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 // TODO: All of these parsers deserve docs, but since they're somewhat internal
 180 // parsers, it's less critical than the rest of the API.
 181
 182 #[allow(missing_docs)]
 183 pub fn parse_expression(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 184     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Sexp);
 185
 186     match input.chars().next() {
 187         Some('0'...'9') => parse_number(input, start_loc),
 188         Some('(') => parse_list(input, start_loc),
 189         Some('#') => parse_character(input, start_loc),
 190         Some('"') => parse_string(input, start_loc),
 191         Some(_) => parse_symbol(input, start_loc),
 192         None => unreachable!(),
 193     }
 194 }
 195
 196 #[allow(missing_docs)]
 197 pub fn parse_list(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 198     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::List);
 199
 200     match input.chars().nth(0) {
 201         Some('(') => (),
 202         Some(c) =>
 203             return Error(ParseError::List(
 204                 Box::new(ParseError::Unexpected(c, 0)),
 205                 (0, 0).offset(start_loc))),
 206         None => unreachable!(),
 207     }
 208
 209     let mut input = &input[1..];
 210     let mut loc = start_loc + 1;
 211     let mut members = Vec::new();
 212     loop {
 213         {
 214             let (new_input, new_loc) = consume_whitespace!(input, loc, ParseError::List);
 215             input = new_input;
 216             loc = new_loc;
 217         }
 218
 219         match input.chars().nth(0) {
 220             Some(')') =>
 221                 return Done(&input[1..],
 222                             Sexp::List(members, (start_loc, loc+1))),
 223             Some(_) => (),
 224             None => unreachable!(),
 225         }
 226
 227         match parse_expression(input, loc) {
 228             Done(new_input, member) => {
 229                 loc = member.get_loc().1;
 230                 members.push(member);
 231                 input = new_input;
 232             }
 233             Error(err) =>
 234                 return Error(ParseError::List(
 235                     Box::new(err),
 236                     (0, 0).offset(loc)))
 237         }
 238     }
 239 }
 240
 241 #[allow(missing_docs)]
 242 pub fn parse_number(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 243     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Number);
 244
 245     match input.chars().next() {
 246         Some(c) if !c.is_digit(10) => {
 247             return Error(ParseError::Number(
 248                 Box::new(ParseError::Unexpected(c, start_loc)),
 249                 (0, c.len_utf8()).offset(start_loc)));
 250         }
 251         None => return Error(ParseError::Number(
 252             Box::new(ParseError::UnexpectedEof),
 253             (0, 0).offset(start_loc))),
 254         _ => (),
 255     }
 256
 257     let base = 10;
 258
 259     let mut end = 0;
 260     // Before the decimal point
 261     for (i, c) in input.char_indices() {
 262         if c == '.' {
 263             end = i + 1;
 264             break;
 265         }
 266
 267         if c.is_delimiter() {
 268             return Done(&input[i..],
 269                         Sexp::Int(input[..i].parse().expect("Already matched digits"),
 270                                   (0, i).offset(start_loc)));
 271         }
 272
 273         if !c.is_digit(base) {
 274             return Error(ParseError::Number(
 275                 Box::new(ParseError::Unexpected(c, start_loc + i)),
 276                 (i, i).offset(start_loc)));
 277         }
 278
 279         end = i + c.len_utf8();
 280     }
 281
 282     if input[end..].is_empty() {
 283         return Done(&input[end..],
 284                     Sexp::Int(input.parse().expect("Already matched digits"),
