src/libsyntax/parse/mod.rs

   1 // Copyright 2012-2014 The Rust Project Developers. See the COPYRIGHT
   2 // file at the top-level directory of this distribution and at
   3 // http://rust-lang.org/COPYRIGHT.
   4 //
   5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
   6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
   7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
   8 // option. This file may not be copied, modified, or distributed
   9 // except according to those terms.
  10
  11 //! The main parser interface
  12
  13 use ast;
  14 use codemap::{self, Span, CodeMap, FileMap};
  15 use errors::{Handler, ColorConfig, DiagnosticBuilder};
  16 use parse::parser::Parser;
  17 use parse::token::InternedString;
  18 use ptr::P;
  19 use str::char_at;
  20
  21 use std::cell::RefCell;
  22 use std::io::Read;
  23 use std::iter;
  24 use std::path::{Path, PathBuf};
  25 use std::rc::Rc;
  26 use std::str;
  27
  28 pub type PResult<'a, T> = Result<T, DiagnosticBuilder<'a>>;
  29
  30 #[macro_use]
  31 pub mod parser;
  32
  33 pub mod lexer;
  34 pub mod token;
  35 pub mod attr;
  36
  37 pub mod common;
  38 pub mod classify;
  39 pub mod obsolete;
  40
  41 /// Info about a parsing session.
  42 pub struct ParseSess {
  43     pub span_diagnostic: Handler, // better be the same as the one in the reader!
  44     /// Used to determine and report recursive mod inclusions
  45     included_mod_stack: RefCell<Vec<PathBuf>>,
  46     code_map: Rc<CodeMap>,
  47 }
  48
  49 impl ParseSess {
  50     pub fn new() -> ParseSess {
  51         let cm = Rc::new(CodeMap::new());
  52         let handler = Handler::with_tty_emitter(ColorConfig::Auto, None, true, false, cm.clone());
  53         ParseSess::with_span_handler(handler, cm)
  54     }
  55
  56     pub fn with_span_handler(handler: Handler, code_map: Rc<CodeMap>) -> ParseSess {
  57         ParseSess {
  58             span_diagnostic: handler,
  59             included_mod_stack: RefCell::new(vec![]),
  60             code_map: code_map
  61         }
  62     }
  63
  64     pub fn codemap(&self) -> &CodeMap {
  65         &self.code_map
  66     }
  67 }
  68
  69 // a bunch of utility functions of the form parse_<thing>_from_<source>
  70 // where <thing> includes crate, expr, item, stmt, tts, and one that
  71 // uses a HOF to parse anything, and <source> includes file and
  72 // source_str.
  73
  74 pub fn parse_crate_from_file<'a>(input: &Path,
  75                                  cfg: ast::CrateConfig,
  76                                  sess: &'a ParseSess)
  77                                  -> PResult<'a, ast::Crate> {
  78     let mut parser = new_parser_from_file(sess, cfg, input);
  79     parser.parse_crate_mod()
  80 }
  81
  82 pub fn parse_crate_attrs_from_file<'a>(input: &Path,
  83                                        cfg: ast::CrateConfig,
  84                                        sess: &'a ParseSess)
  85                                        -> PResult<'a, Vec<ast::Attribute>> {
  86     let mut parser = new_parser_from_file(sess, cfg, input);
  87     parser.parse_inner_attributes()
  88 }
  89
  90 pub fn parse_crate_from_source_str<'a>(name: String,
  91                                        source: String,
  92                                        cfg: ast::CrateConfig,
  93                                        sess: &'a ParseSess)
  94                                        -> PResult<'a, ast::Crate> {
  95     let mut p = new_parser_from_source_str(sess,
  96                                            cfg,
  97                                            name,
  98                                            source);
  99     p.parse_crate_mod()
 100 }
 101
 102 pub fn parse_crate_attrs_from_source_str<'a>(name: String,
 103                                              source: String,
 104                                              cfg: ast::CrateConfig,
 105                                              sess: &'a ParseSess)
 106                                              -> PResult<'a, Vec<ast::Attribute>> {
 107     let mut p = new_parser_from_source_str(sess,
 108                                            cfg,
 109                                            name,
 110                                            source);
 111     p.parse_inner_attributes()
 112 }
 113
 114 pub fn parse_expr_from_source_str<'a>(name: String,
 115                                       source: String,
 116                                       cfg: ast::CrateConfig,
 117                                       sess: &'a ParseSess)
 118                                       -> PResult<'a, P<ast::Expr>> {
 119     let mut p = new_parser_from_source_str(sess, cfg, name, source);
 120     p.parse_expr()
 121 }
 122
 123 /// Parses an item.
 124 ///
 125 /// Returns `Ok(Some(item))` when successful, `Ok(None)` when no item was found, and`Err`
 126 /// when a syntax error occurred.
 127 pub fn parse_item_from_source_str<'a>(name: String,
 128                                       source: String,
 129                                       cfg: ast::CrateConfig,
 130                                       sess: &'a ParseSess)
 131                                       -> PResult<'a, Option<P<ast::Item>>> {
 132     let mut p = new_parser_from_source_str(sess, cfg, name, source);
 133     p.parse_item()
 134 }
 135
 136 pub fn parse_meta_from_source_str<'a>(name: String,
 137                                       source: String,
 138                                       cfg: ast::CrateConfig,
 139                                       sess: &'a ParseSess)
 140                                       -> PResult<'a, P<ast::MetaItem>> {
 141     let mut p = new_parser_from_source_str(sess, cfg, name, source);
 142     p.parse_meta_item()
 143 }
 144
 145 pub fn parse_stmt_from_source_str<'a>(name: String,
 146                                       source: String,
 147                                       cfg: ast::CrateConfig,
 148                                       sess: &'a ParseSess)
 149                                       -> PResult<'a, Option<ast::Stmt>> {
 150     let mut p = new_parser_from_source_str(
 151         sess,
 152         cfg,
 153         name,
 154         source
 155     );
 156     p.parse_stmt()
 157 }
 158
 159 // Warning: This parses with quote_depth > 0, which is not the default.
