Witch of Git - ess/blob - src/lib.rs

   1 //! A lightweight S-expression parser intended for language implementation.
   2
   3 // #![warn(missing_docs)]
   4 #![deny(unsafe_code)]
   5
   6 use std::borrow::Cow;
   7
   8 /// A type representing arbitrary symbolic expressions. `Sexp` carries the
   9 /// source code location it came from along with it for later diagnostic
  10 /// purposes.
  11 #[derive(Debug, PartialEq, Clone, PartialOrd)]
  12 pub enum Sexp<'a, Loc=ByteSpan> where Loc: Span {
  13     /// A value representing a symbol.
  14     Sym(Cow<'a, str>, Loc),
  15     /// A value representing a string literal.
  16     Str(Cow<'a, str>, Loc),
  17     /// A value representing a single character.
  18     Char(char, Loc),
  19     /// A value representing an integer. Any number containing no decimal point
  20     /// will be parsed as an `Int`.
  21     Int(i64, Loc),
  22     /// A value representing a floating point number. Any number containing a
  23     /// decimal point will be parsed as a `Float`.
  24     Float(f64, Loc),
  25     /// A list of subexpressions.
  26     List(Vec<Sexp<'a, Loc>>, Loc),
  27 }
  28
  29 impl<'a, Loc> Sexp<'a, Loc> where Loc: Span {
  30     pub fn get_loc(&self) -> &Loc {
  31         match *self {
  32             Sexp::Sym(.., ref l) => l,
  33             Sexp::Str(.., ref l) => l,
  34             Sexp::Char(.., ref l) => l,
  35             Sexp::Int(.., ref l) => l,
  36             Sexp::Float(.., ref l) => l,
  37             Sexp::List(.., ref l) => l,
  38         }
  39     }
  40
  41     pub fn get_loc_mut(&mut self) -> &mut Loc {
  42         match *self {
  43             Sexp::Sym(.., ref mut l) => l,
  44             Sexp::Str(.., ref mut l) => l,
  45             Sexp::Char(.., ref mut l) => l,
  46             Sexp::Int(.., ref mut l) => l,
  47             Sexp::Float(.., ref mut l) => l,
  48             Sexp::List(.., ref mut l) => l,
  49         }
  50     }
  51 }
  52
  53 \f
  54 // General Parsing Types ///////////////////////////////////////////////////////
  55
  56 pub trait Span {
  57     type Begin;
  58
  59     fn offset(&self, begin: Self::Begin) -> Self;
  60     fn begin(&self) -> Self::Begin;
  61     fn union(&self, other: &Self) -> Self;
  62 }
  63
  64 #[derive(Debug, PartialEq, Eq, Clone)]
  65 pub enum ParseResult<'a, T, E> {
  66     Done(&'a str, T),
  67     Error(E),
  68 }
  69
  70 use ParseResult::*;
  71
  72 \f
  73 // Specific Parsing Types (ParseError, ByteSpan) ///////////////////////////////
  74
  75 /// Indicates how parsing failed.
  76 #[derive(Debug, PartialEq, Eq, Clone)]
  77 pub enum ParseError<Loc=ByteSpan> where Loc: Span {
  78     /// We can't explain how the parsing failed.
