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 edition::Edition;
  14 use ext::base::{DummyResult, ExtCtxt, MacResult, SyntaxExtension};
  15 use ext::base::{NormalTT, TTMacroExpander};
  16 use ext::expand::{AstFragment, AstFragmentKind};
  17 use ext::tt::macro_parser::{Success, Error, Failure};
  18 use ext::tt::macro_parser::{MatchedSeq, MatchedNonterminal};
  19 use ext::tt::macro_parser::{parse, parse_failure_msg};
  20 use ext::tt::quoted;
  21 use ext::tt::transcribe::transcribe;
  22 use feature_gate::{self, emit_feature_err, Features, GateIssue};
  23 use parse::{Directory, ParseSess};
  24 use parse::parser::Parser;
  25 use parse::token::{self, NtTT};
  26 use parse::token::Token::*;
  27 use symbol::Symbol;
  28 use tokenstream::{TokenStream, TokenTree};
  29
  30 use std::borrow::Cow;
  31 use std::collections::HashMap;
  32 use std::collections::hash_map::Entry;
  33
  34 use rustc_data_structures::sync::Lrc;
  35
  36 pub struct ParserAnyMacro<'a> {
  37     parser: Parser<'a>,
  38
  39     /// Span of the expansion site of the macro this parser is for
  40     site_span: Span,
  41     /// The ident of the macro we're parsing
  42     macro_ident: ast::Ident
  43 }
  44
  45 impl<'a> ParserAnyMacro<'a> {
  46     pub fn make(mut self: Box<ParserAnyMacro<'a>>, kind: AstFragmentKind) -> AstFragment {
  47         let ParserAnyMacro { site_span, macro_ident, ref mut parser } = *self;
  48         let fragment = panictry!(parser.parse_ast_fragment(kind, true));
  49
  50         // We allow semicolons at the end of expressions -- e.g. the semicolon in
  51         // `macro_rules! m { () => { panic!(); } }` isn't parsed by `.parse_expr()`,
  52         // but `m!()` is allowed in expression positions (c.f. issue #34706).
  53         if kind == AstFragmentKind::Expr && parser.token == token::Semi {
  54             parser.bump();
  55         }
  56
  57         // Make sure we don't have any tokens left to parse so we don't silently drop anything.
  58         let path = ast::Path::from_ident(macro_ident.with_span_pos(site_span));
  59         parser.ensure_complete_parse(&path, kind.name(), site_span);
  60         fragment
  61     }
  62 }
  63
  64 struct MacroRulesMacroExpander {
  65     name: ast::Ident,
  66     lhses: Vec<quoted::TokenTree>,
  67     rhses: Vec<quoted::TokenTree>,
  68     valid: bool,
  69 }
  70
  71 impl TTMacroExpander for MacroRulesMacroExpander {
  72     fn expand<'cx>(&self,
  73                    cx: &'cx mut ExtCtxt,
  74                    sp: Span,
  75                    input: TokenStream)
  76                    -> Box<dyn MacResult+'cx> {
  77         if !self.valid {
  78             return DummyResult::any(sp);
  79         }
  80         generic_extension(cx,
  81                           sp,
  82                           self.name,
  83                           input,
  84                           &self.lhses,
  85                           &self.rhses)
  86     }
  87 }
  88
  89 fn trace_macros_note(cx: &mut ExtCtxt, sp: Span, message: String) {
  90     let sp = sp.macro_backtrace().last().map(|trace| trace.call_site).unwrap_or(sp);
  91     cx.expansions.entry(sp).or_default().push(message);
  92 }
  93
  94 /// Given `lhses` and `rhses`, this is the new macro we create
  95 fn generic_extension<'cx>(cx: &'cx mut ExtCtxt,
  96                           sp: Span,
  97                           name: ast::Ident,
  98                           arg: TokenStream,
  99                           lhses: &[quoted::TokenTree],
 100                           rhses: &[quoted::TokenTree])
 101                           -> Box<dyn MacResult+'cx> {
 102     if cx.trace_macros() {
 103         trace_macros_note(cx, sp, format!("expanding `{}! {{ {} }}`", name, arg));
 104     }
 105
 106     // Which arm's failure should we report? (the one furthest along)
 107     let mut best_fail_spot = DUMMY_SP;
 108     let mut best_fail_tok = None;
 109
 110     for (i, lhs) in lhses.iter().enumerate() { // try each arm's matchers
 111         let lhs_tt = match *lhs {
 112             quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 113             _ => cx.span_bug(sp, "malformed macro lhs")
 114         };
 115
 116         match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 117             Success(named_matches) => {
 118                 let rhs = match rhses[i] {
 119                     // ignore delimiters
 120                     quoted::TokenTree::Delimited(_, ref delimed) => delimed.tts.clone(),
 121                     _ => cx.span_bug(sp, "malformed macro rhs"),
 122                 };
 123
 124                 let rhs_spans = rhs.iter().map(|t| t.span()).collect::<Vec<_>>();
 125                 // rhs has holes ( `$id` and `$(...)` that need filled)
 126                 let mut tts = transcribe(cx, Some(named_matches), rhs);
 127
 128                 // Replace all the tokens for the corresponding positions in the macro, to maintain
 129                 // proper positions in error reporting, while maintaining the macro_backtrace.
