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

   1 //! Functions to parse s-expressions and expression atoms.
   2 //!
   3 //! This module contains the core parsing machinery.
   4 //!
   5 //! * If you're interested in getting a parsed s-expression that you can use,
   6 //!   then looking at [`parse`] and [`parse_one`] are your best bet.
   7 //! * If you want to write your own parsers that contain s-expressions,
   8 //!   [`ParseResult`] and [`parse_expression`] will be the most useful to you.
   9 //!
  10 //! [`parse`]: fn.parse.html
  11 //! [`parse_one`]: fn.parse_one.html
  12 //! [`ParseResult`]: enum.ParseResult.html
  13 //! [`parse_expression`]: fn.parse_expression.html
  14
  15 use sexp::Sexp;
  16 use span::{Span, ByteSpan};
  17
  18 \f
  19 // Parsing Types ///////////////////////////////////////////////////////////////
  20
  21 /// Represents what to do next in partially completed parsing.
  22 ///
  23 /// `ParseResult` is returned from all intermediate parsers. If you just want to
  24 /// get back parsed s-expressions, you won't need to worry about this type since
  25 /// the top level parsers just return a `Result`.
  26 ///
  27 /// If the parser failed to produce a result, it will return `Error`, and if it
  28 /// succeeded we'll get the `Done` variant containing the value produced and the
  29 /// rest of the text to work on.
  30 #[derive(Debug, PartialEq, Eq, Clone)]
  31 pub enum ParseResult<'a, T, E> {
  32     /// The parser succeeded, this contains first the un-consumed portion of the
  33     /// input then the result produced by parsing.
  34     Done(&'a str, T),
  35     /// The parser failed, the `E` represents the reason for the failure.
  36     Error(E),
  37 }
  38
  39 /// Indicates how parsing failed.
  40 ///
  41 /// Most `ParseError` variants contain a `Box<ParseError>` that represents the
  42 /// cause of that error. Using this, `ParseError` variants can be chained to
  43 /// produce a more complete picture of what exactly went wrong during parsing.
  44 #[derive(Debug, PartialEq, Eq, Clone)]
  45 pub enum ParseError<Loc=ByteSpan> where Loc: Span {
  46     /// Parsing reached the end of input where not expecting to, usually this
  47     /// will be contained inside another `ParseError` like `String(box
  48     /// UnexpectedEof, ...)` which indicates that the closing quote was never
  49     /// found.
  50     UnexpectedEof,
  51     /// Some problem occurred while parsing a list, along with the cause of that
  52     /// error.
  53     List(Box<ParseError>, Loc),
  54     /// Some problem occurred while parsing an s-expression. This will only be
  55     /// generated if EOF is reached unexpectedly at the beginning of
  56     /// `parse_expression`, so it should probably be removed.
  57     Sexp(Box<ParseError>, Loc),
  58     /// Some problem occurred while parsing a character literal, along with the
  59     /// cause of the error.
  60     Char(Box<ParseError>, Loc),
  61     /// Some problem occurred while parsing a string literal, along with the
  62     /// cause of the error.
  63     String(Box<ParseError>, Loc),
  64     /// Some problem occurred while parsing a symbol, along with the cause of
  65     /// the error.
  66     Symbol(Box<ParseError>, Loc),
  67     /// Some problem occurred while parsing a number literal, along with the
  68     /// cause of the error.
  69     Number(Box<ParseError>, Loc),
  70     /// An unexpected character was found. This will usually be the root cause
  71     /// in some chain of `ParseError`s.
  72     Unexpected(char, Loc::Begin),
  73 }
  74 use self::ParseResult::*;
  75
  76 \f
  77 // Parsing Utilities ///////////////////////////////////////////////////////////
  78
  79 trait IsDelimeter {
  80     fn is_delimiter(&self) -> bool;
  81 }
  82
  83 impl IsDelimeter for char {
  84     fn is_delimiter(&self) -> bool {
  85         self.is_whitespace() || *self == ';'
  86             || *self == '(' || *self == ')'
  87             || *self == '[' || *self == ']'
  88             || *self == '{' || *self == '}'
  89             || *self == '"' || *self == '\''
  90             || *self == '`' || *self == ','
  91     }
  92 }
  93
  94 macro_rules! consume_whitespace {
  95     ($input:expr, $start_loc:expr, $ErrorFn:expr) => {
  96         if let Some(pos) = $input.find(|c: char| !c.is_whitespace()) {
  97             (&$input[pos..], $start_loc + pos)
  98         } else {
  99             return Error($ErrorFn(
 100                 Box::new(ParseError::UnexpectedEof),
 101                 ($input.len(), $input.len()).offset($start_loc)));
 102         }
 103     }
 104 }
 105
 106 \f
 107 // Top Level Parsers ///////////////////////////////////////////////////////////
 108
 109 /// Parse a sequence of s-expressions.
