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     let values: &mut Vec<String> = cx.expansions.entry(sp).or_insert_with(Vec::new);
  92     values.push(message);
  93 }
  94
  95 /// Given `lhses` and `rhses`, this is the new macro we create
  96 fn generic_extension<'cx>(cx: &'cx mut ExtCtxt,
  97                           sp: Span,
  98                           name: ast::Ident,
  99                           arg: TokenStream,
 100                           lhses: &[quoted::TokenTree],
 101                           rhses: &[quoted::TokenTree])
 102                           -> Box<dyn MacResult+'cx> {
 103     if cx.trace_macros() {
 104         trace_macros_note(cx, sp, format!("expanding `{}! {{ {} }}`", name, arg));
 105     }
 106
 107     // Which arm's failure should we report? (the one furthest along)
 108     let mut best_fail_spot = DUMMY_SP;
 109     let mut best_fail_tok = None;
 110
 111     for (i, lhs) in lhses.iter().enumerate() { // try each arm's matchers
 112         let lhs_tt = match *lhs {
 113             quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 114             _ => cx.span_bug(sp, "malformed macro lhs")
 115         };
 116
 117         match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 118             Success(named_matches) => {
 119                 let rhs = match rhses[i] {
 120                     // ignore delimiters
 121                     quoted::TokenTree::Delimited(_, ref delimed) => delimed.tts.clone(),
 122                     _ => cx.span_bug(sp, "malformed macro rhs"),
 123                 };
 124
 125                 let rhs_spans = rhs.iter().map(|t| t.span()).collect::<Vec<_>>();
 126                 // rhs has holes ( `$id` and `$(...)` that need filled)
 127                 let mut tts = transcribe(cx, Some(named_matches), rhs);
 128
 129                 // Replace all the tokens for the corresponding positions in the macro, to maintain
 130                 // proper positions in error reporting, while maintaining the macro_backtrace.
 131                 if rhs_spans.len() == tts.len() {
 132                     tts = tts.map_enumerated(|i, tt| {
 133                         let mut tt = tt.clone();
 134                         let mut sp = rhs_spans[i];
 135                         sp = sp.with_ctxt(tt.span().ctxt());
 136                         tt.set_span(sp);
 137                         tt
 138                     });
 139                 }
 140
 141                 if cx.trace_macros() {
 142                     trace_macros_note(cx, sp, format!("to `{}`", tts));
 143                 }
 144
 145                 let directory = Directory {
 146                     path: Cow::from(cx.current_expansion.module.directory.as_path()),
 147                     ownership: cx.current_expansion.directory_ownership,
 148                 };
 149                 let mut p = Parser::new(cx.parse_sess(), tts, Some(directory), true, false);
 150                 p.root_module_name = cx.current_expansion.module.mod_path.last()
 151                     .map(|id| id.as_str().to_string());
 152
 153                 p.process_potential_macro_variable();
 154                 // Let the context choose how to interpret the result.
 155                 // Weird, but useful for X-macros.
 156                 return Box::new(ParserAnyMacro {
 157                     parser: p,
 158
 159                     // Pass along the original expansion site and the name of the macro
 160                     // so we can print a useful error message if the parse of the expanded
 161                     // macro leaves unparsed tokens.
 162                     site_span: sp,
 163                     macro_ident: name
 164                 })
 165             }
 166             Failure(sp, tok) => if sp.lo() >= best_fail_spot.lo() {
 167                 best_fail_spot = sp;
 168                 best_fail_tok = Some(tok);
 169             },
 170             Error(err_sp, ref msg) => {
 171                 cx.span_fatal(err_sp.substitute_dummy(sp), &msg[..])
 172             }
 173         }
 174     }
 175
 176     let best_fail_msg = parse_failure_msg(best_fail_tok.expect("ran no matchers"));
 177     let mut err = cx.struct_span_err(best_fail_spot.substitute_dummy(sp), &best_fail_msg);
 178
 179     // Check whether there's a missing comma in this macro call, like `println!("{}" a);`
 180     if let Some((arg, comma_span)) = arg.add_comma() {
 181         for lhs in lhses { // try each arm's matchers
 182             let lhs_tt = match *lhs {
 183                 quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 184                 _ => cx.span_bug(sp, "malformed macro lhs")
 185             };
 186             match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 187                 Success(_) => {
 188                     if comma_span == DUMMY_SP {
 189                         err.note("you might be missing a comma");
 190                     } else {
 191                         err.span_suggestion_short(
 192                             comma_span,
 193                             "missing comma here",
 194                             ",".to_string(),
 195                         );
 196                     }
 197                 }
 198                 _ => {}
 199             }
 200         }
 201     }
 202     err.emit();
 203     cx.trace_macros_diag();
 204     DummyResult::any(sp)
 205 }
 206
 207 // Note that macro-by-example's input is also matched against a token tree:
 208 //                   $( $lhs:tt => $rhs:tt );+
 209 //
 210 // Holy self-referential!
