src/libsyntax/ext/tt/macro_rules.rs

   1 use {ast, attr};
   2 use syntax_pos::{Span, DUMMY_SP};
   3 use edition::Edition;
   4 use errors::FatalError;
   5 use ext::base::{DummyResult, ExtCtxt, MacResult, SyntaxExtension};
   6 use ext::base::{NormalTT, TTMacroExpander};
   7 use ext::expand::{AstFragment, AstFragmentKind};
   8 use ext::tt::macro_parser::{Success, Error, Failure};
   9 use ext::tt::macro_parser::{MatchedSeq, MatchedNonterminal};
  10 use ext::tt::macro_parser::{parse, parse_failure_msg};
  11 use ext::tt::quoted;
  12 use ext::tt::transcribe::transcribe;
  13 use feature_gate::Features;
  14 use parse::{Directory, ParseSess};
  15 use parse::parser::Parser;
  16 use parse::token::{self, NtTT};
  17 use parse::token::Token::*;
  18 use symbol::Symbol;
  19 use tokenstream::{DelimSpan, TokenStream, TokenTree};
  20
  21 use rustc_data_structures::fx::FxHashMap;
  22 use std::borrow::Cow;
  23 use std::collections::hash_map::Entry;
  24
  25 use rustc_data_structures::sync::Lrc;
  26 use errors::Applicability;
  27
  28 const VALID_FRAGMENT_NAMES_MSG: &str = "valid fragment specifiers are \
  29     `ident`, `block`, `stmt`, `expr`, `pat`, `ty`, `lifetime`, `literal`, \
  30     `path`, `meta`, `tt`, `item` and `vis`";
  31
  32 pub struct ParserAnyMacro<'a> {
  33     parser: Parser<'a>,
  34
  35     /// Span of the expansion site of the macro this parser is for
  36     site_span: Span,
  37     /// The ident of the macro we're parsing
  38     macro_ident: ast::Ident,
  39     arm_span: Span,
  40 }
  41
  42 impl<'a> ParserAnyMacro<'a> {
  43     pub fn make(mut self: Box<ParserAnyMacro<'a>>, kind: AstFragmentKind) -> AstFragment {
  44         let ParserAnyMacro { site_span, macro_ident, ref mut parser, arm_span } = *self;
  45         let fragment = panictry!(parser.parse_ast_fragment(kind, true).map_err(|mut e| {
  46             if parser.token == token::Eof && e.message().ends_with(", found `<eof>`") {
  47                 if !e.span.is_dummy() {  // early end of macro arm (#52866)
  48                     e.replace_span_with(parser.sess.source_map().next_point(parser.span));
  49                 }
  50                 let msg = &e.message[0];
  51                 e.message[0] = (
  52                     format!(
  53                         "macro expansion ends with an incomplete expression: {}",
  54                         msg.0.replace(", found `<eof>`", ""),
  55                     ),
  56                     msg.1,
  57                 );
  58             }
  59             if e.span.is_dummy() {  // Get around lack of span in error (#30128)
  60                 e.replace_span_with(site_span);
  61                 if parser.sess.source_map().span_to_filename(arm_span).is_real() {
  62                     e.span_label(arm_span, "in this macro arm");
  63                 }
  64             } else if !parser.sess.source_map().span_to_filename(parser.span).is_real() {
  65                 e.span_label(site_span, "in this macro invocation");
  66             }
  67             e
  68         }));
  69
  70         // We allow semicolons at the end of expressions -- e.g., the semicolon in
  71         // `macro_rules! m { () => { panic!(); } }` isn't parsed by `.parse_expr()`,
  72         // but `m!()` is allowed in expression positions (cf. issue #34706).
  73         if kind == AstFragmentKind::Expr && parser.token == token::Semi {
  74             parser.bump();
  75         }
  76
  77         // Make sure we don't have any tokens left to parse so we don't silently drop anything.
