rust-parser-combinator/src/lib.rs

#![feature(assert_matches)]
#![allow(dead_code)] //TODO eventually turn this off
use std::rc::Rc;

type ParseResult<I, O, E> = Result<(O, I), E>;

trait Parser<I, O, E> {
    fn parse(&self, input: I) -> ParseResult<I, O, E>;

    fn map<'a, F, O2>(self, map_fn: F) -> BoxedParser<'a, I, O2, E>
    where
        Self: Sized + 'a,
        I: 'a,
        E: 'a,
        O: 'a,
        O2: 'a,
        F: Fn(O) -> O2 + 'a,
    {
        BoxedParser::new(map(self, map_fn))
    }

    fn to<'a, O2>(self, item: O2) -> BoxedParser<'a, I, O2, E>
    where
        Self: Sized + 'a,
        I: 'a,
        O: 'a,
        O2: Clone + 'a,
        E: 'a,
    {
        self.map(move |_| item.clone())
    }

    fn then<'a, P, O2>(self, next_parser: P) -> BoxedParser<'a, I, (O, O2), E>
    where
        Self: Sized + 'a,
        I: 'a,
        O: 'a,
        O2: 'a,
        E: 'a,
        P: Parser<I, O2, E> + 'a,
    {
        BoxedParser::new(seq(self, next_parser))
    }
}

struct BoxedParser<'a, I, O, E> {
    inner: Box<dyn Parser<I, O, E> + 'a>,
}

impl<'a, I, O, E> BoxedParser<'a, I, O, E> {
    fn new<P>(inner: P) -> Self
    where
        P: Parser<I, O, E> + 'a,
    {
        BoxedParser {
            inner: Box::new(inner),
        }
    }
}

impl<'a, I, O, E> Parser<I, O, E> for BoxedParser<'a, I, O, E> {
    fn parse(&self, input: I) -> ParseResult<I, O, E> {
        self.inner.parse(input)
    }
}

impl<I, O, E, F> Parser<I, O, E> for F
where
    F: Fn(I) -> ParseResult<I, O, E>,
{
    fn parse(&self, input: I) -> ParseResult<I, O, E> {
        self(input)
    }
}

impl<I, O, E, T> Parser<I, O, E> for Rc<T>
where
    T: Parser<I, O, E>,
{
    fn parse(&self, input: I) -> ParseResult<I, O, E> {
        self.as_ref().parse(input)
    }
}

fn literal(expected: &'static str) -> impl Fn(&str) -> ParseResult<&str, &str, &str> {
    move |input| match input.get(0..expected.len()) {
        Some(next) if next == expected => Ok((expected, &input[expected.len()..])),
        _ => Err(input),
    }
}

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

fn seq<P1, P2, I, O1, O2, E>(parser1: P1, parser2: P2) -> impl Parser<I, (O1, O2), E>
where
    P1: Parser<I, O1, E>,
    P2: Parser<I, O2, E>,
{
    move |input| {
        parser1.parse(input).and_then(|(result1, rest1)| {
            parser2
                .parse(rest1)
                .map(|(result2, rest2)| ((result1, result2), rest2))
        })
    }
}

fn pred<P, F, I, O>(parser: P, pred_fn: F) -> impl Parser<I, O, I>
where
    P: Parser<I, O, I>,
    F: Fn(&O) -> bool,
{
    move |input| {
        parser.parse(input).and_then(|(result, rest)| {
            if pred_fn(&result) {
                Ok((result, rest))
            } else {
                Err(rest)
            }
        })
    }
}

fn zero_or_more<P, I, O>(parser: P) -> impl Parser<I, Vec<O>, I>
where
    P: Parser<I, O, I>,
    I: Copy,
{
    move |mut input| {
        let mut results = Vec::new();

        while let Ok((item, rest)) = parser.parse(input) {
            results.push(item);
            input = rest;
        }

