hyaenidae/content/src/email.rs

use crate::url::domain;
use combine::{
    choice, many, many1,
    parser::char::{alpha_num, char as parsechar},
    Parser, Stream,
};

fn unquoted<Input>() -> impl Parser<Input, Output = char>
where
    Input: Stream<Token = char>,
{
    choice((
        alpha_num(),
        parsechar('!'),
        parsechar('#'),
        parsechar('$'),
        parsechar('%'),
        parsechar('&'),
        parsechar('\''),
        parsechar('*'),
        parsechar('+'),
        parsechar('-'),
        parsechar('/'),
        parsechar('='),
        parsechar('?'),
        parsechar('^'),
        parsechar('`'),
        parsechar('{'),
        parsechar('|'),
        parsechar('}'),
        parsechar('~'),
    ))
}

fn unquoted_middle_segment<Input>() -> impl Parser<Input, Output = String>
where
    Input: Stream<Token = char>,
{
    parsechar('.')
        .and(many1(unquoted()))
        .map(|(c, s): (_, String)| {
            let mut string = String::new();
            string.push(c);
            string += &s;
            string
        })
}

fn unquoted_full<Input>() -> impl Parser<Input, Output = String>
where
    Input: Stream<Token = char>,
{
    many1(unquoted())
        .and(many(unquoted_middle_segment()))
        .map(|(s1, s2): (String, String)| s1 + &s2)
}

pub(crate) fn email<Input>() -> impl Parser<Input, Output = String>
where
    Input: Stream<Token = char>,
{
    unquoted_full()
        .skip(parsechar('@'))
        .and(domain().map(|d| d.0))
        .map(|(local, domain)| format!("{}@{}", local, domain))
}

#[cfg(test)]
mod tests {
    use super::*;
    use combine::EasyParser;

    #[test]
    fn unquoted_parses_chars() {
        for c in &['a', 'b', '$', '#'] {
            let s = c.to_string();
            let (_, rest) = unquoted().easy_parse(s.as_str()).unwrap();
            assert_eq!(rest, "");
        }
    }

    #[test]
    fn unquoted_middle_segment_parses_dots() {
        let (_, rest) = unquoted_middle_segment().easy_parse(".one").unwrap();
        assert_eq!(rest, "");
    }

    #[test]
    fn unquoted_full_parses_dots() {
        let (_, rest) = unquoted_full().easy_parse("one.two.three.four").unwrap();
        assert_eq!(rest, "");
    }

    #[test]
    fn parses_basic_email() {
        let (_, rest) = email().easy_parse("a@b").unwrap();
        assert_eq!(rest, "");
    }

    #[test]
    fn doesnt_parse_invalid_email() {
        assert!(email().easy_parse("@a@b").is_err());
    }

    #[test]
    fn parses_longer_email() {
        let (_, rest) = email()
            .easy_parse("one.two.three.four@sub.domain.tld")
            .unwrap();
        assert_eq!(rest, "");
    }

    #[test]
    fn doesnt_parse_double_dot() {
        assert!(email().easy_parse("bad..email@tld").is_err());
    }

    #[test]
    fn doesnt_parse_dot_local() {
        assert!(email().easy_parse(".local@tld").is_err());
    }

    #[test]
    fn doesnt_parse_end_dot_local() {
        assert!(email().easy_parse("local.@tld").is_err());
    }
}