如何从选项向量创建极系列

Question

我有点困惑。创建带有“简单”选项向量的极系列

Vec<Option<f64>>

起作用..。但是具有复杂值的选项向量(例如向量/数组)不是吗？！

Vec<Option<Vec<f64>>>

为什么？以及如何让它发挥作用？

use polars::prelude::*;

fn main() {

    // This does NOT work
    let v = vec![Some(vec![0.123, 67.12]), None, Some(vec![2.123, 44.98])];

    // This does NOT work
    let v = vec![Some([0.123, 67.12]), None, Some([2.123, 44.98])];

    // This does work
    let v = vec![Some(0.123), None, Some(44.98)];

    let s = Series::new("got nulls", v);

    println!("{:?}", s)
}

当我试图将let v = vec![Some(vec![0.123, 67.12]), None, Some(vec![2.123, 44.98])];变成一个系列时，我会收到以下错误消息：

error[E0277]: the trait bound `polars::prelude::Series: polars::prelude::NamedFrom<Vec<Option<Vec<{float}>>>, _>` is not satisfied
  --> src\main.rs:13:38
   |
13 |     let s = Series::new("got nulls", v);
   |             -----------              ^ the trait `polars::prelude::NamedFrom<Vec<Option<Vec<{float}>>>, _>` is not implemented for `polars::prelude::Series`
   |             |
   |             required by a bound introduced by this call
   |
   = help: the following other types implement trait `polars::prelude::NamedFrom<T, Phantom>`:
             <polars::prelude::Series as polars::prelude::NamedFrom<&polars::prelude::Series, str>>
             <polars::prelude::Series as polars::prelude::NamedFrom<T, ListType>>
             <polars::prelude::Series as polars::prelude::NamedFrom<T, T>>
             <polars::prelude::Series as polars::prelude::NamedFrom<T, [&'a str]>>
             <polars::prelude::Series as polars::prelude::NamedFrom<T, [AnyValue<'a>]>>
             <polars::prelude::Series as polars::prelude::NamedFrom<T, [Cow<'a, str>]>>
             <polars::prelude::Series as polars::prelude::NamedFrom<T, [NaiveDateTime]>>
             <polars::prelude::Series as polars::prelude::NamedFrom<T, [NaiveDate]>>
           and 37 others

Answer 1

这里有一个由this启发的解决方案

创建一个结构-> Json -> DataFrame ->系列。

有点让人费解，但对我有用。

欢迎对更好的解决方案提出任何建议:)

use polars::prelude::*;
use std::io::Cursor;

fn main() {
    #[derive(serde::Serialize)]
    struct ListData {
        data: Option<Vec<f64>>,
    }

    impl ListData {
        fn new(value: Option<Vec<f64>>) -> Self {
            ListData { data: value }
        }
    }

    let test_vec: Vec<ListData> = vec![
        ListData::new(Some(vec![0.123, 67.12])),
        ListData::new(None),
        ListData::new(Some(vec![2.123, 44.98])),
    ];

    let json = serde_json::to_string(&test_vec).unwrap();

    let cursor = Cursor::new(json);

    let df = JsonReader::new(cursor).finish().ok().unwrap();

    let s = df.column("data").unwrap().clone();

}