How do I apply and/or boolean logic on Polars .when conditionals?

Let's start with my dataframe. It has 2 columns, src and tgt. When tgt is not null and src is not "?" , I want to set tgt=src.

┌─────┬──────┐
│ tgt ┆ src  │
│ --- ┆ ---  │
│ str ┆ str  │
╞═════╪══════╡
│ a   ┆ !a   │
│ ?   ┆ b    │
│ ?   ┆ null │
└─────┴──────┘

should then give, with an alias to newtgt

┌─────┬──────┬────────┐
│ tgt ┆ src  ┆ newtgt │
│ --- ┆ ---  ┆ ---    │
│ str ┆ str  ┆ str    │
╞═════╪══════╪════════╡
│ a   ┆ !a   ┆ a      │
│ ?   ┆ b    ┆ b      │
│ ?   ┆ null ┆ ?      │
└─────┴──────┴────────┘

I can check the not null and I can check == "?". How do I combine them? I tried and, & and &&, none of which worked.

What I have so far, including error messages:

import polars as pl

df = pl.from_dict(
    dict(tgt=["a","?","?"],src=["!a","b",None])
)
print("\ndf before:\n",df)

df2 = df.with_columns(
    pl.when(pl.col("src").is_not_null())
    .then(pl.col("src"))
    .otherwise(pl.col("tgt"))
    .alias("newtgt")
)
print("\ndf2 check if src not null:\n",df2)

df2 = df.with_columns(
    pl.when(pl.col("tgt") == "?")
    .then(pl.col("src"))
    .otherwise(pl.col("tgt"))
    .alias("newtgt")
)
print("\ndf2 if check tgt already known:\n",df2)

try:
    print("\n\ncheck both with `and`: ")
    df2 = df.with_columns(
        pl.when(pl.col("tgt") == "?" and pl.col("src").is_not_null())
        .then(pl.col("src"))
        .otherwise(pl.col("tgt"))
        .alias("newtgt")
    )
except (ValueError,) as e: 
    print("\nnot happy with `and`:\n  ", e)

try:
    print("\n\ncheck both with `&`: ")
    df2 = df.with_columns(
        pl.when(pl.col("tgt") == "?" & pl.col("src").is_not_null())
        .then(pl.col("src"))
        .otherwise(pl.col("tgt"))
        .alias("newtgt")
    )
except (pl.exceptions.InvalidOperationError,) as e: 
    print("\nnot happy with `&`:\n  ", e)

output:

df before:
 shape: (3, 2)
┌─────┬──────┐
│ tgt ┆ src  │
│ --- ┆ ---  │
│ str ┆ str  │
╞═════╪══════╡
│ a   ┆ !a   │
│ ?   ┆ b    │
│ ?   ┆ null │
└─────┴──────┘

df2 check if src not null:
 shape: (3, 3)
┌─────┬──────┬────────┐
│ tgt ┆ src  ┆ newtgt │
│ --- ┆ ---  ┆ ---    │
│ str ┆ str  ┆ str    │
╞═════╪══════╪════════╡
│ a   ┆ !a   ┆ !a     │
│ ?   ┆ b    ┆ b      │
│ ?   ┆ null ┆ ?      │
└─────┴──────┴────────┘

df2 if check tgt already known:
 shape: (3, 3)
┌─────┬──────┬────────┐
│ tgt ┆ src  ┆ newtgt │
│ --- ┆ ---  ┆ ---    │
│ str ┆ str  ┆ str    │
╞═════╪══════╪════════╡
│ a   ┆ !a   ┆ a      │
│ ?   ┆ b    ┆ b      │
│ ?   ┆ null ┆ null   │
└─────┴──────┴────────┘


check both with `and`: 

not happy with `and`:
   Since Expr are lazy, the truthiness of an Expr is ambiguous. Hint: use '&' or '|' to logically combine Expr, not 'and'/'or', and use 'x.is_in([y,z])' instead of 'x in [y,z]' to check membership.


check both with `&`: 

not happy with `&`:
   `bitand` operation not supported for dtype `str`

In polars you need to parenthesize each part of a complex and/or expression to avoid that ambiguous error. As that error message implies, & is required as well over and:

df.with_columns(
        pl.when((pl.col("tgt") == "?") & (pl.col("src").is_not_null()))
        .then(pl.col("src"))
        .otherwise(pl.col("tgt"))
        .alias("newtgt")
    )

shape: (3, 3)
┌─────┬──────┬────────┐
│ tgt ┆ src  ┆ newtgt │
│ --- ┆ ---  ┆ ---    │
│ str ┆ str  ┆ str    │
╞═════╪══════╪════════╡
│ a   ┆ !a   ┆ a      │
│ ?   ┆ b    ┆ b      │
│ ?   ┆ null ┆ ?      │
└─────┴──────┴────────┘

Another equivalent option is pl.all(expr1, expr2, ...)