 130                 if rhs_spans.len() == tts.len() {
 131                     tts = tts.map_enumerated(|i, tt| {
 132                         let mut tt = tt.clone();
 133                         let mut sp = rhs_spans[i];
 134                         sp = sp.with_ctxt(tt.span().ctxt());
 135                         tt.set_span(sp);
 136                         tt
 137                     });
 138                 }
 139
 140                 if cx.trace_macros() {
 141                     trace_macros_note(cx, sp, format!("to `{}`", tts));
 142                 }
 143
 144                 let directory = Directory {
 145                     path: Cow::from(cx.current_expansion.module.directory.as_path()),
 146                     ownership: cx.current_expansion.directory_ownership,
 147                 };
 148                 let mut p = Parser::new(cx.parse_sess(), tts, Some(directory), true, false);
 149                 p.root_module_name = cx.current_expansion.module.mod_path.last()
 150                     .map(|id| id.as_str().to_string());
 151
 152                 p.process_potential_macro_variable();
 153                 // Let the context choose how to interpret the result.
 154                 // Weird, but useful for X-macros.
 155                 return Box::new(ParserAnyMacro {
 156                     parser: p,
 157
 158                     // Pass along the original expansion site and the name of the macro
 159                     // so we can print a useful error message if the parse of the expanded
 160                     // macro leaves unparsed tokens.
 161                     site_span: sp,
 162                     macro_ident: name
 163                 })
 164             }
 165             Failure(sp, tok) => if sp.lo() >= best_fail_spot.lo() {
 166                 best_fail_spot = sp;
 167                 best_fail_tok = Some(tok);
 168             },
 169             Error(err_sp, ref msg) => {
 170                 cx.span_fatal(err_sp.substitute_dummy(sp), &msg[..])
 171             }
 172         }
 173     }
 174
 175     let best_fail_msg = parse_failure_msg(best_fail_tok.expect("ran no matchers"));
 176     let mut err = cx.struct_span_err(best_fail_spot.substitute_dummy(sp), &best_fail_msg);
 177
 178     // Check whether there's a missing comma in this macro call, like `println!("{}" a);`
 179     if let Some((arg, comma_span)) = arg.add_comma() {
 180         for lhs in lhses { // try each arm's matchers
 181             let lhs_tt = match *lhs {
 182                 quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 183                 _ => continue,
 184             };
 185             match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 186                 Success(_) => {
 187                     if comma_span == DUMMY_SP {
 188                         err.note("you might be missing a comma");
 189                     } else {
 190                         err.span_suggestion_short(
 191                             comma_span,
 192                             "missing comma here",
 193                             ", ".to_string(),
 194                         );
 195                     }
 196                 }
 197                 _ => {}
 198             }
 199         }
 200     }
 201     err.emit();
 202     cx.trace_macros_diag();
 203     DummyResult::any(sp)
 204 }
 205
 206 // Note that macro-by-example's input is also matched against a token tree:
 207 //                   $( $lhs:tt => $rhs:tt );+
 208 //
 209 // Holy self-referential!
 210
 211 /// Converts a `macro_rules!` invocation into a syntax extension.
 212 pub fn compile(sess: &ParseSess, features: &Features, def: &ast::Item, edition: Edition)
 213                -> SyntaxExtension {
 214     let lhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("lhs"));
 215     let rhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("rhs"));
 216
 217     // Parse the macro_rules! invocation
 218     let body = match def.node {
 219         ast::ItemKind::MacroDef(ref body) => body,
 220         _ => unreachable!(),
 221     };
 222
 223     // The pattern that macro_rules matches.
 224     // The grammar for macro_rules! is:
 225     // $( $lhs:tt => $rhs:tt );+
 226     // ...quasiquoting this would be nice.
