This project is a demonstration of a project generated with FASTAPI-MVC. The generated application has been modified to have the following characteristics:

1. Use Loguru instead of the standard Python logging module.
2. That the logs generated by both Gunicorn and the application had the same format and were controlled by Loguru.
3. Safe operation of the logging system with async (Gunicorn/Uvicorn) and with threads/multiprocessing (Gunicorn), even logging to file.
4. Generate an X-Request-ID for each request received in Fastapi. This was done with middleware created with asgi-correlation-id.
5. Have a development environment where the application server does a hot reload when the source changes.

Below an example of the log generate by the application:

Added these files:

This is an custom config for Gunicorn to use a custom Logger class. The relevant change is:

logger_class  = "mvc_demo.core.loguru_logs.StubbedGunicornLogger"

This file instructs Gunicorn to use a custom Logger class. In this file are the Gunicorn custom Logger class StubbedGunicornLogger and all code needed to configure Loguru to intercept Gunicor/Uvicorn logs and to create an custom log format for the app.

The relevant change is:

• Get extra info to use in the log format

def set_log_extras(record):
    record["extra"]["datetime"] = datetime.now(timezone.utc)   # Log datetime in UTC time zone, even if server is using another timezone
    record["extra"]["host"] = os.getenv('HOSTNAME', os.getenv('COMPUTERNAME', platform.node())).split('.')[0]
    record["extra"]["pid"] = os.getpid()
    record["extra"]["request_id"] = correlation_id.get()
    record["extra"]["app_name"] = settings.PROJECT_NAME

Note: correlation_id.get() is from the asgi-correlation-id middleware.

• Define the format of log lines

def format_record(record: dict) -> str:
    ...
    format_string = "<green>{extra[datetime]}</green> | <green>{extra[app_name]}</green> | <green>{extra[host]}</green> | <green>{extra[pid]}</green> | <green>{extra[request_id]}</green> | <level>{level: <8}</level> | <cyan>{name}</cyan>:<cyan>{function}</cyan>:<cyan>{line}</cyan> - <level>{message}</level>"
    ...

• Define the format of log lines

def format_record(record: dict) -> str:
    ...
    format_string = "<green>{extra[datetime]}</green> | <green>{extra[app_name]}</green> | <green>{extra[host]}</green> | <green>{extra[pid]}</green> | <green>{extra[request_id]}</green> | <level>{level: <8}</level> | <cyan>{name}</cyan>:<cyan>{function}</cyan>:<cyan>{line}</cyan> - <level>{message}</level>"
    ...

• Use Loguru enqueue to work smoothless with async/threads/multiprocessing

def global_log_config(log_level: Union[str, int] = logging.INFO, json: bool = True):
    ...
    else:
        logger.configure(
            handlers=[
                {
                    "sink": stdout,
                    # https://loguru.readthedocs.io/en/stable/api/logger.html#sink
                    # "sink": "./somefile.log",
                    # "rotation": "10 MB",
                    "serialize": False,
                    # "format": "<green>{extra[datetime]}</green> | <green>{extra[correlation_id]}</green> | <green>{extra[request_id]}</green> | <level>{level: <8}</level> | <cyan>{name}</cyan>:<cyan>{function}</cyan>:<cyan>{line}</cyan> - <level>{message}</level>",
                    "format": format_record,
                    "diagnose": True,
                    "backtrace": True,
                    "enqueue": True,
                }
            ]
        )
    ...

and changed:

...
from mvc_demo.core.loguru_logs import global_log_config

global_log_config(
    log_level=logging.getLevelName(settings.LOG_LEVEL),
    json=settings.JSON_LOGS,
)

log = logger
...

changed:

...

    # Register middlewares
    app.add_middleware(CorrelationIdMiddleware, header_name='X-Request-ID')

...

Added this file:

This is an custom module to start a development Uvicorn server for the app with reload enabled.

and changed:

...
[tool.poetry.scripts]
mvc-demo = 'mvc_demo.cli.cli:cli'
mvc-demo-dev = 'mvc_demo.wsgi_uvicorn:run_dev_wsgi'
...

First run poetry install in project root directory to install all dependencies.

$ poetry install
$ poetry shell
$ mvc-demo serve --config mvc_demo/config/gunicorn_loguru.conf.py

Note: this process may take a while on first run.

$ mvc-demo serve --config mvc_demo/config/gunicorn_loguru.conf.py

$ mvc-demo-dev

If You want to go easy way and use provided virtualized environment You'll need to have installed:

• Python >= 3.9
• Poetry

Otherwise, for local complete project environment with k8s infrastructure bootstrapping You'll need to follow these instructions: fastapi-mvc

For application:

• Python 3.9 or later installed How to install python
• Poetry How to install poetry

With make:

make install

You can customize poetry installation with environment variables

export POETRY_HOME=/custom/poetry/path
export POETRY_CACHE_DIR=/custom/poetry/path/cache
export POETRY_VIRTUALENVS_IN_PROJECT=true
make install

Or using poetry directly:

poetry install

This package exposes simple CLI for easier interaction:

$ mvc-demo --help
Usage: mvc-demo [OPTIONS] COMMAND [ARGS]...

  mvc-demo CLI root.

