colback

Column backed lists of structs.

Synopsis

This crate provides procedural macros that generate code to extract data from dataframes so that their column data can be used as a view into a proxy struct. You want to keep contiguous columnar data, but you want to be able to write normal code with type-safety and minimal boilerplate that is also performant.

What it does

This will first create some proxy view and reference structs based on a struct that you use a derive macro on. The derive macro then generates some code that will handle extracting the chunked arrays backing the dataframe columns, as well as proxy/view types for each row. These proxy types are populated using data from the dataframe, so each instance of a proxy struct corresponds directly to a row.

Example

use colback::ColbackView;
use polars::df;

#[derive(ColbackView)]
struct MyRow {
    col_a: u32,
    col_b: bool,
}
let df = df!["col_a" => [0u32, 1u32], "col_b" => [true, false]].unwrap();
let row_view = MyRow::view(&df).unwrap();
let row_ref = row_view.get(0).unwrap();
assert_eq!(row_ref.col_a, 0);
assert_eq!(row_ref.col_b, true);