 227     // These spans won't matter, anyways
 228     let argument_gram = vec![
 229         quoted::TokenTree::Sequence(DUMMY_SP, Lrc::new(quoted::SequenceRepetition {
 230             tts: vec![
 231                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, lhs_nm, ast::Ident::from_str("tt")),
 232                 quoted::TokenTree::Token(DUMMY_SP, token::FatArrow),
 233                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, rhs_nm, ast::Ident::from_str("tt")),
 234             ],
 235             separator: Some(if body.legacy { token::Semi } else { token::Comma }),
 236             op: quoted::KleeneOp::OneOrMore,
 237             num_captures: 2,
 238         })),
 239         // to phase into semicolon-termination instead of semicolon-separation
 240         quoted::TokenTree::Sequence(DUMMY_SP, Lrc::new(quoted::SequenceRepetition {
 241             tts: vec![quoted::TokenTree::Token(DUMMY_SP, token::Semi)],
 242             separator: None,
 243             op: quoted::KleeneOp::ZeroOrMore,
 244             num_captures: 0
 245         })),
 246     ];
 247
 248     let argument_map = match parse(sess, body.stream(), &argument_gram, None, true) {
 249         Success(m) => m,
 250         Failure(sp, tok) => {
 251             let s = parse_failure_msg(tok);
 252             sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s).raise();
 253         }
 254         Error(sp, s) => {
 255             sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s).raise();
 256         }
 257     };
 258
 259     let mut valid = true;
 260
 261     // Extract the arguments:
 262     let lhses = match *argument_map[&lhs_nm] {
 263         MatchedSeq(ref s, _) => {
 264             s.iter().map(|m| {
 265                 if let MatchedNonterminal(ref nt) = *m {
 266                     if let NtTT(ref tt) = **nt {
 267                         let tt = quoted::parse(
 268                             tt.clone().into(),
 269                             true,
 270                             sess,
 271                             features,
 272                             &def.attrs,
 273                             edition,
 274                             def.id,
 275                         )
 276                         .pop()
 277                         .unwrap();
 278                         valid &= check_lhs_nt_follows(sess, features, &def.attrs, &tt);
 279                         return tt;
 280                     }
 281                 }
 282                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 283             }).collect::<Vec<quoted::TokenTree>>()
 284         }
 285         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 286     };
 287
 288     let rhses = match *argument_map[&rhs_nm] {
 289         MatchedSeq(ref s, _) => {
 290             s.iter().map(|m| {
 291                 if let MatchedNonterminal(ref nt) = *m {
 292                     if let NtTT(ref tt) = **nt {
 293                         return quoted::parse(
 294                             tt.clone().into(),
 295                             false,
 296                             sess,
 297                             features,
 298                             &def.attrs,
 299                             edition,
 300                             def.id,
 301                         ).pop()
 302                          .unwrap();
 303                     }
 304                 }
 305                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 306             }).collect::<Vec<quoted::TokenTree>>()
 307         }
 308         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured rhs")
 309     };
 310
 311     for rhs in &rhses {
 312         valid &= check_rhs(sess, rhs);
 313     }
 314
 315     // don't abort iteration early, so that errors for multiple lhses can be reported
 316     for lhs in &lhses {
 317         valid &= check_lhs_no_empty_seq(sess, &[lhs.clone()])
 318     }
 319
 320     let expander: Box<_> = Box::new(MacroRulesMacroExpander {
 321         name: def.ident,
 322         lhses,
 323         rhses,
 324         valid,
 325     });
 326
 327     if body.legacy {
 328         let allow_internal_unstable = attr::contains_name(&def.attrs, "allow_internal_unstable");
 329         let allow_internal_unsafe = attr::contains_name(&def.attrs, "allow_internal_unsafe");
 330         let mut local_inner_macros = false;
 331         if let Some(macro_export) = attr::find_by_name(&def.attrs, "macro_export") {
 332             if let Some(l) = macro_export.meta_item_list() {
 333                 local_inner_macros = attr::list_contains_name(&l, "local_inner_macros");
 334             }
 335         }
 336
 337         let unstable_feature = attr::find_stability(&sess.span_diagnostic,
 338                                                     &def.attrs, def.span).and_then(|stability| {
 339             if let attr::StabilityLevel::Unstable { issue, .. } = stability.level {
 340                 Some((stability.feature, issue))
 341             } else {
 342                 None
 343             }
 344         });
 345
 346         NormalTT {
 347             expander,
 348             def_info: Some((def.id, def.span)),
 349             allow_internal_unstable,
 350             allow_internal_unsafe,
 351             local_inner_macros,
 352             unstable_feature,
 353             edition,
 354         }
 355     } else {
 356         let is_transparent = attr::contains_name(&def.attrs, "rustc_transparent_macro");
 357
 358         SyntaxExtension::DeclMacro {
 359             expander,
 360             def_info: Some((def.id, def.span)),
 361             is_transparent,
 362             edition,
 363         }
 364     }
 365 }
 366
 367 fn check_lhs_nt_follows(sess: &ParseSess,
 368                         features: &Features,
 369                         attrs: &[ast::Attribute],
 370                         lhs: &quoted::TokenTree) -> bool {
 371     // lhs is going to be like TokenTree::Delimited(...), where the
 372     // entire lhs is those tts. Or, it can be a "bare sequence", not wrapped in parens.
 373     if let quoted::TokenTree::Delimited(_, ref tts) = *lhs {
 374         check_matcher(sess, features, attrs, &tts.tts)
 375     } else {
 376         let msg = "invalid macro matcher; matchers must be contained in balanced delimiters";
 377         sess.span_diagnostic.span_err(lhs.span(), msg);
 378         false
 379     }
 380     // we don't abort on errors on rejection, the driver will do that for us
 381     // after parsing/expansion. we can report every error in every macro this way.
 382 }
 383
 384 /// Check that the lhs contains no repetition which could match an empty token
 385 /// tree, because then the matcher would hang indefinitely.
