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::{DelimSpan, TokenStream, TokenTree};
  29
  30 use rustc_data_structures::fx::FxHashMap;
  31 use std::borrow::Cow;
  32 use std::collections::hash_map::Entry;
  33
  34 use rustc_data_structures::sync::Lrc;
  35 use errors::Applicability;
  36
  37 const VALID_FRAGMENT_NAMES_MSG: &str = "valid fragment specifiers are \
  38     `ident`, `block`, `stmt`, `expr`, `pat`, `ty`, `lifetime`, `literal`, \
  39     `path`, `meta`, `tt`, `item` and `vis`";
  40
  41 pub struct ParserAnyMacro<'a> {
  42     parser: Parser<'a>,
  43
  44     /// Span of the expansion site of the macro this parser is for
  45     site_span: Span,
  46     /// The ident of the macro we're parsing
  47     macro_ident: ast::Ident
  48 }
  49
  50 impl<'a> ParserAnyMacro<'a> {
  51     pub fn make(mut self: Box<ParserAnyMacro<'a>>, kind: AstFragmentKind) -> AstFragment {
  52         let ParserAnyMacro { site_span, macro_ident, ref mut parser } = *self;
  53         let fragment = panictry!(parser.parse_ast_fragment(kind, true).map_err(|mut e| {
  54             if e.span.is_dummy() {  // Get around lack of span in error (#30128)
  55                 e.set_span(site_span);
  56             }
  57             e
  58         }));
  59
  60         // We allow semicolons at the end of expressions -- e.g. the semicolon in
  61         // `macro_rules! m { () => { panic!(); } }` isn't parsed by `.parse_expr()`,
  62         // but `m!()` is allowed in expression positions (c.f. issue #34706).
  63         if kind == AstFragmentKind::Expr && parser.token == token::Semi {
  64             parser.bump();
  65         }
  66
  67         // Make sure we don't have any tokens left to parse so we don't silently drop anything.
  68         let path = ast::Path::from_ident(macro_ident.with_span_pos(site_span));
  69         parser.ensure_complete_parse(&path, kind.name(), site_span);
  70         fragment
  71     }
  72 }
  73
  74 struct MacroRulesMacroExpander {
  75     name: ast::Ident,
  76     lhses: Vec<quoted::TokenTree>,
  77     rhses: Vec<quoted::TokenTree>,
  78     valid: bool,
  79 }
  80
  81 impl TTMacroExpander for MacroRulesMacroExpander {
  82     fn expand<'cx>(
  83         &self,
  84         cx: &'cx mut ExtCtxt,
  85         sp: Span,
  86         input: TokenStream,
  87         def_span: Option<Span>,
  88     ) -> Box<dyn MacResult+'cx> {
  89         if !self.valid {
  90             return DummyResult::any(sp);
  91         }
  92         generic_extension(cx,
  93                           sp,
  94                           def_span,
  95                           self.name,
  96                           input,
  97                           &self.lhses,
  98                           &self.rhses)
  99     }
 100 }
 101
 102 fn trace_macros_note(cx: &mut ExtCtxt, sp: Span, message: String) {
 103     let sp = sp.macro_backtrace().last().map(|trace| trace.call_site).unwrap_or(sp);
 104     cx.expansions.entry(sp).or_default().push(message);
 105 }
 106
 107 /// Given `lhses` and `rhses`, this is the new macro we create
 108 fn generic_extension<'cx>(cx: &'cx mut ExtCtxt,
 109                           sp: Span,
 110                           def_span: Option<Span>,
 111                           name: ast::Ident,
 112                           arg: TokenStream,
 113                           lhses: &[quoted::TokenTree],
 114                           rhses: &[quoted::TokenTree])
 115                           -> Box<dyn MacResult+'cx> {
 116     if cx.trace_macros() {
 117         trace_macros_note(cx, sp, format!("expanding `{}! {{ {} }}`", name, arg));
 118     }
 119
 120     // Which arm's failure should we report? (the one furthest along)
 121     let mut best_fail_spot = DUMMY_SP;
 122     let mut best_fail_tok = None;
 123
 124     for (i, lhs) in lhses.iter().enumerate() { // try each arm's matchers
 125         let lhs_tt = match *lhs {
 126             quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 127             _ => cx.span_bug(sp, "malformed macro lhs")
 128         };
 129
 130         match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 131             Success(named_matches) => {
 132                 let rhs = match rhses[i] {
 133                     // ignore delimiters
 134                     quoted::TokenTree::Delimited(_, ref delimed) => delimed.tts.clone(),
 135                     _ => cx.span_bug(sp, "malformed macro rhs"),
 136                 };
 137
 138                 let rhs_spans = rhs.iter().map(|t| t.span()).collect::<Vec<_>>();
 139                 // rhs has holes ( `$id` and `$(...)` that need filled)
 140                 let mut tts = transcribe(cx, Some(named_matches), rhs);
 141
 142                 // Replace all the tokens for the corresponding positions in the macro, to maintain
 143                 // proper positions in error reporting, while maintaining the macro_backtrace.
