⚡ ML RESTful API template using FastAPI ⚡
A dynamic FastAPI application base template is fast, well tested, and production ready.
Documentation: https://github.com/ab-inbev-analytics/ml-restful-api-template
Source Code: https://github.com/ab-inbev-analytics/ml-restful-api-template
ML RESTful API template using FastAPI which includes below feature.
Use pip
pip install -r requirements.txt
➝ 01. Directory structure & files
➝ 02. Create new conda env & usage of Makefile
➝ 03. Contribution guide
➝ 04. Documentation
➝ 05. Micro-service, micro-frontend & cross functional pods\
One has to finalize where this app will be hosted. Note, AKS is not for a service. Hence it will be kept outside of scope. For AKS related deployment contact with [email protected].
➝ 3 service principles
➝ Service email
➝ Access to Sonar Cloud
➝ Access to Snyk
➝ Access to Check Marx
➝ Azure MySQL cloud DB
➝ Azure container registry
➝ Azure key vault
➝ Azure storage with Blob
➝ Deployment instance
➝ Azure web app
➝ Azure container registry
➝ Azure container instance
➝ Azure VM (with SSH access and required ports opened) + Azure application gateway (WAF2 enabled)\
➝ SPN 1 (AKV spn) should have read write permission over the AKV.
➝ SPN 2 (ACR spn) should have pull and push in scope of permission.
➝ SPN 3 (Deployment SPN) should have permission to deploy a image from ACR to deployment instance.
➝ SPN 1 related info should be stored in GitHub secrets as AZURE_CREDENTIALS.
➝ SPN 2 and SPN 3 client id and client secrets has to be stored in AKV.
➝ SNYK_ORGID and SNYK_TOKEN has to be stored in AKV.
➝ SONAR_TOKEN has to be stored in AKV.
➝ CHECKMARX_TEAMS, CHECKMARX_URL, CHECKMARX_USERNAME, CHECKMARX_PASSWORD and CHECKMARX_CLIENT_SECRET has to be stored in AKV.
➝ Deployment instance's and developers IP has to be whitelisted in the development VM or DB if required.
➝ projectKey, organization, projectName has to be updated in sonar-project.properties
➝ site_name, site_description, site_author has to be updated in mkdocs.yml
➝ If you are running in a VM you can use the default image FROM tiangolo/uvicorn-gunicorn-fastapi:python3.8 but to run this in a cloud component or AKS this has to be changed to a python base image or number of workers has to be restricted because auto-scaling in these are take care by cloud component or KEDA depending on the deployment instance.
➝ VERSION file in project root has to be updated with the latest version. Docker image will pickup version from this file as tags.
➝ One env file has to be created in the project root with the following content,\
PROJECT_NAME=AdditionAPI
BACKEND_CORS_ORIGINS=["http://localhost:8000","https://localhost:8000","http://localhost","https://localhost"]
MYSQL_ROOT_PASSWORD=development
MYSQL_USER=user
MYSQL_PASSWORD=password
MYSQL_HOST=database
MYSQL_PORT=3306
MYSQL_DATABASE=app
DATABASE_URI=sqlite+aiosqlite:///app.db➝ Note: If you are running this from local then only pass DATABASE_URI=sqlite+aiosqlite:///app.db. Otherwise remove this.\
➝ No action required.
➝ AZURE_CREDENTIALS need to be updated in GitHub secret.
➝ VERSION file in project root should be updated.
➝ keyvault value need to be updated.
➝ secrets values has to be updated with SPN details
➝ REGISTRY_LOGIN_SERVER has to be updated in GitHub secret
➝ IMAGE_NAME has to be upto date in GitHub secrets
➝ SNYK_TOKEN has to be upto date.
➝ release_version should pickup the latest version from VERSION file.
➝ No action required.
➝ PYTHON_VERSION, ACR_NAME, TEAM, PRODUCT, COMPONENT, COMPONENT_TYPE, IMAGE_NAME, DEPLOYMENT_TYPE, WIKI_CONTAINER_NAME, DB_CONTAINER_NAME, DATA_CONTAINER_NAME & DB_TYPE has to be updated in the github actions. AZURE_STORAGE_CONNECTION_STRING, SONAR_TOKEN and SNYK_TOKEN has to be upto date in github secrets.
➝ No action required.
➝ No action required.
➝ CHECKMARX_TEAMS, CHECKMARX_URL, CHECKMARX_USERNAME, CHECKMARX_PASSWORD and CHECKMARX_CLIENT_SECRET is required.
