BUG: Refering to a local variable in a query within a list comprehension errors

[yuvalwas]

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

# does not work:
def test1():
    df = pd.DataFrame([0, 1], columns=['group'])
    g = 1
    print([df.query('group == @g') for _ in range(1)])
test1()

#works:
def test2():
    df = pd.DataFrame([0, 1], columns=['group'])
    g = 1
    print([df.query(f'group == {g}') for _ in range(1)])
test2()

# works:
def test3():
    df = pd.DataFrame([0, 1], columns=['group'])
    g = 1
    for _ in range(1):
        print(df.query(f'group == @g'))
test3()

# works:
df = pd.DataFrame([0, 1], columns=['group'])
g = 1
[df.query('group == @g') for _ in range(1)]

Issue Description

Referring to a local variable in a list comprehension with query's @ does not work, but it does work using f-strings. I tried on Pandas 1.5.3 and 2.0.1.

Expected Behavior

test1 should work as test2.

Installed Versions

INSTALLED VERSIONS

commit : 37ea63d
python : 3.10.11.final.0
python-bits : 64
OS : Linux
OS-release : 5.15.0-71-generic
Version : #78-Ubuntu SMP Tue Apr 18 09:00:29 UTC 2023
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8

pandas : 2.0.1
numpy : 1.24.3
pytz : 2022.7
dateutil : 2.8.2
setuptools : 66.0.0
pip : 23.0.1
Cython : None
pytest : None
hypothesis : None
sphinx : 5.0.2
blosc : None
feather : None
xlsxwriter : None
lxml.etree : 4.9.2
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 3.1.2
IPython : 8.12.0
pandas_datareader: None
bs4 : 4.12.2
bottleneck : 1.3.5
brotli :
fastparquet : None
fsspec : 2023.3.0
gcsfs : None
matplotlib : 3.7.1
numba : None
numexpr : 2.8.4
odfpy : None
openpyxl : 3.0.10
pandas_gbq : None
pyarrow : None
pyreadstat : None
pyxlsb : None
s3fs : None
scipy : 1.10.1
snappy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
zstandard : None
tzdata : 2023.3
qtpy : 2.2.0
pyqt5 : None
/home/yuval/anaconda3/envs/cltorch/lib/python3.10/site-packages/_distutils_hack/init.py:33: UserWarning: Setuptools is replacing distutils.
warnings.warn("Setuptools is replacing distutils.")

[topper-123]

df.query only accepts local variables inside a query string, see here.

Inside a list comprehension the result from calling locals() is the local scope inside the list comprehension only, not the function. This is a limitation of python itself, so I don't think we can do so much about it in pandas. So you've hit a limitation of python here, AFAIK.

Whether this can be mitigated somehow I don't know, but that would be an enhancement to pandas, not a bug fix.

So I think this will be closed as wontfix, unfortunately. You can check if the docs can be made clearer in this regard.

[topper-123]

A small cleanup of the above explanation: the dict comprehension tries first the locals(), then globals(). So the below will work and the g variable is 0:

import pandas as pd

g = 0

def test1():
    df = pd.DataFrame([0, 1], columns=['group'])
    g = 1
    list(*(df.query('group == @g') for _ in range(1)))
test1()

This is all according to the python scoping rules, and can't be fixed in pandas.

[yuvalwas]

Thank you for your answers. Why doesn't it apply for f-strings as well (test2 in my example)? Shouldn't it be the same as a result of the list comprehension scoping?

[topper-123]

The variables that stored in 'locals()' in local to the list comprehension, not the function:

def test1():
    g = 1
    print([locals() for _ in range(1)])
test1()

Notice that g isn't stored in the list comprehension, meaning it won't be available for use later...

AFAIK test2 works because list comprehensions are weird: scope when calling locals() and when accessing variables directly are different in list comprehension. I don't have a better explanation, sorry. There's this SO discussion about it