 144                 if rhs_spans.len() == tts.len() {
 145                     tts = tts.map_enumerated(|i, tt| {
 146                         let mut tt = tt.clone();
 147                         let mut sp = rhs_spans[i];
 148                         sp = sp.with_ctxt(tt.span().ctxt());
 149                         tt.set_span(sp);
 150                         tt
 151                     });
 152                 }
 153
 154                 if cx.trace_macros() {
 155                     trace_macros_note(cx, sp, format!("to `{}`", tts));
 156                 }
 157
 158                 let directory = Directory {
 159                     path: Cow::from(cx.current_expansion.module.directory.as_path()),
 160                     ownership: cx.current_expansion.directory_ownership,
 161                 };
 162                 let mut p = Parser::new(cx.parse_sess(), tts, Some(directory), true, false);
 163                 p.root_module_name = cx.current_expansion.module.mod_path.last()
 164                     .map(|id| id.as_str().to_string());
 165
 166                 p.process_potential_macro_variable();
 167                 // Let the context choose how to interpret the result.
 168                 // Weird, but useful for X-macros.
 169                 return Box::new(ParserAnyMacro {
 170                     parser: p,
 171
 172                     // Pass along the original expansion site and the name of the macro
 173                     // so we can print a useful error message if the parse of the expanded
 174                     // macro leaves unparsed tokens.
 175                     site_span: sp,
 176                     macro_ident: name
 177                 })
 178             }
 179             Failure(sp, tok) => if sp.lo() >= best_fail_spot.lo() {
 180                 best_fail_spot = sp;
 181                 best_fail_tok = Some(tok);
 182             },
 183             Error(err_sp, ref msg) => {
 184                 cx.span_fatal(err_sp.substitute_dummy(sp), &msg[..])
 185             }
 186         }
 187     }
 188
 189     let best_fail_msg = parse_failure_msg(best_fail_tok.expect("ran no matchers"));
 190     let span = best_fail_spot.substitute_dummy(sp);
 191     let mut err = cx.struct_span_err(span, &best_fail_msg);
 192     err.span_label(span, best_fail_msg);
 193     if let Some(sp) = def_span {
 194         if cx.source_map().span_to_filename(sp).is_real() && !sp.is_dummy() {
 195             err.span_label(cx.source_map().def_span(sp), "when calling this macro");
 196         }
 197     }
 198
 199     // Check whether there's a missing comma in this macro call, like `println!("{}" a);`
 200     if let Some((arg, comma_span)) = arg.add_comma() {
 201         for lhs in lhses { // try each arm's matchers
 202             let lhs_tt = match *lhs {
 203                 quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 204                 _ => continue,
 205             };
 206             match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 207                 Success(_) => {
 208                     if comma_span.is_dummy() {
 209                         err.note("you might be missing a comma");
 210                     } else {
 211                         err.span_suggestion_short_with_applicability(
 212                             comma_span,
 213                             "missing comma here",
 214                             ", ".to_string(),
 215                             Applicability::MachineApplicable,
 216                         );
 217                     }
 218                 }
 219                 _ => {}
 220             }
 221         }
 222     }
 223     err.emit();
 224     cx.trace_macros_diag();
 225     DummyResult::any(sp)
 226 }
 227
 228 // Note that macro-by-example's input is also matched against a token tree:
 229 //                   $( $lhs:tt => $rhs:tt );+
 230 //
 231 // Holy self-referential!
 232
 233 /// Converts a `macro_rules!` invocation into a syntax extension.
 234 pub fn compile(sess: &ParseSess, features: &Features, def: &ast::Item, edition: Edition)
 235                -> SyntaxExtension {
 236     let lhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("lhs"));
 237     let rhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("rhs"));
 238
 239     // Parse the macro_rules! invocation
 240     let body = match def.node {
 241         ast::ItemKind::MacroDef(ref body) => body,
 242         _ => unreachable!(),
 243     };
 244
 245     // The pattern that macro_rules matches.
 246     // The grammar for macro_rules! is:
 247     // $( $lhs:tt => $rhs:tt );+
 248     // ...quasiquoting this would be nice.