 160 pub fn parse_tts_from_source_str<'a>(name: String,
 161                                      source: String,
 162                                      cfg: ast::CrateConfig,
 163                                      sess: &'a ParseSess)
 164                                      -> PResult<'a, Vec<ast::TokenTree>> {
 165     let mut p = new_parser_from_source_str(
 166         sess,
 167         cfg,
 168         name,
 169         source
 170     );
 171     p.quote_depth += 1;
 172     // right now this is re-creating the token trees from ... token trees.
 173     p.parse_all_token_trees()
 174 }
 175
 176 // Create a new parser from a source string
 177 pub fn new_parser_from_source_str<'a>(sess: &'a ParseSess,
 178                                       cfg: ast::CrateConfig,
 179                                       name: String,
 180                                       source: String)
 181                                       -> Parser<'a> {
 182     filemap_to_parser(sess, sess.codemap().new_filemap(name, source), cfg)
 183 }
 184
 185 /// Create a new parser, handling errors as appropriate
 186 /// if the file doesn't exist
 187 pub fn new_parser_from_file<'a>(sess: &'a ParseSess,
 188                                 cfg: ast::CrateConfig,
 189                                 path: &Path) -> Parser<'a> {
 190     filemap_to_parser(sess, file_to_filemap(sess, path, None), cfg)
 191 }
 192
 193 /// Given a session, a crate config, a path, and a span, add
 194 /// the file at the given path to the codemap, and return a parser.
 195 /// On an error, use the given span as the source of the problem.
 196 pub fn new_sub_parser_from_file<'a>(sess: &'a ParseSess,
 197                                     cfg: ast::CrateConfig,
 198                                     path: &Path,
 199                                     owns_directory: bool,
 200                                     module_name: Option<String>,
 201                                     sp: Span) -> Parser<'a> {
 202     let mut p = filemap_to_parser(sess, file_to_filemap(sess, path, Some(sp)), cfg);
 203     p.owns_directory = owns_directory;
 204     p.root_module_name = module_name;
 205     p
 206 }
 207
 208 /// Given a filemap and config, return a parser
 209 pub fn filemap_to_parser<'a>(sess: &'a ParseSess,
 210                              filemap: Rc<FileMap>,
 211                              cfg: ast::CrateConfig) -> Parser<'a> {
 212     let end_pos = filemap.end_pos;
 213     let mut parser = tts_to_parser(sess, filemap_to_tts(sess, filemap), cfg);
 214
 215     if parser.token == token::Eof && parser.span == codemap::DUMMY_SP {
 216         parser.span = codemap::mk_sp(end_pos, end_pos);
 217     }
 218
 219     parser
 220 }
 221
 222 // must preserve old name for now, because quote! from the *existing*
 223 // compiler expands into it
 224 pub fn new_parser_from_tts<'a>(sess: &'a ParseSess,
 225                                cfg: ast::CrateConfig,
 226                                tts: Vec<ast::TokenTree>) -> Parser<'a> {
 227     tts_to_parser(sess, tts, cfg)
 228 }
 229
 230
 231 // base abstractions
 232
 233 /// Given a session and a path and an optional span (for error reporting),
 234 /// add the path to the session's codemap and return the new filemap.
 235 fn file_to_filemap(sess: &ParseSess, path: &Path, spanopt: Option<Span>)
 236                    -> Rc<FileMap> {
 237     match sess.codemap().load_file(path) {
 238         Ok(filemap) => filemap,
 239         Err(e) => {
 240             let msg = format!("couldn't read {:?}: {}", path.display(), e);
 241             match spanopt {
 242                 Some(sp) => panic!(sess.span_diagnostic.span_fatal(sp, &msg)),
 243                 None => panic!(sess.span_diagnostic.fatal(&msg))
 244             }
 245         }
 246     }
 247 }
 248
 249 /// Given a filemap, produce a sequence of token-trees
 250 pub fn filemap_to_tts(sess: &ParseSess, filemap: Rc<FileMap>)
 251     -> Vec<ast::TokenTree> {
 252     // it appears to me that the cfg doesn't matter here... indeed,
 253     // parsing tt's probably shouldn't require a parser at all.
 254     let cfg = Vec::new();
 255     let srdr = lexer::StringReader::new(&sess.span_diagnostic, filemap);
 256     let mut p1 = Parser::new(sess, cfg, Box::new(srdr));
 257     panictry!(p1.parse_all_token_trees())
 258 }
 259
 260 /// Given tts and cfg, produce a parser
 261 pub fn tts_to_parser<'a>(sess: &'a ParseSess,
 262                          tts: Vec<ast::TokenTree>,
 263                          cfg: ast::CrateConfig) -> Parser<'a> {
 264     let trdr = lexer::new_tt_reader(&sess.span_diagnostic, None, None, tts);
 265     let mut p = Parser::new(sess, cfg, Box::new(trdr));
 266     p.check_unknown_macro_variable();
 267     p
 268 }
 269
 270 /// Parse a string representing a character literal into its final form.
 271 /// Rather than just accepting/rejecting a given literal, unescapes it as
 272 /// well. Can take any slice prefixed by a character escape. Returns the
 273 /// character and the number of characters consumed.
