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) | Sexp::Str(.., ref l) |
  33             Sexp::Char(.., ref l) | Sexp::Int(.., ref l) |
  34             Sexp::Float(.., ref l) | Sexp::List(.., ref l) => l,
  35         }
  36     }
  37
  38     pub fn get_loc_mut(&mut self) -> &mut Loc {
  39         match *self {
  40             Sexp::Sym(.., ref mut l) | Sexp::Str(.., ref mut l) |
  41             Sexp::Char(.., ref mut l) | Sexp::Int(.., ref mut l) |
  42             Sexp::Float(.., ref mut l) | Sexp::List(.., ref mut l) => l,
  43         }
  44     }
  45 }
  46
  47 \f
  48 // General Parsing Types ///////////////////////////////////////////////////////
  49
  50 pub trait Span {
  51     type Begin;
  52
  53     fn offset(&self, begin: Self::Begin) -> Self;
  54     fn begin(&self) -> Self::Begin;
  55     fn union(&self, other: &Self) -> Self;
  56 }
  57
  58 #[derive(Debug, PartialEq, Eq, Clone)]
  59 pub enum ParseResult<'a, T, E> {
  60     Done(&'a str, T),
  61     Error(E),
  62 }
  63
  64 use ParseResult::*;
  65
  66 \f
  67 // Specific Parsing Types (ParseError, ByteSpan) ///////////////////////////////
  68
  69 /// Indicates how parsing failed.
  70 #[derive(Debug, PartialEq, Eq, Clone)]
  71 pub enum ParseError<Loc=ByteSpan> where Loc: Span {
  72     UnexpectedEof,
  73     List(Box<ParseError>, Loc),
  74     Sexp(Box<ParseError>, Loc),
  75     Char(Box<ParseError>, Loc),
  76     String(Box<ParseError>, Loc),
  77     Symbol(Box<ParseError>, Loc),
  78     Number(Box<ParseError>, Loc),
  79     Unexpected(char, Loc::Begin),
  80     Unimplemented,
  81 }
  82
  83 type ByteSpan = (usize, usize);
  84
  85 impl Span for ByteSpan {
  86     type Begin = usize;
  87
  88     fn offset(&self, begin: Self::Begin) -> Self {
  89         (self.0 + begin, self.1 + begin)
  90     }
  91
  92     fn begin(&self) -> Self::Begin {
  93         self.0
  94     }
  95
  96     fn union(&self, other: &Self) -> Self {
  97         use std::cmp::{min, max};
  98         (min(self.0, other.0), max(self.1, other.1))
  99     }
 100 }
 101
 102
 103 \f
 104 // Parsing Utilities ///////////////////////////////////////////////////////////
 105
 106 trait IsDelimeter {
 107     fn is_delimiter(&self) -> bool;
 108 }
 109
 110 impl IsDelimeter for char {
 111     fn is_delimiter(&self) -> bool {
 112         self.is_whitespace() || *self == ';'
 113             || *self == '(' || *self == ')'
 114             || *self == '[' || *self == ']'
 115             || *self == '{' || *self == '}'
 116             || *self == '"' || *self == '\''
 117             || *self == '`' || *self == ','
 118     }
 119 }
 120
 121 macro_rules! consume_whitespace {
 122     ($input:expr, $start_loc:expr, $ErrorFn:expr) => {
 123         if let Some(pos) = $input.find(|c: char| !c.is_whitespace()) {
 124             (&$input[pos..], $start_loc + pos)
 125         } else {
 126             return Error($ErrorFn(
 127                 Box::new(ParseError::UnexpectedEof),
 128                 ($input.len(), $input.len()).offset($start_loc)));
 129         }
 130     }
 131 }
 132
 133 \f
 134 // Top Level Parsers ///////////////////////////////////////////////////////////
 135
 136 pub fn parse_one(input: &str) -> Result<(Sexp, &str), ParseError> {
 137     match parse_sexp(input, 0) {
 138         Done(rest, result) => Ok((result, rest)),
 139         Error(err) => Err(err),
 140     }
 141 }
 142
 143 pub fn parse(mut input: &str) -> (Vec<Sexp>, Option<ParseError>) {
 144     let mut start_loc = 0;
 145     let mut results = Vec::new();
 146     loop {
 147         match parse_sexp(input, start_loc) {
 148             Done(rest, result) => {
 149                 input = rest;
 150                 start_loc = result.get_loc().1;
 151                 results.push(result);
