src/libsyntax/ext/tt/macro_rules.rs

   1 // Copyright 2015 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 use {ast, attr};
  12 use syntax_pos::{Span, DUMMY_SP};
  13 use ext::base::{DummyResult, ExtCtxt, MacResult, SyntaxExtension};
  14 use ext::base::{NormalTT, TTMacroExpander};
  15 use ext::expand::{Expansion, ExpansionKind};
  16 use ext::tt::macro_parser::{Success, Error, Failure};
  17 use ext::tt::macro_parser::{MatchedSeq, MatchedNonterminal};
  18 use ext::tt::macro_parser::{parse, parse_failure_msg};
  19 use ext::tt::quoted;
  20 use ext::tt::transcribe::transcribe;
  21 use feature_gate::{self, emit_feature_err, Features, GateIssue};
  22 use parse::{Directory, ParseSess};
  23 use parse::parser::Parser;
  24 use parse::token::{self, NtTT};
  25 use parse::token::Token::*;
  26 use symbol::Symbol;
  27 use tokenstream::{TokenStream, TokenTree};
  28
  29 use std::cell::RefCell;
  30 use std::collections::HashMap;
  31 use std::collections::hash_map::Entry;
  32 use std::rc::Rc;
  33
  34 pub struct ParserAnyMacro<'a> {
  35     parser: Parser<'a>,
  36
  37     /// Span of the expansion site of the macro this parser is for
  38     site_span: Span,
  39     /// The ident of the macro we're parsing
  40     macro_ident: ast::Ident
  41 }
  42
  43 impl<'a> ParserAnyMacro<'a> {
  44     pub fn make(mut self: Box<ParserAnyMacro<'a>>, kind: ExpansionKind) -> Expansion {
  45         let ParserAnyMacro { site_span, macro_ident, ref mut parser } = *self;
  46         let expansion = panictry!(parser.parse_expansion(kind, true));
  47
  48         // We allow semicolons at the end of expressions -- e.g. the semicolon in
  49         // `macro_rules! m { () => { panic!(); } }` isn't parsed by `.parse_expr()`,
  50         // but `m!()` is allowed in expression positions (c.f. issue #34706).
  51         if kind == ExpansionKind::Expr && parser.token == token::Semi {
  52             parser.bump();
  53         }
  54
  55         // Make sure we don't have any tokens left to parse so we don't silently drop anything.
  56         let path = ast::Path::from_ident(site_span, macro_ident);
  57         parser.ensure_complete_parse(&path, kind.name(), site_span);
  58         expansion
  59     }
  60 }
  61
  62 struct MacroRulesMacroExpander {
  63     name: ast::Ident,
  64     lhses: Vec<quoted::TokenTree>,
  65     rhses: Vec<quoted::TokenTree>,
  66     valid: bool,
  67 }
  68
  69 impl TTMacroExpander for MacroRulesMacroExpander {
  70     fn expand<'cx>(&self,
  71                    cx: &'cx mut ExtCtxt,
  72                    sp: Span,
  73                    input: TokenStream)
  74                    -> Box<MacResult+'cx> {
  75         if !self.valid {
  76             return DummyResult::any(sp);
  77         }
  78         generic_extension(cx,
  79                           sp,
  80                           self.name,
  81                           input,
  82                           &self.lhses,
  83                           &self.rhses)
  84     }
  85 }
  86
  87 fn trace_macros_note(cx: &mut ExtCtxt, sp: Span, message: String) {
  88     let sp = sp.macro_backtrace().last().map(|trace| trace.call_site).unwrap_or(sp);
  89     let mut values: &mut Vec<String> = cx.expansions.entry(sp).or_insert_with(Vec::new);
  90     values.push(message);
  91 }
  92
  93 /// Given `lhses` and `rhses`, this is the new macro we create
  94 fn generic_extension<'cx>(cx: &'cx mut ExtCtxt,
  95                           sp: Span,
  96                           name: ast::Ident,
  97                           arg: TokenStream,
  98                           lhses: &[quoted::TokenTree],
  99                           rhses: &[quoted::TokenTree])
 100                           -> Box<MacResult+'cx> {
 101     if cx.trace_macros() {
 102         trace_macros_note(cx, sp, format!("expanding `{}! {{ {} }}`", name, arg));
 103     }
 104
 105     // Which arm's failure should we report? (the one furthest along)
 106     let mut best_fail_spot = DUMMY_SP;
 107     let mut best_fail_tok = None;
 108
 109     for (i, lhs) in lhses.iter().enumerate() { // try each arm's matchers
 110         let lhs_tt = match *lhs {
 111             quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 112             _ => cx.span_bug(sp, "malformed macro lhs")
 113         };
 114
 115         match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 116             Success(named_matches) => {
 117                 let rhs = match rhses[i] {
 118                     // ignore delimiters
 119                     quoted::TokenTree::Delimited(_, ref delimed) => delimed.tts.clone(),
 120                     _ => cx.span_bug(sp, "malformed macro rhs"),
 121                 };
 122
 123                 let rhs_spans = rhs.iter().map(|t| t.span()).collect::<Vec<_>>();
 124                 // rhs has holes ( `$id` and `$(...)` that need filled)
 125                 let mut tts = transcribe(cx, Some(named_matches), rhs);
 126
 127                 // Replace all the tokens for the corresponding positions in the macro, to maintain
 128                 // proper positions in error reporting, while maintaining the macro_backtrace.
