Check if Series consists of strings only, instead of casting to unicode

[henrifroese]

Currently, some functions do not check if the Series they get as input really consists of strings only, and they give unexpected results, e.g. if there are missing values.

Example:

import texthero as hero
import pandas as pd
import numpy as np

s = pd.Series(["Test", np.nan])
hero.noun_chunks(s)
>>0                   []
>>1    [(nan, NP, 0, 3)]

This could be fixed by stopping to use s.astype('unicode') which e.g. converts np.nan -> "nan". Instead, a function should check whether the Series consists of strings only. Something along the lines of

def _check_series_strings(s):
    if not df.map(type).eq(str).all():
        raise TypeError("Non-string values in series. Use hero.drop_no_content(s) to drop those values.")

[jbesomi]

Amazing!

We might want to have a different name for this function. If we agree on the name of the kinds of pandas series defined in #60, we could call it _check_is_text_series or something like that.

Can you work on this?

[henrifroese]

Yes, I'll think about / make some comments at #60 and work on this one.

[henrifroese]

See #60 and related PRs