 249     // These spans won't matter, anyways
 250     let argument_gram = vec![
 251         quoted::TokenTree::Sequence(DelimSpan::dummy(), Lrc::new(quoted::SequenceRepetition {
 252             tts: vec![
 253                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, lhs_nm, ast::Ident::from_str("tt")),
 254                 quoted::TokenTree::Token(DUMMY_SP, token::FatArrow),
 255                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, rhs_nm, ast::Ident::from_str("tt")),
 256             ],
 257             separator: Some(if body.legacy { token::Semi } else { token::Comma }),
 258             op: quoted::KleeneOp::OneOrMore,
 259             num_captures: 2,
 260         })),
 261         // to phase into semicolon-termination instead of semicolon-separation
 262         quoted::TokenTree::Sequence(DelimSpan::dummy(), Lrc::new(quoted::SequenceRepetition {
 263             tts: vec![quoted::TokenTree::Token(DUMMY_SP, token::Semi)],
 264             separator: None,
 265             op: quoted::KleeneOp::ZeroOrMore,
 266             num_captures: 0
 267         })),
 268     ];
 269
 270     let argument_map = match parse(sess, body.stream(), &argument_gram, None, true) {
 271         Success(m) => m,
 272         Failure(sp, tok) => {
 273             let s = parse_failure_msg(tok);
 274             sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s).raise();
 275         }
 276         Error(sp, s) => {
 277             sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s).raise();
 278         }
 279     };
 280
 281     let mut valid = true;
 282
 283     // Extract the arguments:
 284     let lhses = match *argument_map[&lhs_nm] {
 285         MatchedSeq(ref s, _) => {
 286             s.iter().map(|m| {
 287                 if let MatchedNonterminal(ref nt) = *m {
 288                     if let NtTT(ref tt) = **nt {
 289                         let tt = quoted::parse(
 290                             tt.clone().into(),
 291                             true,
 292                             sess,
 293                             features,
 294                             &def.attrs,
 295                             edition,
 296                             def.id,
 297                         )
 298                         .pop()
 299                         .unwrap();
 300                         valid &= check_lhs_nt_follows(sess, features, &def.attrs, &tt);
 301                         return tt;
 302                     }
 303                 }
 304                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 305             }).collect::<Vec<quoted::TokenTree>>()
 306         }
 307         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 308     };
 309
 310     let rhses = match *argument_map[&rhs_nm] {
 311         MatchedSeq(ref s, _) => {
 312             s.iter().map(|m| {
 313                 if let MatchedNonterminal(ref nt) = *m {
 314                     if let NtTT(ref tt) = **nt {
 315                         return quoted::parse(
 316                             tt.clone().into(),
 317                             false,
 318                             sess,
 319                             features,
 320                             &def.attrs,
 321                             edition,
 322                             def.id,
 323                         ).pop()
 324                          .unwrap();
 325                     }
 326                 }
 327                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 328             }).collect::<Vec<quoted::TokenTree>>()
 329         }
 330         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured rhs")
 331     };
 332
 333     for rhs in &rhses {
 334         valid &= check_rhs(sess, rhs);
 335     }
 336
 337     // don't abort iteration early, so that errors for multiple lhses can be reported
 338     for lhs in &lhses {
 339         valid &= check_lhs_no_empty_seq(sess, &[lhs.clone()])
 340     }
 341
 342     let expander: Box<_> = Box::new(MacroRulesMacroExpander {
 343         name: def.ident,
 344         lhses,
 345         rhses,
 346         valid,
 347     });
 348
 349     if body.legacy {
 350         let allow_internal_unstable = attr::contains_name(&def.attrs, "allow_internal_unstable");
 351         let allow_internal_unsafe = attr::contains_name(&def.attrs, "allow_internal_unsafe");
 352         let mut local_inner_macros = false;
 353         if let Some(macro_export) = attr::find_by_name(&def.attrs, "macro_export") {
 354             if let Some(l) = macro_export.meta_item_list() {
 355                 local_inner_macros = attr::list_contains_name(&l, "local_inner_macros");
 356             }
 357         }
 358
 359         let unstable_feature = attr::find_stability(&sess.span_diagnostic,
 360                                                     &def.attrs, def.span).and_then(|stability| {
 361             if let attr::StabilityLevel::Unstable { issue, .. } = stability.level {
 362                 Some((stability.feature, issue))
 363             } else {
 364                 None
 365             }
 366         });
 367
 368         NormalTT {
 369             expander,
 370             def_info: Some((def.id, def.span)),
 371             allow_internal_unstable,
 372             allow_internal_unsafe,
 373             local_inner_macros,
 374             unstable_feature,
 375             edition,
 376         }
 377     } else {
 378         let is_transparent = attr::contains_name(&def.attrs, "rustc_transparent_macro");
 379
 380         SyntaxExtension::DeclMacro {
 381             expander,
 382             def_info: Some((def.id, def.span)),
 383             is_transparent,
 384             edition,
 385         }
 386     }
 387 }
 388
 389 fn check_lhs_nt_follows(sess: &ParseSess,
 390                         features: &Features,
 391                         attrs: &[ast::Attribute],
 392                         lhs: &quoted::TokenTree) -> bool {
 393     // lhs is going to be like TokenTree::Delimited(...), where the
 394     // entire lhs is those tts. Or, it can be a "bare sequence", not wrapped in parens.
 395     if let quoted::TokenTree::Delimited(_, ref tts) = *lhs {
 396         check_matcher(sess, features, attrs, &tts.tts)
 397     } else {
 398         let msg = "invalid macro matcher; matchers must be contained in balanced delimiters";
 399         sess.span_diagnostic.span_err(lhs.span(), msg);
 400         false
 401     }
 402     // we don't abort on errors on rejection, the driver will do that for us
 403     // after parsing/expansion. we can report every error in every macro this way.
 404 }
 405
 406 /// Check that the lhs contains no repetition which could match an empty token
 407 /// tree, because then the matcher would hang indefinitely.