  79     UnexpectedEof,
  80     Symbol(Box<ParseError>, Loc),
  81     Number(Box<ParseError>, Loc),
  82     Unexpected(char, Loc::Begin),
  83     Unimplemented,
  84 }
  85
  86 type ByteSpan = (usize, usize);
  87
  88 impl Span for ByteSpan {
  89     type Begin = usize;
  90
  91     fn offset(&self, begin: Self::Begin) -> Self {
  92         (self.0 + begin, self.1 + begin)
  93     }
  94
  95     fn begin(&self) -> Self::Begin {
  96         self.0
  97     }
  98
  99     fn union(&self, other: &Self) -> Self {
 100         use std::cmp::{min, max};
 101         (min(self.0, other.0), max(self.1, other.1))
 102     }
 103 }
 104
 105
 106 \f
 107 // Parsing Utilities ///////////////////////////////////////////////////////////
 108
 109 trait IsDelimeter {
 110     fn is_delimiter(&self) -> bool;
 111 }
 112
 113 impl IsDelimeter for char {
 114     fn is_delimiter(&self) -> bool {
 115         self.is_whitespace() || *self == ';'
 116             || *self == '(' || *self == ')'
 117             || *self == '[' || *self == ']'
 118             || *self == '{' || *self == '}'
 119             || *self == '"' || *self == '\''
 120             || *self == '`' || *self == ','
 121     }
 122 }
 123
 124 \f
 125 // Parsers /////////////////////////////////////////////////////////////////////
 126
 127 // pub fn parse_one(input: &str) -> Result<Sexp, ParseError>;
 128
 129 // pub fn parse(input: &str) -> Result<Vec<Sexp>, ParseError>;
 130
 131 pub fn parse_number(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 132     // Consume all the whitespace at the beginning of the string
 133     let end_of_white = if let Some(pos) = input.find(|c: char| !c.is_whitespace()) {
 134         pos
 135     } else {
 136         return Error(ParseError::Number(
 137             Box::new(ParseError::UnexpectedEof),
 138             (input.len(), input.len()).offset(start_loc)));
 139     };
 140
 141     let input = &input[end_of_white..];
 142     let start_loc = start_loc + end_of_white;
 143
 144     match input.chars().next() {
 145         Some(c) if !c.is_digit(10) => {
 146             return Error(ParseError::Number(
 147                 Box::new(ParseError::Unexpected(c, 0)),
 148                 (0, c.len_utf8()).offset(start_loc)));
 149         }
 150         None => return Error(ParseError::Number(
 151             Box::new(ParseError::UnexpectedEof),
 152             (0, 0).offset(start_loc))),
 153         _ => (),
 154     }
 155
 156     let base = 10;
 157
 158     let mut end = 0;
 159     // Before the decimal point
 160     for (i, c) in input.char_indices() {
 161         if c == '.' {
 162             end = i + 1;
 163             break;
 164         }
 165
 166         if c.is_delimiter() {
 167             return Done(&input[i..],
 168                         Sexp::Int(input[..i].parse().expect("Already matched digits"),
 169                                   (0, i).offset(start_loc)));
 170         }
 171
 172         if !c.is_digit(base) {
 173             return Error(ParseError::Number(
 174                 Box::new(ParseError::Unexpected(c, i)),
 175                 (i, i).offset(start_loc)));
 176         }
 177
 178         end = i + c.len_utf8();
 179     }
 180
 181     if input[end..].is_empty() {
 182         return Done(&input[end..],
 183                     Sexp::Int(input.parse().expect("Already matched digits"),
 184                               (0, end).offset(start_loc)));
 185     }
 186
 187     // After the decimal point
 188     for (i, c) in input[end..].char_indices() {
 189         if c.is_delimiter() {
 190             return Done(&input[i+end..],
