last left-on expression shows up as "literal" column in join output

[mcrumiller]

Polars version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of Polars.

Issue description

When a join occurs with expressions in the left_on, the last expression gets added to the dataframe as "literal". Because we are not allowed to alias these expressions, it doesn't make sense to include them in the final output, and because all expressions default to literal, only one of the expressions ends up in the dataframe, and you are unable to choose which.

Reproducible example

import polars as pl
from polars import col, when

df1 = pl.DataFrame({
    'a': [1, 2, 3, 4, 5]
})

df2 = pl.DataFrame({
    'b': [5, 4, 3, 2, 1]
})

df1.join(
    df2,
    left_on=when(False).then(0).otherwise(col('a')),
    right_on=when(False).then(0).otherwise(col('b')),
    how="left",
)

shape: (5, 3)
┌─────┬─────────┬─────┐
│ a   ┆ literal ┆ b   │
│ --- ┆ ---     ┆ --- │
│ i64 ┆ i64     ┆ i64 │
╞═════╪═════════╪═════╡
│ 1   ┆ 1       ┆ 1   │
│ 2   ┆ 2       ┆ 2   │
│ 3   ┆ 3       ┆ 3   │
│ 4   ┆ 4       ┆ 4   │
│ 5   ┆ 5       ┆ 5   │
└─────┴─────────┴─────┘

Expected behavior

literal column should not be inserted into resulting dataframe.

Installed versions

--------Version info---------
Polars:      0.18.1
Index type:  UInt32
Platform:    Windows-10-10.0.19045-SP0
Python:      3.11.2 (tags/v3.11.2:878ead1, Feb  7 2023, 16:38:35) [MSC v.1934 64 bit (AMD64)]

----Optional dependencies----
numpy:       1.24.3
pandas:      2.0.0
pyarrow:     11.0.0
connectorx:  0.3.2a3
deltalake:   <not installed>
fsspec:      <not installed>
matplotlib:  3.7.1
xlsx2csv:    0.8.1
xlsxwriter:  3.1.0

[ritchie46]

We follow a simple rule for determining the name of the output expresison (unless you provide an alias or keep_name) and that is the left-most expression. That is a literal, hence the name.

[mcrumiller]

@ritchie46 -- my point is that this column shouldn't be in the output whatsoever. Using an expression via a join means "join on this calculation"--not necessary "include the calculation in the output."

[ritchie46]

Oh, right! Now I see it ^^