 408 fn check_lhs_no_empty_seq(sess: &ParseSess, tts: &[quoted::TokenTree]) -> bool {
 409     use self::quoted::TokenTree;
 410     for tt in tts {
 411         match *tt {
 412             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => (),
 413             TokenTree::Delimited(_, ref del) => if !check_lhs_no_empty_seq(sess, &del.tts) {
 414                 return false;
 415             },
 416             TokenTree::Sequence(span, ref seq) => {
 417                 if seq.separator.is_none() && seq.tts.iter().all(|seq_tt| {
 418                     match *seq_tt {
 419                         TokenTree::MetaVarDecl(_, _, id) => id.name == "vis",
 420                         TokenTree::Sequence(_, ref sub_seq) =>
 421                             sub_seq.op == quoted::KleeneOp::ZeroOrMore,
 422                         _ => false,
 423                     }
 424                 }) {
 425                     let sp = span.entire();
 426                     sess.span_diagnostic.span_err(sp, "repetition matches empty token tree");
 427                     return false;
 428                 }
 429                 if !check_lhs_no_empty_seq(sess, &seq.tts) {
 430                     return false;
 431                 }
 432             }
 433         }
 434     }
 435
 436     true
 437 }
 438
 439 fn check_rhs(sess: &ParseSess, rhs: &quoted::TokenTree) -> bool {
 440     match *rhs {
 441         quoted::TokenTree::Delimited(..) => return true,
 442         _ => sess.span_diagnostic.span_err(rhs.span(), "macro rhs must be delimited")
 443     }
 444     false
 445 }
 446
 447 fn check_matcher(sess: &ParseSess,
 448                  features: &Features,
 449                  attrs: &[ast::Attribute],
 450                  matcher: &[quoted::TokenTree]) -> bool {
 451     let first_sets = FirstSets::new(matcher);
 452     let empty_suffix = TokenSet::empty();
 453     let err = sess.span_diagnostic.err_count();
 454     check_matcher_core(sess, features, attrs, &first_sets, matcher, &empty_suffix);
 455     err == sess.span_diagnostic.err_count()
 456 }
 457
 458 // The FirstSets for a matcher is a mapping from subsequences in the
 459 // matcher to the FIRST set for that subsequence.
 460 //
 461 // This mapping is partially precomputed via a backwards scan over the
 462 // token trees of the matcher, which provides a mapping from each
 463 // repetition sequence to its FIRST set.
 464 //
 465 // (Hypothetically sequences should be uniquely identifiable via their
 466 // spans, though perhaps that is false e.g. for macro-generated macros
 467 // that do not try to inject artificial span information. My plan is
 468 // to try to catch such cases ahead of time and not include them in
 469 // the precomputed mapping.)
 470 struct FirstSets {
 471     // this maps each TokenTree::Sequence `$(tt ...) SEP OP` that is uniquely identified by its
 472     // span in the original matcher to the First set for the inner sequence `tt ...`.
 473     //
 474     // If two sequences have the same span in a matcher, then map that
 475     // span to None (invalidating the mapping here and forcing the code to
 476     // use a slow path).
 477     first: FxHashMap<Span, Option<TokenSet>>,
 478 }
 479
 480 impl FirstSets {
 481     fn new(tts: &[quoted::TokenTree]) -> FirstSets {
 482         use self::quoted::TokenTree;
 483
 484         let mut sets = FirstSets { first: FxHashMap::default() };
 485         build_recur(&mut sets, tts);
 486         return sets;
 487
 488         // walks backward over `tts`, returning the FIRST for `tts`
 489         // and updating `sets` at the same time for all sequence
 490         // substructure we find within `tts`.
 491         fn build_recur(sets: &mut FirstSets, tts: &[TokenTree]) -> TokenSet {
 492             let mut first = TokenSet::empty();
 493             for tt in tts.iter().rev() {
 494                 match *tt {
 495                     TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 496                         first.replace_with(tt.clone());
 497                     }
 498                     TokenTree::Delimited(span, ref delimited) => {
 499                         build_recur(sets, &delimited.tts[..]);
 500                         first.replace_with(delimited.open_tt(span.open));
 501                     }
 502                     TokenTree::Sequence(sp, ref seq_rep) => {
 503                         let subfirst = build_recur(sets, &seq_rep.tts[..]);
 504
 505                         match sets.first.entry(sp.entire()) {
 506                             Entry::Vacant(vac) => {
 507                                 vac.insert(Some(subfirst.clone()));
 508                             }
 509                             Entry::Occupied(mut occ) => {
 510                                 // if there is already an entry, then a span must have collided.
 511                                 // This should not happen with typical macro_rules macros,
 512                                 // but syntax extensions need not maintain distinct spans,
 513                                 // so distinct syntax trees can be assigned the same span.
 514                                 // In such a case, the map cannot be trusted; so mark this
 515                                 // entry as unusable.
 516                                 occ.insert(None);
 517                             }
 518                         }
 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.entire(), sep.clone()));
 526                         }
 527
 528                         // Reverse scan: Sequence comes before `first`.
 529                         if subfirst.maybe_empty || seq_rep.op == quoted::KleeneOp::ZeroOrMore {
 530                             // If sequence is potentially empty, then
 531                             // union them (preserving first emptiness).
 532                             first.add_all(&TokenSet { maybe_empty: true, ..subfirst });
 533                         } else {
 534                             // Otherwise, sequence guaranteed
 535                             // non-empty; replace first.
 536                             first = subfirst;
 537                         }
 538                     }
 539                 }
 540             }
 541
 542             first
 543         }
 544     }
 545
 546     // walks forward over `tts` until all potential FIRST tokens are
 547     // identified.