 285                               (0, end).offset(start_loc)));
 286     }
 287
 288     // After the decimal point
 289     for (i, c) in input[end..].char_indices() {
 290         if c.is_delimiter() {
 291             return Done(&input[i+end..],
 292                         Sexp::Float(input[..end+i].parse().expect("Already matched digits.digits"),
 293                                     (0, end+i).offset(start_loc)));
 294         }
 295
 296         if !c.is_digit(base) {
 297             return Error(ParseError::Number(
 298                 Box::new(ParseError::Unexpected(c, start_loc + i + end)),
 299                 (i+end, i+end).offset(start_loc)));
 300         }
 301     }
 302
 303     Done(&input[input.len()..],
 304          Sexp::Float(input.parse().expect("Already matched digits.digits"),
 305                      (0, input.len()).offset(start_loc)))
 306 }
 307
 308 #[allow(missing_docs)]
 309 pub fn parse_symbol(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 310     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Symbol);
 311
 312     match input.chars().next() {
 313         Some(c@'#') | Some(c@':') | Some(c@'0'...'9') =>
 314             return Error(ParseError::Symbol(
 315                 Box::new(ParseError::Unexpected(c, start_loc)),
 316                 (0, 0).offset(start_loc))),
 317         Some(c) if c.is_delimiter() =>
 318             return Error(ParseError::Symbol(
 319                 Box::new(ParseError::Unexpected(c, start_loc)),
 320                 (0, 0).offset(start_loc))),
 321         Some(_) => (),
 322         None => unreachable!(),
 323     }
 324
 325     for (i, c) in input.char_indices() {
 326         if c.is_delimiter() {
 327             return Done(&input[i..],
 328                         Sexp::Sym(input[..i].into(), (0, i).offset(start_loc)));
 329         }
 330     }
 331
 332     Done(&input[input.len()..],
 333          Sexp::Sym(input.into(), (0, input.len()).offset(start_loc)))
 334 }
 335
 336 #[allow(missing_docs)]
 337 pub fn parse_string(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 338     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::String);
 339
 340     match input.chars().next() {
 341         Some('"') => (),
 342         Some(c) =>
 343             return Error(ParseError::String(
 344                 Box::new(ParseError::Unexpected(c, start_loc)),
 345             (0, 0).offset(start_loc))),
 346         None => unreachable!(),
 347     }
 348
 349     for (i, c) in input[1..].char_indices() {
 350         if c == '"' {
 351             return Done(&input[2+i..],
 352             Sexp::Str(input[1..i+1].into(), (0, i+2).offset(start_loc)));
 353         }
 354     }
 355
 356     Error(ParseError::String(
 357         Box::new(ParseError::UnexpectedEof),
 358         (0, input.len()).offset(start_loc)))
 359 }
 360
 361 #[allow(missing_docs)]
 362 pub fn parse_character(input: &str, start_loc: usize) -> ParseResult<Sexp, ParseError> {
 363     let (input, start_loc) = consume_whitespace!(input, start_loc, ParseError::Char);
 364
 365     match input.chars().nth(0) {
 366         Some('#') => (),
 367         Some(c) =>
 368             return Error(ParseError::Char(
 369                 Box::new(ParseError::Unexpected(c, start_loc)),
 370                 (0, 0).offset(start_loc))),
 371         None =>
 372             return Error(ParseError::Char(
 373                 Box::new(ParseError::UnexpectedEof),
 374                 (0, 0).offset(start_loc))),
 375     }
 376
 377     match input.chars().nth(1) {
 378         Some('\\') => (),
 379         Some(c) =>
 380             return Error(ParseError::Char(
 381                 Box::new(ParseError::Unexpected(c, start_loc + 1)),
 382                 (1, 1).offset(start_loc))),
 383         None =>