 274 pub fn char_lit(lit: &str) -> (char, isize) {
 275     use std::char;
 276
 277     let mut chars = lit.chars();
 278     let c = match (chars.next(), chars.next()) {
 279         (Some(c), None) if c != '\\' => return (c, 1),
 280         (Some('\\'), Some(c)) => match c {
 281             '"' => Some('"'),
 282             'n' => Some('\n'),
 283             'r' => Some('\r'),
 284             't' => Some('\t'),
 285             '\\' => Some('\\'),
 286             '\'' => Some('\''),
 287             '0' => Some('\0'),
 288             _ => { None }
 289         },
 290         _ => panic!("lexer accepted invalid char escape `{}`", lit)
 291     };
 292
 293     match c {
 294         Some(x) => return (x, 2),
 295         None => { }
 296     }
 297
 298     let msg = format!("lexer should have rejected a bad character escape {}", lit);
 299     let msg2 = &msg[..];
 300
 301     fn esc(len: usize, lit: &str) -> Option<(char, isize)> {
 302         u32::from_str_radix(&lit[2..len], 16).ok()
 303         .and_then(char::from_u32)
 304         .map(|x| (x, len as isize))
 305     }
 306
 307     let unicode_escape = || -> Option<(char, isize)> {
 308         if lit.as_bytes()[2] == b'{' {
 309             let idx = lit.find('}').expect(msg2);
 310             let subslice = &lit[3..idx];
 311             u32::from_str_radix(subslice, 16).ok()
 312                 .and_then(char::from_u32)
 313                 .map(|x| (x, subslice.chars().count() as isize + 4))
 314         } else {
 315             esc(6, lit)
 316         }
 317     };
 318
 319     // Unicode escapes
 320     return match lit.as_bytes()[1] as char {
 321         'x' | 'X' => esc(4, lit),
 322         'u' => unicode_escape(),
 323         'U' => esc(10, lit),
 324         _ => None,
 325     }.expect(msg2);
 326 }
 327
 328 /// Parse a string representing a string literal into its final form. Does
 329 /// unescaping.
 330 pub fn str_lit(lit: &str) -> String {
 331     debug!("parse_str_lit: given {}", lit.escape_default());
 332     let mut res = String::with_capacity(lit.len());
 333
 334     // FIXME #8372: This could be a for-loop if it didn't borrow the iterator
 335     let error = |i| format!("lexer should have rejected {} at {}", lit, i);
 336
 337     /// Eat everything up to a non-whitespace
 338     fn eat<'a>(it: &mut iter::Peekable<str::CharIndices<'a>>) {
 339         loop {
 340             match it.peek().map(|x| x.1) {
 341                 Some(' ') | Some('\n') | Some('\r') | Some('\t') => {
 342                     it.next();
 343                 },
 344                 _ => { break; }
 345             }
 346         }
 347     }
 348
 349     let mut chars = lit.char_indices().peekable();
 350     loop {
 351         match chars.next() {
 352             Some((i, c)) => {
 353                 match c {
 354                     '\\' => {
 355                         let ch = chars.peek().unwrap_or_else(|| {
 356                             panic!("{}", error(i))
 357                         }).1;
 358
 359                         if ch == '\n' {
 360                             eat(&mut chars);
 361                         } else if ch == '\r' {
 362                             chars.next();
 363                             let ch = chars.peek().unwrap_or_else(|| {
 364                                 panic!("{}", error(i))
 365                             }).1;
 366
 367                             if ch != '\n' {
 368                                 panic!("lexer accepted bare CR");
 369                             }
 370                             eat(&mut chars);
 371                         } else {
 372                             // otherwise, a normal escape
 373                             let (c, n) = char_lit(&lit[i..]);
 374                             for _ in 0..n - 1 { // we don't need to move past the first \
 375                                 chars.next();
 376                             }
 377                             res.push(c);
 378                         }
 379                     },
 380                     '\r' => {
 381                         let ch = chars.peek().unwrap_or_else(|| {
 382                             panic!("{}", error(i))
 383                         }).1;
 384
 385                         if ch != '\n' {
 386                             panic!("lexer accepted bare CR");
 387                         }
 388                         chars.next();
 389                         res.push('\n');
 390                     }
 391                     c => res.push(c),
 392                 }
 393             },
 394             None => break
 395         }
 396     }
 397
 398     res.shrink_to_fit(); // probably not going to do anything, unless there was an escape.
 399     debug!("parse_str_lit: returning {}", res);
 400     res
 401 }
 402
 403 /// Parse a string representing a raw string literal into its final form. The
 404 /// only operation this does is convert embedded CRLF into a single LF.
 405 pub fn raw_str_lit(lit: &str) -> String {
 406     debug!("raw_str_lit: given {}", lit.escape_default());
 407     let mut res = String::with_capacity(lit.len());
 408
 409     // FIXME #8372: This could be a for-loop if it didn't borrow the iterator
 410     let mut chars = lit.chars().peekable();
 411     loop {
 412         match chars.next() {
 413             Some(c) => {
 414                 if c == '\r' {
 415                     if *chars.peek().unwrap() != '\n' {
 416                         panic!("lexer accepted bare CR");
 417                     }
 418                     chars.next();
 419                     res.push('\n');
 420                 } else {
 421                     res.push(c);
 422                 }
 423             },
 424             None => break
 425         }
 426     }
 427
 428     res.shrink_to_fit();
 429     res
 430 }
 431
 432 // check if `s` looks like i32 or u1234 etc.
