Setting charset in FastAPI Response

FastAPI

Recently, I've been using FastAPI package to try out REST API. As its name implies, FastAPI is one of the most popular and a go-to package among Python developers to create APIs.

Encodings

As it happens, for non-Western languages (or all but English?), the original characters are shown as weird for encoding issues.

Recall encoding option in pandas's read_csv() or charset in HTTP Response?

With default setting, this is how non-Western languages are displayed.

[{"region": "KR", "symbol": "005930", "name": "ì‚¼ì„±ì „ìž", "valid": "OK"}]

The original data was supposed to be like this.

[{"region": "KR", "symbol": "005930", "name": "삼성전자", "valid": "OK"}]

Default operation

By default, FastAPI parses the Python dict data structure into JSON ouput.

In that way, it cannot display the output in UTF-8 charset.

from fastapi import FastAPI
import pandas as pd
import uvicorn
 
# Initialize FastAPI app
app = FastAPI()
 
# Front API
@app.get('/')
async def welcome():
    return {'msg':f'Welcome to {title}'}
 
# Sample ticker
@app.get('/tickers/')
async def tickers():
    sample_data = pd.DataFrame([{"region": "KR", "symbol": "005930", "name": "삼성전자", "valid": "OK"}])
    content = sample_data.to_dict('records')
    return content

Response() to the rescue

To enable custom charset of the HTTP Response, Response() function from FastAPI can be used along with json.dumps().

To make an HTTP output, original Python data structure (e.g., dict) should be first converted into str by json.dumps() (with ensure_ascii=False) and then it is transmitted with FastAPI's Response() function.

Here, encoding charset option can be added in the media_type input param of Response() function.

Specifically,

sample_dict = {'결과':'정상'}
output = json.dumps(sample_dict,ensure_ascii=False)
return Response(content=output,media_type='application/json;charset=utf-8')

Here, to display the non-Western characters correctly, I added the conversion of Python dict to str and Response() function to return the result rather than directly returning the dict.

If we back to the original sample, it can be rewritten as this.

from fastapi import FastAPI, Response
import pandas as pd
import uvicorn
 
# Initialize FastAPI app
app = FastAPI()
 
# On startup
@app.on_event('startup')
async def startup_event():
    global content_type
    content_type = f'application/json;charset=utf-8'
 
# Front API
@app.get('/')
async def welcome():
    content_dict = {'msg':f'Welcome to {title}'}
    content = json.dumps(content_dict,ensure_ascii=False)
    return Response(content=content, media_type=content_type)
 
# Sample ticker
@app.get('/tickers/')
async def tickers():
    sample_data = pd.DataFrame([{"region": "KR", "symbol": "005930", "name": "삼성전자", "valid": "OK"}])
    content_dict = sample_data.to_dict('records')
    content = json.dumps(content_dict,ensure_ascii=False)
    return Response(content=content, media_type=content_type)

Unlike in the original code, I added @app.on_event('startup') to add global variable content_type which is being used for every API call functions.