 129                 if rhs_spans.len() == tts.len() {
 130                     tts = tts.map_enumerated(|i, tt| {
 131                         let mut tt = tt.clone();
 132                         let mut sp = rhs_spans[i];
 133                         sp.ctxt = tt.span().ctxt;
 134                         tt.set_span(sp);
 135                         tt
 136                     });
 137                 }
 138
 139                 if cx.trace_macros() {
 140                     trace_macros_note(cx, sp, format!("to `{}`", tts));
 141                 }
 142
 143                 let directory = Directory {
 144                     path: cx.current_expansion.module.directory.clone(),
 145                     ownership: cx.current_expansion.directory_ownership,
 146                 };
 147                 let mut p = Parser::new(cx.parse_sess(), tts, Some(directory), true, false);
 148                 p.root_module_name = cx.current_expansion.module.mod_path.last()
 149                     .map(|id| id.name.as_str().to_string());
 150
 151                 p.process_potential_macro_variable();
 152                 // Let the context choose how to interpret the result.
 153                 // Weird, but useful for X-macros.
 154                 return Box::new(ParserAnyMacro {
 155                     parser: p,
 156
 157                     // Pass along the original expansion site and the name of the macro
 158                     // so we can print a useful error message if the parse of the expanded
 159                     // macro leaves unparsed tokens.
 160                     site_span: sp,
 161                     macro_ident: name
 162                 })
 163             }
 164             Failure(sp, tok) => if sp.lo >= best_fail_spot.lo {
 165                 best_fail_spot = sp;
 166                 best_fail_tok = Some(tok);
 167             },
 168             Error(err_sp, ref msg) => {
 169                 cx.span_fatal(err_sp.substitute_dummy(sp), &msg[..])
 170             }
 171         }
 172     }
 173
 174     let best_fail_msg = parse_failure_msg(best_fail_tok.expect("ran no matchers"));
 175     cx.span_fatal(best_fail_spot.substitute_dummy(sp), &best_fail_msg);
 176 }
 177
 178 // Note that macro-by-example's input is also matched against a token tree:
 179 //                   $( $lhs:tt => $rhs:tt );+
 180 //
 181 // Holy self-referential!
 182
 183 /// Converts a `macro_rules!` invocation into a syntax extension.
 184 pub fn compile(sess: &ParseSess, features: &RefCell<Features>, def: &ast::Item) -> SyntaxExtension {
 185     let lhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("lhs"));
 186     let rhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("rhs"));
 187
 188     // Parse the macro_rules! invocation
 189     let body = match def.node {
 190         ast::ItemKind::MacroDef(ref body) => body,
 191         _ => unreachable!(),
 192     };
 193
 194     // The pattern that macro_rules matches.
 195     // The grammar for macro_rules! is:
 196     // $( $lhs:tt => $rhs:tt );+
 197     // ...quasiquoting this would be nice.
