BUG: json_normalize fails with empty arrays/lists

[mateuszboryn]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

# This works fine:
data = [ {
        "id": 1, "name": "Cole Volk", "fitness": [{"height": 130, "weight": 60}],
        "someList": []
    },
    {
        "id": 2, "name": "Faye Raker", "fitness": [{"height": 130, "weight": 60}],
        "someList": [1]
    },]

pd.json_normalize(data, "fitness", ["id", "name", "someList"])

# This doesn't work
data = [ {
        "id": 1, "name": "Cole Volk", "fitness": [{"height": 130, "weight": 60}],
        "someList": []
    },
    {
        "id": 2, "name": "Faye Raker", "fitness": [{"height": 130, "weight": 60}],
        "someList": []
    },]

pd.json_normalize(data, "fitness", ["id", "name", "someList"])

# fails with ValueError: operands could not be broadcast together with shape (0,) (2,)

Issue Description

when normalizing nested lists, which are empty in all the objects, json_normalize fails with "ValueError: operands could not be broadcast together with shape".

It may be related to the behavior of np.repeat(), but I report the issue to pandas, because it looks like an overlooked corner case:

v = [ [], [], [234] ]
lengths = [1,1,3]
np.array(v, dtype=object).repeat(lengths)

and

v = [ [], [], [] ]
lengths = [1,1,3]
np.array(v, dtype=object).repeat(lengths)

Expected Behavior

Return correct data structure and not fail.

Installed Versions

INSTALLED VERSIONS

commit : 4bfe3d0
python : 3.10.2.final.0
python-bits : 64
OS : Windows
OS-release : 10
Version : 10.0.19044
machine : AMD64
processor : Intel64 Family 6 Model 141 Stepping 1, GenuineIntel
byteorder : little
LC_ALL : en_US.UTF-8
LANG : en_US.UTF-8

[simonjayhawkins]

Thanks @mateuszboryn for the report.

This is the same underlying issue as #37782

i've posted a possible fix there #37782 (comment) that also fixes this issue.

PRs welcome.

[topper-123]

This works as expected now.

>>> data = [ {
...        "id": 1, "name": "Cole Volk", "fitness": [{"height": 130, "weight": 60}],
...        "someList": []
...    },
...    {
...        "id": 2, "name": "Faye Raker", "fitness": [{"height": 130, "weight": 60}],
...        "someList": []
...    },]
>>> pd.json_normalize(data, record_path="fitness", meta=["id", "name", "someList"])
   height  weight id        name someList
0     130      60  1   Cole Volk       []
1     130      60  2  Faye Raker       []

Closing as fixed.