 433 fn looks_like_width_suffix(first_chars: &[char], s: &str) -> bool {
 434     s.len() > 1 &&
 435         first_chars.contains(&char_at(s, 0)) &&
 436         s[1..].chars().all(|c| '0' <= c && c <= '9')
 437 }
 438
 439 fn filtered_float_lit(data: token::InternedString, suffix: Option<&str>,
 440                       sd: &Handler, sp: Span) -> ast::LitKind {
 441     debug!("filtered_float_lit: {}, {:?}", data, suffix);
 442     match suffix.as_ref().map(|s| &**s) {
 443         Some("f32") => ast::LitKind::Float(data, ast::FloatTy::F32),
 444         Some("f64") => ast::LitKind::Float(data, ast::FloatTy::F64),
 445         Some(suf) => {
 446             if suf.len() >= 2 && looks_like_width_suffix(&['f'], suf) {
 447                 // if it looks like a width, lets try to be helpful.
 448                 sd.struct_span_err(sp, &format!("invalid width `{}` for float literal", &suf[1..]))
 449                  .fileline_help(sp, "valid widths are 32 and 64")
 450                  .emit();
 451             } else {
 452                 sd.struct_span_err(sp, &format!("invalid suffix `{}` for float literal", suf))
 453                   .fileline_help(sp, "valid suffixes are `f32` and `f64`")
 454                   .emit();
 455             }
 456
 457             ast::LitKind::FloatUnsuffixed(data)
 458         }
 459         None => ast::LitKind::FloatUnsuffixed(data)
 460     }
 461 }
 462 pub fn float_lit(s: &str, suffix: Option<InternedString>,
 463                  sd: &Handler, sp: Span) -> ast::LitKind {
 464     debug!("float_lit: {:?}, {:?}", s, suffix);
 465     // FIXME #2252: bounds checking float literals is deferred until trans
 466     let s = s.chars().filter(|&c| c != '_').collect::<String>();
 467     let data = token::intern_and_get_ident(&s);
 468     filtered_float_lit(data, suffix.as_ref().map(|s| &**s), sd, sp)
 469 }
 470
 471 /// Parse a string representing a byte literal into its final form. Similar to `char_lit`
 472 pub fn byte_lit(lit: &str) -> (u8, usize) {
 473     let err = |i| format!("lexer accepted invalid byte literal {} step {}", lit, i);
 474
 475     if lit.len() == 1 {
 476         (lit.as_bytes()[0], 1)
 477     } else {
 478         assert!(lit.as_bytes()[0] == b'\\', err(0));
 479         let b = match lit.as_bytes()[1] {
 480             b'"' => b'"',
 481             b'n' => b'\n',
 482             b'r' => b'\r',
 483             b't' => b'\t',
 484             b'\\' => b'\\',
 485             b'\'' => b'\'',
 486             b'0' => b'\0',
 487             _ => {
 488                 match u64::from_str_radix(&lit[2..4], 16).ok() {
 489                     Some(c) =>
 490                         if c > 0xFF {
 491                             panic!(err(2))
 492                         } else {
 493                             return (c as u8, 4)
 494                         },
 495                     None => panic!(err(3))
 496                 }
 497             }
 498         };
 499         return (b, 2);
 500     }
 501 }
 502
 503 pub fn byte_str_lit(lit: &str) -> Rc<Vec<u8>> {
 504     let mut res = Vec::with_capacity(lit.len());
 505
 506     // FIXME #8372: This could be a for-loop if it didn't borrow the iterator
 507     let error = |i| format!("lexer should have rejected {} at {}", lit, i);
 508
 509     /// Eat everything up to a non-whitespace
 510     fn eat<'a, I: Iterator<Item=(usize, u8)>>(it: &mut iter::Peekable<I>) {
 511         loop {
 512             match it.peek().map(|x| x.1) {
 513                 Some(b' ') | Some(b'\n') | Some(b'\r') | Some(b'\t') => {
 514                     it.next();
 515                 },
 516                 _ => { break; }
 517             }
 518         }
 519     }
 520
 521     // byte string literals *must* be ASCII, but the escapes don't have to be
 522     let mut chars = lit.bytes().enumerate().peekable();
 523     loop {
 524         match chars.next() {
 525             Some((i, b'\\')) => {
 526                 let em = error(i);
 527                 match chars.peek().expect(&em).1 {
 528                     b'\n' => eat(&mut chars),
 529                     b'\r' => {
 530                         chars.next();
 531                         if chars.peek().expect(&em).1 != b'\n' {
 532                             panic!("lexer accepted bare CR");
 533                         }
 534                         eat(&mut chars);
 535                     }
 536                     _ => {
 537                         // otherwise, a normal escape
 538                         let (c, n) = byte_lit(&lit[i..]);
 539                         // we don't need to move past the first \
 540                         for _ in 0..n - 1 {
 541                             chars.next();
 542                         }
 543                         res.push(c);
 544                     }
 545                 }
 546             },
 547             Some((i, b'\r')) => {
 548                 let em = error(i);
 549                 if chars.peek().expect(&em).1 != b'\n' {
 550                     panic!("lexer accepted bare CR");
 551                 }
 552                 chars.next();
 553                 res.push(b'\n');
 554             }
 555             Some((_, c)) => res.push(c),
 556             None => break,
 557         }
 558     }
 559
 560     Rc::new(res)
 561 }
 562
 563 pub fn integer_lit(s: &str,
 564                    suffix: Option<InternedString>,
 565                    sd: &Handler,
 566                    sp: Span)
 567                    -> ast::LitKind {
 568     // s can only be ascii, byte indexing is fine
 569
 570     let s2 = s.chars().filter(|&c| c != '_').collect::<String>();
 571     let mut s = &s2[..];
 572
 573     debug!("integer_lit: {}, {:?}", s, suffix);
 574
 575     let mut base = 10;
 576     let orig = s;
 577     let mut ty = ast::LitIntType::Unsuffixed;
 578
 579     if char_at(s, 0) == '0' && s.len() > 1 {
 580         match char_at(s, 1) {
 581             'x' => base = 16,
 582             'o' => base = 8,
 583             'b' => base = 2,
 584             _ => { }
 585         }
 586     }
 587
 588     // 1f64 and 2f32 etc. are valid float literals.
