[BUG]pandas == condition fails

[letpat]

Bug description

Actual behavior

With a very simple dataframe (one column col1 and 3 rows),
res = df[df['col1'] == 2]
throws exception :
Unhandled exception caught during execution: False
mojo: error: execution exited with a non-zero result: 1
or
Unhandled exception caught during execution: <cyfunction item_from_zerodim at 0x7f50ea1cdf20> returned a result with an exception set
mojo: error: execution exited with a non-zero result: 1

Expected behavior

With same dataframe,
res = df[(df['col1'] >= 2) & (df['col1'] <= 2)]

works perfectly and return expected rows

Steps to reproduce

from python import Python

def main():
    pd = Python.import_module("pandas")
    data = Python.dict()
    data['col1'] = [1,2,3]
    df = pd.DataFrame(data)
    res = df[df['col1'] == 2]
    print(res)

System information

• magic 07.2
• mojo 25.2.0 (26172dfe)
• pandas 2.2.3

[rcghpge]

I started probing around to get an idea of Mojo’s compiler design with your use case. A generic Python object in Python does not translate 1:1 in Mojo yet. Until the compiler is open sourced, I would build helper functions as a workaround or leverage Mojo’s capabilities to do the same thing.

Example maybe of some utils for Python

from python import Python, PythonObject, TypedPythonObject

fn py_int(val: Int) raises -> PythonObject:
    var builtins = Python.import_module("builtins")
    return builtins.__getattr__("int").__call__(val)

fn py_str(text: String) raises -> PythonObject:
    return builtins.__getattr__("str").__call__(text)

[letpat]

Thanks for the workaround.