json_normalize does not normalize subrecords properly if any subrecords values are NoneType

[aerymilts]

Code Sample, a copy-pastable example if possible

data_fail_to_normalize = \
        [{'info': None}, 
         
        {'info': 
         {'created_at': '11/08/1993', 'last_updated': '26/05/2012'},
        'author_name': 
         {'first': 'Jane', 'last_name': 'Doe'}
        }]

data_partial_fail = \
        [{'info': None, 
         'author_name': 
         {'first': 'Smith', 'last_name': 'Appleseed'}
        }, 
        
        {'info': 
         {'created_at': '11/08/1993', 'last_updated': '26/05/2012'},
        'author_name': 
         {'first': 'Jane', 'last_name': 'Doe'}
        }]

>>> import pandas as pd
>>> pd.io.json.json_normalize(data_fail_to_normalize)

Output 1

	author_name	info
0	nan	None
1	{'first': 'Jane', 'last_name': 'Doe'}	{'created_at': '11/08/1993', 'last_updated': '26/05/2012'}

>>> pd.io.json.json_normalize(data_partial_fail)

Output 2

	author_name.first	author_name.last_name	info	info.created_at	info.last_updated
0	Smith	Appleseed	nan	nan	nan
1	Jane	Doe	nan	11/08/1993	26/05/2012

Problem description

I expected that the json_normalize function takes into account the presence of NoneTypes in the dictionaries. This leads to 2 separate issues (If I should open this as 2 separate issues, let me know).

I have already written a fix that solves this issue - if anyone else can validate that this is not working as intended, I can set up a PR.

Output 1

Does not unnest json after encountering NoneType at first instance of subrecord, line 192 of pandas/io/json/normalize.py

Output 2

Keeps the None value when encountered, [{k: {'alpha': 'foo', 'beta': 'bar'}}, {k: None}], see nested_to_record function. Creates additional column of nans which would not otherwise occur if that particular key was removed.

Expected Output

Output 1

	author_name.first	author_name.last_name	info.created_at	info.last_updated
0	nan	nan	nan	nan
1	Jane	Doe	11/08/1993	26/05/2012

Output 2

	author_name.first	author_name.last_name	info.created_at	info.last_updated
0	Smith	Appleseed	nan	nan
1	Jane	Doe	11/08/1993	26/05/2012

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.6.4.final.0
python-bits: 64
OS: Linux
OS-release: 4.4.0-104-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: en_US.UTF-8
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8

pandas: 0.22.0
pytest: None
pip: 9.0.1
setuptools: 38.4.0
Cython: None
numpy: 1.13.1
scipy: None
pyarrow: None
xarray: None
IPython: 6.2.1
sphinx: None
patsy: None
dateutil: 2.6.1
pytz: 2017.3
blosc: None
bottleneck: None
tables: None
numexpr: None
feather: None
matplotlib: 2.1.2
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: 1.0.1
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[gfyoung]

I would personally prefer the second option. Feel free to post the PR!

[aerymilts]

@gfyoung Were you referring to opening this as two separate issues?

[gfyoung]

@aerymilts : I was referring to the second output being preferable to the first.