Incorrect `PolarsInefficientMapWarning` for string containment with `in`

[henryharbeck]

Incorrect `PolarsInefficientMapWarning` for string containment with `in`

[henryharbeck]

Also somewhat related, but I don't think warrants its own issue, is the missing use of pl.lit in these suggestion as well.

string = "qwerty"
substring = "we"

# only the substring
df_1 = pl.DataFrame({"substring": [substring]})

# Suggests to replace with `pl.col("substring").is_in(string)`
df_1.select(
    pl.col("substring").map_elements(
        lambda substring: substring in string,
        return_dtype=pl.Boolean,
    )
)
df_1.select(pl.col("substring").is_in(string)) # ColumnNotFoundError: qwerty


# only the string
df_2 = pl.DataFrame({"string": [string]})

# Suggests to replace with `substring.is_in(pl.col("string"))`
df_2.select(
    pl.col("string").map_elements(
        lambda string: substring in string,
        return_dtype=pl.Boolean,
    )
)
df_2.select(substring.is_in(pl.col("string"))) # AttributeError: 'str' object has no attribute 'is_in'

In both cases adding pl.lit(...) around the string literal makes the code not raise, but produces the incorrect output.

It is possible that the Expr in Expr.str.contains(...) (as the correct suggestion) would not be a column in the frame and need to be suggested as a literal.

[MarcoGorelli]

thanks @henryharbeck for the report - i think it should be possible to correct these without too much complexity