crsn/lib/spanned_sexp/src/lib.rs

//! A lightweight, self-contained s-expression parser and data format.
//! Use `parse` to get an s-expression from its string representation, and the
//! `Display` trait to serialize it, potentially by doing `sexp.to_string()`.

#![deny(unsafe_code)]

#[macro_use]
extern crate log;

use std::borrow::Cow;
use std::fmt;
use std::str::{self, FromStr};

use error::{ERes, err, spos};
pub use error::Error;
pub use error::SourcePosition;

#[cfg(test)]
mod test;

mod error;

/// A single data element in an s-expression. Floats are excluded to ensure
/// atoms may be used as keys in ordered and hashed data structures.
///
/// All strings must be valid utf-8.
#[derive(PartialEq, Clone, PartialOrd)]
#[allow(missing_docs)]
pub enum Atom {
    S(String),
    I(i64),
    F(f64),
}

/// An s-expression is either an atom or a list of s-expressions. This is
/// similar to the data format used by lisp.
#[derive(Clone)]
pub enum Sexp {
    /// Atom
    Atom(Atom, SourcePosition),
    /// List of expressions
    List(Vec<Sexp>, SourcePosition),
}

impl Sexp {
    pub fn pos(&self) -> &SourcePosition {
        match self {
            Sexp::List(_, pos) | Sexp::Atom(_, pos) => pos
        }
    }

    /// Check fi thsi Sexp is an atom
    pub fn is_atom(&self) -> bool {
        match self {
            Sexp::Atom(_, _) => true,
            _ => false,
        }
    }

    /// Check fi thsi Sexp is a list
    pub fn is_list(&self) -> bool {
        match self {
            Sexp::List(_, _) => true,
            _ => false,
        }
    }
}

impl PartialEq for Sexp {
    fn eq(&self, other: &Self) -> bool {
        match (self, other) {
            (Sexp::Atom(a, _), Sexp::Atom(b, _)) => {
                a == b
            }
            (Sexp::List(a, _), Sexp::List(b, _)) => {
                a == b
            }
            _ => false
        }
    }
}


fn atom_of_string(s: String) -> Atom {
    match FromStr::from_str(&s) {
        Ok(i) => return Atom::I(i),
        Err(_) => {}
    };

    match FromStr::from_str(&s) {
        Ok(f) => return Atom::F(f),
        Err(_) => {}
    };

    Atom::S(s)
}

// returns the char it found, and the new size if you wish to consume that char
fn peek(s: &str, pos: &usize) -> ERes<(char, usize)> {
    trace!("peek {}", pos);
    if *pos == s.len() { return err("unexpected eof", s, pos); }
    if s.is_char_boundary(*pos) {
        let ch = s[*pos..].chars().next().unwrap();
        let next = *pos + ch.len_utf8();
        Ok((ch, next))
    } else {
        // strings must be composed of valid utf-8 chars.
        unreachable!()
    }
}

fn expect(s: &str, pos: &mut usize, c: char) -> ERes<()> {
    trace!("expect {}", pos);
    let (ch, next) = peek(s, pos)?;
    *pos = next;
    if ch == c { Ok(()) } else { err("unexpected character", s, pos) }
}

fn consume_until_newline(s: &str, pos: &mut usize) -> ERes<()> {
    loop {
        if *pos == s.len() { return Ok(()); }
        let (ch, next) = peek(s, pos)?;
        *pos = next;
        if ch == '\n' { return Ok(()); }
    }
}

// zero or more spaces
fn zspace(s: &str, pos: &mut usize) -> ERes<()> {
    trace!("zspace {}", pos);
    loop {
        if *pos == s.len() { return Ok(()); }
        let (ch, next) = peek(s, pos)?;

        if ch == ';' { consume_until_newline(s, pos)? } else if ch.is_whitespace() { *pos = next; } else { return Ok(()); }
    }
}

fn parse_quoted_atom(s: &str, pos: &mut usize) -> ERes<Atom> {
    trace!("parse_quoted_atom {}", pos);
    let mut cs: String = String::new();

    expect(s, pos, '"')?;

    loop {
        let (ch, next) = peek(s, pos)?;
        if ch == '"' {
            *pos = next;
            break;
        } else if ch == '\\' {
            let (postslash, nextnext) = peek(s, &next)?;
            if postslash == '"' || postslash == '\\' {
                cs.push(postslash);
            } else {
                cs.push(ch);
                cs.push(postslash);
            }
            *pos = nextnext;
        } else {
            cs.push(ch);
            *pos = next;
        }
    }