 110 ///
 111 /// This function returns `(Vec<Sexp>, Option<ParseError>)` so that it can
 112 /// return partial results, for when some component parses successfully and a
 113 /// later part fails.
 114 ///
 115 /// # Errors
 116 ///
 117 /// If the text contains an invalid s-expression (imbalanced parenthesis,
 118 /// quotes, invalid numbers like 123q, etc.) then the parser will stop and
 119 /// return an error. Every s-expression before that point that successfully
 120 /// parsed will still be returned.
 121 ///
 122 /// # Examples
 123 ///
 124 /// We can get useful partial results
 125 ///
 126 /// ```rust
 127 /// # use ess::parser::parse;
 128 /// let (exprs, err) = parse("1 2 3 ( 4");
 129 /// assert_eq!(exprs.len(), 3);
 130 /// assert!(err.is_some());
 131 /// ```
 132 pub fn parse(mut input: &str) -> (Vec<Sexp>, Option<ParseError>) {
 133     let mut start_loc = 0;
 134     let mut results = Vec::new();
 135     loop {
 136         match parse_expression(input, start_loc) {
 137             Done(rest, result) => {
 138                 input = rest;
 139                 start_loc = result.get_loc().1;
 140                 results.push(result);
 141                 if rest.trim() == "" {
 142                     return (results, None);
 143                 }
 144             }
 145             Error(err) => {
 146                 return (results, Some(err));
 147             }
 148         }
 149     }
 150 }
 151
 152 /// Parses a single s-expression, ignoring any trailing text.
 153 ///
 154 /// This function returns a pair of the parsed s-expression and the tail of the text.
 155 ///
 156 /// # Errors
 157 ///
 158 /// If the text begins with an invalid s-expression (imbalanced parenthesis,
 159 /// quotes, invalid numbers like 123q, etc.) then the parser will return an
 160 /// error. Any text after the first s-expression doesn't affect the parsing.
 161 ///
 162 /// # Examples
 163 ///
 164 /// ```rust
 165 /// # use ess::parser::parse_one;
 166 /// let (expr, rest) = parse_one("1 (").unwrap();
 167 /// assert_eq!(rest, " (");
 168 /// ```
 169 pub fn parse_one(input: &str) -> Result<(Sexp, &str), ParseError> {
 170     match parse_expression(input, 0) {
 171         Done(rest, result) => Ok((result, rest)),
 172         Error(err) => Err(err),
 173     }
 174 }
 175
 176 \f
 177 // Core Parsers ////////////////////////////////////////////////////////////////
 178
 179 pub fn parse_expression(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 180     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Sexp);
 181
 182     match input.chars().next() {
 183         Some('0'...'9') => parse_number(input, start_loc),
 184         Some('(') => parse_list(input, start_loc),
 185         Some('#') => parse_character(input, start_loc),
 186         Some('"') => parse_string(input, start_loc),
 187         Some(_) => parse_symbol(input, start_loc),
 188         None => unreachable!(),
 189     }
 190 }
 191
 192 pub fn parse_list(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 193     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::List);
 194
 195     match input.chars().nth(0) {
 196         Some('(') => (),
 197         Some(c) =>
 198             return Error(ParseError::List(
 199                 Box::new(ParseError::Unexpected(c, 0)),
 200                 (0, 0).offset(start_loc))),
 201         None => unreachable!(),
 202     }
 203
 204     let mut input = &input[1..];
 205     let mut loc = start_loc + 1;
 206     let mut members = Vec::new();
 207     loop {
 208         {
 209             let (new_input, new_loc) = consume_whitespace!(input, loc, ParseError::List);
 210             input = new_input;
 211             loc = new_loc;
 212         }
 213
 214         match input.chars().nth(0) {
 215             Some(')') =>
 216                 return Done(&input[1..],
 217                             Sexp::List(members, (start_loc, loc+1))),
 218             Some(_) => (),
 219             None => unreachable!(),
 220         }
 221
 222         match parse_expression(input, loc) {
 223             Done(new_input, member) => {
 224                 loc = member.get_loc().1;
 225                 members.push(member);
 226                 input = new_input;
 227             }
 228             Error(err) =>
 229                 return Error(ParseError::List(