 198     // These spans won't matter, anyways
 199     let argument_gram = vec![
 200         quoted::TokenTree::Sequence(DUMMY_SP, Rc::new(quoted::SequenceRepetition {
 201             tts: vec![
 202                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, lhs_nm, ast::Ident::from_str("tt")),
 203                 quoted::TokenTree::Token(DUMMY_SP, token::FatArrow),
 204                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, rhs_nm, ast::Ident::from_str("tt")),
 205             ],
 206             separator: Some(if body.legacy { token::Semi } else { token::Comma }),
 207             op: quoted::KleeneOp::OneOrMore,
 208             num_captures: 2,
 209         })),
 210         // to phase into semicolon-termination instead of semicolon-separation
 211         quoted::TokenTree::Sequence(DUMMY_SP, Rc::new(quoted::SequenceRepetition {
 212             tts: vec![quoted::TokenTree::Token(DUMMY_SP, token::Semi)],
 213             separator: None,
 214             op: quoted::KleeneOp::ZeroOrMore,
 215             num_captures: 0
 216         })),
 217     ];
 218
 219     let argument_map = match parse(sess, body.stream(), &argument_gram, None, true) {
 220         Success(m) => m,
 221         Failure(sp, tok) => {
 222             let s = parse_failure_msg(tok);
 223             panic!(sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s));
 224         }
 225         Error(sp, s) => {
 226             panic!(sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s));
 227         }
 228     };
 229
 230     let mut valid = true;
 231
 232     // Extract the arguments:
 233     let lhses = match *argument_map[&lhs_nm] {
 234         MatchedSeq(ref s, _) => {
 235             s.iter().map(|m| {
 236                 if let MatchedNonterminal(ref nt) = *m {
 237                     if let NtTT(ref tt) = **nt {
 238                         let tt = quoted::parse(tt.clone().into(), true, sess).pop().unwrap();
 239                         valid &= check_lhs_nt_follows(sess, features, &def.attrs, &tt);
 240                         return tt;
 241                     }
 242                 }
 243                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 244             }).collect::<Vec<quoted::TokenTree>>()
 245         }
 246         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 247     };
 248
 249     let rhses = match *argument_map[&rhs_nm] {
 250         MatchedSeq(ref s, _) => {
 251             s.iter().map(|m| {
 252                 if let MatchedNonterminal(ref nt) = *m {
 253                     if let NtTT(ref tt) = **nt {
 254                         return quoted::parse(tt.clone().into(), false, sess).pop().unwrap();
 255                     }
 256                 }
 257                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 258             }).collect::<Vec<quoted::TokenTree>>()
 259         }
 260         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured rhs")
 261     };
 262
 263     for rhs in &rhses {
 264         valid &= check_rhs(sess, rhs);
 265     }
 266
 267     // don't abort iteration early, so that errors for multiple lhses can be reported
 268     for lhs in &lhses {
 269         valid &= check_lhs_no_empty_seq(sess, &[lhs.clone()])
 270     }
 271
 272     let exp: Box<_> = Box::new(MacroRulesMacroExpander {
 273         name: def.ident,
 274         lhses: lhses,
 275         rhses: rhses,
 276         valid: valid,
 277     });
 278
 279     if body.legacy {
 280         let allow_internal_unstable = attr::contains_name(&def.attrs, "allow_internal_unstable");
 281         NormalTT(exp, Some((def.id, def.span)), allow_internal_unstable)
 282     } else {
 283         SyntaxExtension::DeclMacro(exp, Some((def.id, def.span)))
 284     }
 285 }
 286
 287 fn check_lhs_nt_follows(sess: &ParseSess,
 288                         features: &RefCell<Features>,
 289                         attrs: &[ast::Attribute],
 290                         lhs: &quoted::TokenTree) -> bool {
 291     // lhs is going to be like TokenTree::Delimited(...), where the
 292     // entire lhs is those tts. Or, it can be a "bare sequence", not wrapped in parens.
 293     if let quoted::TokenTree::Delimited(_, ref tts) = *lhs {
 294         check_matcher(sess, features, attrs, &tts.tts)
 295     } else {
 296         let msg = "invalid macro matcher; matchers must be contained in balanced delimiters";
 297         sess.span_diagnostic.span_err(lhs.span(), msg);
 298         false
 299     }
 300     // we don't abort on errors on rejection, the driver will do that for us
 301     // after parsing/expansion. we can report every error in every macro this way.
 302 }
 303
 304 /// Check that the lhs contains no repetition which could match an empty token
 305 /// tree, because then the matcher would hang indefinitely.
 306 fn check_lhs_no_empty_seq(sess: &ParseSess, tts: &[quoted::TokenTree]) -> bool {
 307     use self::quoted::TokenTree;
 308     for tt in tts {
 309         match *tt {
 310             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => (),
 311             TokenTree::Delimited(_, ref del) => if !check_lhs_no_empty_seq(sess, &del.tts) {
 312                 return false;
 313             },
 314             TokenTree::Sequence(span, ref seq) => {
 315                 if seq.separator.is_none() && seq.tts.iter().all(|seq_tt| {
 316                     match *seq_tt {
 317                         TokenTree::MetaVarDecl(_, _, id) => id.name == "vis",
 318                         TokenTree::Sequence(_, ref sub_seq) =>
 319                             sub_seq.op == quoted::KleeneOp::ZeroOrMore,
 320                         _ => false,
 321                     }
 322                 }) {
 323                     sess.span_diagnostic.span_err(span, "repetition matches empty token tree");
 324                     return false;
 325                 }
 326                 if !check_lhs_no_empty_seq(sess, &seq.tts) {
 327                     return false;
 328                 }
 329             }
 330         }
 331     }
 332
 333     true
 334 }
 335
 336 fn check_rhs(sess: &ParseSess, rhs: &quoted::TokenTree) -> bool {
 337     match *rhs {
 338         quoted::TokenTree::Delimited(..) => return true,
 339         _ => sess.span_diagnostic.span_err(rhs.span(), "macro rhs must be delimited")
 340     }
 341     false
 342 }
 343
 344 fn check_matcher(sess: &ParseSess,
 345                  features: &RefCell<Features>,
 346                  attrs: &[ast::Attribute],
 347                  matcher: &[quoted::TokenTree]) -> bool {
 348     let first_sets = FirstSets::new(matcher);
 349     let empty_suffix = TokenSet::empty();
 350     let err = sess.span_diagnostic.err_count();
 351     check_matcher_core(sess, features, attrs, &first_sets, matcher, &empty_suffix);
 352     err == sess.span_diagnostic.err_count()
 353 }
 354
 355 // The FirstSets for a matcher is a mapping from subsequences in the
 356 // matcher to the FIRST set for that subsequence.
