#!/usr/bin/env python
# coding: utf-8

# In[ ]:


get_ipython().run_line_magic('pip', "install 'whylogs>=1.5.0'")


# In[ ]:


from typing import Any
import pandas as pd
import numpy as np

def load_testing_data() -> pd.DataFrame:  # TODO remove after user testing
    data = pd.read_csv("https://guest-session-testing-public.s3.us-west-2.amazonaws.com/adult_income_m.csv")

    def convert_random_values(value: Any) -> Any:
        if isinstance(value, int) and np.random.random() < 1 / 100:
            return str(value)
        return value

    data["capital-gain"] = data["capital-loss"].apply(convert_random_values)
    data["capital-loss"] = data["capital-loss"].apply(convert_random_values)
    return data


df = load_testing_data()


# In[ ]:


import whylogs as why

why.init(upload_on_log=True, allow_local=False)


# In[ ]:


profile = why.log(df, name="foo")


# In[ ]:


# Upload the same data as a batch profile by leaving out the name
profile = why.log(df)


# In[ ]:


why.log(multiple={'foo': df, 'bar': df})


# # Switch to an autheneticated session

# In[ ]:


why.init(reinit=True, allow_anonymous=False, upload_on_log=True, whylabs_api_key="MPq7Hg002z.Na5VweqsJfu5ArGILjQTlGAyPyOhtOnEVEtqY2b5PXNGJLZLjHscT:org-JpsdM6", default_dataset_id="model-62")


# In[ ]:


profile = why.log(df, name="real_dataset")


# ## Or upload via the whylabs writer
# This will use the session for credentials as well, it just won't have all of the fancy output.

# In[ ]:


profile.writer('whylabs').write()


# In[ ]:


# as a reference profile
profile.writer('whylabs').option(reference_profile_name="authenticated_ref").write()