 230                     Box::new(err),
 231                     (0, 0).offset(loc)))
 232         }
 233     }
 234 }
 235
 236 pub fn parse_number(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 237     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Number);
 238
 239     match input.chars().next() {
 240         Some(c) if !c.is_digit(10) => {
 241             return Error(ParseError::Number(
 242                 Box::new(ParseError::Unexpected(c, start_loc)),
 243                 (0, c.len_utf8()).offset(start_loc)));
 244         }
 245         None => return Error(ParseError::Number(
 246             Box::new(ParseError::UnexpectedEof),
 247             (0, 0).offset(start_loc))),
 248         _ => (),
 249     }
 250
 251     let base = 10;
 252
 253     let mut end = 0;
 254     // Before the decimal point
 255     for (i, c) in input.char_indices() {
 256         if c == '.' {
 257             end = i + 1;
 258             break;
 259         }
 260
 261         if c.is_delimiter() {
 262             return Done(&input[i..],
 263                         Sexp::Int(input[..i].parse().expect("Already matched digits"),
 264                                   (0, i).offset(start_loc)));
 265         }
 266
 267         if !c.is_digit(base) {
 268             return Error(ParseError::Number(
 269                 Box::new(ParseError::Unexpected(c, start_loc + i)),
 270                 (i, i).offset(start_loc)));
 271         }
 272
 273         end = i + c.len_utf8();
 274     }
 275
 276     if input[end..].is_empty() {
 277         return Done(&input[end..],
 278                     Sexp::Int(input.parse().expect("Already matched digits"),
 279                               (0, end).offset(start_loc)));
 280     }
 281
 282     // After the decimal point
 283     for (i, c) in input[end..].char_indices() {
 284         if c.is_delimiter() {
 285             return Done(&input[i+end..],
 286                         Sexp::Float(input[..end+i].parse().expect("Already matched digits.digits"),
 287                                     (0, end+i).offset(start_loc)));
 288         }
 289
 290         if !c.is_digit(base) {
 291             return Error(ParseError::Number(
 292                 Box::new(ParseError::Unexpected(c, start_loc + i + end)),
 293                 (i+end, i+end).offset(start_loc)));
 294         }
 295     }
 296
 297     Done(&input[input.len()..],
 298          Sexp::Float(input.parse().expect("Already matched digits.digits"),
 299                      (0, input.len()).offset(start_loc)))
 300 }
 301
 302 pub fn parse_symbol(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 303     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Symbol);
 304
 305     match input.chars().next() {
 306         Some(c@'#') | Some(c@':') | Some(c@'0'...'9') =>
 307             return Error(ParseError::Symbol(
 308                 Box::new(ParseError::Unexpected(c, start_loc)),
 309                 (0, 0).offset(start_loc))),
 310         Some(c) if c.is_delimiter() =>
 311             return Error(ParseError::Symbol(
 312                 Box::new(ParseError::Unexpected(c, start_loc)),
 313                 (0, 0).offset(start_loc))),
 314         Some(_) => (),
 315         None => unreachable!(),
 316     }
 317
 318     for (i, c) in input.char_indices() {
 319         if c.is_delimiter() {
 320             return Done(&input[i..],
 321                         Sexp::Sym(input[..i].into(), (0, i).offset(start_loc)));
 322         }
 323     }
 324
 325     Done(&input[input.len()..],
 326          Sexp::Sym(input.into(), (0, input.len()).offset(start_loc)))
 327 }
 328
 329 pub fn parse_string(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 330     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::String);
 331
 332     match input.chars().next() {
 333         Some('"') => (),
 334         Some(c) =>
 335             return Error(ParseError::String(
 336                 Box::new(ParseError::Unexpected(c, start_loc)),
 337             (0, 0).offset(start_loc))),
 338         None => unreachable!(),
 339     }
 340
 341     for (i, c) in input[1..].char_indices() {
 342         if c == '"' {
 343             return Done(&input[2+i..],
 344             Sexp::Str(input[1..i+1].into(), (0, i+2).offset(start_loc)));
 345         }
 346     }
 347
 348     Error(ParseError::String(