 211
 212 /// Converts a `macro_rules!` invocation into a syntax extension.
 213 pub fn compile(sess: &ParseSess, features: &Features, def: &ast::Item, edition: Edition)
 214                -> SyntaxExtension {
 215     let lhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("lhs"));
 216     let rhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("rhs"));
 217
 218     // Parse the macro_rules! invocation
 219     let body = match def.node {
 220         ast::ItemKind::MacroDef(ref body) => body,
 221         _ => unreachable!(),
 222     };
 223
 224     // The pattern that macro_rules matches.
 225     // The grammar for macro_rules! is:
 226     // $( $lhs:tt => $rhs:tt );+
 227     // ...quasiquoting this would be nice.
 228     // These spans won't matter, anyways
 229     let argument_gram = vec![
 230         quoted::TokenTree::Sequence(DUMMY_SP, Lrc::new(quoted::SequenceRepetition {
 231             tts: vec![
 232                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, lhs_nm, ast::Ident::from_str("tt")),
 233                 quoted::TokenTree::Token(DUMMY_SP, token::FatArrow),
 234                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, rhs_nm, ast::Ident::from_str("tt")),
 235             ],
 236             separator: Some(if body.legacy { token::Semi } else { token::Comma }),
 237             op: quoted::KleeneOp::OneOrMore,
 238             num_captures: 2,
 239         })),
 240         // to phase into semicolon-termination instead of semicolon-separation
 241         quoted::TokenTree::Sequence(DUMMY_SP, Lrc::new(quoted::SequenceRepetition {
 242             tts: vec![quoted::TokenTree::Token(DUMMY_SP, token::Semi)],
 243             separator: None,
 244             op: quoted::KleeneOp::ZeroOrMore,
 245             num_captures: 0
 246         })),
 247     ];
 248
 249     let argument_map = match parse(sess, body.stream(), &argument_gram, None, true) {
 250         Success(m) => m,
 251         Failure(sp, tok) => {
 252             let s = parse_failure_msg(tok);
 253             sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s).raise();
 254         }
 255         Error(sp, s) => {
 256             sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s).raise();
 257         }
 258     };
 259
 260     let mut valid = true;
 261
 262     // Extract the arguments:
 263     let lhses = match *argument_map[&lhs_nm] {
 264         MatchedSeq(ref s, _) => {
 265             s.iter().map(|m| {
 266                 if let MatchedNonterminal(ref nt) = *m {
 267                     if let NtTT(ref tt) = **nt {
 268                         let tt = quoted::parse(
 269                             tt.clone().into(),
 270                             true,
 271                             sess,
 272                             features,
 273                             &def.attrs,
 274                             edition,
 275                             def.id,
 276                         )
 277                         .pop()
 278                         .unwrap();
 279                         valid &= check_lhs_nt_follows(sess, features, &def.attrs, &tt);
 280                         return tt;
 281                     }
 282                 }
 283                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 284             }).collect::<Vec<quoted::TokenTree>>()
 285         }
 286         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 287     };
 288
 289     let rhses = match *argument_map[&rhs_nm] {
 290         MatchedSeq(ref s, _) => {
 291             s.iter().map(|m| {
 292                 if let MatchedNonterminal(ref nt) = *m {
 293                     if let NtTT(ref tt) = **nt {
 294                         return quoted::parse(
 295                             tt.clone().into(),
 296                             false,
 297                             sess,
 298                             features,
 299                             &def.attrs,
 300                             edition,
 301                             def.id,
 302                         ).pop()
 303                          .unwrap();
 304                     }
 305                 }
 306                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 307             }).collect::<Vec<quoted::TokenTree>>()
 308         }
 309         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured rhs")
 310     };
 311
 312     for rhs in &rhses {
 313         valid &= check_rhs(sess, rhs);
 314     }
 315
 316     // don't abort iteration early, so that errors for multiple lhses can be reported
 317     for lhs in &lhses {
 318         valid &= check_lhs_no_empty_seq(sess, &[lhs.clone()])
 319     }
 320
 321     let expander: Box<_> = Box::new(MacroRulesMacroExpander {
 322         name: def.ident,
 323         lhses,
 324         rhses,
 325         valid,
 326     });
 327
 328     if body.legacy {
