BUG (Regression): Using a tuple as first column name in empty dataframe creates a non functional dataframe and memory errors

[jenno-verdonck]

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

import pandas as pd

df1 = pd.DataFrame()
df1[(0, 0)] = [1, 2, 3]

df2 = pd.DataFrame()
df2['a'] = [1, 2, 3]
df2[(0, 0)] = [4, 5, 6]

print(df2.columns)
print(f'df2=\n{df2}')

print(df1.columns)
print(f'df1=\n{df1}')

Issue Description

A dataframe becomes completely unusable when the first column was assigned using a tuple as key. (example df1)
The tuple gets transformed in 2 different columns while data for only one is given. This cause print statements to return "ValueError: Length of values (1) does not match length of index (2)".

When a dataframe already has an existing column (example df2) the assignment works as expected.

Running the assignment on the df1 dataframe 5 times results in a wide range of memory related errors such as:
malloc(): unaligned tcache chunk detected
IndexError: index 94573497622112 is out of bounds for axis 0 with size 5
corrupted size vs. prev_size
double free or corruption (out)

Expected Behavior

The assignment should behave the same in both examples and use the tuple as single header.

Installed Versions

INSTALLED VERSIONS

commit : 0f43794
python : 3.10.12.final.0
python-bits : 64
OS : Linux
OS-release : 5.19.0-50-generic
Version : #50-Ubuntu SMP PREEMPT_DYNAMIC Mon Jul 10 18:24:29 UTC 2023
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8

pandas : 2.0.3
numpy : 1.25.1
pytz : 2022.7.1
dateutil : 2.8.2
setuptools : 68.0.0
pip : 23.2.1
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 3.1.2
IPython : 8.14.0
pandas_datareader: None
bs4 : 4.12.2
bottleneck : None
brotli :
fastparquet : None
fsspec : None
gcsfs : None
matplotlib : 3.7.2
numba : None
numexpr : None
odfpy : None
openpyxl : 3.1.2
pandas_gbq : None
pyarrow : None
pyreadstat : None
pyxlsb : None
s3fs : None
scipy : 1.11.1
snappy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
zstandard : None
tzdata : 2023.3
qtpy : 2.3.1
pyqt5 : None

[jenno-verdonck]

This seems to be a regression.
Running this in version 2.0.0 results in an InvalidIndexError.
Running this in version 1.5.3 works as expected without any errors.

[Tialo]

I also experienced it.

import pandas as pd

df = pd.DataFrame([1, 2, 3])
df[0, 0] = 4

if you will try to print this dataframe it will fail with this error

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/core/frame.py", line 1094, in __repr__
    return self.to_string(**repr_params)
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/core/frame.py", line 1271, in to_string
    return fmt.DataFrameRenderer(formatter).to_string(
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/format.py", line 1134, in to_string
    string = string_formatter.to_string()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/string.py", line 30, in to_string
    text = self._get_string_representation()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/string.py", line 45, in _get_string_representation
    strcols = self._get_strcols()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/string.py", line 36, in _get_strcols
    strcols = self.fmt.get_strcols()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/format.py", line 615, in get_strcols
    strcols = self._get_strcols_without_index()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/format.py", line 870, in _get_strcols_without_index
    str_columns = self._get_formatted_column_labels(self.tr_frame)
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/format.py", line 949, in _get_formatted_column_labels
    dtypes = self.frame.dtypes
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/core/generic.py", line 6159, in dtypes
    return self._constructor_sliced(data, index=self._info_axis, dtype=np.object_)
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/core/series.py", line 500, in __init__
    com.require_length_match(data, index)
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/core/common.py", line 576, in require_length_match
    raise ValueError(
ValueError: Length of values (2) does not match length of index (3)

But if you set another index

import pandas as pd

df = pd.DataFrame([1, 2, 3])
df[1, 0] = 4

print will fail with another exception

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/core/frame.py", line 1094, in __repr__
    return self.to_string(**repr_params)
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/core/frame.py", line 1271, in to_string
    return fmt.DataFrameRenderer(formatter).to_string(
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/format.py", line 1134, in to_string
    string = string_formatter.to_string()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/string.py", line 30, in to_string
    text = self._get_string_representation()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/string.py", line 45, in _get_string_representation
    strcols = self._get_strcols()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/string.py", line 36, in _get_strcols
    strcols = self.fmt.get_strcols()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/format.py", line 615, in get_strcols
    strcols = self._get_strcols_without_index()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/format.py", line 870, in _get_strcols_without_index
    str_columns = self._get_formatted_column_labels(self.tr_frame)
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/io/formats/format.py", line 949, in _get_formatted_column_labels
    dtypes = self.frame.dtypes
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/core/generic.py", line 6158, in dtypes
    data = self._mgr.get_dtypes()
  File "/Users/y/PycharmProjects/tamli/.venv/lib/python3.8/site-packages/pandas/core/internals/managers.py", line 278, in get_dtypes
    return dtypes.take(self.blknos)
IndexError: index 3 is out of bounds for axis 0 with size 3

What is really important here, that exception does not occur while setting values, but only while working with the dataframe.

[jenno-verdonck]

I get the same ValueError in both cases using python 3.9.

It seems that the double index is split into 2 columns while only having values for 1. This results in ValueErrors when trying to print the dataframe with undefined values. Your second example (df[1,0]) reveals to me that the assignment is only done on the first given index. Displaying df[1] seems to work and returns a series containing only the value 4. Displaying the full dataframe results in ValueError while displaying df[0] causes python to fully crash without any exceptions. Eventviewer on windows does shows an application error saying that python.exe failed with Exception code: 0xc0000005.

[jenno-verdonck]

The bug still exists in the latest python and pandas release. Especially the memory errors seem concerning.