  78         let path = ast::Path::from_ident(macro_ident.with_span_pos(site_span));
  79         parser.ensure_complete_parse(&path, kind.name(), site_span);
  80         fragment
  81     }
  82 }
  83
  84 struct MacroRulesMacroExpander {
  85     name: ast::Ident,
  86     lhses: Vec<quoted::TokenTree>,
  87     rhses: Vec<quoted::TokenTree>,
  88     valid: bool,
  89 }
  90
  91 impl TTMacroExpander for MacroRulesMacroExpander {
  92     fn expand<'cx>(
  93         &self,
  94         cx: &'cx mut ExtCtxt,
  95         sp: Span,
  96         input: TokenStream,
  97         def_span: Option<Span>,
  98     ) -> Box<dyn MacResult+'cx> {
  99         if !self.valid {
 100             return DummyResult::any(sp);
 101         }
 102         generic_extension(cx,
 103                           sp,
 104                           def_span,
 105                           self.name,
 106                           input,
 107                           &self.lhses,
 108                           &self.rhses)
 109     }
 110 }
 111
 112 fn trace_macros_note(cx: &mut ExtCtxt, sp: Span, message: String) {
 113     let sp = sp.macro_backtrace().last().map(|trace| trace.call_site).unwrap_or(sp);
 114     cx.expansions.entry(sp).or_default().push(message);
 115 }
 116
 117 /// Given `lhses` and `rhses`, this is the new macro we create
 118 fn generic_extension<'cx>(cx: &'cx mut ExtCtxt,
 119                           sp: Span,
 120                           def_span: Option<Span>,
 121                           name: ast::Ident,
 122                           arg: TokenStream,
 123                           lhses: &[quoted::TokenTree],
 124                           rhses: &[quoted::TokenTree])
 125                           -> Box<dyn MacResult+'cx> {
 126     if cx.trace_macros() {
 127         trace_macros_note(cx, sp, format!("expanding `{}! {{ {} }}`", name, arg));
 128     }
 129
 130     // Which arm's failure should we report? (the one furthest along)
 131     let mut best_fail_spot = DUMMY_SP;
 132     let mut best_fail_tok = None;
 133     let mut best_fail_text = None;
 134
 135     for (i, lhs) in lhses.iter().enumerate() { // try each arm's matchers
 136         let lhs_tt = match *lhs {
 137             quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 138             _ => cx.span_bug(sp, "malformed macro lhs")
 139         };
 140
 141         match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 142             Success(named_matches) => {
 143                 let rhs = match rhses[i] {
 144                     // ignore delimiters
 145                     quoted::TokenTree::Delimited(_, ref delimed) => delimed.tts.clone(),
 146                     _ => cx.span_bug(sp, "malformed macro rhs"),
 147                 };
 148                 let arm_span = rhses[i].span();
 149
 150                 let rhs_spans = rhs.iter().map(|t| t.span()).collect::<Vec<_>>();
 151                 // rhs has holes ( `$id` and `$(...)` that need filled)
 152                 let mut tts = transcribe(cx, Some(named_matches), rhs);
 153
 154                 // Replace all the tokens for the corresponding positions in the macro, to maintain
 155                 // proper positions in error reporting, while maintaining the macro_backtrace.
 156                 if rhs_spans.len() == tts.len() {
 157                     tts = tts.map_enumerated(|i, mut tt| {
 158                         let mut sp = rhs_spans[i];
 159                         sp = sp.with_ctxt(tt.span().ctxt());
 160                         tt.set_span(sp);
 161                         tt
 162                     });
 163                 }
 164
 165                 if cx.trace_macros() {
 166                     trace_macros_note(cx, sp, format!("to `{}`", tts));
 167                 }
 168
 169                 let directory = Directory {
 170                     path: Cow::from(cx.current_expansion.module.directory.as_path()),
 171                     ownership: cx.current_expansion.directory_ownership,
 172                 };
 173                 let mut p = Parser::new(cx.parse_sess(), tts, Some(directory), true, false);
 174                 p.root_module_name = cx.current_expansion.module.mod_path.last()
 175                     .map(|id| id.as_str().to_string());
 176
 177                 p.process_potential_macro_variable();
 178                 // Let the context choose how to interpret the result.
 179                 // Weird, but useful for X-macros.
 180                 return Box::new(ParserAnyMacro {
 181                     parser: p,
 182
 183                     // Pass along the original expansion site and the name of the macro
 184                     // so we can print a useful error message if the parse of the expanded
 185                     // macro leaves unparsed tokens.
 186                     site_span: sp,
 187                     macro_ident: name,
 188                     arm_span,
 189                 })
 190             }
 191             Failure(sp, tok, t) => if sp.lo() >= best_fail_spot.lo() {
 192                 best_fail_spot = sp;
 193                 best_fail_tok = Some(tok);
 194                 best_fail_text = Some(t);
 195             },
 196             Error(err_sp, ref msg) => {
 197                 cx.span_fatal(err_sp.substitute_dummy(sp), &msg[..])
 198             }
 199         }
 200     }
 201
 202     let best_fail_msg = parse_failure_msg(best_fail_tok.expect("ran no matchers"));
 203     let span = best_fail_spot.substitute_dummy(sp);
 204     let mut err = cx.struct_span_err(span, &best_fail_msg);
 205     err.span_label(span, best_fail_text.unwrap_or(&best_fail_msg));
 206     if let Some(sp) = def_span {
 207         if cx.source_map().span_to_filename(sp).is_real() && !sp.is_dummy() {
 208             err.span_label(cx.source_map().def_span(sp), "when calling this macro");
 209         }
 210     }
 211
 212     // Check whether there's a missing comma in this macro call, like `println!("{}" a);`
 213     if let Some((arg, comma_span)) = arg.add_comma() {
 214         for lhs in lhses { // try each arm's matchers
 215             let lhs_tt = match *lhs {
 216                 quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 217                 _ => continue,
 218             };
 219             match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 220                 Success(_) => {
 221                     if comma_span.is_dummy() {
 222                         err.note("you might be missing a comma");
 223                     } else {
 224                         err.span_suggestion_short(
 225                             comma_span,
 226                             "missing comma here",
 227                             ", ".to_string(),
 228                             Applicability::MachineApplicable,
 229                         );
 230                     }
 231                 }
 232                 _ => {}
 233             }
 234         }
 235     }
 236     err.emit();
 237     cx.trace_macros_diag();
 238     DummyResult::any(sp)
 239 }
 240
 241 // Note that macro-by-example's input is also matched against a token tree:
 242 //                   $( $lhs:tt => $rhs:tt );+
 243 //
 244 // Holy self-referential!
 245
 246 /// Converts a `macro_rules!` invocation into a syntax extension.