 329         let allow_internal_unstable = attr::contains_name(&def.attrs, "allow_internal_unstable");
 330         let allow_internal_unsafe = attr::contains_name(&def.attrs, "allow_internal_unsafe");
 331         let mut local_inner_macros = false;
 332         if let Some(macro_export) = attr::find_by_name(&def.attrs, "macro_export") {
 333             if let Some(l) = macro_export.meta_item_list() {
 334                 local_inner_macros = attr::list_contains_name(&l, "local_inner_macros");
 335             }
 336         }
 337
 338         let unstable_feature = attr::find_stability(&sess.span_diagnostic,
 339                                                     &def.attrs, def.span).and_then(|stability| {
 340             if let attr::StabilityLevel::Unstable { issue, .. } = stability.level {
 341                 Some((stability.feature, issue))
 342             } else {
 343                 None
 344             }
 345         });
 346
 347         NormalTT {
 348             expander,
 349             def_info: Some((def.id, def.span)),
 350             allow_internal_unstable,
 351             allow_internal_unsafe,
 352             local_inner_macros,
 353             unstable_feature,
 354             edition,
 355         }
 356     } else {
 357         let is_transparent = attr::contains_name(&def.attrs, "rustc_transparent_macro");
 358
 359         SyntaxExtension::DeclMacro {
 360             expander,
 361             def_info: Some((def.id, def.span)),
 362             is_transparent,
 363             edition,
 364         }
 365     }
 366 }
 367
 368 fn check_lhs_nt_follows(sess: &ParseSess,
 369                         features: &Features,
 370                         attrs: &[ast::Attribute],
 371                         lhs: &quoted::TokenTree) -> bool {
 372     // lhs is going to be like TokenTree::Delimited(...), where the
 373     // entire lhs is those tts. Or, it can be a "bare sequence", not wrapped in parens.
 374     if let quoted::TokenTree::Delimited(_, ref tts) = *lhs {
 375         check_matcher(sess, features, attrs, &tts.tts)
 376     } else {
 377         let msg = "invalid macro matcher; matchers must be contained in balanced delimiters";
 378         sess.span_diagnostic.span_err(lhs.span(), msg);
 379         false
 380     }
 381     // we don't abort on errors on rejection, the driver will do that for us
 382     // after parsing/expansion. we can report every error in every macro this way.
 383 }
 384
 385 /// Check that the lhs contains no repetition which could match an empty token
 386 /// tree, because then the matcher would hang indefinitely.
 387 fn check_lhs_no_empty_seq(sess: &ParseSess, tts: &[quoted::TokenTree]) -> bool {
 388     use self::quoted::TokenTree;
 389     for tt in tts {
 390         match *tt {
 391             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => (),
 392             TokenTree::Delimited(_, ref del) => if !check_lhs_no_empty_seq(sess, &del.tts) {
 393                 return false;
 394             },
 395             TokenTree::Sequence(span, ref seq) => {
 396                 if seq.separator.is_none() && seq.tts.iter().all(|seq_tt| {
 397                     match *seq_tt {
 398                         TokenTree::MetaVarDecl(_, _, id) => id.name == "vis",
 399                         TokenTree::Sequence(_, ref sub_seq) =>
 400                             sub_seq.op == quoted::KleeneOp::ZeroOrMore,
 401                         _ => false,
 402                     }
 403                 }) {
 404                     sess.span_diagnostic.span_err(span, "repetition matches empty token tree");
 405                     return false;
 406                 }
 407                 if !check_lhs_no_empty_seq(sess, &seq.tts) {
 408                     return false;
 409                 }
 410             }
 411         }
 412     }
 413
 414     true
 415 }
 416
 417 fn check_rhs(sess: &ParseSess, rhs: &quoted::TokenTree) -> bool {
 418     match *rhs {
 419         quoted::TokenTree::Delimited(..) => return true,
 420         _ => sess.span_diagnostic.span_err(rhs.span(), "macro rhs must be delimited")
 421     }
 422     false
 423 }
 424
 425 fn check_matcher(sess: &ParseSess,
 426                  features: &Features,
 427                  attrs: &[ast::Attribute],
 428                  matcher: &[quoted::TokenTree]) -> bool {
 429     let first_sets = FirstSets::new(matcher);
 430     let empty_suffix = TokenSet::empty();
 431     let err = sess.span_diagnostic.err_count();
 432     check_matcher_core(sess, features, attrs, &first_sets, matcher, &empty_suffix);
 433     err == sess.span_diagnostic.err_count()
 434 }
 435
 436 // The FirstSets for a matcher is a mapping from subsequences in the
 437 // matcher to the FIRST set for that subsequence.
 438 //
 439 // This mapping is partially precomputed via a backwards scan over the
 440 // token trees of the matcher, which provides a mapping from each
 441 // repetition sequence to its FIRST set.
