Pandas import error when merging tables

[trhaarman]

Environment

Local

Delta-rs version:
v0.15.1

Binding:
Python

Environment:

• OS: Windows

---

Bug

What happened:
Trying to get the merging example from the docs, but using pandas instead of pyarrow for the tabkle. This gives a NameError that pandas is not defined, as in the table.py pandas is only imported when type checking:

delta-rs/python/deltalake/table.py

Line 941 in 1a984ce

elif isinstance(source, pandas.DataFrame):

As per my limited understanding of this construct, pandas is only imported for static type checking. But seeing as it is actually used in the code, should it be imported normally?

What you expected to happen:
That the pandas dataframe gets merged in the existing delta table :).

How to reproduce it:
Using the example from the docs, but with a pandas DataFrame instead of a pyarrow table:

from deltalake import DeltaTable, write_deltalake
import pandas as pd

data = pd.DataFrame({"x": [1, 2, 3], "y": [4, 5, 6]})
write_deltalake("tmp", data)
dt = DeltaTable("tmp")
new_data = pd.DataFrame({"x": [1], "y": [7]})

(
     dt.merge(
         source=new_data,
         predicate="target.x = source.x",
         source_alias="source",
         target_alias="target")
     .when_matched_update(updates={"x": "source.x", "y": "source.y"})
     .execute()
)

dt.to_pandas()

results in:
NameError: name 'pandas' is not defined

More details:
I'm not very familiar with static type checking or this code, so please let me know if I did something wrong and this is not a bug!

[ion-elgreco]

Yeah, the top-level import is missing apparently so if you import pandas without alias it works, but this needs to be fixed by adding the same lines as in the writer and then if _has_pandas and isinstance(df, pl.Dataframe):

try:
    import pandas as pd  # noqa: F811
except ModuleNotFoundError:
    _has_pandas = False
else:
    _has_pandas = True

If you want, you can open a PR for it :)

[trhaarman]

Alright, thanks! I'll have a quick look at it tonight.