 548     fn first(&self, tts: &[quoted::TokenTree]) -> TokenSet {
 549         use self::quoted::TokenTree;
 550
 551         let mut first = TokenSet::empty();
 552         for tt in tts.iter() {
 553             assert!(first.maybe_empty);
 554             match *tt {
 555                 TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 556                     first.add_one(tt.clone());
 557                     return first;
 558                 }
 559                 TokenTree::Delimited(span, ref delimited) => {
 560                     first.add_one(delimited.open_tt(span.open));
 561                     return first;
 562                 }
 563                 TokenTree::Sequence(sp, ref seq_rep) => {
 564                     match self.first.get(&sp.entire()) {
 565                         Some(&Some(ref subfirst)) => {
 566
 567                             // If the sequence contents can be empty, then the first
 568                             // token could be the separator token itself.
 569
 570                             if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 571                                                             subfirst.maybe_empty) {
 572                                 first.add_one_maybe(TokenTree::Token(sp.entire(), sep.clone()));
 573                             }
 574
 575                             assert!(first.maybe_empty);
 576                             first.add_all(subfirst);
 577                             if subfirst.maybe_empty ||
 578                                seq_rep.op == quoted::KleeneOp::ZeroOrMore {
 579                                 // continue scanning for more first
 580                                 // tokens, but also make sure we
 581                                 // restore empty-tracking state
 582                                 first.maybe_empty = true;
 583                                 continue;
 584                             } else {
 585                                 return first;
 586                             }
 587                         }
 588
 589                         Some(&None) => {
 590                             panic!("assume all sequences have (unique) spans for now");
 591                         }
 592
 593                         None => {
 594                             panic!("We missed a sequence during FirstSets construction");
 595                         }
 596                     }
 597                 }
 598             }
 599         }
 600
 601         // we only exit the loop if `tts` was empty or if every
 602         // element of `tts` matches the empty sequence.
 603         assert!(first.maybe_empty);
 604         first
 605     }
 606 }
 607
 608 // A set of `quoted::TokenTree`s, which may include `TokenTree::Match`s
 609 // (for macro-by-example syntactic variables). It also carries the
 610 // `maybe_empty` flag; that is true if and only if the matcher can
 611 // match an empty token sequence.
 612 //
 613 // The First set is computed on submatchers like `$($a:expr b),* $(c)* d`,
 614 // which has corresponding FIRST = {$a:expr, c, d}.
 615 // Likewise, `$($a:expr b),* $(c)+ d` has FIRST = {$a:expr, c}.
 616 //
 617 // (Notably, we must allow for *-op to occur zero times.)
 618 #[derive(Clone, Debug)]
 619 struct TokenSet {
 620     tokens: Vec<quoted::TokenTree>,
 621     maybe_empty: bool,
 622 }
 623
 624 impl TokenSet {
 625     // Returns a set for the empty sequence.
 626     fn empty() -> Self { TokenSet { tokens: Vec::new(), maybe_empty: true } }
 627
 628     // Returns the set `{ tok }` for the single-token (and thus
 629     // non-empty) sequence [tok].
 630     fn singleton(tok: quoted::TokenTree) -> Self {
 631         TokenSet { tokens: vec![tok], maybe_empty: false }
 632     }
 633
 634     // Changes self to be the set `{ tok }`.
 635     // Since `tok` is always present, marks self as non-empty.
 636     fn replace_with(&mut self, tok: quoted::TokenTree) {
 637         self.tokens.clear();
 638         self.tokens.push(tok);
 639         self.maybe_empty = false;
 640     }
 641
 642     // Changes self to be the empty set `{}`; meant for use when
 643     // the particular token does not matter, but we want to
 644     // record that it occurs.
 645     fn replace_with_irrelevant(&mut self) {
 646         self.tokens.clear();
 647         self.maybe_empty = false;
 648     }
 649
 650     // Adds `tok` to the set for `self`, marking sequence as non-empy.
 651     fn add_one(&mut self, tok: quoted::TokenTree) {
 652         if !self.tokens.contains(&tok) {
 653             self.tokens.push(tok);
 654         }
 655         self.maybe_empty = false;
 656     }
 657
 658     // Adds `tok` to the set for `self`. (Leaves `maybe_empty` flag alone.)
 659     fn add_one_maybe(&mut self, tok: quoted::TokenTree) {
 660         if !self.tokens.contains(&tok) {
 661             self.tokens.push(tok);
 662         }
 663     }
 664
 665     // Adds all elements of `other` to this.
 666     //
 667     // (Since this is a set, we filter out duplicates.)
 668     //
 669     // If `other` is potentially empty, then preserves the previous
 670     // setting of the empty flag of `self`. If `other` is guaranteed
 671     // non-empty, then `self` is marked non-empty.
 672     fn add_all(&mut self, other: &Self) {
 673         for tok in &other.tokens {
 674             if !self.tokens.contains(tok) {
 675                 self.tokens.push(tok.clone());
 676             }
 677         }
 678         if !other.maybe_empty {
 679             self.maybe_empty = false;
 680         }
 681     }
 682 }
 683
 684 // Checks that `matcher` is internally consistent and that it
 685 // can legally by followed by a token N, for all N in `follow`.
 686 // (If `follow` is empty, then it imposes no constraint on
 687 // the `matcher`.)