 191                         Sexp::Float(input[..end+i].parse().expect("Already matched digits.digits"),
 192                                     (0, end+i).offset(start_loc)));
 193         }
 194
 195         if !c.is_digit(base) {
 196             return Error(ParseError::Number(
 197                 Box::new(ParseError::Unexpected(c, i + end)),
 198                 (i+end, i+end).offset(start_loc)));
 199         }
 200     }
 201
 202     Done(&input[input.len()..],
 203          Sexp::Float(input.parse().expect("Already matched digits.digits"),
 204                      (0, input.len()).offset(start_loc)))
 205 }
 206
 207 pub fn parse_symbol(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 208     let end_of_white = if let Some(pos) = input.find(|c: char| !c.is_whitespace()) {
 209         pos
 210     } else {
 211         return Error(ParseError::Symbol(
 212             Box::new(ParseError::UnexpectedEof),
 213             (input.len(), input.len()).offset(start_loc)));
 214     };
 215
 216     let input = &input[end_of_white..];
 217     let start_loc = start_loc + end_of_white;
 218
 219     match input.chars().next() {
 220         Some(c@'#') | Some(c@':') | Some(c@'0'...'9') =>
 221             return Error(ParseError::Symbol(
 222                 Box::new(ParseError::Unexpected(c, start_loc)),
 223                 (0, 0).offset(start_loc))),
 224         Some(c) if c.is_delimiter() =>
 225             return Error(ParseError::Symbol(
 226                 Box::new(ParseError::Unexpected(c, start_loc)),
 227                 (0, 0).offset(start_loc))),
 228         Some(_) => (),
 229         None => unreachable!(),
 230     }
 231
 232     for (i, c) in input.char_indices() {
 233         if c.is_delimiter() {
 234             return Done(&input[i..],
 235                         Sexp::Sym(input[..i].into(), (0, i).offset(start_loc)));
 236         }
 237     }
 238
 239     Done(&input[input.len()..],
 240          Sexp::Sym(input.into(), (0, input.len()).offset(start_loc)))
 241 }
 242
 243 \f
 244 // Tests ///////////////////////////////////////////////////////////////////////
 245
 246 #[cfg(test)]
 247 mod test {
 248     use super::*;
 249     use super::ParseResult::*;
 250
 251     #[test]
 252     fn test_parse_number() {
 253         assert_eq!(parse_number("1", 0), Done("", Sexp::Int(1, (0, 1))));
 254         assert_eq!(parse_number(" 13", 0), Done("", Sexp::Int(13, (1, 3))));
 255         assert_eq!(parse_number("1.2", 0), Done("", Sexp::Float(1.2, (0, 3))));
 256         assert_eq!(parse_number("\u{3000}4.2", 0), Done("", Sexp::Float(4.2, (0, 3).offset('\u{3000}'.len_utf8()))));
 257         assert_eq!(parse_number("       42 ", 0), Done(" ", Sexp::Int(42, (2, 4))));
 258         assert_eq!(parse_number(" 4.2  ", 0), Done("  ", Sexp::Float(4.2, (1, 4))));
 259         assert_eq!(parse_number("1()", 0), Done("()", Sexp::Int(1, (0, 1))));
 260         assert_eq!(parse_number("3.6()", 0), Done("()", Sexp::Float(3.6, (0, 3))));
 261
 262         assert_eq!(parse_number("", 0), Error(ParseError::Number(Box::new(ParseError::UnexpectedEof), (0, 0))));
 263         assert_eq!(parse_number("123a", 0), Error(ParseError::Number(Box::new(ParseError::Unexpected('a', 3)), (3, 3))));
 264         assert_eq!(parse_number("66.6+", 0), Error(ParseError::Number(Box::new(ParseError::Unexpected('+', 4)), (4, 4))));
 265     }
 266
 267     #[test]
 268     fn test_parse_ident() {
 269         assert_eq!(parse_symbol("+", 0), Done("", Sexp::Sym("+".into(), (0, 1))));
 270         assert_eq!(parse_symbol(" nil?", 0), Done("", Sexp::Sym("nil?".into(), (1, 5))));
 271         assert_eq!(parse_symbol(" ->socket", 0), Done("", Sexp::Sym("->socket".into(), (1, 9))));
