BUG: AttributeError: 'StringArray' object has no attribute 'tolist'

[aiqc]

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of pandas.
• (optional) I have confirmed this bug exists on the master branch of pandas.

---

Code Sample, a copy-pastable example

$ pip show pandas

Name: pandas
Version: 1.2.4

import pandas as pd
import numpy as np

lzt = ["a","b","c"]
arr = np.array(lzt)
df = pd.DataFrame(arr).astype('string')

srs = df[0]
u = srs.unique()


type(u)
<class 'pandas.core.arrays.string_.StringArray'>


u.dtype == 'string'
True


u.tolist()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: 'StringArray' object has no attribute 'tolist'

Problem description

So I tried changing my code to use list(srs.unique()), but then all of my numeric dtypes become unserializable for JSON.

Output of pd.show_versions()

INSTALLED VERSIONS ------------------ commit : 2cb9652 python : 3.7.6.final.0 python-bits : 64 OS : Darwin OS-release : 20.3.0 Version : Darwin Kernel Version 20.3.0: Thu Jan 21 00:07:06 PST 2021; root:xnu-7195.81.3~1/RELEASE_X86_64 machine : x86_64 processor : i386 byteorder : little LC_ALL : None LANG : en_US.UTF-8 LOCALE : en_US.UTF-8

pandas : 1.2.4
numpy : 1.18.5
pytz : 2020.1
dateutil : 2.8.1
pip : 20.2.2
setuptools : 41.2.0
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 2.11.2
IPython : 7.17.0
pandas_datareader: None
bs4 : None
bottleneck : None
fsspec : None
fastparquet : None
gcsfs : None
matplotlib : 3.3.1
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : 2.0.0
pyxlsb : None
s3fs : None
scipy : 1.5.2
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
numba : None

[isaac-chung]

@aiqc Was able to use json.dumps(list(srs.unique())) which gave an output '["a", "b", "c"]'. I'm a bit confused by your reference to numeric dtypes when your example only had strings/list of strings. Does this resolve your issue?

[aiqc]

@aiqc Was able to use json.dumps(list(srs.unique())) which gave an output '["a", "b", "c"]'. Does this resolve your issue?

I'd say it's not really resolved until StringArray has a tolist() method like the other dtypes. Part of making the new dtype feature complete.

I'm a bit confused by your reference to numeric dtypes when your example only had strings/list of strings.

I run many non-float dtypes through unique(), not just strings, so it's hard to get a manual workaround (e.g. list()) that fits all dtypes

[isaac-chung]

To add to your point, I spot-checked another class, and tried a something similar on Pandas 1.2.4 with BooleanArray class and grabbed the unique() from a DataFrame column. Found that when running json.dumps(list(u))

• if u is of <class 'pandas.core.arrays.string_.StringArray'>, the line runs without error
• if u is of <class 'pandas.core.arrays.boolean.BooleanArray'>, the line returns a not JSON serializable error

Both types do not have tolist() methods, like how the numpy.ndarray does.

[dsaxton]

@aiqc The tolist method is part of the Series and not the PandasArray API so this is behaving as expected.

Regarding data being JSON serializable you can pass a custom serializer to cls when using json.dumps to convert from numpy types to primitive Python types (there are probably several other ways to do this).

EDIT: Actually, a pandas one-liner to do this would be: srs.drop_duplicates().to_json(orient="records").

[mroeschke]

Yes this is expected as explained above and I don't think tolist is in scope for PandasArray therefore closing