 688 //
 689 // Returns the set of NT tokens that could possibly come last in
 690 // `matcher`. (If `matcher` matches the empty sequence, then
 691 // `maybe_empty` will be set to true.)
 692 //
 693 // Requires that `first_sets` is pre-computed for `matcher`;
 694 // see `FirstSets::new`.
 695 fn check_matcher_core(sess: &ParseSess,
 696                       features: &Features,
 697                       attrs: &[ast::Attribute],
 698                       first_sets: &FirstSets,
 699                       matcher: &[quoted::TokenTree],
 700                       follow: &TokenSet) -> TokenSet {
 701     use self::quoted::TokenTree;
 702
 703     let mut last = TokenSet::empty();
 704
 705     // 2. For each token and suffix  [T, SUFFIX] in M:
 706     // ensure that T can be followed by SUFFIX, and if SUFFIX may be empty,
 707     // then ensure T can also be followed by any element of FOLLOW.
 708     'each_token: for i in 0..matcher.len() {
 709         let token = &matcher[i];
 710         let suffix = &matcher[i+1..];
 711
 712         let build_suffix_first = || {
 713             let mut s = first_sets.first(suffix);
 714             if s.maybe_empty { s.add_all(follow); }
 715             s
 716         };
 717
 718         // (we build `suffix_first` on demand below; you can tell
 719         // which cases are supposed to fall through by looking for the
 720         // initialization of this variable.)
 721         let suffix_first;
 722
 723         // First, update `last` so that it corresponds to the set
 724         // of NT tokens that might end the sequence `... token`.
 725         match *token {
 726             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 727                 let can_be_followed_by_any;
 728                 if let Err(bad_frag) = has_legal_fragment_specifier(sess, features, attrs, token) {
 729                     let msg = format!("invalid fragment specifier `{}`", bad_frag);
 730                     sess.span_diagnostic.struct_span_err(token.span(), &msg)
 731                         .help(VALID_FRAGMENT_NAMES_MSG)
 732                         .emit();
 733                     // (This eliminates false positives and duplicates
 734                     // from error messages.)
 735                     can_be_followed_by_any = true;
 736                 } else {
 737                     can_be_followed_by_any = token_can_be_followed_by_any(token);
 738                 }
 739
 740                 if can_be_followed_by_any {
 741                     // don't need to track tokens that work with any,
 742                     last.replace_with_irrelevant();
 743                     // ... and don't need to check tokens that can be
 744                     // followed by anything against SUFFIX.
 745                     continue 'each_token;
 746                 } else {
 747                     last.replace_with(token.clone());
 748                     suffix_first = build_suffix_first();
 749                 }
 750             }
 751             TokenTree::Delimited(span, ref d) => {
 752                 let my_suffix = TokenSet::singleton(d.close_tt(span.close));
 753                 check_matcher_core(sess, features, attrs, first_sets, &d.tts, &my_suffix);
 754                 // don't track non NT tokens
 755                 last.replace_with_irrelevant();
 756
 757                 // also, we don't need to check delimited sequences
 758                 // against SUFFIX
 759                 continue 'each_token;
 760             }
 761             TokenTree::Sequence(sp, ref seq_rep) => {
 762                 suffix_first = build_suffix_first();
 763                 // The trick here: when we check the interior, we want
 764                 // to include the separator (if any) as a potential
 765                 // (but not guaranteed) element of FOLLOW. So in that
 766                 // case, we make a temp copy of suffix and stuff
 767                 // delimiter in there.
 768                 //
 769                 // FIXME: Should I first scan suffix_first to see if
 770                 // delimiter is already in it before I go through the
 771                 // work of cloning it? But then again, this way I may
 772                 // get a "tighter" span?
 773                 let mut new;
 774                 let my_suffix = if let Some(ref u) = seq_rep.separator {
 775                     new = suffix_first.clone();
 776                     new.add_one_maybe(TokenTree::Token(sp.entire(), u.clone()));
 777                     &new
 778                 } else {
 779                     &suffix_first
 780                 };
 781
 782                 // At this point, `suffix_first` is built, and
 783                 // `my_suffix` is some TokenSet that we can use
 784                 // for checking the interior of `seq_rep`.
 785                 let next = check_matcher_core(sess,
 786                                               features,
 787                                               attrs,
 788                                               first_sets,
 789                                               &seq_rep.tts,
 790                                               my_suffix);
 791                 if next.maybe_empty {
 792                     last.add_all(&next);
 793                 } else {
 794                     last = next;
 795                 }
 796
 797                 // the recursive call to check_matcher_core already ran the 'each_last
 798                 // check below, so we can just keep going forward here.
 799                 continue 'each_token;
 800             }
 801         }
 802
 803         // (`suffix_first` guaranteed initialized once reaching here.)
 804
 805         // Now `last` holds the complete set of NT tokens that could
 806         // end the sequence before SUFFIX. Check that every one works with `suffix`.
 807         'each_last: for token in &last.tokens {
 808             if let TokenTree::MetaVarDecl(_, ref name, ref frag_spec) = *token {
 809                 for next_token in &suffix_first.tokens {
 810                     match is_in_follow(next_token, &frag_spec.as_str()) {
 811                         IsInFollow::Invalid(msg, help) => {
 812                             sess.span_diagnostic.struct_span_err(next_token.span(), &msg)
 813                                 .help(help).emit();
 814                             // don't bother reporting every source of
 815                             // conflict for a particular element of `last`.
