futilehdl/src/parser/tokens.rs

//! convert text into a token stream

use super::{
    literals::{identifier, ws0},
    IResult, Span,
};
use nom::{
    branch::alt,
    bytes::complete::tag,
    character::complete::{anychar, digit1},
    combinator::{consumed, map, recognize},
    error::ParseError,
    multi::many0,
};
use std::fmt;

pub struct Token<'a> {
    span: Span<'a>,
    kind: TokenKind,
}

impl fmt::Debug for Token<'_> {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        write!(
            f,
            "{:?} @{} {:?}",
            self.kind,
            self.span.location_offset(),
            self.span.fragment()
        )?;
        Ok(())
    }
}

impl<'a> Token<'a> {
    fn new(span: Span<'a>, kind: TokenKind) -> Self {
        Self { span, kind }
    }

    pub fn span(&self) -> Span {
        self.span
    }
}

#[derive(Debug, PartialEq, Clone)]
pub enum TokenKind {
    // no whitespace, for now
    // no token trees either, for now
    // Braces
    LParen,
    RParen,
    LAngle,
    RAngle,
    LBrace,
    RBrace,
    LSquare,
    RSquare,
    // single chars
    Colon,
    Semicolon,
    Comma,
    BitNot,
    BitAnd,
    BitOr,
    BitXor,
    EqAssign,
    // Multi Chars
    FatArrow,
    RArrow,
    // Literals
    Ident,
    Number,
    // Keywords
    Module,
    Assign,
    Match,
    Proc,
    // Error
    Error,
}

#[derive(Debug, Clone)]
pub struct TokenSpan<'a> {
    rest: &'a [Token<'a>],
    pos: usize,
}

impl<'a> TokenSpan<'a> {
    pub fn new(rest: &'a [Token<'a>]) -> Self {
        Self { rest, pos: 0 }
    }
    pub fn with_pos(rest: &'a [Token<'a>], pos: usize) -> Self {
        Self { rest, pos }
    }
}

impl nom::InputTake for TokenSpan<'_> {
    fn take(&self, count: usize) -> Self {
        TokenSpan::with_pos(&self.rest[..count], self.pos + count)
    }

    fn take_split(&self, count: usize) -> (Self, Self) {
        let (head, tail) = &self.rest.split_at(count);
        (
            TokenSpan::with_pos(head, self.pos),
            TokenSpan::with_pos(tail, self.pos + count),
        )
    }
}

impl nom::InputLength for TokenSpan<'_> {
    fn input_len(&self) -> usize {
        self.rest.len()
    }
}

impl nom_greedyerror::Position for TokenSpan<'_> {
    fn position(&self) -> usize {
        self.pos
    }
}

/// combinator that matches a token kind
pub fn token<'a, E>(
    kind: TokenKind,
) -> impl FnMut(TokenSpan<'a>) -> nom::IResult<TokenSpan, &Token, E>
where
    E: ParseError<TokenSpan<'a>>,
{
    move |input: TokenSpan| {
        let next = &input.rest[0];
        if next.kind == kind.clone() {
            let rest = TokenSpan::with_pos(&input.rest[1..], input.pos + 1);
            Ok((rest, next))
        } else {
            Err(nom::Err::Error(E::from_error_kind(
                input,
                // TODO: Proper errors here
                nom::error::ErrorKind::Tag,
            )))
        }
    }
}

pub fn lex(input: Span) -> IResult<Span, Vec<Token>> {
    many0(ws0(alt((
        lex_keywords,
        lex_literals,
        lex_braces,
        lex_punctuation,
        map(recognize(anychar), |span| {
            Token::new(span, TokenKind::Error)
        }),
    ))))(input)
}

fn lex_braces(input: Span) -> IResult<Span, Token> {
    map(
        consumed(alt((
            map(tag("("), |_| TokenKind::LParen),
            map(tag(")"), |_| TokenKind::RParen),
            map(tag("<"), |_| TokenKind::LAngle),
            map(tag(">"), |_| TokenKind::RAngle),
            map(tag("{"), |_| TokenKind::LBrace),
            map(tag("}"), |_| TokenKind::RBrace),
            map(tag("["), |_| TokenKind::LSquare),
            map(tag("]"), |_| TokenKind::RSquare),
        ))),
        |(span, kind)| Token::new(span, kind),
    )(input)
}

fn lex_literals(input: Span) -> IResult<Span, Token> {
    map(
        consumed(alt((
            map(identifier, |_| TokenKind::Ident),
            map(digit1, |_| TokenKind::Number),
        ))),
        |(span, kind)| Token::new(span, kind),
    )(input)
}

fn lex_punctuation(input: Span) -> IResult<Span, Token> {
    map(
        consumed(alt((
            map(tag(":"), |_| TokenKind::Colon),
            map(tag(";"), |_| TokenKind::Semicolon),
            map(tag(","), |_| TokenKind::Comma),
            map(tag("->"), |_| TokenKind::RArrow),
            map(tag("=>"), |_| TokenKind::FatArrow),
            map(tag("~"), |_| TokenKind::BitNot),
            map(tag("&"), |_| TokenKind::BitAnd),
            map(tag("^"), |_| TokenKind::BitXor),
            map(tag("|"), |_| TokenKind::BitOr),
            map(tag("="), |_| TokenKind::EqAssign),
        ))),
        |(span, kind)| Token::new(span, kind),
    )(input)
}

fn lex_keywords(input: Span) -> IResult<Span, Token> {
    map(
        consumed(alt((
            map(tag("module"), |_| TokenKind::Module),
            map(tag("assign"), |_| TokenKind::Assign),
            map(tag("match"), |_| TokenKind::Match),
            map(tag("proc"), |_| TokenKind::Proc),
        ))),
        |(span, kind)| Token::new(span, kind),
    )(input)
}