 247 pub fn compile(sess: &ParseSess, features: &Features, def: &ast::Item, edition: Edition)
 248                -> SyntaxExtension {
 249     let lhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("lhs"));
 250     let rhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("rhs"));
 251
 252     // Parse the macro_rules! invocation
 253     let body = match def.node {
 254         ast::ItemKind::MacroDef(ref body) => body,
 255         _ => unreachable!(),
 256     };
 257
 258     // The pattern that macro_rules matches.
 259     // The grammar for macro_rules! is:
 260     // $( $lhs:tt => $rhs:tt );+
 261     // ...quasiquoting this would be nice.
 262     // These spans won't matter, anyways
 263     let argument_gram = vec![
 264         quoted::TokenTree::Sequence(DelimSpan::dummy(), Lrc::new(quoted::SequenceRepetition {
 265             tts: vec![
 266                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, lhs_nm, ast::Ident::from_str("tt")),
 267                 quoted::TokenTree::Token(DUMMY_SP, token::FatArrow),
 268                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, rhs_nm, ast::Ident::from_str("tt")),
 269             ],
 270             separator: Some(if body.legacy { token::Semi } else { token::Comma }),
 271             op: quoted::KleeneOp::OneOrMore,
 272             num_captures: 2,
 273         })),
 274         // to phase into semicolon-termination instead of semicolon-separation
 275         quoted::TokenTree::Sequence(DelimSpan::dummy(), Lrc::new(quoted::SequenceRepetition {
 276             tts: vec![quoted::TokenTree::Token(DUMMY_SP, token::Semi)],
 277             separator: None,
 278             op: quoted::KleeneOp::ZeroOrMore,
 279             num_captures: 0
 280         })),
 281     ];
 282
 283     let argument_map = match parse(sess, body.stream(), &argument_gram, None, true) {
 284         Success(m) => m,
 285         Failure(sp, tok, t) => {
 286             let s = parse_failure_msg(tok);
 287             let sp = sp.substitute_dummy(def.span);
 288             let mut err = sess.span_diagnostic.struct_span_fatal(sp, &s);
 289             err.span_label(sp, t);
 290             err.emit();
 291             FatalError.raise();
 292         }
 293         Error(sp, s) => {
 294             sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s).raise();
 295         }
 296     };
 297
 298     let mut valid = true;
 299
 300     // Extract the arguments:
 301     let lhses = match *argument_map[&lhs_nm] {
 302         MatchedSeq(ref s, _) => {
 303             s.iter().map(|m| {
 304                 if let MatchedNonterminal(ref nt) = *m {
 305                     if let NtTT(ref tt) = **nt {
 306                         let tt = quoted::parse(
 307                             tt.clone().into(),
 308                             true,
 309                             sess,
 310                             features,
 311                             &def.attrs,
 312                             edition,
 313                             def.id,
 314                         )
 315                         .pop()
 316                         .unwrap();
 317                         valid &= check_lhs_nt_follows(sess, features, &def.attrs, &tt);
 318                         return tt;
 319                     }
 320                 }
 321                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 322             }).collect::<Vec<quoted::TokenTree>>()
 323         }
 324         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 325     };
 326
 327     let rhses = match *argument_map[&rhs_nm] {
 328         MatchedSeq(ref s, _) => {
 329             s.iter().map(|m| {
 330                 if let MatchedNonterminal(ref nt) = *m {
 331                     if let NtTT(ref tt) = **nt {
 332                         return quoted::parse(
 333                             tt.clone().into(),
 334                             false,
 335                             sess,
 336                             features,
 337                             &def.attrs,
 338                             edition,
 339                             def.id,
 340                         ).pop()
 341                          .unwrap();
 342                     }
 343                 }
 344                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 345             }).collect::<Vec<quoted::TokenTree>>()
 346         }
 347         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured rhs")
 348     };
 349
 350     for rhs in &rhses {
 351         valid &= check_rhs(sess, rhs);
 352     }
 353
 354     // don't abort iteration early, so that errors for multiple lhses can be reported
 355     for lhs in &lhses {
 356         valid &= check_lhs_no_empty_seq(sess, &[lhs.clone()])
 357     }
 358
 359     let expander: Box<_> = Box::new(MacroRulesMacroExpander {
 360         name: def.ident,
 361         lhses,
 362         rhses,
 363         valid,
 364     });
 365
 366     if body.legacy {
 367         let allow_internal_unstable = attr::contains_name(&def.attrs, "allow_internal_unstable");
 368         let allow_internal_unsafe = attr::contains_name(&def.attrs, "allow_internal_unsafe");
 369         let mut local_inner_macros = false;
 370         if let Some(macro_export) = attr::find_by_name(&def.attrs, "macro_export") {
 371             if let Some(l) = macro_export.meta_item_list() {
 372                 local_inner_macros = attr::list_contains_name(&l, "local_inner_macros");
 373             }
 374         }
 375
 376         let unstable_feature = attr::find_stability(&sess,
 377                                                     &def.attrs, def.span).and_then(|stability| {
 378             if let attr::StabilityLevel::Unstable { issue, .. } = stability.level {
 379                 Some((stability.feature, issue))
 380             } else {
 381                 None
 382             }
 383         });
 384
 385         NormalTT {
 386             expander,
 387             def_info: Some((def.id, def.span)),
 388             allow_internal_unstable,
 389             allow_internal_unsafe,
 390             local_inner_macros,
 391             unstable_feature,
 392             edition,
 393         }
 394     } else {
 395         let is_transparent = attr::contains_name(&def.attrs, "rustc_transparent_macro");
 396
 397         SyntaxExtension::DeclMacro {
 398             expander,
 399             def_info: Some((def.id, def.span)),
 400             is_transparent,
 401             edition,
 402         }
 403     }
 404 }
 405
 406 fn check_lhs_nt_follows(sess: &ParseSess,
 407                         features: &Features,
 408                         attrs: &[ast::Attribute],
 409                         lhs: &quoted::TokenTree) -> bool {
 410     // lhs is going to be like TokenTree::Delimited(...), where the
 411     // entire lhs is those tts. Or, it can be a "bare sequence", not wrapped in parens.
