How to pipe expressions from a list of expressions in Python Polars

Question

I cannot get my head around this. Say that I have a dictionary of regexes and replacement strings that I want to replace, and if none of these regexes are matched (when returns False) I want to resume with the next when statement.

So, instead of this way, that does not check for data in "new data"...

all_items = pl.DataFrame(
    {
        "data": ["Swedish fish", "English tea", "", "", ""],
        "ISO_codes": ["fin", "nor", "eng", "eng", "swe"],           
    })


replacement_rules = {
    r"^Swe.*": "Svenska",
    r"^Eng.*": "English",
    
}

iso_tranlation = {
   "swe": "Svenska",
   "eng": "English",
   "nor": "Norsk",
   "fin": "Finska på finska",

}

for pattern, replacement in replacement_rules.items():
    all_items = (
        all_items.lazy()
        .with_columns(
            pl.when(pl.col("data").str.contains(pattern))
            .then(pl.lit(replacement))
            .alias("new_data")
        )
        .collect()
    ) 

all_items = (
    all_items.lazy()
    .with_columns(
        pl.when(pl.col("ISO_codes").str.len_chars() > 0)
        .then(
            pl.col("ISO_codes")
            .replace(iso_tranlation , default="Unknown ISO Code")
        )
        .alias("new_data")
    )
    .collect()
)

...i would like to do something along the lines of this:

expressions = [
    pl.when(pl.col("data").str.contains(pattern))
    .then(pl.lit(replacement))
     for pattern, replacement in replacement_rules.items()]

all_items = (
    all_items.lazy()
    .with_columns(
        expressions.explode_and_pipe()
        .when(pl.col("ISO_codes").str.len_chars() > 0)
        .then(
            pl.col("ISO_codes")
            .replace(iso_tranlation , default="Unknown ISO Code")
        )
        .alias("new_data")
    )
    .collect()
)

Is there a way to achieve that expressions.explode_and_pipe()?

EDIT: This is the resulting dataframe i'm after:

shape: (5, 3)
┌──────────────┬───────────┬──────────┐
│ data         ┆ ISO_codes ┆ new_data │
│ ---          ┆ ---       ┆ ---      │
│ str          ┆ str       ┆ str      │
╞══════════════╪═══════════╪══════════╡
│ Swedish fish ┆ fin       ┆ Svenska  │
│ English tea  ┆ nor       ┆ English  │
│              ┆ eng       ┆ English  │
│              ┆ eng       ┆ English  │
│              ┆ swe       ┆ Svenska  │
└──────────────┴───────────┴──────────┘

Answer 1

Maybe you are looking for something like this? (coalesce may be the function you want.)

replacements = pl.lit(None)
for pattern, replacement in replacement_rules.items():
    replacements = (
        pl.when(pl.col("data").str.contains(pattern))
        .then(pl.lit(replacement))
        .otherwise(replacements)
    )

iso_translations = pl.col("ISO_codes").replace(
    iso_translation, default="Unknown ISO Code"
)

all_items.with_columns(new_data=pl.coalesce(replacements, iso_translations))

shape: (5, 3)
┌──────────────┬───────────┬──────────┐
│ data         ┆ ISO_codes ┆ new_data │
│ ---          ┆ ---       ┆ ---      │
│ str          ┆ str       ┆ str      │
╞══════════════╪═══════════╪══════════╡
│ Swedish fish ┆ fin       ┆ Svenska  │
│ English tea  ┆ nor       ┆ English  │
│              ┆ eng       ┆ English  │
│              ┆ eng       ┆ English  │
│              ┆ swe       ┆ Svenska  │
└──────────────┴───────────┴──────────┘