 357 //
 358 // This mapping is partially precomputed via a backwards scan over the
 359 // token trees of the matcher, which provides a mapping from each
 360 // repetition sequence to its FIRST set.
 361 //
 362 // (Hypothetically sequences should be uniquely identifiable via their
 363 // spans, though perhaps that is false e.g. for macro-generated macros
 364 // that do not try to inject artificial span information. My plan is
 365 // to try to catch such cases ahead of time and not include them in
 366 // the precomputed mapping.)
 367 struct FirstSets {
 368     // this maps each TokenTree::Sequence `$(tt ...) SEP OP` that is uniquely identified by its
 369     // span in the original matcher to the First set for the inner sequence `tt ...`.
 370     //
 371     // If two sequences have the same span in a matcher, then map that
 372     // span to None (invalidating the mapping here and forcing the code to
 373     // use a slow path).
 374     first: HashMap<Span, Option<TokenSet>>,
 375 }
 376
 377 impl FirstSets {
 378     fn new(tts: &[quoted::TokenTree]) -> FirstSets {
 379         use self::quoted::TokenTree;
 380
 381         let mut sets = FirstSets { first: HashMap::new() };
 382         build_recur(&mut sets, tts);
 383         return sets;
 384
 385         // walks backward over `tts`, returning the FIRST for `tts`
 386         // and updating `sets` at the same time for all sequence
 387         // substructure we find within `tts`.
 388         fn build_recur(sets: &mut FirstSets, tts: &[TokenTree]) -> TokenSet {
 389             let mut first = TokenSet::empty();
 390             for tt in tts.iter().rev() {
 391                 match *tt {
 392                     TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 393                         first.replace_with(tt.clone());
 394                     }
 395                     TokenTree::Delimited(span, ref delimited) => {
 396                         build_recur(sets, &delimited.tts[..]);
 397                         first.replace_with(delimited.open_tt(span));
 398                     }
 399                     TokenTree::Sequence(sp, ref seq_rep) => {
 400                         let subfirst = build_recur(sets, &seq_rep.tts[..]);
 401
 402                         match sets.first.entry(sp) {
 403                             Entry::Vacant(vac) => {
 404                                 vac.insert(Some(subfirst.clone()));
 405                             }
 406                             Entry::Occupied(mut occ) => {
 407                                 // if there is already an entry, then a span must have collided.
 408                                 // This should not happen with typical macro_rules macros,
 409                                 // but syntax extensions need not maintain distinct spans,
 410                                 // so distinct syntax trees can be assigned the same span.
 411                                 // In such a case, the map cannot be trusted; so mark this
 412                                 // entry as unusable.
 413                                 occ.insert(None);
 414                             }
 415                         }
 416
 417                         // If the sequence contents can be empty, then the first
 418                         // token could be the separator token itself.
 419
 420                         if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 421                                                         subfirst.maybe_empty) {
 422                             first.add_one_maybe(TokenTree::Token(sp, sep.clone()));
 423                         }
 424
 425                         // Reverse scan: Sequence comes before `first`.
 426                         if subfirst.maybe_empty || seq_rep.op == quoted::KleeneOp::ZeroOrMore {
 427                             // If sequence is potentially empty, then
 428                             // union them (preserving first emptiness).
 429                             first.add_all(&TokenSet { maybe_empty: true, ..subfirst });
 430                         } else {
 431                             // Otherwise, sequence guaranteed
 432                             // non-empty; replace first.
 433                             first = subfirst;
 434                         }
 435                     }
 436                 }
 437             }
 438
 439             first
 440         }
 441     }
 442
 443     // walks forward over `tts` until all potential FIRST tokens are
 444     // identified.
