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