CODEWITHSUNDEEP
rustrust-polars
lucaszane
lucaszane

Reputation: 3

Rust Polars Dataframe String replacement

I am writing a Rust using Polars. I would like to know how can I manipulate one string dataframe column.

For example, I have the following dataframe:

Id Text
1 Some foo text
2 Other text

And I would like to replace all values that has "foo" by some other value like "new foo" and I wasn't able to find a way to do that. Can anyone help me on this?

I tried to use the function with_column, but I didn't manage how to do it.

Upvotes: 0

Views: 559

Answers (2)

user459872
user459872

Reputation: 24602

You can use replace_literal_all method. Note that this feature is flagged behind strings flag.

[dependencies]
polars = { version = "0.38.1", features = ["strings"] }

Then in your Rust code, you can do the transformation like this:

use polars::prelude::*;

fn main() {
    let mut df = df! [
        "Id" => [1,2],
        "Text" => ["Some foo text", "Other text",],
    ]
    .unwrap();

    println!("Before = {:?}", df);
    let new_column = df["Text"]
        .str()
        .unwrap()
        .replace_literal_all("foo", "new_foo")
        .unwrap();

    let modified = df.with_column(new_column).unwrap();
    println!("After = {:?}", modified);
}

This will produce the following output:

Before = shape: (2, 2)
┌─────┬───────────────┐
│ Id  ┆ Text          │
│ --- ┆ ---           │
│ i32 ┆ str           │
╞═════╪═══════════════╡
│ 1   ┆ Some foo text │
│ 2   ┆ Other text    │
└─────┴───────────────┘
After = shape: (2, 2)
┌─────┬───────────────────┐
│ Id  ┆ Text              │
│ --- ┆ ---               │
│ i32 ┆ str               │
╞═════╪═══════════════════╡
│ 1   ┆ Some new_foo text │
│ 2   ┆ Other text        │
└─────┴───────────────────┘

You can use replace_all API if you want regex based replacement(And please note that it requires regex feature flag to be enabled).

Upvotes: 1

Crispy13
Crispy13

Reputation: 369

You can do like the following:

use polars::prelude::*;
use polars::df;
use regex::Regex;

fn main() {
    // use macro
    let mut df = df! [
        "Id" => [1,2],
        "Text" => ["Some foo text", "Other text",],
    ].unwrap();

    println!("before = {:?}", df);
    
    let re = Regex::new(r"foo").unwrap();
    let target = "new foo";

    let new_foo_series = df.column("Text").unwrap()
        .utf8()
        .unwrap()
        .apply(|e| re.replace(e, target));
    
    let df2 = df.with_column(new_foo_series).unwrap();
    
    println!("after = {:?}", df2);

}

That will show:

before = shape: (2, 2)
┌─────┬───────────────┐
│ Id  ┆ Text          │
│ --- ┆ ---           │
│ i32 ┆ str           │
╞═════╪═══════════════╡
│ 1   ┆ Some foo text │
│ 2   ┆ Other text    │
└─────┴───────────────┘
after = shape: (2, 2)
┌─────┬───────────────────┐
│ Id  ┆ Text              │
│ --- ┆ ---               │
│ i32 ┆ str               │
╞═════╪═══════════════════╡
│ 1   ┆ Some new foo text │
│ 2   ┆ Other text        │
└─────┴───────────────────┘

Ref = https://docs.rs/polars/latest/polars/docs/eager/index.html#apply-functions-closures

Upvotes: 0

Related Questions