➝ PARENT_REPOSITORY,CHECKOUT_BRANCH,PR_AGAINST_BRANCH,OWNER these end variables needs to be updated in github actions. github_token, parent_repository, checkout_branch, pr_against_branch and owner arguments needs to be updated in github actions
➝ We need to define a valid branch tag in branch_pattern argument in ci-branch-naming-checks.yml files. By default branch tags are,
➝ feat
➝ fix
➝ docs
➝ style
➝ docs
➝ refactor
➝ perf
➝ test
➝ build
➝ ci
➝ chore
➝ revert
➝ Valid example of a feature branch is feat/vehicle-mix-chart-integration-AB#1237. Here AB#<ticket id> will like the feature with the Azure DevOps.
Whenever there is a push to the develop branch this action will do the following things in a github actions runner,
➝ Download the latest code
➝ Login to Azure using Azure CLI
➝ Extract the latest version from VERSION file in id getbuildversion as variable version. This will be used later.
➝ The runner will login to the ACR using SPN's client id and secrets.
➝ This will build two images one with version tag and another with latest tags.
➝ This will run a snyk scan of vulnerability.
➝ If the snyk scan is working this will push the image with both the tags to the azure container registry.
This workflow will delete all the older workflow run from github actions section. Check the parameters of Mattraks/delete-workflow-runs@v2, do modify the default behavior.
➝ No action required.
.
├── app # Fast API project directory
│ ├── core # All the core component can be stored here
│ │ ├── config.py # Application level config will be here
│ │ └── __init__.py # __init__ file
│ ├── database.py # All the database related code will be here
│ ├── __init__.py # __init__ file
│ ├── main.py # Fast API main entrypoint file
│ ├── models.py # All the database will be defined in this file
│ ├── routers # All the routers or sub-modules can be stored here
│ ├── schemas # All the request and response models will be defined here
│ ├── services # Any functions related to a service can be stored here
│ ├── settings.py # Application level settings can be configured here
│ └── utils # Any common utility script can be stored here
├── changelog-ci-config.yml # This file is for generating change logs
├── CHANGELOG.md # Automatically generated change log file
├── CODE_OF_CONDUCT.md # Code of conduct
├── CODEOWNERS # Define your code owners here
├── CONTRIBUTING.md # How to contribute to this project
├── docker-compose.yml # Docker compose file to run the service in local with MySQL DB
├── Dockerfile # Standalone docker file for the respective service
├── docs # All the raw documents for wiki should be stored here
│ ├── index.md # Wiki docs
│ ├── javascripts # Additional JS files for wiki
│ │ └── extra.js # Additional JS files for wiki
│ └── stylesheets # Additional CSS files for wiki
│ └── extra.css # Additional CSS files for wiki
├── LICENSE # LICENSE
├── Makefile # Makefile with all helper functions for the project
├── MANIFEST.in # MANIFEST file
├── pyproject.toml # Project level meta-data and tool's configuration
├── README.md # README file
├── requirements.txt # Python dependency file
├── SECURITY.md # SECURITY policy file
├── setup.cfg # Python project setup, other meta-data and configs
├── sonar-project.properties # Sonar Cloud configurations
├── tests # Project tests directory
│ ├── conftest.py # Test related configuration file
│ ├── __init__.py # __init__ file
│ └── test_main.py # Test script for unit tests related to `main.py`
└── VERSION # API versionpytestscd ml-restful-api-template
uvicorn app.main:app --reload
cd ml-restful-api-template
alembic init alembic
vi alembic/script.py.mako
-> Add import statement
import sqlmodel # NEW
vi alembic/env.py
-> Replace target_metadata=None statement
from sqlmodel import SQLModel # NEW
from app.models import ➝ # NEW
target_metadata = SQLModel.metadata # NEW
import os # NEW
from dotenv import load_dotenv # NEW
load_dotenv() # NEW
config.set_main_option('sqlalchemy.url', os.getenv("DATABASE_URL")) # NEWalembic revision --autogenerate -m "First migration."
alembic upgrade head
python cliapp.py ping
python cliapp.py heartbeat
docker build -t abimtmroigbdevacr.azurecr.io/ml-restful-api-template:latest .
docker-compose -f docker-compose.yml down
docker-compose -f docker-compose.yml build
docker-compose -f docker-compose.yml up -d --remove-orphan
https://testdriven.io/blog/fastapi-sqlmodel/
➝ Restructure the folders
➝ Added SQLModel Models
➝ Added Async alembic MYSQL Migrations
➝ Updated Model Parameters
➝ Added CRUD REST APIs
➝ Added Background Tasks
➝ Added CLI App to trigger the APIs
➝ Added DockerFile
➝ Added DockerCompose File
➝ Seggregate CRUD Services from REST APIs
➝ Converted Sync Calls to Async Calls
➝ Integrate SQLAdmin(Fastapisqladmin)
➝ Integrate fastapi basic auth(user name & password for app)
➝ Integrate client id & client secret(fastapi simple security)\
➝ Versioning(fastapi versioning)
➝ fastapi utils(middleware)
➝ locust
➝ https://github.com/aminalaee/sqladmin\
➝ Add index to the model
➝ db integration docker compose
➝ Locust & coverage report\