 386 fn check_lhs_no_empty_seq(sess: &ParseSess, tts: &[quoted::TokenTree]) -> bool {
 387     use self::quoted::TokenTree;
 388     for tt in tts {
 389         match *tt {
 390             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => (),
 391             TokenTree::Delimited(_, ref del) => if !check_lhs_no_empty_seq(sess, &del.tts) {
 392                 return false;
 393             },
 394             TokenTree::Sequence(span, ref seq) => {
 395                 if seq.separator.is_none() && seq.tts.iter().all(|seq_tt| {
 396                     match *seq_tt {
 397                         TokenTree::MetaVarDecl(_, _, id) => id.name == "vis",
 398                         TokenTree::Sequence(_, ref sub_seq) =>
 399                             sub_seq.op == quoted::KleeneOp::ZeroOrMore,
 400                         _ => false,
 401                     }
 402                 }) {
 403                     sess.span_diagnostic.span_err(span, "repetition matches empty token tree");
 404                     return false;
 405                 }
 406                 if !check_lhs_no_empty_seq(sess, &seq.tts) {
 407                     return false;
 408                 }
 409             }
 410         }
 411     }
 412
 413     true
 414 }
 415
 416 fn check_rhs(sess: &ParseSess, rhs: &quoted::TokenTree) -> bool {
 417     match *rhs {
 418         quoted::TokenTree::Delimited(..) => return true,
 419         _ => sess.span_diagnostic.span_err(rhs.span(), "macro rhs must be delimited")
 420     }
 421     false
 422 }
 423
 424 fn check_matcher(sess: &ParseSess,
 425                  features: &Features,
 426                  attrs: &[ast::Attribute],
 427                  matcher: &[quoted::TokenTree]) -> bool {
 428     let first_sets = FirstSets::new(matcher);
 429     let empty_suffix = TokenSet::empty();
 430     let err = sess.span_diagnostic.err_count();
 431     check_matcher_core(sess, features, attrs, &first_sets, matcher, &empty_suffix);
 432     err == sess.span_diagnostic.err_count()
 433 }
 434
 435 // The FirstSets for a matcher is a mapping from subsequences in the
 436 // matcher to the FIRST set for that subsequence.
 437 //
 438 // This mapping is partially precomputed via a backwards scan over the
 439 // token trees of the matcher, which provides a mapping from each
 440 // repetition sequence to its FIRST set.
 441 //
 442 // (Hypothetically sequences should be uniquely identifiable via their
 443 // spans, though perhaps that is false e.g. for macro-generated macros
 444 // that do not try to inject artificial span information. My plan is
 445 // to try to catch such cases ahead of time and not include them in
 446 // the precomputed mapping.)
 447 struct FirstSets {
 448     // this maps each TokenTree::Sequence `$(tt ...) SEP OP` that is uniquely identified by its
 449     // span in the original matcher to the First set for the inner sequence `tt ...`.
 450     //
 451     // If two sequences have the same span in a matcher, then map that
 452     // span to None (invalidating the mapping here and forcing the code to
 453     // use a slow path).
 454     first: HashMap<Span, Option<TokenSet>>,
 455 }
 456
 457 impl FirstSets {
 458     fn new(tts: &[quoted::TokenTree]) -> FirstSets {
 459         use self::quoted::TokenTree;
 460
 461         let mut sets = FirstSets { first: HashMap::new() };
 462         build_recur(&mut sets, tts);
 463         return sets;
 464
 465         // walks backward over `tts`, returning the FIRST for `tts`
 466         // and updating `sets` at the same time for all sequence
 467         // substructure we find within `tts`.
 468         fn build_recur(sets: &mut FirstSets, tts: &[TokenTree]) -> TokenSet {
 469             let mut first = TokenSet::empty();
 470             for tt in tts.iter().rev() {
 471                 match *tt {
 472                     TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 473                         first.replace_with(tt.clone());
 474                     }
 475                     TokenTree::Delimited(span, ref delimited) => {
 476                         build_recur(sets, &delimited.tts[..]);
 477                         first.replace_with(delimited.open_tt(span));
 478                     }
 479                     TokenTree::Sequence(sp, ref seq_rep) => {
 480                         let subfirst = build_recur(sets, &seq_rep.tts[..]);
 481
 482                         match sets.first.entry(sp) {
 483                             Entry::Vacant(vac) => {
 484                                 vac.insert(Some(subfirst.clone()));
 485                             }
 486                             Entry::Occupied(mut occ) => {
 487                                 // if there is already an entry, then a span must have collided.
 488                                 // This should not happen with typical macro_rules macros,
 489                                 // but syntax extensions need not maintain distinct spans,
 490                                 // so distinct syntax trees can be assigned the same span.
 491                                 // In such a case, the map cannot be trusted; so mark this
 492                                 // entry as unusable.
 493                                 occ.insert(None);
 494                             }
 495                         }
 496
 497                         // If the sequence contents can be empty, then the first
 498                         // token could be the separator token itself.
 499
 500                         if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 501                                                         subfirst.maybe_empty) {
 502                             first.add_one_maybe(TokenTree::Token(sp, sep.clone()));
 503                         }
 504
 505                         // Reverse scan: Sequence comes before `first`.