 442 //
 443 // (Hypothetically sequences should be uniquely identifiable via their
 444 // spans, though perhaps that is false e.g. for macro-generated macros
 445 // that do not try to inject artificial span information. My plan is
 446 // to try to catch such cases ahead of time and not include them in
 447 // the precomputed mapping.)
 448 struct FirstSets {
 449     // this maps each TokenTree::Sequence `$(tt ...) SEP OP` that is uniquely identified by its
 450     // span in the original matcher to the First set for the inner sequence `tt ...`.
 451     //
 452     // If two sequences have the same span in a matcher, then map that
 453     // span to None (invalidating the mapping here and forcing the code to
 454     // use a slow path).
 455     first: HashMap<Span, Option<TokenSet>>,
 456 }
 457
 458 impl FirstSets {
 459     fn new(tts: &[quoted::TokenTree]) -> FirstSets {
 460         use self::quoted::TokenTree;
 461
 462         let mut sets = FirstSets { first: HashMap::new() };
 463         build_recur(&mut sets, tts);
 464         return sets;
 465
 466         // walks backward over `tts`, returning the FIRST for `tts`
 467         // and updating `sets` at the same time for all sequence
 468         // substructure we find within `tts`.
 469         fn build_recur(sets: &mut FirstSets, tts: &[TokenTree]) -> TokenSet {
 470             let mut first = TokenSet::empty();
 471             for tt in tts.iter().rev() {
 472                 match *tt {
 473                     TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 474                         first.replace_with(tt.clone());
 475                     }
 476                     TokenTree::Delimited(span, ref delimited) => {
 477                         build_recur(sets, &delimited.tts[..]);
 478                         first.replace_with(delimited.open_tt(span));
 479                     }
 480                     TokenTree::Sequence(sp, ref seq_rep) => {
 481                         let subfirst = build_recur(sets, &seq_rep.tts[..]);
 482
 483                         match sets.first.entry(sp) {
 484                             Entry::Vacant(vac) => {
 485                                 vac.insert(Some(subfirst.clone()));
 486                             }
 487                             Entry::Occupied(mut occ) => {
 488                                 // if there is already an entry, then a span must have collided.
 489                                 // This should not happen with typical macro_rules macros,
 490                                 // but syntax extensions need not maintain distinct spans,
 491                                 // so distinct syntax trees can be assigned the same span.
 492                                 // In such a case, the map cannot be trusted; so mark this
 493                                 // entry as unusable.
 494                                 occ.insert(None);
 495                             }
 496                         }
 497
 498                         // If the sequence contents can be empty, then the first
 499                         // token could be the separator token itself.
 500
 501                         if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 502                                                         subfirst.maybe_empty) {
 503                             first.add_one_maybe(TokenTree::Token(sp, sep.clone()));
 504                         }
 505
 506                         // Reverse scan: Sequence comes before `first`.
 507                         if subfirst.maybe_empty || seq_rep.op == quoted::KleeneOp::ZeroOrMore {
 508                             // If sequence is potentially empty, then
 509                             // union them (preserving first emptiness).
 510                             first.add_all(&TokenSet { maybe_empty: true, ..subfirst });
 511                         } else {
 512                             // Otherwise, sequence guaranteed
 513                             // non-empty; replace first.
 514                             first = subfirst;
 515                         }
 516                     }
 517                 }
 518             }
 519
 520             first
 521         }
 522     }
 523
 524     // walks forward over `tts` until all potential FIRST tokens are
 525     // identified.
 526     fn first(&self, tts: &[quoted::TokenTree]) -> TokenSet {
 527         use self::quoted::TokenTree;
 528
 529         let mut first = TokenSet::empty();
 530         for tt in tts.iter() {
 531             assert!(first.maybe_empty);
 532             match *tt {
 533                 TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 534                     first.add_one(tt.clone());
 535                     return first;
 536                 }
 537                 TokenTree::Delimited(span, ref delimited) => {
 538                     first.add_one(delimited.open_tt(span));
 539                     return first;
 540                 }
 541                 TokenTree::Sequence(sp, ref seq_rep) => {
 542                     match self.first.get(&sp) {
 543                         Some(&Some(ref subfirst)) => {
 544
 545                             // If the sequence contents can be empty, then the first
 546                             // token could be the separator token itself.