 349         Box::new(ParseError::UnexpectedEof),
 350         (0, input.len()).offset(start_loc)))
 351 }
 352
 353 pub fn parse_character(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 354     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Char);
 355
 356     match input.chars().nth(0) {
 357         Some('#') => (),
 358         Some(c) =>
 359             return Error(ParseError::Char(
 360                 Box::new(ParseError::Unexpected(c, start_loc)),
 361                 (0, 0).offset(start_loc))),
 362         None =>
 363             return Error(ParseError::Char(
 364                 Box::new(ParseError::UnexpectedEof),
 365                 (0, 0).offset(start_loc))),
 366     }
 367
 368     match input.chars().nth(1) {
 369         Some('\\') => (),
 370         Some(c) =>
 371             return Error(ParseError::Char(
 372                 Box::new(ParseError::Unexpected(c, start_loc + 1)),
 373                 (1, 1).offset(start_loc))),
 374         None =>
 375             return Error(ParseError::Char(
 376                 Box::new(ParseError::UnexpectedEof),
 377                 (1, 1).offset(start_loc)))
 378     }
 379
 380     match input.chars().nth(2) {
 381         Some(c) =>
 382             Done(&input[3..], Sexp::Char(c, (0, 3).offset(start_loc))),
 383         None =>
 384             Error(ParseError::Char(
 385                 Box::new(ParseError::UnexpectedEof),
 386                 (2, 2).offset(start_loc)))
 387     }
 388 }
 389
 390 \f
 391 // Tests ///////////////////////////////////////////////////////////////////////
 392
 393 #[cfg(test)]
 394 mod test {
 395     use sexp::Sexp;
 396     use span::Span;
 397     use parser::*;
 398     use parser::ParseResult::*;
 399
 400     #[test]
 401     fn test_parse() {
 402         assert_eq!(parse("1 2 3"), (vec![
 403             Sexp::Int(1, (0, 1)), Sexp::Int(2, (2, 3)), Sexp::Int(3, (4, 5))
 404         ], None));
 405         assert_eq!(parse("1 2 )"), (vec![
 406             Sexp::Int(1, (0, 1)), Sexp::Int(2, (2, 3))
 407         ], Some(ParseError::Symbol(Box::new(ParseError::Unexpected(')', 4)), (4, 4)))));
 408     }
 409
 410     #[test]
 411     fn test_parse_one() {
 412         assert_eq!(parse_one("1 2"),
 413                    Ok((Sexp::Int(1, (0, 1)), " 2")));
 414     }
 415
 416     #[test]
 417     fn test_parse_expression() {
 418         assert_eq!(parse_expression("   1", 0),
 419                    Done("", Sexp::Int(1, (1, 2))));
 420         assert_eq!(parse_expression("2.2", 0),
 421                    Done("", Sexp::Float(2.2, (0, 3))));
 422         assert_eq!(parse_expression(" a", 0),
 423                    Done("", Sexp::Sym("a".into(), (1, 2))));
 424         assert_eq!(parse_expression("#\\c", 0),
 425                    Done("", Sexp::Char('c', (0, 3))));
 426         assert_eq!(parse_expression(r#""hi""#, 0),
 427                    Done("", Sexp::Str("hi".into(), (0, 4))));
 428         assert_eq!(parse_expression("()", 0),
 429                    Done("", Sexp::List(vec![], (0, 2))));
 430         assert_eq!(parse_expression("( 1 2 3 )", 0),
 431                    Done("", Sexp::List(vec![
 432             Sexp::Int(1, (2, 3)),
 433             Sexp::Int(2, (4, 5)),
 434             Sexp::Int(3, (6, 7)),
 435         ], (0, 9))));
 436
 437         assert_eq!(parse_expression("", 0),
 438                    Error(ParseError::Sexp(Box::new(ParseError::UnexpectedEof), (0, 0))));
 439     }
 440
 441     #[test]
 442     fn test_parse_list() {
 443         assert_eq!(parse_list("()", 0),
 444                    Done("", Sexp::List(vec![], (0, 2))));
 445         assert_eq!(parse_list("(1)", 0),
 446                    Done("", Sexp::List(vec![Sexp::Int(1, (1, 2))], (0, 3))));
 447         assert_eq!(parse_list("  ( 1    2  3 a )", 0), Done("", Sexp::List(vec![
 448             Sexp::Int(1, (4, 5)),
 449             Sexp::Int(2, (9, 10)),
 450             Sexp::Int(3, (12, 13)),
 451             Sexp::Sym("a".into(), (14, 15)),
 452         ], (2, 17))));
 453     }
 454
 455     #[test]
 456     fn test_parse_number() {
 457         assert_eq!(parse_number("1", 0),
 458                    Done("", Sexp::Int(1, (0, 1))));
 459         assert_eq!(parse_number(" 13", 0),
 460                    Done("", Sexp::Int(13, (1, 3))));
 461         assert_eq!(parse_number("1.2", 0),