 412     if let quoted::TokenTree::Delimited(_, ref tts) = *lhs {
 413         check_matcher(sess, features, attrs, &tts.tts)
 414     } else {
 415         let msg = "invalid macro matcher; matchers must be contained in balanced delimiters";
 416         sess.span_diagnostic.span_err(lhs.span(), msg);
 417         false
 418     }
 419     // we don't abort on errors on rejection, the driver will do that for us
 420     // after parsing/expansion. we can report every error in every macro this way.
 421 }
 422
 423 /// Check that the lhs contains no repetition which could match an empty token
 424 /// tree, because then the matcher would hang indefinitely.
 425 fn check_lhs_no_empty_seq(sess: &ParseSess, tts: &[quoted::TokenTree]) -> bool {
 426     use self::quoted::TokenTree;
 427     for tt in tts {
 428         match *tt {
 429             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => (),
 430             TokenTree::Delimited(_, ref del) => if !check_lhs_no_empty_seq(sess, &del.tts) {
 431                 return false;
 432             },
 433             TokenTree::Sequence(span, ref seq) => {
 434                 if seq.separator.is_none() && seq.tts.iter().all(|seq_tt| {
 435                     match *seq_tt {
 436                         TokenTree::MetaVarDecl(_, _, id) => id.name == "vis",
 437                         TokenTree::Sequence(_, ref sub_seq) =>
 438                             sub_seq.op == quoted::KleeneOp::ZeroOrMore
 439                             || sub_seq.op == quoted::KleeneOp::ZeroOrOne,
 440                         _ => false,
 441                     }
 442                 }) {
 443                     let sp = span.entire();
 444                     sess.span_diagnostic.span_err(sp, "repetition matches empty token tree");
 445                     return false;
 446                 }
 447                 if !check_lhs_no_empty_seq(sess, &seq.tts) {
 448                     return false;
 449                 }
 450             }
 451         }
 452     }
 453
 454     true
 455 }
 456
 457 fn check_rhs(sess: &ParseSess, rhs: &quoted::TokenTree) -> bool {
 458     match *rhs {
 459         quoted::TokenTree::Delimited(..) => return true,
 460         _ => sess.span_diagnostic.span_err(rhs.span(), "macro rhs must be delimited")
 461     }
 462     false
 463 }
 464
 465 fn check_matcher(sess: &ParseSess,
 466                  features: &Features,
 467                  attrs: &[ast::Attribute],
 468                  matcher: &[quoted::TokenTree]) -> bool {
 469     let first_sets = FirstSets::new(matcher);
 470     let empty_suffix = TokenSet::empty();
 471     let err = sess.span_diagnostic.err_count();
 472     check_matcher_core(sess, features, attrs, &first_sets, matcher, &empty_suffix);
 473     err == sess.span_diagnostic.err_count()
 474 }
 475
 476 // `The FirstSets` for a matcher is a mapping from subsequences in the
 477 // matcher to the FIRST set for that subsequence.
 478 //
 479 // This mapping is partially precomputed via a backwards scan over the
 480 // token trees of the matcher, which provides a mapping from each
 481 // repetition sequence to its *first* set.
 482 //
 483 // (Hypothetically, sequences should be uniquely identifiable via their
 484 // spans, though perhaps that is false, e.g., for macro-generated macros
 485 // that do not try to inject artificial span information. My plan is
 486 // to try to catch such cases ahead of time and not include them in
 487 // the precomputed mapping.)
 488 struct FirstSets {
 489     // this maps each TokenTree::Sequence `$(tt ...) SEP OP` that is uniquely identified by its
 490     // span in the original matcher to the First set for the inner sequence `tt ...`.
 491     //
 492     // If two sequences have the same span in a matcher, then map that
 493     // span to None (invalidating the mapping here and forcing the code to
 494     // use a slow path).
 495     first: FxHashMap<Span, Option<TokenSet>>,
 496 }
 497
 498 impl FirstSets {
 499     fn new(tts: &[quoted::TokenTree]) -> FirstSets {
 500         use self::quoted::TokenTree;
 501
 502         let mut sets = FirstSets { first: FxHashMap::default() };
 503         build_recur(&mut sets, tts);
 504         return sets;
 505
 506         // walks backward over `tts`, returning the FIRST for `tts`
 507         // and updating `sets` at the same time for all sequence
 508         // substructure we find within `tts`.
