rust-parser-combinator/src/combinators.rs

use crate::parser::{BoxedParser, ParseResult, Parser, ParserInput};

pub fn optional<P, I, O, E>(parser: P) -> impl Parser<I, Option<O>, E>
where
    P: Parser<I, O, E>,
    I: ParserInput + Clone,
{
    move |input: I| match parser.parse(input.clone()) {
        Ok((output, rest)) => Ok((Some(output), rest)),
        Err(_e) => Ok((None, input)),
    }
}

pub fn map<P, F, I, O1, O2, E>(parser: P, map_fn: F) -> impl Parser<I, O2, E>
where
    I: ParserInput,
    P: Parser<I, O1, E>,
    F: Fn(O1) -> O2,
{
    move |input| {
        parser
            .parse(input)
            .map(|(result, rest)| (map_fn(result), rest))
    }
}

pub struct Repeated<'a, I, O>
where
    I: ParserInput + Clone,
{
    inner_parser: BoxedParser<'a, I, O, I>,
    at_least: Option<u16>,
    at_most: Option<u16>,
}

impl<'a, I, O> Repeated<'a, I, O>
where
    I: ParserInput + Clone,
{
    pub fn at_least(self, n: u16) -> Self {
        Self {
            at_least: Some(n),
            ..self
        }
    }
    pub fn at_most(self, n: u16) -> Self {
        Self {
            at_most: Some(n),
            ..self
        }
    }

    pub fn separated_by<D, O2>(self, delimiter: D, allow_trailing: bool) -> SeparatedBy<'a, I, O>
    where
        D: Parser<I, O2, I> + 'a,
        O2: 'a,
        I: 'a,
    {
        SeparatedBy {
            inner_repeated: self,
            delimiter: BoxedParser::new(delimiter.to(())),
            allow_trailing,
        }
    }
}

impl<'a, I, O> Parser<I, Vec<O>, I> for Repeated<'a, I, O>
where
    I: ParserInput + Clone + 'a,
{
    fn parse(&self, input: I) -> ParseResult<I, Vec<O>, I> {
        let at_least = self.at_least.unwrap_or(0);
        let at_most = self.at_most.unwrap_or(u16::MAX);

        if at_most == 0 {
            return Ok((vec![], input));
        }

        let mut results = Vec::new();
        let mut count: u16 = 0;
        let mut further_input = input.clone();

        while let Ok((item, rest)) = self.inner_parser.parse(further_input.clone()) {
            results.push(item);
            further_input = rest;
            count += 1;
            if count >= at_most {
                break;
            }
        }
        if count < at_least {
            return Err(input);
        }

        Ok((results, further_input))
    }
}

pub struct SeparatedBy<'a, I, O>
where
    I: ParserInput + Clone,
{
    inner_repeated: Repeated<'a, I, O>,
    delimiter: BoxedParser<'a, I, (), I>,
    allow_trailing: bool,
}

impl<'a, I, O> Parser<I, Vec<O>, I> for SeparatedBy<'a, I, O>
where
    I: ParserInput + Clone + 'a,
{
    fn parse(&self, input: I) -> ParseResult<I, Vec<O>, I> {
        let at_least = self.inner_repeated.at_least.unwrap_or(0);
        let at_most = self.inner_repeated.at_most.unwrap_or(u16::MAX);
        let parser = &self.inner_repeated.inner_parser;
        let delimiter = &self.delimiter;

        if at_most == 0 {
            return Ok((vec![], input));
        }

        let mut results = Vec::new();
        let mut count: u16 = 0;
        let mut further_input;

        match parser.parse(input.clone()) {
            Ok((item, rest)) => {
                results.push(item);
                further_input = rest;
            }
            Err(_e) => {
                if at_least > 0 {
                    return Err(input);
                } else {
                    return Ok((vec![], input));
                }
            }
        }

        loop {
            match delimiter.parse(further_input.clone()) {
                Ok(((), rest)) => {
                    further_input = rest;
                }
                Err(_e) => {
                    break;
                }
            }

            match parser.parse(further_input.clone()) {
                Ok((item, rest)) => {
                    results.push(item);
                    further_input = rest;
                    count += 1;
                }
                Err(_e) if self.allow_trailing => {
                    break;
                }
                Err(e) => {
                    return Err(e);
                }
            }

            if count >= at_most {
                break;
            }
        }

        if count < at_least {
            return Err(input);
        }

        Ok((results, further_input))
    }
}

pub fn repeated<'a, P, I, O>(parser: P) -> Repeated<'a, I, O>
where
    P: Parser<I, O, I> + 'static,
    I: ParserInput + Clone + 'static,
{
    Repeated {
        inner_parser: BoxedParser::new(parser),
        at_least: None,
        at_most: None,
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::primitives::literal;

    #[test]
    fn test_map() {
        let lit_a = literal("a");
        let output = lit_a.map(|s| s.to_uppercase()).parse("a yolo");
        assert_eq!(output.unwrap(), ("A".to_string(), " yolo"));
    }

    #[test]
    fn test_one_or_more() {
        let p = repeated(literal("bongo ")).at_least(1);
        let input = "bongo bongo bongo bongo bongo ";

        let (output, rest) = p.parse(input).unwrap();
        assert_eq!(rest, "");
        assert_eq!(output.len(), 5);

        let (output, rest) = p.parse("bongo ecks").unwrap();
        assert_eq!(output.len(), 1);
        assert_eq!(rest, "ecks");
    }

    #[test]
    fn test_separated_by() {
        let p = repeated(literal("garb").to(20))
            .separated_by(repeated(literal(" ")).at_least(1), false);

        assert_eq!(
            p.parse("garb garb      garb     garb").unwrap(),
            (vec![20, 20, 20, 20], "")
        );

        assert!(p.parse("garb garb      garb     garb ").is_err());

        let p =
            repeated(literal("garb").to(20)).separated_by(repeated(literal(" ")).at_least(1), true);

        assert_eq!(
            p.parse("garb garb      garb     garb").unwrap(),
            (vec![20, 20, 20, 20], "")
        );

        assert_eq!(
            p.parse("garb garb      garb     garb     ").unwrap(),
            (vec![20, 20, 20, 20], "")
        );
        assert_eq!(
            p.parse("garb garb      garb     garb     q").unwrap(),
            (vec![20, 20, 20, 20], "q")
        );
    }
}