 272         assert_eq!(parse_symbol("fib(", 0), Done("(", Sexp::Sym("fib".into(), (0, 3))));
 273         assert_eq!(parse_symbol("foo2", 0), Done("", Sexp::Sym("foo2".into(), (0, 4))));
 274
 275         // We reserve #foo for the implementation to do as it wishes
 276         assert_eq!(parse_symbol("#hi", 0), Error(ParseError::Symbol(Box::new(ParseError::Unexpected('#', 0)), (0, 0))));
 277         // We reserve :foo for keywords
 278         assert_eq!(parse_symbol(":hi", 0), Error(ParseError::Symbol(Box::new(ParseError::Unexpected(':', 0)), (0, 0))));
 279
 280         assert_eq!(parse_symbol("", 0), Error(ParseError::Symbol(Box::new(ParseError::UnexpectedEof), (0, 0))));
 281         assert_eq!(parse_symbol("0", 0), Error(ParseError::Symbol(Box::new(ParseError::Unexpected('0', 0)), (0, 0))));
 282         assert_eq!(parse_symbol("()", 0), Error(ParseError::Symbol(Box::new(ParseError::Unexpected('(', 0)), (0, 0))));
 283     }
 284 }
 285
 286 // #[cfg(test)]
 287
 288 // #[cfg(test)]
 289 // #[test]
 290 // fn test_parse_string() {
 291 //     assert_eq!(string(r#""hello""#), IResult::Done("", Sexp::Str("hello".into())));
 292 //     assert_eq!(string(r#"  "this is a nice string
 293 // with 0123 things in it""#),
 294 //                IResult::Done("", Sexp::Str("this is a nice string\nwith 0123 things in it".into())));
 295
 296 //     assert!(string(r#""hi"#).is_err());
 297 // }
 298
 299 // #[cfg(test)]
 300 // #[test]
 301 // fn test_parse_char() {
 302 //     assert_eq!(character(r#"#\""#), IResult::Done("", Sexp::Char('"')));
 303 //     assert_eq!(character(r#"#\ "#), IResult::Done("", Sexp::Char(' ')));
 304 //     assert_eq!(character(r#"  #\\"#), IResult::Done("", Sexp::Char('\\')));
 305
 306 //     assert!(character("#").is_incomplete());
 307 //     assert!(character("a").is_err());
 308 // }
 309
 310
 311 // #[cfg(test)]
 312 // #[test]
 313 // fn test_parse_list() {
 314 //     assert_eq!(list("()"), IResult::Done("", vec![]));
 315 //     assert_eq!(list("(1)"), IResult::Done("", vec![Sexp::Int(1)]));
 316 //     assert_eq!(list("  ( 1    2  3 a )"), IResult::Done("", vec![
 317 //         Sexp::Int(1),
 318 //         Sexp::Int(2),
 319 //         Sexp::Int(3),
 320 //         Sexp::Sym("a".into()),
 321 //     ]));
 322 // }
 323
 324 // #[cfg(test)]
 325 // #[test]
 326 // fn test_parse_only_one() {
 327 //     assert!(parse_one("1 2").is_err());
 328 // }
 329
 330 // #[cfg(test)]
 331 // #[test]
 332 // fn test_parse_expression() {
 333 //     assert_eq!(parse_one(r#"
 334 // (def (main)
 335 //   (print (str "say " #\" "Hello, World" #\" " today!")))
 336 // "#),
 337 //                Ok(Sexp::List(vec![
 338 //                    Sexp::Sym("def".into()),
 339 //                    Sexp::List(
 340 //                        vec![Sexp::Sym("main".into())]
 341 //                    ),
 342 //                    Sexp::List(vec![
 343 //                        Sexp::Sym("print".into()),
 344 //                        Sexp::List(vec![
 345 //                            Sexp::Sym("str".into()),
 346 //                            Sexp::Str("say ".into()),
 347 //                            Sexp::Char('"'),
 348 //                            Sexp::Str("Hello, World".into()),
 349 //                            Sexp::Char('"'),
 350 //                            Sexp::Str(" today!".into()),
 351 //                        ])
 352 //                    ])
 353 //                ])));
 354 // }
 355
 356 // #[cfg(test)]
 357 // #[test]
 358 // fn test_parse_multi() {
 359 //     assert_eq!(parse(" 1 2 3 "),
 360 //                Ok(vec![Sexp::Int(1), Sexp::Int(2), Sexp::Int(3)]));
 361 // }