 506                         if subfirst.maybe_empty || seq_rep.op == quoted::KleeneOp::ZeroOrMore {
 507                             // If sequence is potentially empty, then
 508                             // union them (preserving first emptiness).
 509                             first.add_all(&TokenSet { maybe_empty: true, ..subfirst });
 510                         } else {
 511                             // Otherwise, sequence guaranteed
 512                             // non-empty; replace first.
 513                             first = subfirst;
 514                         }
 515                     }
 516                 }
 517             }
 518
 519             first
 520         }
 521     }
 522
 523     // walks forward over `tts` until all potential FIRST tokens are
 524     // identified.
 525     fn first(&self, tts: &[quoted::TokenTree]) -> TokenSet {
 526         use self::quoted::TokenTree;
 527
 528         let mut first = TokenSet::empty();
 529         for tt in tts.iter() {
 530             assert!(first.maybe_empty);
 531             match *tt {
 532                 TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 533                     first.add_one(tt.clone());
 534                     return first;
 535                 }
 536                 TokenTree::Delimited(span, ref delimited) => {
 537                     first.add_one(delimited.open_tt(span));
 538                     return first;
 539                 }
 540                 TokenTree::Sequence(sp, ref seq_rep) => {
 541                     match self.first.get(&sp) {
 542                         Some(&Some(ref subfirst)) => {
 543
 544                             // If the sequence contents can be empty, then the first
 545                             // token could be the separator token itself.
 546
 547                             if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 548                                                             subfirst.maybe_empty) {
 549                                 first.add_one_maybe(TokenTree::Token(sp, sep.clone()));
 550                             }
 551
 552                             assert!(first.maybe_empty);
 553                             first.add_all(subfirst);
 554                             if subfirst.maybe_empty ||
 555                                seq_rep.op == quoted::KleeneOp::ZeroOrMore {
 556                                 // continue scanning for more first
 557                                 // tokens, but also make sure we
 558                                 // restore empty-tracking state
 559                                 first.maybe_empty = true;
 560                                 continue;
 561                             } else {
 562                                 return first;
 563                             }
 564                         }
 565
 566                         Some(&None) => {
 567                             panic!("assume all sequences have (unique) spans for now");
 568                         }
 569
 570                         None => {
 571                             panic!("We missed a sequence during FirstSets construction");
 572                         }
 573                     }
 574                 }
 575             }
 576         }
 577
 578         // we only exit the loop if `tts` was empty or if every
 579         // element of `tts` matches the empty sequence.
 580         assert!(first.maybe_empty);
 581         first
 582     }
 583 }
 584
 585 // A set of `quoted::TokenTree`s, which may include `TokenTree::Match`s
 586 // (for macro-by-example syntactic variables). It also carries the
 587 // `maybe_empty` flag; that is true if and only if the matcher can
 588 // match an empty token sequence.
 589 //
 590 // The First set is computed on submatchers like `$($a:expr b),* $(c)* d`,
 591 // which has corresponding FIRST = {$a:expr, c, d}.
 592 // Likewise, `$($a:expr b),* $(c)+ d` has FIRST = {$a:expr, c}.
 593 //
 594 // (Notably, we must allow for *-op to occur zero times.)
 595 #[derive(Clone, Debug)]
 596 struct TokenSet {
 597     tokens: Vec<quoted::TokenTree>,
 598     maybe_empty: bool,
 599 }
 600
 601 impl TokenSet {
 602     // Returns a set for the empty sequence.
 603     fn empty() -> Self { TokenSet { tokens: Vec::new(), maybe_empty: true } }
 604
 605     // Returns the set `{ tok }` for the single-token (and thus
 606     // non-empty) sequence [tok].
 607     fn singleton(tok: quoted::TokenTree) -> Self {
 608         TokenSet { tokens: vec![tok], maybe_empty: false }
 609     }
 610
 611     // Changes self to be the set `{ tok }`.
 612     // Since `tok` is always present, marks self as non-empty.
 613     fn replace_with(&mut self, tok: quoted::TokenTree) {
 614         self.tokens.clear();
 615         self.tokens.push(tok);
 616         self.maybe_empty = false;
 617     }
 618
 619     // Changes self to be the empty set `{}`; meant for use when
 620     // the particular token does not matter, but we want to
 621     // record that it occurs.
 622     fn replace_with_irrelevant(&mut self) {
 623         self.tokens.clear();
 624         self.maybe_empty = false;
 625     }
 626
 627     // Adds `tok` to the set for `self`, marking sequence as non-empy.
 628     fn add_one(&mut self, tok: quoted::TokenTree) {
 629         if !self.tokens.contains(&tok) {
 630             self.tokens.push(tok);
 631         }
 632         self.maybe_empty = false;
 633     }
 634
 635     // Adds `tok` to the set for `self`. (Leaves `maybe_empty` flag alone.)
 636     fn add_one_maybe(&mut self, tok: quoted::TokenTree) {
 637         if !self.tokens.contains(&tok) {
 638             self.tokens.push(tok);
 639         }
 640     }
 641
 642     // Adds all elements of `other` to this.