 509         fn build_recur(sets: &mut FirstSets, tts: &[TokenTree]) -> TokenSet {
 510             let mut first = TokenSet::empty();
 511             for tt in tts.iter().rev() {
 512                 match *tt {
 513                     TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 514                         first.replace_with(tt.clone());
 515                     }
 516                     TokenTree::Delimited(span, ref delimited) => {
 517                         build_recur(sets, &delimited.tts[..]);
 518                         first.replace_with(delimited.open_tt(span.open));
 519                     }
 520                     TokenTree::Sequence(sp, ref seq_rep) => {
 521                         let subfirst = build_recur(sets, &seq_rep.tts[..]);
 522
 523                         match sets.first.entry(sp.entire()) {
 524                             Entry::Vacant(vac) => {
 525                                 vac.insert(Some(subfirst.clone()));
 526                             }
 527                             Entry::Occupied(mut occ) => {
 528                                 // if there is already an entry, then a span must have collided.
 529                                 // This should not happen with typical macro_rules macros,
 530                                 // but syntax extensions need not maintain distinct spans,
 531                                 // so distinct syntax trees can be assigned the same span.
 532                                 // In such a case, the map cannot be trusted; so mark this
 533                                 // entry as unusable.
 534                                 occ.insert(None);
 535                             }
 536                         }
 537
 538                         // If the sequence contents can be empty, then the first
 539                         // token could be the separator token itself.
 540
 541                         if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 542                                                         subfirst.maybe_empty) {
 543                             first.add_one_maybe(TokenTree::Token(sp.entire(), sep.clone()));
 544                         }
 545
 546                         // Reverse scan: Sequence comes before `first`.
 547                         if subfirst.maybe_empty
 548                            || seq_rep.op == quoted::KleeneOp::ZeroOrMore
 549                            || seq_rep.op == quoted::KleeneOp::ZeroOrOne
 550                         {
 551                             // If sequence is potentially empty, then
 552                             // union them (preserving first emptiness).
 553                             first.add_all(&TokenSet { maybe_empty: true, ..subfirst });
 554                         } else {
 555                             // Otherwise, sequence guaranteed
 556                             // non-empty; replace first.
 557                             first = subfirst;
 558                         }
 559                     }
 560                 }
 561             }
 562
 563             first
 564         }
 565     }
 566
 567     // walks forward over `tts` until all potential FIRST tokens are
 568     // identified.
 569     fn first(&self, tts: &[quoted::TokenTree]) -> TokenSet {
 570         use self::quoted::TokenTree;
 571
 572         let mut first = TokenSet::empty();
 573         for tt in tts.iter() {
 574             assert!(first.maybe_empty);
 575             match *tt {
 576                 TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 577                     first.add_one(tt.clone());
 578                     return first;
 579                 }
 580                 TokenTree::Delimited(span, ref delimited) => {
 581                     first.add_one(delimited.open_tt(span.open));
 582                     return first;
 583                 }
 584                 TokenTree::Sequence(sp, ref seq_rep) => {
 585                     match self.first.get(&sp.entire()) {
 586                         Some(&Some(ref subfirst)) => {
 587
 588                             // If the sequence contents can be empty, then the first
 589                             // token could be the separator token itself.
 590
 591                             if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 592                                                             subfirst.maybe_empty) {
 593                                 first.add_one_maybe(TokenTree::Token(sp.entire(), sep.clone()));
 594                             }
 595
 596                             assert!(first.maybe_empty);
 597                             first.add_all(subfirst);
 598                             if subfirst.maybe_empty
 599                                || seq_rep.op == quoted::KleeneOp::ZeroOrMore
 600                                || seq_rep.op == quoted::KleeneOp::ZeroOrOne
 601                             {
 602                                 // continue scanning for more first
 603                                 // tokens, but also make sure we
 604                                 // restore empty-tracking state
 605                                 first.maybe_empty = true;
 606                                 continue;
 607                             } else {
 608                                 return first;
 609                             }
 610                         }
 611
 612                         Some(&None) => {
 613                             panic!("assume all sequences have (unique) spans for now");
 614                         }
 615
 616                         None => {
 617                             panic!("We missed a sequence during FirstSets construction");
 618                         }
 619                     }
 620                 }
 621             }
 622         }
 623
 624         // we only exit the loop if `tts` was empty or if every
 625         // element of `tts` matches the empty sequence.
 626         assert!(first.maybe_empty);
 627         first
 628     }
 629 }
 630
 631 // A set of `quoted::TokenTree`s, which may include `TokenTree::Match`s
 632 // (for macro-by-example syntactic variables). It also carries the
 633 // `maybe_empty` flag; that is true if and only if the matcher can
 634 // match an empty token sequence.
 635 //
 636 // The First set is computed on submatchers like `$($a:expr b),* $(c)* d`,
 637 // which has corresponding FIRST = {$a:expr, c, d}.
 638 // Likewise, `$($a:expr b),* $(c)+ d` has FIRST = {$a:expr, c}.
 639 //
 640 // (Notably, we must allow for *-op to occur zero times.)
 641 #[derive(Clone, Debug)]
 642 struct TokenSet {
 643     tokens: Vec<quoted::TokenTree>,
 644     maybe_empty: bool,
 645 }
 646
 647 impl TokenSet {
 648     // Returns a set for the empty sequence.
 649     fn empty() -> Self { TokenSet { tokens: Vec::new(), maybe_empty: true } }
 650
 651     // Returns the set `{ tok }` for the single-token (and thus
 652     // non-empty) sequence [tok].
 653     fn singleton(tok: quoted::TokenTree) -> Self {
 654         TokenSet { tokens: vec![tok], maybe_empty: false }
 655     }
 656
 657     // Changes self to be the set `{ tok }`.