 152                 if rest.trim() == "" {
 153                     return (results, None);
 154                 }
 155             }
 156             Error(err) => {
 157                 return (results, Some(err));
 158             }
 159         }
 160     }
 161 }
 162
 163 \f
 164 // Core Parsers ////////////////////////////////////////////////////////////////
 165
 166 pub fn parse_sexp(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 167     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Sexp);
 168
 169     match input.chars().next() {
 170         Some('0'...'9') => parse_number(input, start_loc),
 171         Some('(') => parse_list(input, start_loc),
 172         Some('#') => parse_character(input, start_loc),
 173         Some('"') => parse_string(input, start_loc),
 174         Some(_) => parse_symbol(input, start_loc),
 175         None => unreachable!(),
 176     }
 177 }
 178
 179 pub fn parse_list(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 180     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::List);
 181
 182     match input.chars().nth(0) {
 183         Some('(') => (),
 184         Some(c) =>
 185             return Error(ParseError::List(
 186                 Box::new(ParseError::Unexpected(c, 0)),
 187                 (0, 0).offset(start_loc))),
 188         None => unreachable!(),
 189     }
 190
 191     let mut input = &input[1..];
 192     let mut loc = start_loc + 1;
 193     let mut members = Vec::new();
 194     loop {
 195         {
 196             let (new_input, new_loc) = consume_whitespace!(input, loc, ParseError::List);
 197             input = new_input;
 198             loc = new_loc;
 199         }
 200
 201         match input.chars().nth(0) {
 202             Some(')') =>
 203                 return Done(&input[1..],
 204                             Sexp::List(members, (start_loc, loc+1))),
 205             Some(_) => (),
 206             None => unreachable!(),
 207         }
 208
 209         match parse_sexp(input, loc) {
 210             Done(new_input, member) => {
 211                 loc = member.get_loc().1;
 212                 members.push(member);
 213                 input = new_input;
 214             }
 215             Error(err) =>
 216                 return Error(ParseError::List(
 217                     Box::new(err),
 218                     (0, 0).offset(loc)))
 219         }
 220     }
 221 }
 222
 223 pub fn parse_number(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 224     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Number);
 225
 226     match input.chars().next() {
 227         Some(c) if !c.is_digit(10) => {
 228             return Error(ParseError::Number(
 229                 Box::new(ParseError::Unexpected(c, start_loc)),
 230                 (0, c.len_utf8()).offset(start_loc)));
 231         }
 232         None => return Error(ParseError::Number(
 233             Box::new(ParseError::UnexpectedEof),
 234             (0, 0).offset(start_loc))),
 235         _ => (),
 236     }
 237
 238     let base = 10;
 239
 240     let mut end = 0;
 241     // Before the decimal point
 242     for (i, c) in input.char_indices() {
 243         if c == '.' {
 244             end = i + 1;
 245             break;
 246         }
 247
 248         if c.is_delimiter() {
 249             return Done(&input[i..],
 250                         Sexp::Int(input[..i].parse().expect("Already matched digits"),
 251                                   (0, i).offset(start_loc)));
 252         }
 253
 254         if !c.is_digit(base) {
 255             return Error(ParseError::Number(
 256                 Box::new(ParseError::Unexpected(c, start_loc + i)),
 257                 (i, i).offset(start_loc)));
 258         }
 259
 260         end = i + c.len_utf8();
 261     }
 262