 643     //
 644     // (Since this is a set, we filter out duplicates.)
 645     //
 646     // If `other` is potentially empty, then preserves the previous
 647     // setting of the empty flag of `self`. If `other` is guaranteed
 648     // non-empty, then `self` is marked non-empty.
 649     fn add_all(&mut self, other: &Self) {
 650         for tok in &other.tokens {
 651             if !self.tokens.contains(tok) {
 652                 self.tokens.push(tok.clone());
 653             }
 654         }
 655         if !other.maybe_empty {
 656             self.maybe_empty = false;
 657         }
 658     }
 659 }
 660
 661 // Checks that `matcher` is internally consistent and that it
 662 // can legally by followed by a token N, for all N in `follow`.
 663 // (If `follow` is empty, then it imposes no constraint on
 664 // the `matcher`.)
 665 //
 666 // Returns the set of NT tokens that could possibly come last in
 667 // `matcher`. (If `matcher` matches the empty sequence, then
 668 // `maybe_empty` will be set to true.)
 669 //
 670 // Requires that `first_sets` is pre-computed for `matcher`;
 671 // see `FirstSets::new`.
 672 fn check_matcher_core(sess: &ParseSess,
 673                       features: &Features,
 674                       attrs: &[ast::Attribute],
 675                       first_sets: &FirstSets,
 676                       matcher: &[quoted::TokenTree],
 677                       follow: &TokenSet) -> TokenSet {
 678     use self::quoted::TokenTree;
 679
 680     let mut last = TokenSet::empty();
 681
 682     // 2. For each token and suffix  [T, SUFFIX] in M:
 683     // ensure that T can be followed by SUFFIX, and if SUFFIX may be empty,
 684     // then ensure T can also be followed by any element of FOLLOW.
 685     'each_token: for i in 0..matcher.len() {
 686         let token = &matcher[i];
 687         let suffix = &matcher[i+1..];
 688
 689         let build_suffix_first = || {
 690             let mut s = first_sets.first(suffix);
 691             if s.maybe_empty { s.add_all(follow); }
 692             s
 693         };
 694
 695         // (we build `suffix_first` on demand below; you can tell
 696         // which cases are supposed to fall through by looking for the
 697         // initialization of this variable.)
 698         let suffix_first;
 699
 700         // First, update `last` so that it corresponds to the set
 701         // of NT tokens that might end the sequence `... token`.
 702         match *token {
 703             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 704                 let can_be_followed_by_any;
 705                 if let Err(bad_frag) = has_legal_fragment_specifier(sess, features, attrs, token) {
 706                     let msg = format!("invalid fragment specifier `{}`", bad_frag);
 707                     sess.span_diagnostic.struct_span_err(token.span(), &msg)
 708                         .help("valid fragment specifiers are `ident`, `block`, `stmt`, `expr`, \
 709                               `pat`, `ty`, `literal`, `path`, `meta`, `tt`, `item` and `vis`")
 710                         .emit();
 711                     // (This eliminates false positives and duplicates
 712                     // from error messages.)
 713                     can_be_followed_by_any = true;
 714                 } else {
 715                     can_be_followed_by_any = token_can_be_followed_by_any(token);
 716                 }
 717
 718                 if can_be_followed_by_any {
 719                     // don't need to track tokens that work with any,
 720                     last.replace_with_irrelevant();
 721                     // ... and don't need to check tokens that can be
 722                     // followed by anything against SUFFIX.
 723                     continue 'each_token;
 724                 } else {
 725                     last.replace_with(token.clone());
 726                     suffix_first = build_suffix_first();
 727                 }
 728             }
 729             TokenTree::Delimited(span, ref d) => {
 730                 let my_suffix = TokenSet::singleton(d.close_tt(span));
 731                 check_matcher_core(sess, features, attrs, first_sets, &d.tts, &my_suffix);
 732                 // don't track non NT tokens
 733                 last.replace_with_irrelevant();
 734
 735                 // also, we don't need to check delimited sequences
 736                 // against SUFFIX
 737                 continue 'each_token;
 738             }
 739             TokenTree::Sequence(sp, ref seq_rep) => {
 740                 suffix_first = build_suffix_first();
 741                 // The trick here: when we check the interior, we want
 742                 // to include the separator (if any) as a potential
 743                 // (but not guaranteed) element of FOLLOW. So in that
 744                 // case, we make a temp copy of suffix and stuff
 745                 // delimiter in there.
 746                 //
 747                 // FIXME: Should I first scan suffix_first to see if
 748                 // delimiter is already in it before I go through the
 749                 // work of cloning it? But then again, this way I may
 750                 // get a "tighter" span?
 751                 let mut new;
 752                 let my_suffix = if let Some(ref u) = seq_rep.separator {
 753                     new = suffix_first.clone();
 754                     new.add_one_maybe(TokenTree::Token(sp, u.clone()));
 755                     &new
 756                 } else {
 757                     &suffix_first
 758                 };
 759
 760                 // At this point, `suffix_first` is built, and
 761                 // `my_suffix` is some TokenSet that we can use
 762                 // for checking the interior of `seq_rep`.
