Async requests

[JR-1991]

Changes this PR introduces

This pull request adds async support to pyDataverse based on the previously merged switch from requests to httpx. Asynchronous requests significantly improve pyDataverse's performance by using concurrent requests. The implementation provides a seamless usage to work either sync or async by the usage of a context manager.

Small benchmark

• 100 calls to api/info/version to demo.dataverse.org
• Synchronous --> ~35s
• Asynchronous --> ~0.5s 🏎️

Overview

• Added _sync_request and _async_request base methods
• Added Context manager to API to enable async requests
• Small test to verify async requests are working

Example

Synchronous

native_api = NativeApi("https://demo.dataverse.org")
response = native_api.get_info_version()

Asynchronous

import asyncio
from pyDataverse.api import NativeApi


async def run_async(native_api):
    async with native_api:
        tasks = [native_api.get_info_version() for _ in range(100)]
        responses = await asyncio.gather(*tasks)

if __name__ == "__main__":
    native_api = NativeApi("https://demo.dataverse.org")
    asyncio.run(run_async(native_api))

[haeussma]

Tested the example with python=3.11.

Notices missing dependency urllib3 which is not specified in the requirements but imported in src/pyDataverse/__init__.py.

Except for that the example works as expected.

[haeussma]

Tested the example with python=3.11.

Notices missing dependency urllib3 which is not specified in the requirements but imported in src/pyDataverse/__init__.py.

Except for that the example works as expected.

[JR-1991]

@haeussma thanks for the review! I have added urllib3 to the list of dependencies.