test_middleware.py 文件源码-python代码片段

def test_splash_request():
    mw = _get_mw()
    cookie_mw = _get_cookie_mw()

    req = SplashRequest("http://example.com?foo=bar&url=1&wait=100")
    assert repr(req) == "<GET http://example.com?foo=bar&url=1&wait=100>"

    # check request preprocessing
    req2 = cookie_mw.process_request(req, None) or req
    req2 = mw.process_request(req2, None) or req2
    assert req2 is not None
    assert req2 is not req
    assert req2.url == "http://127.0.0.1:8050/render.html"
    assert req2.headers == {b'Content-Type': [b'application/json']}
    assert req2.method == 'POST'
    assert isinstance(req2, SplashRequest)
    assert repr(req2) == "<GET http://example.com?foo=bar&url=1&wait=100 via http://127.0.0.1:8050/render.html>"

    expected_body = {'url': req.url}
    assert json.loads(to_native_str(req2.body)) == expected_body

    # check response post-processing
    response = TextResponse("http://127.0.0.1:8050/render.html",
                            # Scrapy doesn't pass request to constructor
                            # request=req2,
                            headers={b'Content-Type': b'text/html'},
                            body=b"<html><body>Hello</body></html>")
    response2 = mw.process_response(req2, response, None)
    response2 = cookie_mw.process_response(req2, response2, None)
    assert isinstance(response2, scrapy_splash.SplashTextResponse)
    assert response2 is not response
    assert response2.real_url == req2.url
    assert response2.url == req.url
    assert response2.body == b"<html><body>Hello</body></html>"
    assert response2.css("body").extract_first() == "<body>Hello</body>"
    assert response2.headers == {b'Content-Type': [b'text/html']}

    # check .replace method
    response3 = response2.replace(status=404)
    assert response3.status == 404
    assert isinstance(response3, scrapy_splash.SplashTextResponse)
    for attr in ['url', 'real_url', 'headers', 'body']:
        assert getattr(response3, attr) == getattr(response2, attr)