 763                 let next = check_matcher_core(sess,
 764                                               features,
 765                                               attrs,
 766                                               first_sets,
 767                                               &seq_rep.tts,
 768                                               my_suffix);
 769                 if next.maybe_empty {
 770                     last.add_all(&next);
 771                 } else {
 772                     last = next;
 773                 }
 774
 775                 // the recursive call to check_matcher_core already ran the 'each_last
 776                 // check below, so we can just keep going forward here.
 777                 continue 'each_token;
 778             }
 779         }
 780
 781         // (`suffix_first` guaranteed initialized once reaching here.)
 782
 783         // Now `last` holds the complete set of NT tokens that could
 784         // end the sequence before SUFFIX. Check that every one works with `suffix`.
 785         'each_last: for token in &last.tokens {
 786             if let TokenTree::MetaVarDecl(_, ref name, ref frag_spec) = *token {
 787                 for next_token in &suffix_first.tokens {
 788                     match is_in_follow(next_token, &frag_spec.as_str()) {
 789                         Err((msg, help)) => {
 790                             sess.span_diagnostic.struct_span_err(next_token.span(), &msg)
 791                                 .help(help).emit();
 792                             // don't bother reporting every source of
 793                             // conflict for a particular element of `last`.
 794                             continue 'each_last;
 795                         }
 796                         Ok(true) => {}
 797                         Ok(false) => {
 798                             let may_be = if last.tokens.len() == 1 &&
 799                                 suffix_first.tokens.len() == 1
 800                             {
 801                                 "is"
 802                             } else {
 803                                 "may be"
 804                             };
 805
 806                             sess.span_diagnostic.span_err(
 807                                 next_token.span(),
 808                                 &format!("`${name}:{frag}` {may_be} followed by `{next}`, which \
 809                                           is not allowed for `{frag}` fragments",
 810                                          name=name,
 811                                          frag=frag_spec,
 812                                          next=quoted_tt_to_string(next_token),
 813                                          may_be=may_be)
 814                             );
 815                         }
 816                     }
 817                 }
 818             }
 819         }
 820     }
 821     last
 822 }
 823
 824 fn token_can_be_followed_by_any(tok: &quoted::TokenTree) -> bool {
 825     if let quoted::TokenTree::MetaVarDecl(_, _, frag_spec) = *tok {
 826         frag_can_be_followed_by_any(&frag_spec.as_str())
 827     } else {
 828         // (Non NT's can always be followed by anthing in matchers.)
 829         true
 830     }
 831 }
 832
 833 /// True if a fragment of type `frag` can be followed by any sort of
 834 /// token.  We use this (among other things) as a useful approximation
 835 /// for when `frag` can be followed by a repetition like `$(...)*` or
 836 /// `$(...)+`. In general, these can be a bit tricky to reason about,
 837 /// so we adopt a conservative position that says that any fragment
 838 /// specifier which consumes at most one token tree can be followed by
 839 /// a fragment specifier (indeed, these fragments can be followed by
 840 /// ANYTHING without fear of future compatibility hazards).
 841 fn frag_can_be_followed_by_any(frag: &str) -> bool {
 842     match frag {
 843         "item"     | // always terminated by `}` or `;`
 844         "block"    | // exactly one token tree
 845         "ident"    | // exactly one token tree
 846         "literal"  | // exactly one token tree
 847         "meta"     | // exactly one token tree
 848         "lifetime" | // exactly one token tree
 849         "tt" =>   // exactly one token tree
 850             true,
 851
 852         _ =>
 853             false,
 854     }
 855 }
 856
 857 /// True if `frag` can legally be followed by the token `tok`. For
 858 /// fragments that can consume an unbounded number of tokens, `tok`
 859 /// must be within a well-defined follow set. This is intended to
 860 /// guarantee future compatibility: for example, without this rule, if
 861 /// we expanded `expr` to include a new binary operator, we might
 862 /// break macros that were relying on that binary operator as a
 863 /// separator.
 864 // when changing this do not forget to update doc/book/macros.md!
 865 fn is_in_follow(tok: &quoted::TokenTree, frag: &str) -> Result<bool, (String, &'static str)> {
 866     use self::quoted::TokenTree;
 867
 868     if let TokenTree::Token(_, token::CloseDelim(_)) = *tok {
 869         // closing a token tree can never be matched by any fragment;
 870         // iow, we always require that `(` and `)` match, etc.