 462                    Done("", Sexp::Float(1.2, (0, 3))));
 463         assert_eq!(parse_number("\u{3000}4.2", 0),
 464                    Done("", Sexp::Float(4.2, (0, 3).offset('\u{3000}'.len_utf8()))));
 465         assert_eq!(parse_number("       42 ", 0),
 466                    Done(" ", Sexp::Int(42, (2, 4))));
 467         assert_eq!(parse_number(" 4.2  ", 0),
 468                    Done("  ", Sexp::Float(4.2, (1, 4))));
 469         assert_eq!(parse_number("1()", 0),
 470                    Done("()", Sexp::Int(1, (0, 1))));
 471         assert_eq!(parse_number("3.6()", 0),
 472                    Done("()", Sexp::Float(3.6, (0, 3))));
 473
 474         assert_eq!(parse_number("", 0),
 475                    Error(ParseError::Number(Box::new(ParseError::UnexpectedEof), (0, 0))));
 476         assert_eq!(parse_number("123a", 0),
 477                    Error(ParseError::Number(Box::new(ParseError::Unexpected('a', 3)), (3, 3))));
 478         assert_eq!(parse_number("66.6+", 0),
 479                    Error(ParseError::Number(Box::new(ParseError::Unexpected('+', 4)), (4, 4))));
 480     }
 481
 482     #[test]
 483     fn test_parse_ident() {
 484         assert_eq!(parse_symbol("+", 0),
 485                    Done("", Sexp::Sym("+".into(), (0, 1))));
 486         assert_eq!(parse_symbol(" nil?", 0),
 487                    Done("", Sexp::Sym("nil?".into(), (1, 5))));
 488         assert_eq!(parse_symbol(" ->socket", 0),
 489                    Done("", Sexp::Sym("->socket".into(), (1, 9))));
 490         assert_eq!(parse_symbol("fib(", 0),
 491                    Done("(", Sexp::Sym("fib".into(), (0, 3))));
 492         assert_eq!(parse_symbol("foo2", 0),
 493                    Done("", Sexp::Sym("foo2".into(), (0, 4))));
 494
 495         // We reserve #foo for the implementation to do as it wishes
 496         assert_eq!(parse_symbol("#hi", 0),
 497                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected('#', 0)), (0, 0))));
 498         // We reserve :foo for keywords
 499         assert_eq!(parse_symbol(":hi", 0),
 500                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected(':', 0)), (0, 0))));
 501
 502         assert_eq!(parse_symbol("", 0),
 503                    Error(ParseError::Symbol(Box::new(ParseError::UnexpectedEof), (0, 0))));
 504         assert_eq!(parse_symbol("0", 0),
 505                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected('0', 0)), (0, 0))));
 506         assert_eq!(parse_symbol("()", 0),
 507                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected('(', 0)), (0, 0))));
 508     }
 509
 510     #[test]
 511     fn test_parse_string() {
 512         assert_eq!(parse_string(r#""""#, 0),
 513                    Done("", Sexp::Str("".into(), (0, 2))));
 514         assert_eq!(parse_string(r#""hello""#, 0),
 515                    Done("", Sexp::Str("hello".into(), (0, 7))));
 516         assert_eq!(parse_string(r#"  "this is a nice string
 517 with 0123 things in it""#, 0),
 518                    Done("", Sexp::Str("this is a nice string\nwith 0123 things in it".into(), (2, 48))));
 519
 520         assert_eq!(parse_string("", 0),
 521                    Error(ParseError::String(Box::new(ParseError::UnexpectedEof), (0, 0))));
 522         assert_eq!(parse_string(r#""hi"#, 0),
 523                    Error(ParseError::String(Box::new(ParseError::UnexpectedEof), (0, 3))));
 524     }
 525
 526     #[test]
 527     fn test_parse_char() {
 528         assert_eq!(parse_character(r#"#\""#, 0), Done("", Sexp::Char('"', (0, 3))));
 529         assert_eq!(parse_character(r#"#\ "#, 0), Done("", Sexp::Char(' ', (0, 3))));
 530         assert_eq!(parse_character(r#"  #\\"#, 0), Done("", Sexp::Char('\\', (2, 5))));
 531
 532         assert_eq!(parse_character("", 0),
 533                    Error(ParseError::Char(Box::new(ParseError::UnexpectedEof), (0, 0))));
 534         assert_eq!(parse_character("#", 0),
 535                    Error(ParseError::Char(Box::new(ParseError::UnexpectedEof), (1, 1))));
 536         assert_eq!(parse_character("#\\", 0),
 537                    Error(ParseError::Char(Box::new(ParseError::UnexpectedEof), (2, 2))));
 538         assert_eq!(parse_character("a", 0),
 539                    Error(ParseError::Char(Box::new(ParseError::Unexpected('a', 0)), (0, 0))));
 540     }
 541 }