deepl-fastapi-pw

Your own deepl server via fastapi and playwright, cross-platform (Windows/Linux/MacOs) with API for OmegaT

Installation

• Create a virual environment: optional but recommended e.g.,

  # Linux and friends
  python3.8 -m venv .venv
  source .venv/bin/activate
  
  # Windows
  # py -3.8 -m venv .venv
  # .venv\Scripts\activate

pip install deepl-fastapi-pw

or (if your use poetry)

poetry add deepl-fastapi-pw

or

 pip install git+https://github.com/ffreemt/deepl-fastapi-playwright.git

or

• Clone the repo https://github.com/ffreemt/deepl-fastapi-playwright.git

  git clone https://github.com/ffreemt/deepl-fastapi-playwright.git

  and cd deepl-fastapi-playwright
• `pip install -r requirements.txt
  • or poetry install

Usage

• Start the server

python  -m deepl_fastapi_pw

Or use uvicorn directly (note the deepl_server module, not run_uvicorn)

uvicorn deepl_fastapi_pw.deepl_server_async:app

or

python -m deepl_fastapi_pw.deepl_server_async

or run the server on the external net, for example at port 9888

uvicorn deepl_fastapi_pw.deepl_server:app --reload --host 0.0.0.0 --port 9888

Explore and consume

Point your browser to http://127.0.0.1:8001/text/?q=test&to_lang=zh

Or in command line:

python -c "import httpx; print(httpx.get('http://127.0.0.1:8001/text/?to_lang=zh&q=test 1\ntest 2').json())"
# output: {'q': 'test 1\ntest 2', 'from_lang': None, 'to_lang': 'zh', 'trtext': '测试 1\n测试 2', 'translation': '测试 1\n测试 2'}

Or in python code (pip install requests first)

import requests

# get
url = "http://127.0.0.1:8001/text/?q=test me&to_lang=zh"
print(requests.get(url).json())
# {'q': 'test me', 'from_lang': None, 'to_lang':
# 'zh', 'trtext': '考我', 'translation': '考我'}

'translation' is there for OmegaT plugin. Refer to the OmegaT Fake MT plugin setup part in https://github.com/ffreemt/deepl-fastapi

# post
text = "test me \n and him"
data = {"text": text, "to_lang": "zh"}
resp = requests.post("http://127.0.0.1:8001/text", json=data)
print(resp.json())
# {'q': {'text': 'test me \n and him', 'from_lang': None,
# 'to_lang': 'zh', 'description': None}, 'result': '考验我 \n  我和他'}

NB

• Text longer than 1500 characters will be trimmed to 1500. This is what the deepl.com webpage can handle (used to be 5000 characters).
• The package deepl-fastapi-pw is in essence a browser from your local IP (although a proxy can be used, refer to get-pwbroser's doc for setting up). Hence if you send too many requests too frequently, your IP may be blocked by deepl.com temporarily, which of course will also disable deepl-fastapi-pw.

Interactive Docs (Swagger UI)

http://127.0.0.1:8001/docs