 547
 548                             if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 549                                                             subfirst.maybe_empty) {
 550                                 first.add_one_maybe(TokenTree::Token(sp, sep.clone()));
 551                             }
 552
 553                             assert!(first.maybe_empty);
 554                             first.add_all(subfirst);
 555                             if subfirst.maybe_empty ||
 556                                seq_rep.op == quoted::KleeneOp::ZeroOrMore {
 557                                 // continue scanning for more first
 558                                 // tokens, but also make sure we
 559                                 // restore empty-tracking state
 560                                 first.maybe_empty = true;
 561                                 continue;
 562                             } else {
 563                                 return first;
 564                             }
 565                         }
 566
 567                         Some(&None) => {
 568                             panic!("assume all sequences have (unique) spans for now");
 569                         }
 570
 571                         None => {
 572                             panic!("We missed a sequence during FirstSets construction");
 573                         }
 574                     }
 575                 }
 576             }
 577         }
 578
 579         // we only exit the loop if `tts` was empty or if every
 580         // element of `tts` matches the empty sequence.
 581         assert!(first.maybe_empty);
 582         first
 583     }
 584 }
 585
 586 // A set of `quoted::TokenTree`s, which may include `TokenTree::Match`s
 587 // (for macro-by-example syntactic variables). It also carries the
 588 // `maybe_empty` flag; that is true if and only if the matcher can
 589 // match an empty token sequence.
 590 //
 591 // The First set is computed on submatchers like `$($a:expr b),* $(c)* d`,
 592 // which has corresponding FIRST = {$a:expr, c, d}.
 593 // Likewise, `$($a:expr b),* $(c)+ d` has FIRST = {$a:expr, c}.
 594 //
 595 // (Notably, we must allow for *-op to occur zero times.)
 596 #[derive(Clone, Debug)]
 597 struct TokenSet {
 598     tokens: Vec<quoted::TokenTree>,
 599     maybe_empty: bool,
 600 }
 601
 602 impl TokenSet {
 603     // Returns a set for the empty sequence.
 604     fn empty() -> Self { TokenSet { tokens: Vec::new(), maybe_empty: true } }
 605
 606     // Returns the set `{ tok }` for the single-token (and thus
 607     // non-empty) sequence [tok].
 608     fn singleton(tok: quoted::TokenTree) -> Self {
 609         TokenSet { tokens: vec![tok], maybe_empty: false }
 610     }
 611
 612     // Changes self to be the set `{ tok }`.
 613     // Since `tok` is always present, marks self as non-empty.
 614     fn replace_with(&mut self, tok: quoted::TokenTree) {
 615         self.tokens.clear();
 616         self.tokens.push(tok);
 617         self.maybe_empty = false;
 618     }
 619
 620     // Changes self to be the empty set `{}`; meant for use when
 621     // the particular token does not matter, but we want to
 622     // record that it occurs.
 623     fn replace_with_irrelevant(&mut self) {
 624         self.tokens.clear();
 625         self.maybe_empty = false;
 626     }
 627
 628     // Adds `tok` to the set for `self`, marking sequence as non-empy.
 629     fn add_one(&mut self, tok: quoted::TokenTree) {
 630         if !self.tokens.contains(&tok) {
 631             self.tokens.push(tok);
 632         }
 633         self.maybe_empty = false;
 634     }
 635
 636     // Adds `tok` to the set for `self`. (Leaves `maybe_empty` flag alone.)
 637     fn add_one_maybe(&mut self, tok: quoted::TokenTree) {
 638         if !self.tokens.contains(&tok) {
 639             self.tokens.push(tok);
 640         }
 641     }
 642
 643     // Adds all elements of `other` to this.
 644     //
 645     // (Since this is a set, we filter out duplicates.)
 646     //
 647     // If `other` is potentially empty, then preserves the previous
 648     // setting of the empty flag of `self`. If `other` is guaranteed
 649     // non-empty, then `self` is marked non-empty.
 650     fn add_all(&mut self, other: &Self) {
 651         for tok in &other.tokens {
 652             if !self.tokens.contains(tok) {
 653                 self.tokens.push(tok.clone());
 654             }
 655         }
 656         if !other.maybe_empty {
 657             self.maybe_empty = false;
 658         }
 659     }
 660 }
 661
 662 // Checks that `matcher` is internally consistent and that it
 663 // can legally by followed by a token N, for all N in `follow`.
 664 // (If `follow` is empty, then it imposes no constraint on
 665 // the `matcher`.)
 666 //
 667 // Returns the set of NT tokens that could possibly come last in
 668 // `matcher`. (If `matcher` matches the empty sequence, then
 669 // `maybe_empty` will be set to true.)
 670 //
 671 // Requires that `first_sets` is pre-computed for `matcher`;
 672 // see `FirstSets::new`.