 263     if input[end..].is_empty() {
 264         return Done(&input[end..],
 265                     Sexp::Int(input.parse().expect("Already matched digits"),
 266                               (0, end).offset(start_loc)));
 267     }
 268
 269     // After the decimal point
 270     for (i, c) in input[end..].char_indices() {
 271         if c.is_delimiter() {
 272             return Done(&input[i+end..],
 273                         Sexp::Float(input[..end+i].parse().expect("Already matched digits.digits"),
 274                                     (0, end+i).offset(start_loc)));
 275         }
 276
 277         if !c.is_digit(base) {
 278             return Error(ParseError::Number(
 279                 Box::new(ParseError::Unexpected(c, start_loc + i + end)),
 280                 (i+end, i+end).offset(start_loc)));
 281         }
 282     }
 283
 284     Done(&input[input.len()..],
 285          Sexp::Float(input.parse().expect("Already matched digits.digits"),
 286                      (0, input.len()).offset(start_loc)))
 287 }
 288
 289 pub fn parse_symbol(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 290     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Symbol);
 291
 292     match input.chars().next() {
 293         Some(c@'#') | Some(c@':') | Some(c@'0'...'9') =>
 294             return Error(ParseError::Symbol(
 295                 Box::new(ParseError::Unexpected(c, start_loc)),
 296                 (0, 0).offset(start_loc))),
 297         Some(c) if c.is_delimiter() =>
 298             return Error(ParseError::Symbol(
 299                 Box::new(ParseError::Unexpected(c, start_loc)),
 300                 (0, 0).offset(start_loc))),
 301         Some(_) => (),
 302         None => unreachable!(),
 303     }
 304
 305     for (i, c) in input.char_indices() {
 306         if c.is_delimiter() {
 307             return Done(&input[i..],
 308                         Sexp::Sym(input[..i].into(), (0, i).offset(start_loc)));
 309         }
 310     }
 311
 312     Done(&input[input.len()..],
 313          Sexp::Sym(input.into(), (0, input.len()).offset(start_loc)))
 314 }
 315
 316 pub fn parse_string(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 317     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::String);
 318
 319     match input.chars().next() {
 320         Some('"') => (),
 321         Some(c) =>
 322             return Error(ParseError::String(
 323                 Box::new(ParseError::Unexpected(c, start_loc)),
 324             (0, 0).offset(start_loc))),
 325         None => unreachable!(),
 326     }
 327
 328     for (i, c) in input[1..].char_indices() {
 329         if c == '"' {
 330             return Done(&input[2+i..],
 331             Sexp::Str(input[1..i+1].into(), (0, i+2).offset(start_loc)));
 332         }
 333     }
 334
 335     Error(ParseError::String(
 336         Box::new(ParseError::UnexpectedEof),
 337         (0, input.len()).offset(start_loc)))
 338 }
 339
 340 pub fn parse_character(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 341     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Char);
 342
 343     match input.chars().nth(0) {
 344         Some('#') => (),
 345         Some(c) =>
 346             return Error(ParseError::Char(
 347                 Box::new(ParseError::Unexpected(c, start_loc)),
 348                 (0, 0).offset(start_loc))),
 349         None =>
 350             return Error(ParseError::Char(
 351                 Box::new(ParseError::UnexpectedEof),
 352                 (0, 0).offset(start_loc))),
 353     }
 354
 355     match input.chars().nth(1) {
 356         Some('\\') => (),
 357         Some(c) =>
 358             return Error(ParseError::Char(
 359                 Box::new(ParseError::Unexpected(c, start_loc + 1)),
 360                 (1, 1).offset(start_loc))),