 445     fn first(&self, tts: &[quoted::TokenTree]) -> TokenSet {
 446         use self::quoted::TokenTree;
 447
 448         let mut first = TokenSet::empty();
 449         for tt in tts.iter() {
 450             assert!(first.maybe_empty);
 451             match *tt {
 452                 TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 453                     first.add_one(tt.clone());
 454                     return first;
 455                 }
 456                 TokenTree::Delimited(span, ref delimited) => {
 457                     first.add_one(delimited.open_tt(span));
 458                     return first;
 459                 }
 460                 TokenTree::Sequence(sp, ref seq_rep) => {
 461                     match self.first.get(&sp) {
 462                         Some(&Some(ref subfirst)) => {
 463
 464                             // If the sequence contents can be empty, then the first
 465                             // token could be the separator token itself.
 466
 467                             if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 468                                                             subfirst.maybe_empty) {
 469                                 first.add_one_maybe(TokenTree::Token(sp, sep.clone()));
 470                             }
 471
 472                             assert!(first.maybe_empty);
 473                             first.add_all(subfirst);
 474                             if subfirst.maybe_empty ||
 475                                seq_rep.op == quoted::KleeneOp::ZeroOrMore {
 476                                 // continue scanning for more first
 477                                 // tokens, but also make sure we
 478                                 // restore empty-tracking state
 479                                 first.maybe_empty = true;
 480                                 continue;
 481                             } else {
 482                                 return first;
 483                             }
 484                         }
 485
 486                         Some(&None) => {
 487                             panic!("assume all sequences have (unique) spans for now");
 488                         }
 489
 490                         None => {
 491                             panic!("We missed a sequence during FirstSets construction");
 492                         }
 493                     }
 494                 }
 495             }
 496         }
 497
 498         // we only exit the loop if `tts` was empty or if every
 499         // element of `tts` matches the empty sequence.
 500         assert!(first.maybe_empty);
 501         first
 502     }
 503 }
 504
 505 // A set of `quoted::TokenTree`s, which may include `TokenTree::Match`s
 506 // (for macro-by-example syntactic variables). It also carries the
 507 // `maybe_empty` flag; that is true if and only if the matcher can
 508 // match an empty token sequence.
 509 //
 510 // The First set is computed on submatchers like `$($a:expr b),* $(c)* d`,
 511 // which has corresponding FIRST = {$a:expr, c, d}.
 512 // Likewise, `$($a:expr b),* $(c)+ d` has FIRST = {$a:expr, c}.
 513 //
 514 // (Notably, we must allow for *-op to occur zero times.)
 515 #[derive(Clone, Debug)]
 516 struct TokenSet {
 517     tokens: Vec<quoted::TokenTree>,
 518     maybe_empty: bool,
 519 }
 520
 521 impl TokenSet {
 522     // Returns a set for the empty sequence.
 523     fn empty() -> Self { TokenSet { tokens: Vec::new(), maybe_empty: true } }
 524
 525     // Returns the set `{ tok }` for the single-token (and thus
 526     // non-empty) sequence [tok].
 527     fn singleton(tok: quoted::TokenTree) -> Self {
 528         TokenSet { tokens: vec![tok], maybe_empty: false }
 529     }
 530
 531     // Changes self to be the set `{ tok }`.
 532     // Since `tok` is always present, marks self as non-empty.
 533     fn replace_with(&mut self, tok: quoted::TokenTree) {
 534         self.tokens.clear();
 535         self.tokens.push(tok);
 536         self.maybe_empty = false;
 537     }
 538
 539     // Changes self to be the empty set `{}`; meant for use when
 540     // the particular token does not matter, but we want to
 541     // record that it occurs.
 542     fn replace_with_irrelevant(&mut self) {
 543         self.tokens.clear();
 544         self.maybe_empty = false;
 545     }
 546
 547     // Adds `tok` to the set for `self`, marking sequence as non-empy.
 548     fn add_one(&mut self, tok: quoted::TokenTree) {
 549         if !self.tokens.contains(&tok) {
 550             self.tokens.push(tok);
 551         }
 552         self.maybe_empty = false;
 553     }
 554
 555     // Adds `tok` to the set for `self`. (Leaves `maybe_empty` flag alone.)
 556     fn add_one_maybe(&mut self, tok: quoted::TokenTree) {
 557         if !self.tokens.contains(&tok) {
 558             self.tokens.push(tok);
 559         }
 560     }
 561
 562     // Adds all elements of `other` to this.
 563     //
 564     // (Since this is a set, we filter out duplicates.)
 565     //
 566     // If `other` is potentially empty, then preserves the previous
 567     // setting of the empty flag of `self`. If `other` is guaranteed
 568     // non-empty, then `self` is marked non-empty.