 673 fn check_matcher_core(sess: &ParseSess,
 674                       features: &Features,
 675                       attrs: &[ast::Attribute],
 676                       first_sets: &FirstSets,
 677                       matcher: &[quoted::TokenTree],
 678                       follow: &TokenSet) -> TokenSet {
 679     use self::quoted::TokenTree;
 680
 681     let mut last = TokenSet::empty();
 682
 683     // 2. For each token and suffix  [T, SUFFIX] in M:
 684     // ensure that T can be followed by SUFFIX, and if SUFFIX may be empty,
 685     // then ensure T can also be followed by any element of FOLLOW.
 686     'each_token: for i in 0..matcher.len() {
 687         let token = &matcher[i];
 688         let suffix = &matcher[i+1..];
 689
 690         let build_suffix_first = || {
 691             let mut s = first_sets.first(suffix);
 692             if s.maybe_empty { s.add_all(follow); }
 693             s
 694         };
 695
 696         // (we build `suffix_first` on demand below; you can tell
 697         // which cases are supposed to fall through by looking for the
 698         // initialization of this variable.)
 699         let suffix_first;
 700
 701         // First, update `last` so that it corresponds to the set
 702         // of NT tokens that might end the sequence `... token`.
 703         match *token {
 704             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 705                 let can_be_followed_by_any;
 706                 if let Err(bad_frag) = has_legal_fragment_specifier(sess, features, attrs, token) {
 707                     let msg = format!("invalid fragment specifier `{}`", bad_frag);
 708                     sess.span_diagnostic.struct_span_err(token.span(), &msg)
 709                         .help("valid fragment specifiers are `ident`, `block`, `stmt`, `expr`, \
 710                               `pat`, `ty`, `literal`, `path`, `meta`, `tt`, `item` and `vis`")
 711                         .emit();
 712                     // (This eliminates false positives and duplicates
 713                     // from error messages.)
 714                     can_be_followed_by_any = true;
 715                 } else {
 716                     can_be_followed_by_any = token_can_be_followed_by_any(token);
 717                 }
 718
 719                 if can_be_followed_by_any {
 720                     // don't need to track tokens that work with any,
 721                     last.replace_with_irrelevant();
 722                     // ... and don't need to check tokens that can be
 723                     // followed by anything against SUFFIX.
 724                     continue 'each_token;
 725                 } else {
 726                     last.replace_with(token.clone());
 727                     suffix_first = build_suffix_first();
 728                 }
 729             }
 730             TokenTree::Delimited(span, ref d) => {
 731                 let my_suffix = TokenSet::singleton(d.close_tt(span));
 732                 check_matcher_core(sess, features, attrs, first_sets, &d.tts, &my_suffix);
 733                 // don't track non NT tokens
 734                 last.replace_with_irrelevant();
 735
 736                 // also, we don't need to check delimited sequences
 737                 // against SUFFIX
 738                 continue 'each_token;
 739             }
 740             TokenTree::Sequence(sp, ref seq_rep) => {
 741                 suffix_first = build_suffix_first();
 742                 // The trick here: when we check the interior, we want
 743                 // to include the separator (if any) as a potential
 744                 // (but not guaranteed) element of FOLLOW. So in that
 745                 // case, we make a temp copy of suffix and stuff
 746                 // delimiter in there.
 747                 //
 748                 // FIXME: Should I first scan suffix_first to see if
 749                 // delimiter is already in it before I go through the
 750                 // work of cloning it? But then again, this way I may
 751                 // get a "tighter" span?
 752                 let mut new;
 753                 let my_suffix = if let Some(ref u) = seq_rep.separator {
 754                     new = suffix_first.clone();
 755                     new.add_one_maybe(TokenTree::Token(sp, u.clone()));
 756                     &new
 757                 } else {
 758                     &suffix_first
 759                 };
 760
 761                 // At this point, `suffix_first` is built, and
 762                 // `my_suffix` is some TokenSet that we can use
 763                 // for checking the interior of `seq_rep`.
 764                 let next = check_matcher_core(sess,
 765                                               features,
 766                                               attrs,
 767                                               first_sets,
 768                                               &seq_rep.tts,
 769                                               my_suffix);
 770                 if next.maybe_empty {
 771                     last.add_all(&next);
 772                 } else {
 773                     last = next;
 774                 }
 775
 776                 // the recursive call to check_matcher_core already ran the 'each_last
 777                 // check below, so we can just keep going forward here.
 778                 continue 'each_token;
 779             }
 780         }
 781
 782         // (`suffix_first` guaranteed initialized once reaching here.)
 783
 784         // Now `last` holds the complete set of NT tokens that could
 785         // end the sequence before SUFFIX. Check that every one works with `suffix`.