 658     // Since `tok` is always present, marks self as non-empty.
 659     fn replace_with(&mut self, tok: quoted::TokenTree) {
 660         self.tokens.clear();
 661         self.tokens.push(tok);
 662         self.maybe_empty = false;
 663     }
 664
 665     // Changes self to be the empty set `{}`; meant for use when
 666     // the particular token does not matter, but we want to
 667     // record that it occurs.
 668     fn replace_with_irrelevant(&mut self) {
 669         self.tokens.clear();
 670         self.maybe_empty = false;
 671     }
 672
 673     // Adds `tok` to the set for `self`, marking sequence as non-empy.
 674     fn add_one(&mut self, tok: quoted::TokenTree) {
 675         if !self.tokens.contains(&tok) {
 676             self.tokens.push(tok);
 677         }
 678         self.maybe_empty = false;
 679     }
 680
 681     // Adds `tok` to the set for `self`. (Leaves `maybe_empty` flag alone.)
 682     fn add_one_maybe(&mut self, tok: quoted::TokenTree) {
 683         if !self.tokens.contains(&tok) {
 684             self.tokens.push(tok);
 685         }
 686     }
 687
 688     // Adds all elements of `other` to this.
 689     //
 690     // (Since this is a set, we filter out duplicates.)
 691     //
 692     // If `other` is potentially empty, then preserves the previous
 693     // setting of the empty flag of `self`. If `other` is guaranteed
 694     // non-empty, then `self` is marked non-empty.
 695     fn add_all(&mut self, other: &Self) {
 696         for tok in &other.tokens {
 697             if !self.tokens.contains(tok) {
 698                 self.tokens.push(tok.clone());
 699             }
 700         }
 701         if !other.maybe_empty {
 702             self.maybe_empty = false;
 703         }
 704     }
 705 }
 706
 707 // Checks that `matcher` is internally consistent and that it
 708 // can legally by followed by a token N, for all N in `follow`.
 709 // (If `follow` is empty, then it imposes no constraint on
 710 // the `matcher`.)
 711 //
 712 // Returns the set of NT tokens that could possibly come last in
 713 // `matcher`. (If `matcher` matches the empty sequence, then
 714 // `maybe_empty` will be set to true.)
 715 //
 716 // Requires that `first_sets` is pre-computed for `matcher`;
 717 // see `FirstSets::new`.
 718 fn check_matcher_core(sess: &ParseSess,
 719                       features: &Features,
 720                       attrs: &[ast::Attribute],
 721                       first_sets: &FirstSets,
 722                       matcher: &[quoted::TokenTree],
 723                       follow: &TokenSet) -> TokenSet {
 724     use self::quoted::TokenTree;
 725
 726     let mut last = TokenSet::empty();
 727
 728     // 2. For each token and suffix  [T, SUFFIX] in M:
 729     // ensure that T can be followed by SUFFIX, and if SUFFIX may be empty,
 730     // then ensure T can also be followed by any element of FOLLOW.
 731     'each_token: for i in 0..matcher.len() {
 732         let token = &matcher[i];
 733         let suffix = &matcher[i+1..];
 734
 735         let build_suffix_first = || {
 736             let mut s = first_sets.first(suffix);
 737             if s.maybe_empty { s.add_all(follow); }
 738             s
 739         };
 740
 741         // (we build `suffix_first` on demand below; you can tell
 742         // which cases are supposed to fall through by looking for the
 743         // initialization of this variable.)
 744         let suffix_first;
 745
 746         // First, update `last` so that it corresponds to the set
 747         // of NT tokens that might end the sequence `... token`.
 748         match *token {
 749             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 750                 let can_be_followed_by_any;
 751                 if let Err(bad_frag) = has_legal_fragment_specifier(sess, features, attrs, token) {
 752                     let msg = format!("invalid fragment specifier `{}`", bad_frag);
 753                     sess.span_diagnostic.struct_span_err(token.span(), &msg)
 754                         .help(VALID_FRAGMENT_NAMES_MSG)
 755                         .emit();
 756                     // (This eliminates false positives and duplicates
 757                     // from error messages.)
 758                     can_be_followed_by_any = true;
 759                 } else {
 760                     can_be_followed_by_any = token_can_be_followed_by_any(token);
 761                 }
 762
 763                 if can_be_followed_by_any {
 764                     // don't need to track tokens that work with any,
 765                     last.replace_with_irrelevant();
 766                     // ... and don't need to check tokens that can be
 767                     // followed by anything against SUFFIX.
 768                     continue 'each_token;
 769                 } else {
 770                     last.replace_with(token.clone());
 771                     suffix_first = build_suffix_first();
 772                 }
 773             }
 774             TokenTree::Delimited(span, ref d) => {
 775                 let my_suffix = TokenSet::singleton(d.close_tt(span.close));
 776                 check_matcher_core(sess, features, attrs, first_sets, &d.tts, &my_suffix);
 777                 // don't track non NT tokens
 778                 last.replace_with_irrelevant();
 779
 780                 // also, we don't need to check delimited sequences
 781                 // against SUFFIX
 782                 continue 'each_token;
 783             }
 784             TokenTree::Sequence(sp, ref seq_rep) => {
 785                 suffix_first = build_suffix_first();
 786                 // The trick here: when we check the interior, we want
 787                 // to include the separator (if any) as a potential
 788                 // (but not guaranteed) element of FOLLOW. So in that
 789                 // case, we make a temp copy of suffix and stuff
 790                 // delimiter in there.