 361         None =>
 362             return Error(ParseError::Char(
 363                 Box::new(ParseError::UnexpectedEof),
 364                 (1, 1).offset(start_loc)))
 365     }
 366
 367     match input.chars().nth(2) {
 368         Some(c) =>
 369             Done(&input[3..], Sexp::Char(c, (0, 3).offset(start_loc))),
 370         None =>
 371             Error(ParseError::Char(
 372                 Box::new(ParseError::UnexpectedEof),
 373                 (2, 2).offset(start_loc)))
 374     }
 375 }
 376
 377 \f
 378 // Tests ///////////////////////////////////////////////////////////////////////
 379
 380 #[cfg(test)]
 381 mod test {
 382     use super::*;
 383     use super::ParseResult::*;
 384
 385     #[test]
 386     fn test_parse() {
 387         assert_eq!(parse("1 2 3"), (vec![
 388             Sexp::Int(1, (0, 1)), Sexp::Int(2, (2, 3)), Sexp::Int(3, (4, 5))
 389         ], None));
 390         assert_eq!(parse("1 2 )"), (vec![
 391             Sexp::Int(1, (0, 1)), Sexp::Int(2, (2, 3))
 392         ], Some(ParseError::Symbol(Box::new(ParseError::Unexpected(')', 4)), (4, 4)))));
 393     }
 394
 395     #[test]
 396     fn test_parse_one() {
 397         assert_eq!(parse_one("1 2"),
 398                    Ok((Sexp::Int(1, (0, 1)), " 2")));
 399     }
 400
 401     #[test]
 402     fn test_parse_sexp() {
 403         assert_eq!(parse_sexp(" 1", 0),
 404                    Done("", Sexp::Int(1, (1, 2))));
 405         assert_eq!(parse_sexp("2.2", 0),
 406                    Done("", Sexp::Float(2.2, (0, 3))));
 407         assert_eq!(parse_sexp(" a", 0),
 408                    Done("", Sexp::Sym("a".into(), (1, 2))));
 409         assert_eq!(parse_sexp("#\\c", 0),
 410                    Done("", Sexp::Char('c', (0, 3))));
 411         assert_eq!(parse_sexp(r#""hi""#, 0),
 412                    Done("", Sexp::Str("hi".into(), (0, 4))));
 413         assert_eq!(parse_sexp("()", 0),
 414                    Done("", Sexp::List(vec![], (0, 2))));
 415         assert_eq!(parse_sexp("( 1 2 3 )", 0),
 416                    Done("", Sexp::List(vec![
 417             Sexp::Int(1, (2, 3)),
 418             Sexp::Int(2, (4, 5)),
 419             Sexp::Int(3, (6, 7)),
 420         ], (0, 9))));
 421
 422         assert_eq!(parse_sexp("", 0),
 423                    Error(ParseError::Sexp(Box::new(ParseError::UnexpectedEof), (0, 0))));
 424     }
 425
 426     #[test]
 427     fn test_parse_list() {
 428         assert_eq!(parse_list("()", 0),
 429                    Done("", Sexp::List(vec![], (0, 2))));
 430         assert_eq!(parse_list("(1)", 0),
 431                    Done("", Sexp::List(vec![Sexp::Int(1, (1, 2))], (0, 3))));
 432         assert_eq!(parse_list("  ( 1    2  3 a )", 0), Done("", Sexp::List(vec![
 433             Sexp::Int(1, (4, 5)),
 434             Sexp::Int(2, (9, 10)),
 435             Sexp::Int(3, (12, 13)),
 436             Sexp::Sym("a".into(), (14, 15)),
 437         ], (2, 17))));
 438     }
 439
 440     #[test]
 441     fn test_parse_number() {
 442         assert_eq!(parse_number("1", 0),
 443                    Done("", Sexp::Int(1, (0, 1))));
 444         assert_eq!(parse_number(" 13", 0),
 445                    Done("", Sexp::Int(13, (1, 3))));
 446         assert_eq!(parse_number("1.2", 0),
 447                    Done("", Sexp::Float(1.2, (0, 3))));
 448         assert_eq!(parse_number("\u{3000}4.2", 0),
 449                    Done("", Sexp::Float(4.2, (0, 3).offset('\u{3000}'.len_utf8()))));
 450         assert_eq!(parse_number("       42 ", 0),
 451                    Done(" ", Sexp::Int(42, (2, 4))));
 452         assert_eq!(parse_number(" 4.2  ", 0),
 453                    Done("  ", Sexp::Float(4.2, (1, 4))));
 454         assert_eq!(parse_number("1()", 0),
 455                    Done("()", Sexp::Int(1, (0, 1))));