 589     if let Some(ref suf) = suffix {
 590         if looks_like_width_suffix(&['f'], suf) {
 591             match base {
 592                 16 => sd.span_err(sp, "hexadecimal float literal is not supported"),
 593                 8 => sd.span_err(sp, "octal float literal is not supported"),
 594                 2 => sd.span_err(sp, "binary float literal is not supported"),
 595                 _ => ()
 596             }
 597             let ident = token::intern_and_get_ident(&s);
 598             return filtered_float_lit(ident, Some(&suf), sd, sp)
 599         }
 600     }
 601
 602     if base != 10 {
 603         s = &s[2..];
 604     }
 605
 606     if let Some(ref suf) = suffix {
 607         if suf.is_empty() { sd.span_bug(sp, "found empty literal suffix in Some")}
 608         ty = match &**suf {
 609             "isize" => ast::LitIntType::Signed(ast::IntTy::Is),
 610             "i8"  => ast::LitIntType::Signed(ast::IntTy::I8),
 611             "i16" => ast::LitIntType::Signed(ast::IntTy::I16),
 612             "i32" => ast::LitIntType::Signed(ast::IntTy::I32),
 613             "i64" => ast::LitIntType::Signed(ast::IntTy::I64),
 614             "usize" => ast::LitIntType::Unsigned(ast::UintTy::Us),
 615             "u8"  => ast::LitIntType::Unsigned(ast::UintTy::U8),
 616             "u16" => ast::LitIntType::Unsigned(ast::UintTy::U16),
 617             "u32" => ast::LitIntType::Unsigned(ast::UintTy::U32),
 618             "u64" => ast::LitIntType::Unsigned(ast::UintTy::U64),
 619             _ => {
 620                 // i<digits> and u<digits> look like widths, so lets
 621                 // give an error message along those lines
 622                 if looks_like_width_suffix(&['i', 'u'], suf) {
 623                     sd.struct_span_err(sp, &format!("invalid width `{}` for integer literal",
 624                                              &suf[1..]))
 625                       .fileline_help(sp, "valid widths are 8, 16, 32 and 64")
 626                       .emit();
 627                 } else {
 628                     sd.struct_span_err(sp, &format!("invalid suffix `{}` for numeric literal", suf))
 629                       .fileline_help(sp, "the suffix must be one of the integral types \
 630                                       (`u32`, `isize`, etc)")
 631                       .emit();
 632                 }
 633
 634                 ty
 635             }
 636         }
 637     }
 638
 639     debug!("integer_lit: the type is {:?}, base {:?}, the new string is {:?}, the original \
 640            string was {:?}, the original suffix was {:?}", ty, base, s, orig, suffix);
 641
 642     match u64::from_str_radix(s, base) {
 643         Ok(r) => ast::LitKind::Int(r, ty),
 644         Err(_) => {
 645             // small bases are lexed as if they were base 10, e.g, the string
 646             // might be `0b10201`. This will cause the conversion above to fail,
 647             // but these cases have errors in the lexer: we don't want to emit
 648             // two errors, and we especially don't want to emit this error since
 649             // it isn't necessarily true.
 650             let already_errored = base < 10 &&
 651                 s.chars().any(|c| c.to_digit(10).map_or(false, |d| d >= base));
 652
 653             if !already_errored {
 654                 sd.span_err(sp, "int literal is too large");
 655             }
 656             ast::LitKind::Int(0, ty)
 657         }
 658     }
 659 }
 660
 661 #[cfg(test)]
 662 mod tests {
 663     use super::*;
 664     use std::rc::Rc;
 665     use codemap::{Span, BytePos, Pos, Spanned, NO_EXPANSION};
 666     use ast::{self, TokenTree, PatKind};
 667     use abi::Abi;
 668     use attr::{first_attr_value_str_by_name, AttrMetaMethods};
 669     use parse;
 670     use parse::parser::Parser;
 671     use parse::token::{str_to_ident};
 672     use print::pprust::item_to_string;
 673     use ptr::P;
 674     use util::parser_testing::{string_to_tts, string_to_parser};
 675     use util::parser_testing::{string_to_expr, string_to_item, string_to_stmt};
 676
 677     // produce a codemap::span
 678     fn sp(a: u32, b: u32) -> Span {
 679         Span {lo: BytePos(a), hi: BytePos(b), expn_id: NO_EXPANSION}
 680     }
 681
 682     #[test] fn path_exprs_1() {
 683         assert!(string_to_expr("a".to_string()) ==
 684                    P(ast::Expr{
 685                     id: ast::DUMMY_NODE_ID,
 686                     node: ast::ExprKind::Path(None, ast::Path {
 687                         span: sp(0, 1),
 688                         global: false,
 689                         segments: vec!(
 690                             ast::PathSegment {
 691                                 identifier: str_to_ident("a"),
 692                                 parameters: ast::PathParameters::none(),
 693                             }
 694                         ),
 695                     }),
 696                     span: sp(0, 1),
 697                     attrs: None,
 698                    }))
 699     }
 700
 701     #[test] fn path_exprs_2 () {
 702         assert!(string_to_expr("::a::b".to_string()) ==
 703                    P(ast::Expr {
 704                     id: ast::DUMMY_NODE_ID,
 705                     node: ast::ExprKind::Path(None, ast::Path {
 706                             span: sp(0, 6),
 707                             global: true,
 708                             segments: vec!(
 709                                 ast::PathSegment {
 710                                     identifier: str_to_ident("a"),
 711                                     parameters: ast::PathParameters::none(),
 712                                 },
 713                                 ast::PathSegment {
 714                                     identifier: str_to_ident("b"),
 715                                     parameters: ast::PathParameters::none(),
 716                                 }
 717                             )
 718                         }),
 719                     span: sp(0, 6),
 720                     attrs: None,
 721                    }))
 722     }
 723
 724     #[should_panic]
 725     #[test] fn bad_path_expr_1() {