 791                 //
 792                 // FIXME: Should I first scan suffix_first to see if
 793                 // delimiter is already in it before I go through the
 794                 // work of cloning it? But then again, this way I may
 795                 // get a "tighter" span?
 796                 let mut new;
 797                 let my_suffix = if let Some(ref u) = seq_rep.separator {
 798                     new = suffix_first.clone();
 799                     new.add_one_maybe(TokenTree::Token(sp.entire(), u.clone()));
 800                     &new
 801                 } else {
 802                     &suffix_first
 803                 };
 804
 805                 // At this point, `suffix_first` is built, and
 806                 // `my_suffix` is some TokenSet that we can use
 807                 // for checking the interior of `seq_rep`.
 808                 let next = check_matcher_core(sess,
 809                                               features,
 810                                               attrs,
 811                                               first_sets,
 812                                               &seq_rep.tts,
 813                                               my_suffix);
 814                 if next.maybe_empty {
 815                     last.add_all(&next);
 816                 } else {
 817                     last = next;
 818                 }
 819
 820                 // the recursive call to check_matcher_core already ran the 'each_last
 821                 // check below, so we can just keep going forward here.
 822                 continue 'each_token;
 823             }
 824         }
 825
 826         // (`suffix_first` guaranteed initialized once reaching here.)
 827
 828         // Now `last` holds the complete set of NT tokens that could
 829         // end the sequence before SUFFIX. Check that every one works with `suffix`.
 830         'each_last: for token in &last.tokens {
 831             if let TokenTree::MetaVarDecl(_, ref name, ref frag_spec) = *token {
 832                 for next_token in &suffix_first.tokens {
 833                     match is_in_follow(next_token, &frag_spec.as_str()) {
 834                         IsInFollow::Invalid(msg, help) => {
 835                             sess.span_diagnostic.struct_span_err(next_token.span(), &msg)
 836                                 .help(help).emit();
 837                             // don't bother reporting every source of
 838                             // conflict for a particular element of `last`.
 839                             continue 'each_last;
 840                         }
 841                         IsInFollow::Yes => {}
 842                         IsInFollow::No(ref possible) => {
 843                             let may_be = if last.tokens.len() == 1 &&
 844                                 suffix_first.tokens.len() == 1
 845                             {
 846                                 "is"
 847                             } else {
 848                                 "may be"
 849                             };
 850
 851                             let sp = next_token.span();
 852                             let mut err = sess.span_diagnostic.struct_span_err(
 853                                 sp,
 854                                 &format!("`${name}:{frag}` {may_be} followed by `{next}`, which \
 855                                           is not allowed for `{frag}` fragments",
 856                                          name=name,
 857                                          frag=frag_spec,
 858                                          next=quoted_tt_to_string(next_token),
 859                                          may_be=may_be),
 860                             );
 861                             err.span_label(
 862                                 sp,
 863                                 format!("not allowed after `{}` fragments", frag_spec),
 864                             );
 865                             let msg = "allowed there are: ";
 866                             match &possible[..] {
 867                                 &[] => {}
 868                                 &[t] => {
 869                                     err.note(&format!(
 870                                         "only {} is allowed after `{}` fragments",
 871                                         t,
 872                                         frag_spec,
 873                                     ));
 874                                 }
 875                                 ts => {
 876                                     err.note(&format!(
 877                                         "{}{} or {}",
 878                                         msg,
 879                                         ts[..ts.len() - 1].iter().map(|s| *s)
 880                                             .collect::<Vec<_>>().join(", "),
 881                                         ts[ts.len() - 1],
 882                                     ));
 883                                 }
 884                             }
 885                             err.emit();
 886                         }
 887                     }
 888                 }
 889             }
 890         }
 891     }
 892     last
 893 }
 894
 895 fn token_can_be_followed_by_any(tok: &quoted::TokenTree) -> bool {
 896     if let quoted::TokenTree::MetaVarDecl(_, _, frag_spec) = *tok {
 897         frag_can_be_followed_by_any(&frag_spec.as_str())
 898     } else {
 899         // (Non NT's can always be followed by anthing in matchers.)
 900         true
 901     }
 902 }
 903
 904 /// True if a fragment of type `frag` can be followed by any sort of
 905 /// token.  We use this (among other things) as a useful approximation
 906 /// for when `frag` can be followed by a repetition like `$(...)*` or
 907 /// `$(...)+`. In general, these can be a bit tricky to reason about,
 908 /// so we adopt a conservative position that says that any fragment
 909 /// specifier which consumes at most one token tree can be followed by
 910 /// a fragment specifier (indeed, these fragments can be followed by
 911 /// ANYTHING without fear of future compatibility hazards).
 912 fn frag_can_be_followed_by_any(frag: &str) -> bool {
 913     match frag {
 914         "item"     | // always terminated by `}` or `;`
 915         "block"    | // exactly one token tree
 916         "ident"    | // exactly one token tree
 917         "literal"  | // exactly one token tree
 918         "meta"     | // exactly one token tree
 919         "lifetime" | // exactly one token tree
 920         "tt" =>   // exactly one token tree
 921             true,
 922
 923         _ =>
 924             false,
 925     }
 926 }
 927
 928 enum IsInFollow {
 929     Yes,
 930     No(Vec<&'static str>),
 931     Invalid(String, &'static str),
 932 }
 933
 934 /// True if `frag` can legally be followed by the token `tok`. For
 935 /// fragments that can consume an unbounded number of tokens, `tok`
 936 /// must be within a well-defined follow set. This is intended to
 937 /// guarantee future compatibility: for example, without this rule, if
 938 /// we expanded `expr` to include a new binary operator, we might
 939 /// break macros that were relying on that binary operator as a
 940 /// separator.