 871         Ok(true)
 872     } else {
 873         match frag {
 874             "item" => {
 875                 // since items *must* be followed by either a `;` or a `}`, we can
 876                 // accept anything after them
 877                 Ok(true)
 878             },
 879             "block" => {
 880                 // anything can follow block, the braces provide an easy boundary to
 881                 // maintain
 882                 Ok(true)
 883             },
 884             "stmt" | "expr"  => match *tok {
 885                 TokenTree::Token(_, ref tok) => match *tok {
 886                     FatArrow | Comma | Semi => Ok(true),
 887                     _ => Ok(false)
 888                 },
 889                 _ => Ok(false),
 890             },
 891             "pat" => match *tok {
 892                 TokenTree::Token(_, ref tok) => match *tok {
 893                     FatArrow | Comma | Eq | BinOp(token::Or) => Ok(true),
 894                     Ident(i, false) if i.name == "if" || i.name == "in" => Ok(true),
 895                     _ => Ok(false)
 896                 },
 897                 _ => Ok(false),
 898             },
 899             "path" | "ty" => match *tok {
 900                 TokenTree::Token(_, ref tok) => match *tok {
 901                     OpenDelim(token::DelimToken::Brace) | OpenDelim(token::DelimToken::Bracket) |
 902                     Comma | FatArrow | Colon | Eq | Gt | Semi | BinOp(token::Or) => Ok(true),
 903                     Ident(i, false) if i.name == "as" || i.name == "where" => Ok(true),
 904                     _ => Ok(false)
 905                 },
 906                 TokenTree::MetaVarDecl(_, _, frag) if frag.name == "block" => Ok(true),
 907                 _ => Ok(false),
 908             },
 909             "ident" | "lifetime" => {
 910                 // being a single token, idents and lifetimes are harmless
 911                 Ok(true)
 912             },
 913             "literal" => {
 914                 // literals may be of a single token, or two tokens (negative numbers)
 915                 Ok(true)
 916             },
 917             "meta" | "tt" => {
 918                 // being either a single token or a delimited sequence, tt is
 919                 // harmless
 920                 Ok(true)
 921             },
 922             "vis" => {
 923                 // Explicitly disallow `priv`, on the off chance it comes back.
 924                 match *tok {
 925                     TokenTree::Token(_, ref tok) => match *tok {
 926                         Comma => Ok(true),
 927                         Ident(i, is_raw) if is_raw || i.name != "priv" => Ok(true),
 928                         ref tok => Ok(tok.can_begin_type())
 929                     },
 930                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "ident"
 931                                                        || frag.name == "ty"
 932                                                        || frag.name == "path" => Ok(true),
 933                     _ => Ok(false)
 934                 }
 935             },
 936             "" => Ok(true), // keywords::Invalid
 937             _ => Err((format!("invalid fragment specifier `{}`", frag),
 938                      "valid fragment specifiers are `ident`, `block`, \
 939                       `stmt`, `expr`, `pat`, `ty`, `path`, `meta`, `tt`, \
 940                       `literal`, `item` and `vis`"))
 941         }
 942     }
 943 }
 944
 945 fn has_legal_fragment_specifier(sess: &ParseSess,
 946                                 features: &Features,
 947                                 attrs: &[ast::Attribute],
 948                                 tok: &quoted::TokenTree) -> Result<(), String> {
 949     debug!("has_legal_fragment_specifier({:?})", tok);
 950     if let quoted::TokenTree::MetaVarDecl(_, _, ref frag_spec) = *tok {
 951         let frag_name = frag_spec.as_str();
 952         let frag_span = tok.span();
 953         if !is_legal_fragment_specifier(sess, features, attrs, &frag_name, frag_span) {
 954             return Err(frag_name.to_string());
 955         }
 956     }
 957     Ok(())
 958 }
 959
 960 fn is_legal_fragment_specifier(sess: &ParseSess,
 961                                features: &Features,
 962                                attrs: &[ast::Attribute],
 963                                frag_name: &str,
 964                                frag_span: Span) -> bool {
 965     match frag_name {
 966         "item" | "block" | "stmt" | "expr" | "pat" | "lifetime" |
 967         "path" | "ty" | "ident" | "meta" | "tt" | "" => true,
 968         "literal" => {
 969             if !features.macro_literal_matcher &&
 970                !attr::contains_name(attrs, "allow_internal_unstable") {
 971                 let explain = feature_gate::EXPLAIN_LITERAL_MATCHER;
 972                 emit_feature_err(sess,
 973                                  "macro_literal_matcher",
 974                                  frag_span,
 975                                  GateIssue::Language,
 976                                  explain);
 977             }
 978             true
 979         },
 980         "vis" => {
 981             if !features.macro_vis_matcher &&
 982                !attr::contains_name(attrs, "allow_internal_unstable") {
 983                 let explain = feature_gate::EXPLAIN_VIS_MATCHER;
 984                 emit_feature_err(sess,
 985                                  "macro_vis_matcher",
 986                                  frag_span,
 987                                  GateIssue::Language,
 988                                  explain);
 989             }
 990             true
 991         },
 992         _ => false,
 993     }
 994 }
 995
 996 fn quoted_tt_to_string(tt: &quoted::TokenTree) -> String {
 997     match *tt {
 998         quoted::TokenTree::Token(_, ref tok) => ::print::pprust::token_to_string(tok),
 999         quoted::TokenTree::MetaVar(_, name) => format!("${}", name),
1000         quoted::TokenTree::MetaVarDecl(_, name, kind) => format!("${}:{}", name, kind),
1001         _ => panic!("unexpected quoted::TokenTree::{{Sequence or Delimited}} \
1002                      in follow set checker"),
1003     }
1004 }