 726         string_to_expr("::abc::def::return".to_string());
 727     }
 728
 729     // check the token-tree-ization of macros
 730     #[test]
 731     fn string_to_tts_macro () {
 732         let tts = string_to_tts("macro_rules! zip (($a)=>($a))".to_string());
 733         let tts: &[ast::TokenTree] = &tts[..];
 734
 735         match (tts.len(), tts.get(0), tts.get(1), tts.get(2), tts.get(3)) {
 736             (
 737                 4,
 738                 Some(&TokenTree::Token(_, token::Ident(name_macro_rules, token::Plain))),
 739                 Some(&TokenTree::Token(_, token::Not)),
 740                 Some(&TokenTree::Token(_, token::Ident(name_zip, token::Plain))),
 741                 Some(&TokenTree::Delimited(_, ref macro_delimed)),
 742             )
 743             if name_macro_rules.name.as_str() == "macro_rules"
 744             && name_zip.name.as_str() == "zip" => {
 745                 let tts = &macro_delimed.tts[..];
 746                 match (tts.len(), tts.get(0), tts.get(1), tts.get(2)) {
 747                     (
 748                         3,
 749                         Some(&TokenTree::Delimited(_, ref first_delimed)),
 750                         Some(&TokenTree::Token(_, token::FatArrow)),
 751                         Some(&TokenTree::Delimited(_, ref second_delimed)),
 752                     )
 753                     if macro_delimed.delim == token::Paren => {
 754                         let tts = &first_delimed.tts[..];
 755                         match (tts.len(), tts.get(0), tts.get(1)) {
 756                             (
 757                                 2,
 758                                 Some(&TokenTree::Token(_, token::Dollar)),
 759                                 Some(&TokenTree::Token(_, token::Ident(ident, token::Plain))),
 760                             )
 761                             if first_delimed.delim == token::Paren
 762                             && ident.name.as_str() == "a" => {},
 763                             _ => panic!("value 3: {:?}", **first_delimed),
 764                         }
 765                         let tts = &second_delimed.tts[..];
 766                         match (tts.len(), tts.get(0), tts.get(1)) {
 767                             (
 768                                 2,
 769                                 Some(&TokenTree::Token(_, token::Dollar)),
 770                                 Some(&TokenTree::Token(_, token::Ident(ident, token::Plain))),
 771                             )
 772                             if second_delimed.delim == token::Paren
 773                             && ident.name.as_str() == "a" => {},
 774                             _ => panic!("value 4: {:?}", **second_delimed),
 775                         }
 776                     },
 777                     _ => panic!("value 2: {:?}", **macro_delimed),
 778                 }
 779             },
 780             _ => panic!("value: {:?}",tts),
 781         }
 782     }
 783
 784     #[test]
 785     fn string_to_tts_1() {
 786         let tts = string_to_tts("fn a (b : i32) { b; }".to_string());
 787
 788         let expected = vec![
 789             TokenTree::Token(sp(0, 2),
 790                          token::Ident(str_to_ident("fn"),
 791                          token::IdentStyle::Plain)),
 792             TokenTree::Token(sp(3, 4),
 793                          token::Ident(str_to_ident("a"),
 794                          token::IdentStyle::Plain)),
 795             TokenTree::Delimited(
 796                 sp(5, 14),
 797                 Rc::new(ast::Delimited {
 798                     delim: token::DelimToken::Paren,
 799                     open_span: sp(5, 6),
 800                     tts: vec![
 801                         TokenTree::Token(sp(6, 7),
 802                                      token::Ident(str_to_ident("b"),
 803                                      token::IdentStyle::Plain)),
 804                         TokenTree::Token(sp(8, 9),
 805                                      token::Colon),
 806                         TokenTree::Token(sp(10, 13),
 807                                      token::Ident(str_to_ident("i32"),
 808                                      token::IdentStyle::Plain)),
 809                     ],
 810                     close_span: sp(13, 14),
 811                 })),
 812             TokenTree::Delimited(
 813                 sp(15, 21),
 814                 Rc::new(ast::Delimited {
 815                     delim: token::DelimToken::Brace,
 816                     open_span: sp(15, 16),
 817                     tts: vec![
 818                         TokenTree::Token(sp(17, 18),
 819                                      token::Ident(str_to_ident("b"),
 820                                      token::IdentStyle::Plain)),
 821                         TokenTree::Token(sp(18, 19),
 822                                      token::Semi)
 823                     ],
 824                     close_span: sp(20, 21),
 825                 }))
 826         ];
 827
 828         assert_eq!(tts, expected);
 829     }
 830
 831     #[test] fn ret_expr() {
 832         assert!(string_to_expr("return d".to_string()) ==
 833                    P(ast::Expr{
 834                     id: ast::DUMMY_NODE_ID,
 835                     node:ast::ExprKind::Ret(Some(P(ast::Expr{
 836                         id: ast::DUMMY_NODE_ID,
 837                         node:ast::ExprKind::Path(None, ast::Path{
 838                             span: sp(7, 8),
 839                             global: false,
 840                             segments: vec!(
 841                                 ast::PathSegment {
 842                                     identifier: str_to_ident("d"),
 843                                     parameters: ast::PathParameters::none(),
 844                                 }
 845                             ),
 846                         }),
 847                         span:sp(7,8),
 848                         attrs: None,
 849                     }))),
 850                     span:sp(0,8),
 851                     attrs: None,
 852                    }))
 853     }
 854
 855     #[test] fn parse_stmt_1 () {
 856         assert!(string_to_stmt("b;".to_string()) ==
