[Library] nth option is broken?

[murlakatamenka]

examples/nth.rs:

extern crate skim;
use skim::prelude::*;
use std::io::Cursor;

pub fn main() {
    let input = "foo 123";

    let options = SkimOptionsBuilder::default()
        .nth(Some("2"))
        .query(Some("f"))
        .build()
        .unwrap();
    let item_reader = SkimItemReader::default();

    dbg!(options.nth);

    let items = item_reader.of_bufread(Cursor::new(input));
    let selected_items = Skim::run_with(&options, Some(items))
        .map(|out| out.selected_items)
        .unwrap_or_else(|| Vec::new());

    for item in selected_items.iter() {
        print!("{}{}", item.output(), "\n");
    }
}

cargo run --example nth

It generates a match foo 123 although it shouldn't match anything (2nd field consists of numbers, we match on a letter). Am I doing anything wrong?

skim in terminal works as expected

> echo "foo 123" | sk --nth=1 --filter f
foo 123
> echo "foo 123" | sk --nth=2 --filter f
# no output, exit code 1

[lotabout]

This issue is somehow a technical inconsistency.

• Previously skim provides a all in one API that accepts strings. So the options could control how the lines are parsed and matched(e.g. nth).
• Current API could receive a stream of user customized items. It allows more customization but relies on user to provide item-specific info(e.g. nth)
• Currently SkimItemReader is only a simple wrapper that turns a BufRead into lines of string and nothing more.
• Thus not only nth, but also with-nth, ansi and other item specific options will stop working.

As for how to solve this problem, I now realized that maybe customization of items won't be common? I'll try to provide more helpful reader helpers(e.g. better SkimItemReader)

[murlakatamenka]

Thanks, I see it's a work in progress and things need time to stabilize and get shaped.

To be honest I quite struggle using skim as library once thing get a bit different from given examples.

For instance, slightly modified custom_item.rs, where we have a struct with id: u32 and inner: String, want to match just on inner but output id:

use skim::prelude::*;
use std::fmt;
struct MyItem {
    id: u32,
    inner: String,
}

impl fmt::Display for MyItem {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        write!(f, "{}\t{}", self.id, self.inner)
    }
}

impl SkimItem for MyItem {
    fn display(&self) -> Cow<AnsiString> {
        Cow::Owned(self.to_string().into())
    }

    fn text(&self) -> Cow<str> {
        // commented lines will produce out of bound
        // Cow::Borrowed(&self.inner)
        // Cow::Owned(self.inner.to_string())

        Cow::Owned(self.to_string())
    }
}

pub fn main() {
    let options = SkimOptionsBuilder::default().query(Some("7b")).build().unwrap();

    let vec = vec![
        MyItem {
            inner: "foo".to_string(),
            id: 123,
        },
        MyItem {
            inner: "bar".to_string(),
            id: 789,
        },
    ];

    let (tx_item, rx_item): (SkimItemSender, SkimItemReceiver) = unbounded();
    for item in vec {
        let _ = tx_item.send(Arc::new(item));
    }
    drop(tx_item); // so that skim could know when to stop waiting for more items.

    let selected_items = Skim::run_with(&options, Some(rx_item))
        .map(|out| out.selected_items)
        .unwrap_or_else(|| Vec::new());

    for item in selected_items {
        if let Some(item) = (*item).as_any().downcast_ref::<MyItem>() {
            println!("{}", item.id);
        }
    }
}

Matching on inner causes out of bound panics in threads, so I use to_string() twice just to make it work.

---

Actually, I found why it panics. SkimItem::get_matching_ranges() gets its range from display() rather than text(). text() is shorter than display() in the code above. I guess it's not intended to be this way.

I think the code above doesn't benefit from working with a struct instead of a just String. I'd be glad to get some feedback or if more complex custom_item.rs example is added.

[murlakatamenka]

I did explore how others use skim but most of them used just String + SkimItemReader::of_bufread() and matched on the whole lines.

Current workaround of how to match on needed range:

• use the same string for text() and display()
• override get_matching_ranges() with required range

String can be wrapped into struct that implements SkimItem trait.

But that's really a workaround, feels somewhat artificial.

---

Also I noticed that Skim::filter got moved into binary. Is it possible to use skim non-interactively now?

[lotabout]

SkimItem::get_matching_ranges() gets its range from display() rather than text(). text() is shorter than display() in the code above. I guess it's not intended to be this way

Thanks for point this out, it is a bug.

Is it possible to run use skim non-interactively now?

Yes. With --filter option.

[murlakatamenka]

I'm talking about library use of Skim::filter (https://docs.rs/skim/0.8.0/skim/struct.Skim.html#method.filter), it's been moved into binary and only Skim::run_with is available now (https://docs.rs/skim/0.8.1/skim/struct.Skim.html) for calling from library.

[lotabout]

@murlakatamenka Misunderstood. No longer available.

[murlakatamenka]

Thank you 👍