Python: 'module' object has no attribute url

[EntilZha]

I am writing some code which uses the generated code libraries and am running into a stack trace that looks like this (on develop_2.0, not using 2.1.3 since it doesn't include the patch for incorrect directory):

Traceback (most recent call last):
  File "/Users/prodriguez/Code/datascience-api/docs-tests/tests/integration/test_discovery_api.py", line 26, in test_offset
    response_0 = self.discovery_api.discovery_v1_photoboards_get()
  File "/Users/prodriguez/Code/datascience-api/docs-tests/tests/integration/datascience_api/apis/discovery_api.py", line 182, in discovery_v1_photoboards_get
    response='DiscoveryResponse', auth_settings=auth_settings)
  File "/Users/prodriguez/Code/datascience-api/docs-tests/tests/integration/datascience_api/api_client.py", line 101, in call_api
    post_params=post_params, body=body)
  File "/Users/prodriguez/Code/datascience-api/docs-tests/tests/integration/datascience_api/api_client.py", line 236, in request
    return RESTClient.GET(url, query_params=query_params, headers=headers)
  File "/Users/prodriguez/Code/datascience-api/docs-tests/tests/integration/datascience_api/rest.py", line 218, in GET
    return cls.IMPL.GET(*n, **kw)
  File "/Users/prodriguez/Code/datascience-api/docs-tests/tests/integration/datascience_api/rest.py", line 148, in GET
    return self.request("GET", url, headers=headers, query_params=query_params)
  File "/Users/prodriguez/Code/datascience-api/docs-tests/tests/integration/datascience_api/rest.py", line 123, in request
    r = self.agent(url).request(method, url,
  File "/Users/prodriguez/Code/datascience-api/docs-tests/tests/integration/datascience_api/rest.py", line 70, in agent
    url = urllib3.util.url.parse_url(url)
AttributeError: 'module' object has no attribute 'url'

Basically the problem is that the code has this call:

urllib3.util.url.parse_url

It should be:

urllib3.util.parse_url

On a different note, this and a couple other bugs I have found (like the incorrect directory name) point to that Python code generation seems second class and not well maintained. How trustworthy can I be of the code to be correct? I am using it in a production system so would like to get a feeling for that, and if it is not super well maintained, are there some low hanging fruit to help out with?

[geekerzp]

Hi @EntilZha I have tested urllib3.util.url.parse_url and urllib3.util.parse_url in ipython of python v2 and python v3, and both methods work fine.

>>> urllib3.util.url.parse_url("http://www.google.com")
Url(scheme='http', auth=None, host='www.google.com', port=None, path=None, query=None, fragment=None)

>>> urllib3.util.parse_url("http://www.google.com")
Url(scheme='http', auth=None, host='www.google.com', port=None, path=None, query=None, fragment=None)

In fact, they are the same method as they have the same memory address:

>>> urllib3.util.url.parse_url
<function parse_url at 0x1057301e0>

>>> urllib3.util.parse_url
<function parse_url at 0x1057301e0>

You also can look into the source code of urllib3: https://github.com/shazow/urllib3/blob/master/urllib3/util/url.py#L121 https://github.com/shazow/urllib3/blob/master/urllib3/util/__init__.py#L21

[EntilZha]

What might explain the error? I was using this within a docker environment so the installations are from latest pypi release. My guess is that somewhere I have a version discrepancy.

[EntilZha]

I think I know what happened. When I ran tests locally, it was using my older 1.7.1 version of urllib3 on my local machine. On docker it was using a newer version, but by the time I ran the tests there, I already had changed the call to urllib3.util.parse_url so didn't see that either would work in the newest version.

My proposal would be to keep urllib3.util.parse_url since it is better for backwards compatibility (seems like the url package was added to organize repo), but in any case my problem would be fixed by upgrading.

[geekerzp]

Thanks @EntilZha , I will change urllib3.util.url.parse_url to urllib3.util.parse_url for backwards compatibility, and would be better you can upgrade your urllib3.