 816                             continue 'each_last;
 817                         }
 818                         IsInFollow::Yes => {}
 819                         IsInFollow::No(ref possible) => {
 820                             let may_be = if last.tokens.len() == 1 &&
 821                                 suffix_first.tokens.len() == 1
 822                             {
 823                                 "is"
 824                             } else {
 825                                 "may be"
 826                             };
 827
 828                             let sp = next_token.span();
 829                             let mut err = sess.span_diagnostic.struct_span_err(
 830                                 sp,
 831                                 &format!("`${name}:{frag}` {may_be} followed by `{next}`, which \
 832                                           is not allowed for `{frag}` fragments",
 833                                          name=name,
 834                                          frag=frag_spec,
 835                                          next=quoted_tt_to_string(next_token),
 836                                          may_be=may_be),
 837                             );
 838                             err.span_label(
 839                                 sp,
 840                                 format!("not allowed after `{}` fragments", frag_spec),
 841                             );
 842                             let msg = "allowed there are: ";
 843                             match &possible[..] {
 844                                 &[] => {}
 845                                 &[t] => {
 846                                     err.note(&format!(
 847                                         "only {} is allowed after `{}` fragments",
 848                                         t,
 849                                         frag_spec,
 850                                     ));
 851                                 }
 852                                 ts => {
 853                                     err.note(&format!(
 854                                         "{}{} or {}",
 855                                         msg,
 856                                         ts[..ts.len() - 1].iter().map(|s| *s)
 857                                             .collect::<Vec<_>>().join(", "),
 858                                         ts[ts.len() - 1],
 859                                     ));
 860                                 }
 861                             }
 862                             err.emit();
 863                         }
 864                     }
 865                 }
 866             }
 867         }
 868     }
 869     last
 870 }
 871
 872 fn token_can_be_followed_by_any(tok: &quoted::TokenTree) -> bool {
 873     if let quoted::TokenTree::MetaVarDecl(_, _, frag_spec) = *tok {
 874         frag_can_be_followed_by_any(&frag_spec.as_str())
 875     } else {
 876         // (Non NT's can always be followed by anthing in matchers.)
 877         true
 878     }
 879 }
 880
 881 /// True if a fragment of type `frag` can be followed by any sort of
 882 /// token.  We use this (among other things) as a useful approximation
 883 /// for when `frag` can be followed by a repetition like `$(...)*` or
 884 /// `$(...)+`. In general, these can be a bit tricky to reason about,
 885 /// so we adopt a conservative position that says that any fragment
 886 /// specifier which consumes at most one token tree can be followed by
 887 /// a fragment specifier (indeed, these fragments can be followed by
 888 /// ANYTHING without fear of future compatibility hazards).
 889 fn frag_can_be_followed_by_any(frag: &str) -> bool {
 890     match frag {
 891         "item"     | // always terminated by `}` or `;`
 892         "block"    | // exactly one token tree
 893         "ident"    | // exactly one token tree
 894         "literal"  | // exactly one token tree
 895         "meta"     | // exactly one token tree
 896         "lifetime" | // exactly one token tree
 897         "tt" =>   // exactly one token tree
 898             true,
 899
 900         _ =>
 901             false,
 902     }
 903 }
 904
 905 enum IsInFollow {
 906     Yes,
 907     No(Vec<&'static str>),
 908     Invalid(String, &'static str),
 909 }
 910
 911 /// True if `frag` can legally be followed by the token `tok`. For
 912 /// fragments that can consume an unbounded number of tokens, `tok`
 913 /// must be within a well-defined follow set. This is intended to
 914 /// guarantee future compatibility: for example, without this rule, if
 915 /// we expanded `expr` to include a new binary operator, we might
 916 /// break macros that were relying on that binary operator as a
 917 /// separator.
 918 // when changing this do not forget to update doc/book/macros.md!
 919 fn is_in_follow(tok: &quoted::TokenTree, frag: &str) -> IsInFollow {
 920     use self::quoted::TokenTree;
 921
 922     if let TokenTree::Token(_, token::CloseDelim(_)) = *tok {
 923         // closing a token tree can never be matched by any fragment;
 924         // iow, we always require that `(` and `)` match, etc.