 569     fn add_all(&mut self, other: &Self) {
 570         for tok in &other.tokens {
 571             if !self.tokens.contains(tok) {
 572                 self.tokens.push(tok.clone());
 573             }
 574         }
 575         if !other.maybe_empty {
 576             self.maybe_empty = false;
 577         }
 578     }
 579 }
 580
 581 // Checks that `matcher` is internally consistent and that it
 582 // can legally by followed by a token N, for all N in `follow`.
 583 // (If `follow` is empty, then it imposes no constraint on
 584 // the `matcher`.)
 585 //
 586 // Returns the set of NT tokens that could possibly come last in
 587 // `matcher`. (If `matcher` matches the empty sequence, then
 588 // `maybe_empty` will be set to true.)
 589 //
 590 // Requires that `first_sets` is pre-computed for `matcher`;
 591 // see `FirstSets::new`.
 592 fn check_matcher_core(sess: &ParseSess,
 593                       features: &RefCell<Features>,
 594                       attrs: &[ast::Attribute],
 595                       first_sets: &FirstSets,
 596                       matcher: &[quoted::TokenTree],
 597                       follow: &TokenSet) -> TokenSet {
 598     use self::quoted::TokenTree;
 599
 600     let mut last = TokenSet::empty();
 601
 602     // 2. For each token and suffix  [T, SUFFIX] in M:
 603     // ensure that T can be followed by SUFFIX, and if SUFFIX may be empty,
 604     // then ensure T can also be followed by any element of FOLLOW.
 605     'each_token: for i in 0..matcher.len() {
 606         let token = &matcher[i];
 607         let suffix = &matcher[i+1..];
 608
 609         let build_suffix_first = || {
 610             let mut s = first_sets.first(suffix);
 611             if s.maybe_empty { s.add_all(follow); }
 612             s
 613         };
 614
 615         // (we build `suffix_first` on demand below; you can tell
 616         // which cases are supposed to fall through by looking for the
 617         // initialization of this variable.)
 618         let suffix_first;
 619
 620         // First, update `last` so that it corresponds to the set
 621         // of NT tokens that might end the sequence `... token`.
 622         match *token {
 623             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 624                 let can_be_followed_by_any;
 625                 if let Err(bad_frag) = has_legal_fragment_specifier(sess, features, attrs, token) {
 626                     let msg = format!("invalid fragment specifier `{}`", bad_frag);
 627                     sess.span_diagnostic.struct_span_err(token.span(), &msg)
 628                         .help("valid fragment specifiers are `ident`, `block`, `stmt`, `expr`, \
 629                               `pat`, `ty`, `path`, `meta`, `tt`, `item` and `vis`")
 630                         .emit();
 631                     // (This eliminates false positives and duplicates
 632                     // from error messages.)
 633                     can_be_followed_by_any = true;
 634                 } else {
 635                     can_be_followed_by_any = token_can_be_followed_by_any(token);
 636                 }
 637
 638                 if can_be_followed_by_any {
 639                     // don't need to track tokens that work with any,
 640                     last.replace_with_irrelevant();
 641                     // ... and don't need to check tokens that can be
 642                     // followed by anything against SUFFIX.
 643                     continue 'each_token;
 644                 } else {
 645                     last.replace_with(token.clone());
 646                     suffix_first = build_suffix_first();
 647                 }
 648             }
 649             TokenTree::Delimited(span, ref d) => {
 650                 let my_suffix = TokenSet::singleton(d.close_tt(span));
 651                 check_matcher_core(sess, features, attrs, first_sets, &d.tts, &my_suffix);
 652                 // don't track non NT tokens
 653                 last.replace_with_irrelevant();
 654
 655                 // also, we don't need to check delimited sequences
 656                 // against SUFFIX
 657                 continue 'each_token;
 658             }
 659             TokenTree::Sequence(sp, ref seq_rep) => {
 660                 suffix_first = build_suffix_first();
 661                 // The trick here: when we check the interior, we want
 662                 // to include the separator (if any) as a potential
 663                 // (but not guaranteed) element of FOLLOW. So in that
 664                 // case, we make a temp copy of suffix and stuff
 665                 // delimiter in there.
 666                 //
 667                 // FIXME: Should I first scan suffix_first to see if
 668                 // delimiter is already in it before I go through the
 669                 // work of cloning it? But then again, this way I may
 670                 // get a "tighter" span?
 671                 let mut new;
 672                 let my_suffix = if let Some(ref u) = seq_rep.separator {
 673                     new = suffix_first.clone();
 674                     new.add_one_maybe(TokenTree::Token(sp, u.clone()));
 675                     &new
 676                 } else {
 677                     &suffix_first
 678                 };
 679
 680                 // At this point, `suffix_first` is built, and
 681                 // `my_suffix` is some TokenSet that we can use
 682                 // for checking the interior of `seq_rep`.
