iowo/crates/pawarser/src/parser/input.rs

use enumset::{EnumSet, EnumSetType};

use super::SyntaxElement;

pub struct Input<'src, SyntaxKind: SyntaxElement> {
    raw: Vec<(SyntaxKind, &'src str)>,
    // enumset of meaningless tokens
    semantically_meaningless: EnumSet<SyntaxKind>,
    // indices of non-meaningless tokens
    meaningful_toks: Vec<usize>,
}

impl<'src, SyntaxKind: SyntaxElement> Input<'src, SyntaxKind> {
    pub fn new(
        raw_toks: Vec<(SyntaxKind, &'src str)>,
        meaningless: Option<EnumSet<SyntaxKind>>,
    ) -> Self {
        let mut meaningful_toks = Vec::new();

        if let Some(meaningless) = meaningless {
            let meaningful_toks = raw_toks
                .iter()
                .enumerate()
                .filter_map(|(i, tok)| (!meaningless.contains(tok.0)).then_some(i))
                .collect_into(&mut meaningful_toks);
        }

        Self {
            raw: raw_toks,
            semantically_meaningless: meaningless.unwrap_or_default(),
            meaningful_toks,
        }
    }

    pub fn kind(&self, idx: usize) -> SyntaxKind {
        let Some(meaningful_idx) = self.meaningful_toks.get(idx) else {
            return SyntaxKind::EOF;
        };

        self.raw.get(*meaningful_idx).unwrap().0
    }

    pub fn preceding_meaningless(&self, idx: usize) -> usize {
        assert!(self.meaningful_toks.len() > idx);

        if idx == 0 {
            // maybe should be `self.meaningful_toks[idx]` instead??
            1
        } else {
            self.meaningful_toks[idx] - self.meaningful_toks[idx - 1]
        }
    }

    /// get the count of meaningless tokens at the end of the file.
    pub fn meaningless_tail_len(&self) -> usize {
        self.raw.len() - (self.meaningful_toks.last().unwrap() + 1)
    }
}