you've been working around, a ⭐ on GitHub goes a long way for a small open-source project.