 456         assert_eq!(parse_number("3.6()", 0),
 457                    Done("()", Sexp::Float(3.6, (0, 3))));
 458
 459         assert_eq!(parse_number("", 0),
 460                    Error(ParseError::Number(Box::new(ParseError::UnexpectedEof), (0, 0))));
 461         assert_eq!(parse_number("123a", 0),
 462                    Error(ParseError::Number(Box::new(ParseError::Unexpected('a', 3)), (3, 3))));
 463         assert_eq!(parse_number("66.6+", 0),
 464                    Error(ParseError::Number(Box::new(ParseError::Unexpected('+', 4)), (4, 4))));
 465     }
 466
 467     #[test]
 468     fn test_parse_ident() {
 469         assert_eq!(parse_symbol("+", 0),
 470                    Done("", Sexp::Sym("+".into(), (0, 1))));
 471         assert_eq!(parse_symbol(" nil?", 0),
 472                    Done("", Sexp::Sym("nil?".into(), (1, 5))));
 473         assert_eq!(parse_symbol(" ->socket", 0),
 474                    Done("", Sexp::Sym("->socket".into(), (1, 9))));
 475         assert_eq!(parse_symbol("fib(", 0),
 476                    Done("(", Sexp::Sym("fib".into(), (0, 3))));
 477         assert_eq!(parse_symbol("foo2", 0),
 478                    Done("", Sexp::Sym("foo2".into(), (0, 4))));
 479
 480         // We reserve #foo for the implementation to do as it wishes
 481         assert_eq!(parse_symbol("#hi", 0),
 482                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected('#', 0)), (0, 0))));
 483         // We reserve :foo for keywords
 484         assert_eq!(parse_symbol(":hi", 0),
 485                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected(':', 0)), (0, 0))));
 486
 487         assert_eq!(parse_symbol("", 0),
 488                    Error(ParseError::Symbol(Box::new(ParseError::UnexpectedEof), (0, 0))));
 489         assert_eq!(parse_symbol("0", 0),
 490                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected('0', 0)), (0, 0))));
 491         assert_eq!(parse_symbol("()", 0),
 492                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected('(', 0)), (0, 0))));
 493     }
 494
 495     #[test]
 496     fn test_parse_string() {
 497         assert_eq!(parse_string(r#""""#, 0),
 498                    Done("", Sexp::Str("".into(), (0, 2))));
 499         assert_eq!(parse_string(r#""hello""#, 0),
 500                    Done("", Sexp::Str("hello".into(), (0, 7))));
 501         assert_eq!(parse_string(r#"  "this is a nice string
 502 with 0123 things in it""#, 0),
 503                    Done("", Sexp::Str("this is a nice string\nwith 0123 things in it".into(), (2, 48))));
 504
 505         assert_eq!(parse_string("", 0),
 506                    Error(ParseError::String(Box::new(ParseError::UnexpectedEof), (0, 0))));
 507         assert_eq!(parse_string(r#""hi"#, 0),
 508                    Error(ParseError::String(Box::new(ParseError::UnexpectedEof), (0, 3))));
 509     }
 510
 511     #[test]
 512     fn test_parse_char() {
 513         assert_eq!(parse_character(r#"#\""#, 0), Done("", Sexp::Char('"', (0, 3))));
 514         assert_eq!(parse_character(r#"#\ "#, 0), Done("", Sexp::Char(' ', (0, 3))));
 515         assert_eq!(parse_character(r#"  #\\"#, 0), Done("", Sexp::Char('\\', (2, 5))));
 516
 517         assert_eq!(parse_character("", 0),
 518                    Error(ParseError::Char(Box::new(ParseError::UnexpectedEof), (0, 0))));
 519         assert_eq!(parse_character("#", 0),
 520                    Error(ParseError::Char(Box::new(ParseError::UnexpectedEof), (1, 1))));
 521         assert_eq!(parse_character("#\\", 0),
 522                    Error(ParseError::Char(Box::new(ParseError::UnexpectedEof), (2, 2))));
 523         assert_eq!(parse_character("a", 0),
 524                    Error(ParseError::Char(Box::new(ParseError::Unexpected('a', 0)), (0, 0))));
 525     }
 526 }