 683                 let next = check_matcher_core(sess,
 684                                               features,
 685                                               attrs,
 686                                               first_sets,
 687                                               &seq_rep.tts,
 688                                               my_suffix);
 689                 if next.maybe_empty {
 690                     last.add_all(&next);
 691                 } else {
 692                     last = next;
 693                 }
 694
 695                 // the recursive call to check_matcher_core already ran the 'each_last
 696                 // check below, so we can just keep going forward here.
 697                 continue 'each_token;
 698             }
 699         }
 700
 701         // (`suffix_first` guaranteed initialized once reaching here.)
 702
 703         // Now `last` holds the complete set of NT tokens that could
 704         // end the sequence before SUFFIX. Check that every one works with `suffix`.
 705         'each_last: for token in &last.tokens {
 706             if let TokenTree::MetaVarDecl(_, ref name, ref frag_spec) = *token {
 707                 for next_token in &suffix_first.tokens {
 708                     match is_in_follow(next_token, &frag_spec.name.as_str()) {
 709                         Err((msg, help)) => {
 710                             sess.span_diagnostic.struct_span_err(next_token.span(), &msg)
 711                                 .help(help).emit();
 712                             // don't bother reporting every source of
 713                             // conflict for a particular element of `last`.
 714                             continue 'each_last;
 715                         }
 716                         Ok(true) => {}
 717                         Ok(false) => {
 718                             let may_be = if last.tokens.len() == 1 &&
 719                                 suffix_first.tokens.len() == 1
 720                             {
 721                                 "is"
 722                             } else {
 723                                 "may be"
 724                             };
 725
 726                             sess.span_diagnostic.span_err(
 727                                 next_token.span(),
 728                                 &format!("`${name}:{frag}` {may_be} followed by `{next}`, which \
 729                                           is not allowed for `{frag}` fragments",
 730                                          name=name,
 731                                          frag=frag_spec,
 732                                          next=quoted_tt_to_string(next_token),
 733                                          may_be=may_be)
 734                             );
 735                         }
 736                     }
 737                 }
 738             }
 739         }
 740     }
 741     last
 742 }
 743
 744 fn token_can_be_followed_by_any(tok: &quoted::TokenTree) -> bool {
 745     if let quoted::TokenTree::MetaVarDecl(_, _, frag_spec) = *tok {
 746         frag_can_be_followed_by_any(&frag_spec.name.as_str())
 747     } else {
 748         // (Non NT's can always be followed by anthing in matchers.)
 749         true
 750     }
 751 }
 752
 753 /// True if a fragment of type `frag` can be followed by any sort of
 754 /// token.  We use this (among other things) as a useful approximation
 755 /// for when `frag` can be followed by a repetition like `$(...)*` or
 756 /// `$(...)+`. In general, these can be a bit tricky to reason about,
 757 /// so we adopt a conservative position that says that any fragment
 758 /// specifier which consumes at most one token tree can be followed by
 759 /// a fragment specifier (indeed, these fragments can be followed by
 760 /// ANYTHING without fear of future compatibility hazards).
 761 fn frag_can_be_followed_by_any(frag: &str) -> bool {
 762     match frag {
 763         "item"  | // always terminated by `}` or `;`
 764         "block" | // exactly one token tree
 765         "ident" | // exactly one token tree
 766         "meta"  | // exactly one token tree
 767         "tt" =>   // exactly one token tree
 768             true,
 769
 770         _ =>
 771             false,
 772     }
 773 }
 774
 775 /// True if `frag` can legally be followed by the token `tok`. For
 776 /// fragments that can consume an unbounded number of tokens, `tok`
 777 /// must be within a well-defined follow set. This is intended to
 778 /// guarantee future compatibility: for example, without this rule, if
 779 /// we expanded `expr` to include a new binary operator, we might
 780 /// break macros that were relying on that binary operator as a
 781 /// separator.
 782 // when changing this do not forget to update doc/book/macros.md!
 783 fn is_in_follow(tok: &quoted::TokenTree, frag: &str) -> Result<bool, (String, &'static str)> {
 784     use self::quoted::TokenTree;
 785
 786     if let TokenTree::Token(_, token::CloseDelim(_)) = *tok {
 787         // closing a token tree can never be matched by any fragment;
 788         // iow, we always require that `(` and `)` match, etc.