 857                    Some(Spanned{
 858                        node: ast::StmtKind::Expr(P(ast::Expr {
 859                            id: ast::DUMMY_NODE_ID,
 860                            node: ast::ExprKind::Path(None, ast::Path {
 861                                span:sp(0,1),
 862                                global:false,
 863                                segments: vec!(
 864                                 ast::PathSegment {
 865                                     identifier: str_to_ident("b"),
 866                                     parameters: ast::PathParameters::none(),
 867                                 }
 868                                ),
 869                             }),
 870                            span: sp(0,1),
 871                            attrs: None}),
 872                                            ast::DUMMY_NODE_ID),
 873                        span: sp(0,1)}))
 874
 875     }
 876
 877     fn parser_done(p: Parser){
 878         assert_eq!(p.token.clone(), token::Eof);
 879     }
 880
 881     #[test] fn parse_ident_pat () {
 882         let sess = ParseSess::new();
 883         let mut parser = string_to_parser(&sess, "b".to_string());
 884         assert!(panictry!(parser.parse_pat())
 885                 == P(ast::Pat{
 886                 id: ast::DUMMY_NODE_ID,
 887                 node: PatKind::Ident(ast::BindingMode::ByValue(ast::Mutability::Immutable),
 888                                     Spanned{ span:sp(0, 1),
 889                                              node: str_to_ident("b")
 890                     },
 891                                     None),
 892                 span: sp(0,1)}));
 893         parser_done(parser);
 894     }
 895
 896     // check the contents of the tt manually:
 897     #[test] fn parse_fundecl () {
 898         // this test depends on the intern order of "fn" and "i32"
 899         assert_eq!(string_to_item("fn a (b : i32) { b; }".to_string()),
 900                   Some(
 901                       P(ast::Item{ident:str_to_ident("a"),
 902                             attrs:Vec::new(),
 903                             id: ast::DUMMY_NODE_ID,
 904                             node: ast::ItemKind::Fn(P(ast::FnDecl {
 905                                 inputs: vec!(ast::Arg{
 906                                     ty: P(ast::Ty{id: ast::DUMMY_NODE_ID,
 907                                                   node: ast::TyKind::Path(None, ast::Path{
 908                                         span:sp(10,13),
 909                                         global:false,
 910                                         segments: vec!(
 911                                             ast::PathSegment {
 912                                                 identifier:
 913                                                     str_to_ident("i32"),
 914                                                 parameters: ast::PathParameters::none(),
 915                                             }
 916                                         ),
 917                                         }),
 918                                         span:sp(10,13)
 919                                     }),
 920                                     pat: P(ast::Pat {
 921                                         id: ast::DUMMY_NODE_ID,
 922                                         node: PatKind::Ident(
 923                                             ast::BindingMode::ByValue(ast::Mutability::Immutable),
 924                                                 Spanned{
 925                                                     span: sp(6,7),
 926                                                     node: str_to_ident("b")},
 927                                                 None
 928                                                     ),
 929                                             span: sp(6,7)
 930                                     }),
 931                                         id: ast::DUMMY_NODE_ID
 932                                     }),
 933                                 output: ast::FunctionRetTy::Default(sp(15, 15)),
 934                                 variadic: false
 935                             }),
 936                                     ast::Unsafety::Normal,
 937                                     ast::Constness::NotConst,
 938                                     Abi::Rust,
 939                                     ast::Generics{ // no idea on either of these:
 940                                         lifetimes: Vec::new(),
 941                                         ty_params: P::empty(),
 942                                         where_clause: ast::WhereClause {
 943                                             id: ast::DUMMY_NODE_ID,
 944                                             predicates: Vec::new(),
 945                                         }
 946                                     },
 947                                     P(ast::Block {
 948                                         stmts: vec!(Spanned{
 949                                             node: ast::StmtKind::Semi(P(ast::Expr{
 950                                                 id: ast::DUMMY_NODE_ID,
 951                                                 node: ast::ExprKind::Path(None,
 952                                                       ast::Path{
 953                                                         span:sp(17,18),
 954                                                         global:false,
 955                                                         segments: vec!(
 956                                                             ast::PathSegment {
 957                                                                 identifier:
 958                                                                 str_to_ident(
 959                                                                     "b"),
 960                                                                 parameters:
 961                                                                 ast::PathParameters::none(),
 962                                                             }
 963                                                         ),