 384             return Error(ParseError::Char(
 385                 Box::new(ParseError::UnexpectedEof),
 386                 (1, 1).offset(start_loc)))
 387     }
 388
 389     match input.chars().nth(2) {
 390         Some(c) =>
 391             Done(&input[3..], Sexp::Char(c, (0, 3).offset(start_loc))),
 392         None =>
 393             Error(ParseError::Char(
 394                 Box::new(ParseError::UnexpectedEof),
 395                 (2, 2).offset(start_loc)))
 396     }
 397 }
 398
 399 \f
 400 // Tests ///////////////////////////////////////////////////////////////////////
 401
 402 #[cfg(test)]
 403 mod test {
 404     use sexp::Sexp;
 405     use span::Span;
 406     use parser::*;
 407     use parser::ParseResult::*;
 408
 409     #[test]
 410     fn test_parse() {
 411         assert_eq!(parse("1 2 3"), (vec![
 412             Sexp::Int(1, (0, 1)), Sexp::Int(2, (2, 3)), Sexp::Int(3, (4, 5))
 413         ], None));
 414         assert_eq!(parse("1 2 )"), (vec![
 415             Sexp::Int(1, (0, 1)), Sexp::Int(2, (2, 3))
 416         ], Some(ParseError::Symbol(Box::new(ParseError::Unexpected(')', 4)), (4, 4)))));
 417     }
 418
 419     #[test]
 420     fn test_parse_one() {
 421         assert_eq!(parse_one("1 2"),
 422                    Ok((Sexp::Int(1, (0, 1)), " 2")));
 423     }
 424
 425     #[test]
 426     fn test_parse_expression() {
 427         assert_eq!(parse_expression("   1", 0),
 428                    Done("", Sexp::Int(1, (1, 2))));
 429         assert_eq!(parse_expression("2.2", 0),
 430                    Done("", Sexp::Float(2.2, (0, 3))));
 431         assert_eq!(parse_expression(" a", 0),
 432                    Done("", Sexp::Sym("a".into(), (1, 2))));
 433         assert_eq!(parse_expression("#\\c", 0),
 434                    Done("", Sexp::Char('c', (0, 3))));
 435         assert_eq!(parse_expression(r#""hi""#, 0),
 436                    Done("", Sexp::Str("hi".into(), (0, 4))));
 437         assert_eq!(parse_expression("()", 0),
 438                    Done("", Sexp::List(vec![], (0, 2))));
 439         assert_eq!(parse_expression("( 1 2 3 )", 0),
 440                    Done("", Sexp::List(vec![
 441             Sexp::Int(1, (2, 3)),
 442             Sexp::Int(2, (4, 5)),
 443             Sexp::Int(3, (6, 7)),
 444         ], (0, 9))));
 445
 446         assert_eq!(parse_expression("", 0),
 447                    Error(ParseError::Sexp(Box::new(ParseError::UnexpectedEof), (0, 0))));
 448     }
 449
 450     #[test]
 451     fn test_parse_list() {
 452         assert_eq!(parse_list("()", 0),
 453                    Done("", Sexp::List(vec![], (0, 2))));
 454         assert_eq!(parse_list("(1)", 0),
 455                    Done("", Sexp::List(vec![Sexp::Int(1, (1, 2))], (0, 3))));
 456         assert_eq!(parse_list("  ( 1    2  3 a )", 0), Done("", Sexp::List(vec![
 457             Sexp::Int(1, (4, 5)),
 458             Sexp::Int(2, (9, 10)),
 459             Sexp::Int(3, (12, 13)),
 460             Sexp::Sym("a".into(), (14, 15)),
 461         ], (2, 17))));
 462     }
 463
 464     #[test]
 465     fn test_parse_number() {
 466         assert_eq!(parse_number("1", 0),
 467                    Done("", Sexp::Int(1, (0, 1))));
 468         assert_eq!(parse_number(" 13", 0),
 469                    Done("", Sexp::Int(13, (1, 3))));
 470         assert_eq!(parse_number("1.2", 0),
 471                    Done("", Sexp::Float(1.2, (0, 3))));
 472         assert_eq!(parse_number("\u{3000}4.2", 0),
 473                    Done("", Sexp::Float(4.2, (0, 3).offset('\u{3000}'.len_utf8()))));
 474         assert_eq!(parse_number("       42 ", 0),
 475                    Done(" ", Sexp::Int(42, (2, 4))));
 476         assert_eq!(parse_number(" 4.2  ", 0),
 477                    Done("  ", Sexp::Float(4.2, (1, 4))));
 478         assert_eq!(parse_number("1()", 0),
