tests/vsp/app/scrapers/test_integration_linkedin_downloader.py

import asyncio

import pytest

from vsp.app.scrapers.linkedin_downloader import LinkedinDownloader
from vsp.shared import logger_factory

logger = logger_factory.get_logger(__name__)


@pytest.mark.asyncio
async def test_fetch_linkedin_data_integration():
    """
    Integration test for LinkedinDownloader.

    This test makes an actual network request to fetch the Linkedin profile of 'navkast'.
    It requires a valid RapidAPI key to be set in the AWS Parameter Store.

    Note: This test should be run sparingly to avoid unnecessary API calls and potential rate limiting.
    """
    downloader = LinkedinDownloader()
    linkedin_url = "https://www.linkedin.com/in/navkast/"

    try:
        profile = await downloader.fetch_linkedin_data(linkedin_url)

        # Log the fetched profile data
        logger.info("Fetched Linkedin profile", first_name=profile.first_name, last_name=profile.last_name)

        # Assertions to verify the fetched data
        assert profile.first_name == "Naveen"
        assert profile.last_name == "K."
        assert "University of Pennsylvania" in [edu.school_name for edu in profile.educations]
        logger.info("Integration test passed successfully")
    except Exception as e:
        logger.error("Integration test failed", error=str(e))
        raise


if __name__ == "__main__":
    asyncio.run(test_fetch_linkedin_data_integration())