        Ok((results, input))
    }
}

fn one_or_more<P, I, O>(parser: P) -> impl Parser<I, Vec<O>, I>
where
    P: Parser<I, O, I>,
    I: Copy,
{
    let parser = std::rc::Rc::new(parser);
    map(
        seq(parser.clone(), zero_or_more(parser)),
        |(first, rest)| {
            let mut output = vec![first];
            output.extend(rest.into_iter());
            output
        },
    )
}

/// Parses a standard identifier in a programming language
fn identifier(input: &str) -> ParseResult<&str, String, &str> {
    let mut chars = input.chars();
    let mut buf = String::new();

    match chars.next() {
        Some(ch) if ch.is_alphabetic() => buf.push(ch),
        _ => return Err(input),
    }

    while let Some(next) = chars.next() {
        if next.is_alphanumeric() {
            buf.push(next);
        } else {
            break;
        }
    }

    let next_index = buf.len();
    Ok((buf, &input[next_index..]))
}

fn any_char(input: &str) -> ParseResult<&str, char, &str> {
    match input.chars().next() {
        Some(ch) => Ok((ch, &input[ch.len_utf8()..])),
        None => Err(input),
    }
}

fn choice<P1, P2, I, O, E>(parser1: P1, parser2: P2) -> impl Parser<I, O, E>
where
    P1: Parser<I, O, E>,
    P2: Parser<I, O, E>,
    I: Copy,
{
    move |input| match parser1.parse(input) {
        ok @ Ok(..) => ok,
        Err(_e) => parser2.parse(input),
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use std::assert_matches::assert_matches;

    #[test]
    fn test_parsing() {
        let output = literal("a")("a yolo");
        assert_matches!(output.unwrap(), ("a", " yolo"));
    }

    #[test]
    fn test_identifier() {
        assert_matches!(identifier("bongo1beans"), Ok((s, "")) if s == "bongo1beans");
        assert_matches!(identifier("2bongo1beans"), Err("2bongo1beans"));
    }

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

    #[test]
    fn test_seq() {
        let p = seq(identifier, seq(literal(" "), literal("ruts")));
        assert_matches!(p.parse("fort1 ruts"), Ok((r, "")) if r.0 == "fort1" && r.1 == (" ", "ruts") );

        let p = identifier.then(literal(" ")).then(literal("ruts"));
        assert_matches!(p.parse("fort1 ruts"), Ok((r, "")) if r.0.0 == "fort1" && r.0.1== " " && r.1 == "ruts");
    }

    #[test]
    fn test_one_or_more() {
        let p = one_or_more(literal("bongo "));
        let input = "bongo bongo bongo bongo bongo ";
        assert_matches!(p.parse(input), Ok((v, "")) if v.len() == 5);
        let input = "bongo ecks";
        assert_matches!(p.parse(input), Ok((v, "ecks")) if v.len() == 1);
    }

    #[test]
    fn test_pred() {
        let p = pred(any_char, |c| *c == 'f');
        assert_eq!(p.parse("frog"), Ok(('f', "rog")));
    }

    #[test]
    fn test_choice() {
        let p = choice(literal("gnostika").to(1), one_or_more(literal(" ")).to(2));
        assert_eq!(p.parse("gnostika twentynine"), Ok((1, " twentynine")));
    }

    /*
     * JSON BNF
     * <JSON>     ::= <value>
    <value>    ::= <object> | <array> | <boolean> | <string> | <number> | <null>
    <array>    ::= "[" [<value>] {"," <value>}* "]"
    <object>   ::= "{" [<property>] {"," <property>}* "}"
    <property> ::= <string> ":" <value>
    */

    #[test]
    fn parse_json() {
        let json_null = literal("null");
        let json_true = literal("true");
        let json_false = literal("false");

        let json_value = choice(json_null, choice(json_true, json_false));

        assert_matches!(json_value.parse("true"), Ok(("true", "")));
    }
}