 925         IsInFollow::Yes
 926     } else {
 927         match frag {
 928             "item" => {
 929                 // since items *must* be followed by either a `;` or a `}`, we can
 930                 // accept anything after them
 931                 IsInFollow::Yes
 932             },
 933             "block" => {
 934                 // anything can follow block, the braces provide an easy boundary to
 935                 // maintain
 936                 IsInFollow::Yes
 937             },
 938             "stmt" | "expr"  => {
 939                 let tokens = vec!["`=>`", "`,`", "`;`"];
 940                 match *tok {
 941                     TokenTree::Token(_, ref tok) => match *tok {
 942                         FatArrow | Comma | Semi => IsInFollow::Yes,
 943                         _ => IsInFollow::No(tokens),
 944                     },
 945                     _ => IsInFollow::No(tokens),
 946                 }
 947             },
 948             "pat" => {
 949                 let tokens = vec!["`=>`", "`,`", "`=`", "`|`", "`if`", "`in`"];
 950                 match *tok {
 951                     TokenTree::Token(_, ref tok) => match *tok {
 952                         FatArrow | Comma | Eq | BinOp(token::Or) => IsInFollow::Yes,
 953                         Ident(i, false) if i.name == "if" || i.name == "in" => IsInFollow::Yes,
 954                         _ => IsInFollow::No(tokens),
 955                     },
 956                     _ => IsInFollow::No(tokens),
 957                 }
 958             },
 959             "path" | "ty" => {
 960                 let tokens = vec![
 961                     "`{`", "`[`", "`=>`", "`,`", "`>`","`=`", "`:`", "`;`", "`|`", "`as`",
 962                     "`where`",
 963                 ];
 964                 match *tok {
 965                     TokenTree::Token(_, ref tok) => match *tok {
 966                         OpenDelim(token::DelimToken::Brace) |
 967                         OpenDelim(token::DelimToken::Bracket) |
 968                         Comma | FatArrow | Colon | Eq | Gt | BinOp(token::Shr) | Semi |
 969                         BinOp(token::Or) => IsInFollow::Yes,
 970                         Ident(i, false) if i.name == "as" || i.name == "where" => IsInFollow::Yes,
 971                         _ => IsInFollow::No(tokens),
 972                     },
 973                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "block" => IsInFollow::Yes,
 974                     _ => IsInFollow::No(tokens),
 975                 }
 976             },
 977             "ident" | "lifetime" => {
 978                 // being a single token, idents and lifetimes are harmless
 979                 IsInFollow::Yes
 980             },
 981             "literal" => {
 982                 // literals may be of a single token, or two tokens (negative numbers)
 983                 IsInFollow::Yes
 984             },
 985             "meta" | "tt" => {
 986                 // being either a single token or a delimited sequence, tt is
 987                 // harmless
 988                 IsInFollow::Yes
 989             },
 990             "vis" => {
 991                 // Explicitly disallow `priv`, on the off chance it comes back.
 992                 let tokens = vec!["`,`", "an ident", "a type"];
 993                 match *tok {
 994                     TokenTree::Token(_, ref tok) => match *tok {
 995                         Comma => IsInFollow::Yes,
 996                         Ident(i, is_raw) if is_raw || i.name != "priv" => IsInFollow::Yes,
 997                         ref tok => if tok.can_begin_type() {
 998                             IsInFollow::Yes
 999                         } else {
1000                             IsInFollow::No(tokens)
1001                         }
1002                     },
1003                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "ident"
1004                                                        || frag.name == "ty"
1005                                                        || frag.name == "path" => IsInFollow::Yes,
1006                     _ => IsInFollow::No(tokens),
1007                 }
1008             },
1009             "" => IsInFollow::Yes, // keywords::Invalid
1010             _ => IsInFollow::Invalid(format!("invalid fragment specifier `{}`", frag),
1011                                      VALID_FRAGMENT_NAMES_MSG),
1012         }
1013     }
1014 }
1015
1016 fn has_legal_fragment_specifier(sess: &ParseSess,
1017                                 features: &Features,
1018                                 attrs: &[ast::Attribute],
1019                                 tok: &quoted::TokenTree) -> Result<(), String> {
1020     debug!("has_legal_fragment_specifier({:?})", tok);
1021     if let quoted::TokenTree::MetaVarDecl(_, _, ref frag_spec) = *tok {
1022         let frag_name = frag_spec.as_str();
1023         let frag_span = tok.span();
1024         if !is_legal_fragment_specifier(sess, features, attrs, &frag_name, frag_span) {
1025             return Err(frag_name.to_string());
1026         }
1027     }
1028     Ok(())
1029 }
1030
1031 fn is_legal_fragment_specifier(sess: &ParseSess,
1032                                features: &Features,
1033                                attrs: &[ast::Attribute],
1034                                frag_name: &str,
1035                                frag_span: Span) -> bool {
1036     match frag_name {
1037         "item" | "block" | "stmt" | "expr" | "pat" | "lifetime" |
1038         "path" | "ty" | "ident" | "meta" | "tt" | "vis" | "" => true,
1039         "literal" => {
1040             if !features.macro_literal_matcher &&
1041                !attr::contains_name(attrs, "allow_internal_unstable") {
1042                 let explain = feature_gate::EXPLAIN_LITERAL_MATCHER;
1043                 emit_feature_err(sess,
1044                                  "macro_literal_matcher",
1045                                  frag_span,
1046                                  GateIssue::Language,
1047                                  explain);
1048             }
1049             true
1050         },
1051         _ => false,
1052     }
1053 }
1054
1055 fn quoted_tt_to_string(tt: &quoted::TokenTree) -> String {
1056     match *tt {
1057         quoted::TokenTree::Token(_, ref tok) => ::print::pprust::token_to_string(tok),
1058         quoted::TokenTree::MetaVar(_, name) => format!("${}", name),
1059         quoted::TokenTree::MetaVarDecl(_, name, kind) => format!("${}:{}", name, kind),
1060         _ => panic!("unexpected quoted::TokenTree::{{Sequence or Delimited}} \
1061                      in follow set checker"),
1062     }
1063 }