 964                                                       }),
 965                                                 span: sp(17,18),
 966                                                 attrs: None,}),
 967                                                 ast::DUMMY_NODE_ID),
 968                                             span: sp(17,19)}),
 969                                         expr: None,
 970                                         id: ast::DUMMY_NODE_ID,
 971                                         rules: ast::BlockCheckMode::Default, // no idea
 972                                         span: sp(15,21),
 973                                     })),
 974                             vis: ast::Visibility::Inherited,
 975                             span: sp(0,21)})));
 976     }
 977
 978     #[test] fn parse_use() {
 979         let use_s = "use foo::bar::baz;";
 980         let vitem = string_to_item(use_s.to_string()).unwrap();
 981         let vitem_s = item_to_string(&vitem);
 982         assert_eq!(&vitem_s[..], use_s);
 983
 984         let use_s = "use foo::bar as baz;";
 985         let vitem = string_to_item(use_s.to_string()).unwrap();
 986         let vitem_s = item_to_string(&vitem);
 987         assert_eq!(&vitem_s[..], use_s);
 988     }
 989
 990     #[test] fn parse_extern_crate() {
 991         let ex_s = "extern crate foo;";
 992         let vitem = string_to_item(ex_s.to_string()).unwrap();
 993         let vitem_s = item_to_string(&vitem);
 994         assert_eq!(&vitem_s[..], ex_s);
 995
 996         let ex_s = "extern crate foo as bar;";
 997         let vitem = string_to_item(ex_s.to_string()).unwrap();
 998         let vitem_s = item_to_string(&vitem);
 999         assert_eq!(&vitem_s[..], ex_s);
1000     }
1001
1002     fn get_spans_of_pat_idents(src: &str) -> Vec<Span> {
1003         let item = string_to_item(src.to_string()).unwrap();
1004
1005         struct PatIdentVisitor {
1006             spans: Vec<Span>
1007         }
1008         impl<'v> ::visit::Visitor<'v> for PatIdentVisitor {
1009             fn visit_pat(&mut self, p: &'v ast::Pat) {
1010                 match p.node {
1011                     PatKind::Ident(_ , ref spannedident, _) => {
1012                         self.spans.push(spannedident.span.clone());
1013                     }
1014                     _ => {
1015                         ::visit::walk_pat(self, p);
1016                     }
1017                 }
1018             }
1019         }
1020         let mut v = PatIdentVisitor { spans: Vec::new() };
1021         ::visit::walk_item(&mut v, &item);
1022         return v.spans;
1023     }
1024
1025     #[test] fn span_of_self_arg_pat_idents_are_correct() {
1026
1027         let srcs = ["impl z { fn a (&self, &myarg: i32) {} }",
1028                     "impl z { fn a (&mut self, &myarg: i32) {} }",
1029                     "impl z { fn a (&'a self, &myarg: i32) {} }",
1030                     "impl z { fn a (self, &myarg: i32) {} }",
1031                     "impl z { fn a (self: Foo, &myarg: i32) {} }",
1032                     ];
1033
1034         for &src in &srcs {
1035             let spans = get_spans_of_pat_idents(src);
1036             let Span{ lo, hi, .. } = spans[0];
1037             assert!("self" == &src[lo.to_usize()..hi.to_usize()],
1038                     "\"{}\" != \"self\". src=\"{}\"",
1039                     &src[lo.to_usize()..hi.to_usize()], src)
1040         }
1041     }
1042
1043     #[test] fn parse_exprs () {
1044         // just make sure that they parse....
1045         string_to_expr("3 + 4".to_string());
1046         string_to_expr("a::z.froob(b,&(987+3))".to_string());
1047     }
1048
1049     #[test] fn attrs_fix_bug () {
1050         string_to_item("pub fn mk_file_writer(path: &Path, flags: &[FileFlag])
1051                    -> Result<Box<Writer>, String> {
1052     #[cfg(windows)]
1053     fn wb() -> c_int {
1054       (O_WRONLY | libc::consts::os::extra::O_BINARY) as c_int
1055     }
1056
1057     #[cfg(unix)]
1058     fn wb() -> c_int { O_WRONLY as c_int }
1059
1060     let mut fflags: c_int = wb();
1061 }".to_string());
1062     }
1063
1064     #[test] fn crlf_doc_comments() {
1065         let sess = ParseSess::new();
1066
1067         let name = "<source>".to_string();
1068         let source = "/// doc comment\r\nfn foo() {}".to_string();
1069         let item = parse_item_from_source_str(name.clone(), source, Vec::new(), &sess)
1070             .unwrap().unwrap();
1071         let doc = first_attr_value_str_by_name(&item.attrs, "doc").unwrap();
1072         assert_eq!(&doc[..], "/// doc comment");
1073
1074         let source = "/// doc comment\r\n/// line 2\r\nfn foo() {}".to_string();
1075         let item = parse_item_from_source_str(name.clone(), source, Vec::new(), &sess)
1076             .unwrap().unwrap();
1077         let docs = item.attrs.iter().filter(|a| &*a.name() == "doc")
1078                     .map(|a| a.value_str().unwrap().to_string()).collect::<Vec<_>>();
1079         let b: &[_] = &["/// doc comment".to_string(), "/// line 2".to_string()];
1080         assert_eq!(&docs[..], b);
1081
1082         let source = "/** doc comment\r\n *  with CRLF */\r\nfn foo() {}".to_string();
1083         let item = parse_item_from_source_str(name, source, Vec::new(), &sess).unwrap().unwrap();
1084         let doc = first_attr_value_str_by_name(&item.attrs, "doc").unwrap();
1085         assert_eq!(&doc[..], "/** doc comment\n *  with CRLF */");
1086     }
1087
1088     #[test]
1089     fn ttdelim_span() {
1090         let sess = ParseSess::new();
1091         let expr = parse::parse_expr_from_source_str("foo".to_string(),
1092             "foo!( fn main() { body } )".to_string(), vec![], &sess).unwrap();
1093
1094         let tts = match expr.node {
1095             ast::ExprKind::Mac(ref mac) => mac.node.tts.clone(),
1096             _ => panic!("not a macro"),
1097         };
1098
1099         let span = tts.iter().rev().next().unwrap().get_span();
1100
1101         match sess.codemap().span_to_snippet(span) {
1102             Ok(s) => assert_eq!(&s[..], "{ body }"),
1103             Err(_) => panic!("could not get snippet"),
1104         }
1105     }
1106 }