Options:
  -v, --verbose  Enable verbose logging.
  --help         Show this message and exit.

Commands:
  serve  mvc-demo CLI serve command.
$ mvc-demo serve --help
Usage: mvc-demo serve [OPTIONS]

  mvc-demo CLI serve command.

Options:
  --host TEXT                  Host to bind.  [default: 127.0.0.1]
  -p, --port INTEGER           Port to bind.  [default: 8000]
  -w, --workers INTEGER RANGE  The number of worker processes for handling
                               requests.  [default: 2]

  -D, --daemon                 Daemonize the Gunicorn process.
  -e, --env TEXT               Set environment variables in the execution
                               environment.

  -c, --config PATH            Uses a custom gunicorn.conf.py configuration
  --pid PATH                   Specifies the PID file.
  --help                       Show this message and exit.

NOTE: Maximum number of workers may be different in your case, it's limited to multiprocessing.cpu_count()

To serve application simply run:

$ mvc-demo serve

To confirm it's working:

$ curl localhost:8000/api/ready
{"status":"ok"}

This repository provides Dockerfile for virtualized environment.

NOTE: Replace podman with docker if it's yours containerization engine.

$ make image
$ podman run -dit --name mvc-demo -p 8000:8000 mvc-demo:$(cat TAG)
f41e5fa7ffd512aea8f1aad1c12157bf1e66f961aeb707f51993e9ac343f7a4b
$ podman ps
CONTAINER ID  IMAGE                                 COMMAND               CREATED        STATUS            PORTS                   NAMES
f41e5fa7ffd5  localhost/mvc-demo:0.1.0  /usr/bin/fastapi ...  2 seconds ago  Up 3 seconds ago  0.0.0.0:8000->8000/tcp  mvc-demo
$ curl localhost:8000/api/ready
{"status":"ok"}

This application provides flexibility of configuration. All significant settings are defined by the environment variables, each with the default value. Moreover, package CLI allows overriding core ones: host, port, workers. You can modify all other available configuration settings in the gunicorn.conf.py file.

Priority of overriding configuration:

1. cli
2. environment variables
3. gunicorn.conf.py

All application configuration is available in mvc_demo.config submodule.

1. Source: mvc_demo.config/gunicorn.conf.py
2. Gunicorn configuration file documentation

Endpoints are defined in mvc_demo.config.router. Just simply import your controller and include it to FastAPI router:

from fastapi import APIRouter
from mvc_demo.app.controllers.api.v1 import ready

router = APIRouter(
    prefix="/api"
)

router.include_router(ready.router, tags=["ready"])

You can implement your own web routes logic straight away in mvc_demo.app.controllers.api.v1 submodule. For more information please see FastAPI documentation.

For your discretion, I've provided some basic utilities:

• RedisClient mvc_demo.app.utils.redis
• AiohttpClient mvc_demo.app.utils.aiohttp_client

They're initialized in asgi.py on FastAPI startup event handler:

async def on_startup():
    """Fastapi startup event handler.

    Creates RedisClient and AiohttpClient session.

    """
    log.debug("Execute FastAPI startup event handler.")
    # Initialize utilities for whole FastAPI application without passing object
    # instances within the logic. Feel free to disable it if you don't need it.
    if settings.USE_REDIS:
        await RedisClient.open_redis_client()

    AiohttpClient.get_aiohttp_client()


async def on_shutdown():
    """Fastapi shutdown event handler.

    Destroys RedisClient and AiohttpClient session.

    """
    log.debug("Execute FastAPI shutdown event handler.")
    # Gracefully close utilities.
    if settings.USE_REDIS:
        await RedisClient.close_redis_client()

    await AiohttpClient.close_aiohttp_client()

and are available for whole application scope without passing object instances. In order to utilize it just execute classmethods directly.

Example:

from mvc_demo.app.utils import RedisClient

response = RedisClient.get("Key")

from mvc_demo.app.utils import AiohttpClient

response = AiohttpClient.get("http://foo.bar")

Source: mvc_demo.app.exceptions.http.py

This exception combined with http_exception_handler method allows you to use it the same manner as you'd use FastAPI.HTTPException with one difference. You have freedom to define returned response body, whereas in FastAPI.HTTPException content is returned under "detail" JSON key. In this application custom handler is added in asgi.py while initializing FastAPI application. This is needed in order to handle it globally.

All routes documentation is available on:

• / with Swagger
• /redoc or ReDoc.

This project is licensed under the terms of the MIT license.

• https://github.com/rszamszur/fastapi-mvc
• https://pawamoy.github.io/posts/unify-logging-for-a-gunicorn-uvicorn-app/
• https://github.com/pahntanapat/Unified-FastAPI-Gunicorn-Log
• Delgan/loguru#365
• https://loguru.readthedocs.io/en/stable/api/logger.html#sink
• https://github.com/snok/asgi-correlation-id