 789         Ok(true)
 790     } else {
 791         match frag {
 792             "item" => {
 793                 // since items *must* be followed by either a `;` or a `}`, we can
 794                 // accept anything after them
 795                 Ok(true)
 796             },
 797             "block" => {
 798                 // anything can follow block, the braces provide an easy boundary to
 799                 // maintain
 800                 Ok(true)
 801             },
 802             "stmt" | "expr"  => match *tok {
 803                 TokenTree::Token(_, ref tok) => match *tok {
 804                     FatArrow | Comma | Semi => Ok(true),
 805                     _ => Ok(false)
 806                 },
 807                 _ => Ok(false),
 808             },
 809             "pat" => match *tok {
 810                 TokenTree::Token(_, ref tok) => match *tok {
 811                     FatArrow | Comma | Eq | BinOp(token::Or) => Ok(true),
 812                     Ident(i) if i.name == "if" || i.name == "in" => Ok(true),
 813                     _ => Ok(false)
 814                 },
 815                 _ => Ok(false),
 816             },
 817             "path" | "ty" => match *tok {
 818                 TokenTree::Token(_, ref tok) => match *tok {
 819                     OpenDelim(token::DelimToken::Brace) | OpenDelim(token::DelimToken::Bracket) |
 820                     Comma | FatArrow | Colon | Eq | Gt | Semi | BinOp(token::Or) => Ok(true),
 821                     Ident(i) if i.name == "as" || i.name == "where" => Ok(true),
 822                     _ => Ok(false)
 823                 },
 824                 TokenTree::MetaVarDecl(_, _, frag) if frag.name == "block" => Ok(true),
 825                 _ => Ok(false),
 826             },
 827             "ident" => {
 828                 // being a single token, idents are harmless
 829                 Ok(true)
 830             },
 831             "meta" | "tt" => {
 832                 // being either a single token or a delimited sequence, tt is
 833                 // harmless
 834                 Ok(true)
 835             },
 836             "vis" => {
 837                 // Explicitly disallow `priv`, on the off chance it comes back.
 838                 match *tok {
 839                     TokenTree::Token(_, ref tok) => match *tok {
 840                         Comma => Ok(true),
 841                         Ident(i) if i.name != "priv" => Ok(true),
 842                         ref tok => Ok(tok.can_begin_type())
 843                     },
 844                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "ident"
 845                                                        || frag.name == "ty"
 846                                                        || frag.name == "path" => Ok(true),
 847                     _ => Ok(false)
 848                 }
 849             },
 850             "" => Ok(true), // keywords::Invalid
 851             _ => Err((format!("invalid fragment specifier `{}`", frag),
 852                      "valid fragment specifiers are `ident`, `block`, \
 853                       `stmt`, `expr`, `pat`, `ty`, `path`, `meta`, `tt`, \
 854                       `item` and `vis`"))
 855         }
 856     }
 857 }
 858
 859 fn has_legal_fragment_specifier(sess: &ParseSess,
 860                                 features: &RefCell<Features>,
 861                                 attrs: &[ast::Attribute],
 862                                 tok: &quoted::TokenTree) -> Result<(), String> {
 863     debug!("has_legal_fragment_specifier({:?})", tok);
 864     if let quoted::TokenTree::MetaVarDecl(_, _, ref frag_spec) = *tok {
 865         let frag_name = frag_spec.name.as_str();
 866         let frag_span = tok.span();
 867         if !is_legal_fragment_specifier(sess, features, attrs, &frag_name, frag_span) {
 868             return Err(frag_name.to_string());
 869         }
 870     }
 871     Ok(())
 872 }
 873
 874 fn is_legal_fragment_specifier(sess: &ParseSess,
 875                                features: &RefCell<Features>,
 876                                attrs: &[ast::Attribute],
 877                                frag_name: &str,
 878                                frag_span: Span) -> bool {
 879     match frag_name {
 880         "item" | "block" | "stmt" | "expr" | "pat" |
 881         "path" | "ty" | "ident" | "meta" | "tt" | "" => true,
 882         "vis" => {
 883             if     !features.borrow().macro_vis_matcher
 884                 && !attr::contains_name(attrs, "allow_internal_unstable") {
 885                 let explain = feature_gate::EXPLAIN_VIS_MATCHER;
 886                 emit_feature_err(sess,
 887                                  "macro_vis_matcher",
 888                                  frag_span,
 889                                  GateIssue::Language,
 890                                  explain);
 891             }
 892             true
 893         },
 894         _ => false,
 895     }
 896 }
 897
 898 fn quoted_tt_to_string(tt: &quoted::TokenTree) -> String {
 899     match *tt {
 900         quoted::TokenTree::Token(_, ref tok) => ::print::pprust::token_to_string(tok),
 901         quoted::TokenTree::MetaVar(_, name) => format!("${}", name),
 902         quoted::TokenTree::MetaVarDecl(_, name, kind) => format!("${}:{}", name, kind),
 903         _ => panic!("unexpected quoted::TokenTree::{{Sequence or Delimited}} \
 904                      in follow set checker"),
 905     }
 906 }