    // Do not try i64 conversion, since this atom was explicitly quoted.
    Ok(Atom::S(cs))
}

fn parse_unquoted_atom(s: &str, pos: &mut usize) -> ERes<Atom> {
    trace!("parse_unquoted_atom {}", pos);
    let mut cs: String = String::new();

    loop {
        if *pos == s.len() { break; }
        let (c, next) = peek(s, pos)?;

        if c == ';' {
            consume_until_newline(s, pos)?;
            break;
        }
        if c.is_whitespace() || c == '(' || c == ')' { break; }
        cs.push(c);
        *pos = next;
    }

    Ok(atom_of_string(cs))
}

fn parse_atom(s: &str, pos: &mut usize) -> ERes<Atom> {
    trace!("parse_atom {}", pos);
    let (ch, _) = peek(s, pos)?;

    if ch == '"' { parse_quoted_atom(s, pos) } else { parse_unquoted_atom(s, pos) }
}

fn parse_list(s: &str, pos: &mut usize) -> ERes<Vec<Sexp>> {
    trace!("parse_list {}", pos);
    zspace(s, pos)?;
    expect(s, pos, '(')?;

    let mut sexps: Vec<Sexp> = Vec::new();

    loop {
        zspace(s, pos)?;
        let (c, next) = peek(s, pos)?;
        if c == ')' {
            *pos = next;
            break;
        }
        sexps.push(parse_sexp(s, pos)?);
    }

    zspace(s, pos)?;

    Ok(sexps)
}

fn parse_sexp(s: &str, pos: &mut usize) -> ERes<Sexp> {
    trace!("parse_sexp {}", pos);
    zspace(s, pos)?;
    let (c, _) = peek(s, pos)?;
    let r = if c == '(' {
        Ok(Sexp::List(parse_list(s, pos)?, spos(s, pos)))
    } else {
        Ok(Sexp::Atom(parse_atom(s, pos)?, spos(s, pos)))
    };
    zspace(s, pos)?;
    r
}

/// Constructs an atomic s-expression from a string.
pub fn atom_s(s: &str) -> Sexp {
    Sexp::Atom(Atom::S(s.to_owned()), Default::default())
}

/// Constructs an atomic s-expression from an int.
pub fn atom_i(i: i64) -> Sexp {
    Sexp::Atom(Atom::I(i), Default::default())
}

/// Constructs an atomic s-expression from a float.
pub fn atom_f(f: f64) -> Sexp {
    Sexp::Atom(Atom::F(f), Default::default())
}

/// Constructs a list s-expression given a slice of s-expressions.
pub fn list(xs: &[Sexp]) -> Sexp {
    Sexp::List(xs.to_owned(), Default::default())
}

/// Reads an s-expression out of a `&str`.
#[inline(never)]
pub fn parse(s: &str) -> Result<Sexp, Box<Error>> {
    let mut pos = 0;
    let ret = parse_sexp(s, &mut pos)?;
    if pos == s.len() { Ok(ret) } else { err("unrecognized post-s-expression data", s, &pos) }
}

// TODO: Pretty print in lisp convention, instead of all on the same line,
//  packed as tightly as possible. It's kinda ugly.

fn is_num_string(s: &str) -> bool {
    let x: Result<i64, _> = FromStr::from_str(&s);
    let y: Result<f64, _> = FromStr::from_str(&s);
    x.is_ok() || y.is_ok()
}

fn string_contains_whitespace(s: &str) -> bool {
    for c in s.chars() {
        if c.is_whitespace() { return true; }
    }
    false
}

fn quote(s: &str) -> Cow<str> {
    if !s.contains("\"")
        && !string_contains_whitespace(s)
        && !is_num_string(s) {
        Cow::Borrowed(s)
    } else {
        let mut r: String = "\"".to_string();
        r.push_str(&s.replace("\\", "\\\\").replace("\"", "\\\""));
        r.push_str("\"");
        Cow::Owned(r)
    }
}

impl fmt::Display for Atom {
    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
        match *self {
            Atom::S(ref s) => write!(f, "{}", quote(s)),
            Atom::I(i) => write!(f, "{}", i),
            Atom::F(d) => write!(f, "{}", d),
        }
    }
}

impl fmt::Display for Sexp {
    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
        match *self {
            Sexp::Atom(ref a, _) => write!(f, "{}", a),
            Sexp::List(ref xs, _) => {
                write!(f, "(")?;
                for (i, x) in xs.iter().enumerate() {
                    let s = if i == 0 { "" } else { " " };
                    write!(f, "{}{}", s, x)?;
                }
                write!(f, ")")
            }
        }
    }
}

impl fmt::Debug for Atom {
    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
        write!(f, "{}", self)
    }
}

impl fmt::Debug for Sexp {
    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
        write!(f, "{}", self)
    }
}