 786         'each_last: for token in &last.tokens {
 787             if let TokenTree::MetaVarDecl(_, ref name, ref frag_spec) = *token {
 788                 for next_token in &suffix_first.tokens {
 789                     match is_in_follow(next_token, &frag_spec.as_str()) {
 790                         Err((msg, help)) => {
 791                             sess.span_diagnostic.struct_span_err(next_token.span(), &msg)
 792                                 .help(help).emit();
 793                             // don't bother reporting every source of
 794                             // conflict for a particular element of `last`.
 795                             continue 'each_last;
 796                         }
 797                         Ok(true) => {}
 798                         Ok(false) => {
 799                             let may_be = if last.tokens.len() == 1 &&
 800                                 suffix_first.tokens.len() == 1
 801                             {
 802                                 "is"
 803                             } else {
 804                                 "may be"
 805                             };
 806
 807                             sess.span_diagnostic.span_err(
 808                                 next_token.span(),
 809                                 &format!("`${name}:{frag}` {may_be} followed by `{next}`, which \
 810                                           is not allowed for `{frag}` fragments",
 811                                          name=name,
 812                                          frag=frag_spec,
 813                                          next=quoted_tt_to_string(next_token),
 814                                          may_be=may_be)
 815                             );
 816                         }
 817                     }
 818                 }
 819             }
 820         }
 821     }
 822     last
 823 }
 824
 825 fn token_can_be_followed_by_any(tok: &quoted::TokenTree) -> bool {
 826     if let quoted::TokenTree::MetaVarDecl(_, _, frag_spec) = *tok {
 827         frag_can_be_followed_by_any(&frag_spec.as_str())
 828     } else {
 829         // (Non NT's can always be followed by anthing in matchers.)
 830         true
 831     }
 832 }
 833
 834 /// True if a fragment of type `frag` can be followed by any sort of
 835 /// token.  We use this (among other things) as a useful approximation
 836 /// for when `frag` can be followed by a repetition like `$(...)*` or
 837 /// `$(...)+`. In general, these can be a bit tricky to reason about,
 838 /// so we adopt a conservative position that says that any fragment
 839 /// specifier which consumes at most one token tree can be followed by
 840 /// a fragment specifier (indeed, these fragments can be followed by
 841 /// ANYTHING without fear of future compatibility hazards).
 842 fn frag_can_be_followed_by_any(frag: &str) -> bool {
 843     match frag {
 844         "item"     | // always terminated by `}` or `;`
 845         "block"    | // exactly one token tree
 846         "ident"    | // exactly one token tree
 847         "literal"  | // exactly one token tree
 848         "meta"     | // exactly one token tree
 849         "lifetime" | // exactly one token tree
 850         "tt" =>   // exactly one token tree
 851             true,
 852
 853         _ =>
 854             false,
 855     }
 856 }
 857
 858 /// True if `frag` can legally be followed by the token `tok`. For
 859 /// fragments that can consume an unbounded number of tokens, `tok`
 860 /// must be within a well-defined follow set. This is intended to
 861 /// guarantee future compatibility: for example, without this rule, if
 862 /// we expanded `expr` to include a new binary operator, we might
 863 /// break macros that were relying on that binary operator as a
 864 /// separator.
 865 // when changing this do not forget to update doc/book/macros.md!
 866 fn is_in_follow(tok: &quoted::TokenTree, frag: &str) -> Result<bool, (String, &'static str)> {
 867     use self::quoted::TokenTree;
 868
 869     if let TokenTree::Token(_, token::CloseDelim(_)) = *tok {
 870         // closing a token tree can never be matched by any fragment;
 871         // iow, we always require that `(` and `)` match, etc.
