parser.rs

use nom;
use nom::{
    bytes::complete::{tag, take_until, take_while1},
    branch::alt,
    character::complete::{char, multispace0},
    combinator::{map, cut},
    error::VerboseError,
    multi::{many1,separated_list0},
    sequence::{preceded, terminated, tuple, separated_pair},
    IResult,
};

fn spaces(s: &str) -> IResult<&str, (), VerboseError<&str>> {
    alt((
        map(
            many1(tuple((
                multispace0,
                preceded(char(';'), terminated(take_until("\n"), char('\n'))),
                multispace0,
            ))),
            |_| (),
        ),
        map(multispace0, |_| ()),
    ))(s)
}

fn string_literal(s: &str) -> IResult<&str, &str, VerboseError<&str>> {
    preceded(char('\"'), cut(terminated(take_until("\""), char('\"'))))(s)
}

fn identifier(s: &str) -> IResult<&str, &str, VerboseError<&str>> {
    take_while1(|c: char| c.is_alphanumeric() || c == '.' || c == '_')(s)
}

#[test]
fn identifier_test() {
    assert_eq!(identifier("llvm.module.flags"),Ok(("","llvm.module.flags")));
    assert_eq!(identifier("0"),Ok(("","0")));
}

fn parse_metadata(s: &str) -> IResult<&str, (&str, Vec<(&str,&str)>), VerboseError<&str>> {
    fn mid(s: &str) -> IResult<&str, &str, VerboseError<&str>> {
        preceded(spaces,preceded(char('!'), identifier))(s)
    }
    fn mtids(s: &str) -> IResult<&str, Vec<(&str,&str)>, VerboseError<&str>> {
        map(tuple((spaces,char('!'),spaces,char('{'),
                separated_list0(preceded(spaces,tag(",")), mtid),spaces,char('}'))),
            |(_,_,_,_,r,_,_)| r)(s)
    }
    fn mtid(s: &str) -> IResult<&str, (&str,&str), VerboseError<&str>> {
        preceded(spaces,tuple((alt((tag("!"),identifier)),value)))(s)
    }
    fn value(s: &str) -> IResult<&str, &str, VerboseError<&str>> {
        preceded(spaces,alt((identifier,string_literal)))(s)
    }
    separated_pair(mid, preceded(spaces,tag("=")), mtids)(s)
}
#[test]
fn parse_metadata_test() {
    let s = "
        !llvm.module.flags = !{!0}
        !llvm.ident = !{!1}
        !0 = !{i32 1, !\"wchar_size\", i32 4}
        !1 = !{!\"clang version 10.0.0-4ubuntu1 \"}
        !llvm.module.flags = !{!0, !1, !2}
        !0 = !{i32 7, !\"PIC Level\", i32 2}
        !1 = !{i32 7, !\"PIE Level\", i32 2}
        !2 = !{i32 2, !\"RtLibUseGOT\", i32 1}
        !3 = !{}
        !4 = !{i32 2849348}
        !4 = !{i32 2849319}
        !4 = !{i32 2849383}";
    assert_eq!(many1(parse_metadata)(s),Ok(("",vec![
        ("llvm.module.flags", vec![("!", "0")]),
        ("llvm.ident", vec![("!", "1")]),
        ("0", vec![("i32", "1"), ("!", "wchar_size"), ("i32", "4")]),
        ("1", vec![("!", "clang version 10.0.0-4ubuntu1 ")]),
        ("llvm.module.flags", vec![("!", "0"), ("!", "1"), ("!", "2")]),
        ("0", vec![("i32", "7"), ("!", "PIC Level"), ("i32", "2")]),
        ("1", vec![("i32", "7"), ("!", "PIE Level"), ("i32", "2")]),
        ("2", vec![("i32", "2"), ("!", "RtLibUseGOT"), ("i32", "1")]),
        ("3", vec![]),
        ("4", vec![("i32", "2849348")]),
        ("4", vec![("i32", "2849319")]),
        ("4", vec![("i32", "2849383")])
    ])));
}

fn main() {
}