ocr.py

from datetime import datetime

# pip install fastapi uvicorn python-multipart
from fastapi import FastAPI, File, UploadFile
from fastapi.responses import PlainTextResponse

# pip install paddlepaddle paddleocr
# pip install opencv-python-headless
from paddleocr import PaddleOCR, draw_ocr

app = FastAPI()

# https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.7/doc/doc_ch/quickstart.md
ocr = PaddleOCR(use_angle_cls=True, lang="ch")


@app.post("/")
async def create_upload_file(file: UploadFile = File(...)):
    x = await file.read()
    print(len(x))
    fn = datetime.now().strftime('%y%m%d_%H%M%S.%f')
    fn = 'tmp.jpg'
    with open(fn, "wb") as f:
        f.write(x)
    result = ocr.ocr(fn)
    l = []
    for i in result:
        for x in i:
            l.append(x[-1][0])
    txt = "\n".join(l)
    return PlainTextResponse(txt)