 872         Ok(true)
 873     } else {
 874         match frag {
 875             "item" => {
 876                 // since items *must* be followed by either a `;` or a `}`, we can
 877                 // accept anything after them
 878                 Ok(true)
 879             },
 880             "block" => {
 881                 // anything can follow block, the braces provide an easy boundary to
 882                 // maintain
 883                 Ok(true)
 884             },
 885             "stmt" | "expr"  => match *tok {
 886                 TokenTree::Token(_, ref tok) => match *tok {
 887                     FatArrow | Comma | Semi => Ok(true),
 888                     _ => Ok(false)
 889                 },
 890                 _ => Ok(false),
 891             },
 892             "pat" => match *tok {
 893                 TokenTree::Token(_, ref tok) => match *tok {
 894                     FatArrow | Comma | Eq | BinOp(token::Or) => Ok(true),
 895                     Ident(i, false) if i.name == "if" || i.name == "in" => Ok(true),
 896                     _ => Ok(false)
 897                 },
 898                 _ => Ok(false),
 899             },
 900             "path" | "ty" => match *tok {
 901                 TokenTree::Token(_, ref tok) => match *tok {
 902                     OpenDelim(token::DelimToken::Brace) | OpenDelim(token::DelimToken::Bracket) |
 903                     Comma | FatArrow | Colon | Eq | Gt | Semi | BinOp(token::Or) => Ok(true),
 904                     Ident(i, false) if i.name == "as" || i.name == "where" => Ok(true),
 905                     _ => Ok(false)
 906                 },
 907                 TokenTree::MetaVarDecl(_, _, frag) if frag.name == "block" => Ok(true),
 908                 _ => Ok(false),
 909             },
 910             "ident" | "lifetime" => {
 911                 // being a single token, idents and lifetimes are harmless
 912                 Ok(true)
 913             },
 914             "literal" => {
 915                 // literals may be of a single token, or two tokens (negative numbers)
 916                 Ok(true)
 917             },
 918             "meta" | "tt" => {
 919                 // being either a single token or a delimited sequence, tt is
 920                 // harmless
 921                 Ok(true)
 922             },
 923             "vis" => {
 924                 // Explicitly disallow `priv`, on the off chance it comes back.
 925                 match *tok {
 926                     TokenTree::Token(_, ref tok) => match *tok {
 927                         Comma => Ok(true),
 928                         Ident(i, is_raw) if is_raw || i.name != "priv" => Ok(true),
 929                         ref tok => Ok(tok.can_begin_type())
 930                     },
 931                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "ident"
 932                                                        || frag.name == "ty"
 933                                                        || frag.name == "path" => Ok(true),
 934                     _ => Ok(false)
 935                 }
 936             },
 937             "" => Ok(true), // keywords::Invalid
 938             _ => Err((format!("invalid fragment specifier `{}`", frag),
 939                      "valid fragment specifiers are `ident`, `block`, \
 940                       `stmt`, `expr`, `pat`, `ty`, `path`, `meta`, `tt`, \
 941                       `literal`, `item` and `vis`"))
 942         }
 943     }
 944 }
 945
 946 fn has_legal_fragment_specifier(sess: &ParseSess,
 947                                 features: &Features,
 948                                 attrs: &[ast::Attribute],
 949                                 tok: &quoted::TokenTree) -> Result<(), String> {
 950     debug!("has_legal_fragment_specifier({:?})", tok);
 951     if let quoted::TokenTree::MetaVarDecl(_, _, ref frag_spec) = *tok {
 952         let frag_name = frag_spec.as_str();
 953         let frag_span = tok.span();
 954         if !is_legal_fragment_specifier(sess, features, attrs, &frag_name, frag_span) {
 955             return Err(frag_name.to_string());
 956         }
 957     }
 958     Ok(())
 959 }
 960
 961 fn is_legal_fragment_specifier(sess: &ParseSess,
 962                                features: &Features,
 963                                attrs: &[ast::Attribute],
 964                                frag_name: &str,
 965                                frag_span: Span) -> bool {
 966     match frag_name {
 967         "item" | "block" | "stmt" | "expr" | "pat" | "lifetime" |
 968         "path" | "ty" | "ident" | "meta" | "tt" | "" => true,
 969         "literal" => {
 970             if !features.macro_literal_matcher &&
 971                !attr::contains_name(attrs, "allow_internal_unstable") {
 972                 let explain = feature_gate::EXPLAIN_LITERAL_MATCHER;
 973                 emit_feature_err(sess,
 974                                  "macro_literal_matcher",
 975                                  frag_span,
 976                                  GateIssue::Language,
 977                                  explain);
 978             }
 979             true
 980         },
 981         "vis" => {
 982             if !features.macro_vis_matcher &&
 983                !attr::contains_name(attrs, "allow_internal_unstable") {
 984                 let explain = feature_gate::EXPLAIN_VIS_MATCHER;
 985                 emit_feature_err(sess,
 986                                  "macro_vis_matcher",
 987                                  frag_span,
 988                                  GateIssue::Language,
 989                                  explain);
 990             }
 991             true
 992         },
 993         _ => false,
 994     }
 995 }
 996
 997 fn quoted_tt_to_string(tt: &quoted::TokenTree) -> String {
 998     match *tt {
 999         quoted::TokenTree::Token(_, ref tok) => ::print::pprust::token_to_string(tok),
1000         quoted::TokenTree::MetaVar(_, name) => format!("${}", name),
1001         quoted::TokenTree::MetaVarDecl(_, name, kind) => format!("${}:{}", name, kind),
1002         _ => panic!("unexpected quoted::TokenTree::{{Sequence or Delimited}} \
1003                      in follow set checker"),
1004     }
1005 }