 479                    Done("()", Sexp::Int(1, (0, 1))));
 480         assert_eq!(parse_number("3.6()", 0),
 481                    Done("()", Sexp::Float(3.6, (0, 3))));
 482
 483         assert_eq!(parse_number("", 0),
 484                    Error(ParseError::Number(Box::new(ParseError::UnexpectedEof), (0, 0))));
 485         assert_eq!(parse_number("123a", 0),
 486                    Error(ParseError::Number(Box::new(ParseError::Unexpected('a', 3)), (3, 3))));
 487         assert_eq!(parse_number("66.6+", 0),
 488                    Error(ParseError::Number(Box::new(ParseError::Unexpected('+', 4)), (4, 4))));
 489     }
 490
 491     #[test]
 492     fn test_parse_ident() {
 493         assert_eq!(parse_symbol("+", 0),
 494                    Done("", Sexp::Sym("+".into(), (0, 1))));
 495         assert_eq!(parse_symbol(" nil?", 0),
 496                    Done("", Sexp::Sym("nil?".into(), (1, 5))));
 497         assert_eq!(parse_symbol(" ->socket", 0),
 498                    Done("", Sexp::Sym("->socket".into(), (1, 9))));
 499         assert_eq!(parse_symbol("fib(", 0),
 500                    Done("(", Sexp::Sym("fib".into(), (0, 3))));
 501         assert_eq!(parse_symbol("foo2", 0),
 502                    Done("", Sexp::Sym("foo2".into(), (0, 4))));
 503
 504         // We reserve #foo for the implementation to do as it wishes
 505         assert_eq!(parse_symbol("#hi", 0),
 506                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected('#', 0)), (0, 0))));
 507         // We reserve :foo for keywords
 508         assert_eq!(parse_symbol(":hi", 0),
 509                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected(':', 0)), (0, 0))));
 510
 511         assert_eq!(parse_symbol("", 0),
 512                    Error(ParseError::Symbol(Box::new(ParseError::UnexpectedEof), (0, 0))));
 513         assert_eq!(parse_symbol("0", 0),
 514                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected('0', 0)), (0, 0))));
 515         assert_eq!(parse_symbol("()", 0),
 516                    Error(ParseError::Symbol(Box::new(ParseError::Unexpected('(', 0)), (0, 0))));
 517     }
 518
 519     #[test]
 520     fn test_parse_string() {
 521         assert_eq!(parse_string(r#""""#, 0),
 522                    Done("", Sexp::Str("".into(), (0, 2))));
 523         assert_eq!(parse_string(r#""hello""#, 0),
 524                    Done("", Sexp::Str("hello".into(), (0, 7))));
 525         assert_eq!(parse_string(r#"  "this is a nice string
 526 with 0123 things in it""#, 0),
 527                    Done("", Sexp::Str("this is a nice string\nwith 0123 things in it".into(), (2, 48))));
 528
 529         assert_eq!(parse_string("", 0),
 530                    Error(ParseError::String(Box::new(ParseError::UnexpectedEof), (0, 0))));
 531         assert_eq!(parse_string(r#""hi"#, 0),
 532                    Error(ParseError::String(Box::new(ParseError::UnexpectedEof), (0, 3))));
 533     }
 534
 535     #[test]
 536     fn test_parse_char() {
 537         assert_eq!(parse_character(r#"#\""#, 0), Done("", Sexp::Char('"', (0, 3))));
 538         assert_eq!(parse_character(r#"#\ "#, 0), Done("", Sexp::Char(' ', (0, 3))));
 539         assert_eq!(parse_character(r#"  #\\"#, 0), Done("", Sexp::Char('\\', (2, 5))));
 540
 541         assert_eq!(parse_character("", 0),
 542                    Error(ParseError::Char(Box::new(ParseError::UnexpectedEof), (0, 0))));
 543         assert_eq!(parse_character("#", 0),
 544                    Error(ParseError::Char(Box::new(ParseError::UnexpectedEof), (1, 1))));
 545         assert_eq!(parse_character("#\\", 0),
 546                    Error(ParseError::Char(Box::new(ParseError::UnexpectedEof), (2, 2))));
 547         assert_eq!(parse_character("a", 0),
 548                    Error(ParseError::Char(Box::new(ParseError::Unexpected('a', 0)), (0, 0))));
 549     }
 550 }