BUG: Construction of DataFrame works with empty list but not empty array

[pavelkomarov]

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

df = pandas.DataFrame(data=[], columns=['a','b']) # works fine
df = pandas.DataFrame(data=numpy.array([]), columns=['a','b'])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.9/site-packages/pandas/core/frame.py", line 694, in __init__
    mgr = ndarray_to_mgr(
  File "/usr/local/lib/python3.9/site-packages/pandas/core/internals/construction.py", line 351, in ndarray_to_mgr
    _check_values_indices_shape_match(values, index, columns)
  File "/usr/local/lib/python3.9/site-packages/pandas/core/internals/construction.py", line 418, in _check_values_indices_shape_match
    raise ValueError("Empty data passed with indices specified.")
ValueError: Empty data passed with indices specified.

Issue Description

Constructing an empty dataframe with an empty list works fine, but giving an empty numpy array, which intuitively should behave the same as with an empty list, results in a mysterious ValueError: Empty data passed with indices specified.

Expected Behavior

Should create an empty dataframe with whatever columns listed.

Installed Versions

INSTALLED VERSIONS

commit : 4bfe3d0
python : 3.9.7.final.0
python-bits : 64
OS : Darwin
OS-release : 21.4.0
Version : Darwin Kernel Version 21.4.0: Mon Feb 21 20:34:37 PST 2022; root:xnu-8020.101.4~2/RELEASE_X86_64
machine : x86_64
processor : i386
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8

pandas : 1.4.2
numpy : 1.21.4
pytz : 2021.1
dateutil : 2.8.2
pip : 21.3.1
setuptools : 57.4.0
Cython : 0.29.24
pytest : 6.2.5
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : 3.0.1
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 3.0.1
IPython : 7.25.0
pandas_datareader: None
bs4 : None
bottleneck : None
brotli : None
fastparquet : None
fsspec : None
gcsfs : None
markupsafe : 2.0.1
matplotlib : 3.4.3
numba : None
numexpr : None
odfpy : None
openpyxl : 3.0.7
pandas_gbq : 0.15.0
pyarrow : 5.0.0
pyreadstat : None
pyxlsb : None
s3fs : None
scipy : 1.7.1
snappy : None
sqlalchemy : None
tables : None
tabulate : 0.8.9
xarray : None
xlrd : None
xlwt : None
zstandard : None

[phofl]

Hi, thanks for your report.

I think this works as intended. A list does not have a shape while a numpy array does, hence we get a missmatch. The error message is definitely intended

[pavelkomarov]

mmmmmm. If the shape is (0,), then it makes sense just to treat it as an empty. Even if an empty list doesn't explicitly have this shape, an array([]) does, so they are intuitively equivalent. If you give a list of the wrong length or an array of the wrong shape, or a list or array or the right length, the treatment is analogous, so why wouldn't it be in the empty case? Just because one has an explicit .shape and the other doesn't shouldn't really matter.

[jreback]

yeah i agree this works as intended. if you are passing an array then it must match in shape to the passed columns. yes i understand this is special casing a tiny bit, but it helps prevent errors.