 941 // when changing this do not forget to update doc/book/macros.md!
 942 fn is_in_follow(tok: &quoted::TokenTree, frag: &str) -> IsInFollow {
 943     use self::quoted::TokenTree;
 944
 945     if let TokenTree::Token(_, token::CloseDelim(_)) = *tok {
 946         // closing a token tree can never be matched by any fragment;
 947         // iow, we always require that `(` and `)` match, etc.
 948         IsInFollow::Yes
 949     } else {
 950         match frag {
 951             "item" => {
 952                 // since items *must* be followed by either a `;` or a `}`, we can
 953                 // accept anything after them
 954                 IsInFollow::Yes
 955             },
 956             "block" => {
 957                 // anything can follow block, the braces provide an easy boundary to
 958                 // maintain
 959                 IsInFollow::Yes
 960             },
 961             "stmt" | "expr"  => {
 962                 let tokens = vec!["`=>`", "`,`", "`;`"];
 963                 match *tok {
 964                     TokenTree::Token(_, ref tok) => match *tok {
 965                         FatArrow | Comma | Semi => IsInFollow::Yes,
 966                         _ => IsInFollow::No(tokens),
 967                     },
 968                     _ => IsInFollow::No(tokens),
 969                 }
 970             },
 971             "pat" => {
 972                 let tokens = vec!["`=>`", "`,`", "`=`", "`|`", "`if`", "`in`"];
 973                 match *tok {
 974                     TokenTree::Token(_, ref tok) => match *tok {
 975                         FatArrow | Comma | Eq | BinOp(token::Or) => IsInFollow::Yes,
 976                         Ident(i, false) if i.name == "if" || i.name == "in" => IsInFollow::Yes,
 977                         _ => IsInFollow::No(tokens),
 978                     },
 979                     _ => IsInFollow::No(tokens),
 980                 }
 981             },
 982             "path" | "ty" => {
 983                 let tokens = vec![
 984                     "`{`", "`[`", "`=>`", "`,`", "`>`","`=`", "`:`", "`;`", "`|`", "`as`",
 985                     "`where`",
 986                 ];
 987                 match *tok {
 988                     TokenTree::Token(_, ref tok) => match *tok {
 989                         OpenDelim(token::DelimToken::Brace) |
 990                         OpenDelim(token::DelimToken::Bracket) |
 991                         Comma | FatArrow | Colon | Eq | Gt | BinOp(token::Shr) | Semi |
 992                         BinOp(token::Or) => IsInFollow::Yes,
 993                         Ident(i, false) if i.name == "as" || i.name == "where" => IsInFollow::Yes,
 994                         _ => IsInFollow::No(tokens),
 995                     },
 996                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "block" => IsInFollow::Yes,
 997                     _ => IsInFollow::No(tokens),
 998                 }
 999             },
1000             "ident" | "lifetime" => {
1001                 // being a single token, idents and lifetimes are harmless
1002                 IsInFollow::Yes
1003             },
1004             "literal" => {
1005                 // literals may be of a single token, or two tokens (negative numbers)
1006                 IsInFollow::Yes
1007             },
1008             "meta" | "tt" => {
1009                 // being either a single token or a delimited sequence, tt is
1010                 // harmless
1011                 IsInFollow::Yes
1012             },
1013             "vis" => {
1014                 // Explicitly disallow `priv`, on the off chance it comes back.
1015                 let tokens = vec!["`,`", "an ident", "a type"];
1016                 match *tok {
1017                     TokenTree::Token(_, ref tok) => match *tok {
1018                         Comma => IsInFollow::Yes,
1019                         Ident(i, is_raw) if is_raw || i.name != "priv" => IsInFollow::Yes,
1020                         ref tok => if tok.can_begin_type() {
1021                             IsInFollow::Yes
1022                         } else {
1023                             IsInFollow::No(tokens)
1024                         }
1025                     },
1026                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "ident"
1027                                                        || frag.name == "ty"
1028                                                        || frag.name == "path" => IsInFollow::Yes,
1029                     _ => IsInFollow::No(tokens),
1030                 }
1031             },
1032             "" => IsInFollow::Yes, // keywords::Invalid
1033             _ => IsInFollow::Invalid(format!("invalid fragment specifier `{}`", frag),
1034                                      VALID_FRAGMENT_NAMES_MSG),
1035         }
1036     }
1037 }
1038
1039 fn has_legal_fragment_specifier(sess: &ParseSess,
1040                                 features: &Features,
1041                                 attrs: &[ast::Attribute],
1042                                 tok: &quoted::TokenTree) -> Result<(), String> {
1043     debug!("has_legal_fragment_specifier({:?})", tok);
1044     if let quoted::TokenTree::MetaVarDecl(_, _, ref frag_spec) = *tok {
1045         let frag_name = frag_spec.as_str();
1046         let frag_span = tok.span();
1047         if !is_legal_fragment_specifier(sess, features, attrs, &frag_name, frag_span) {
1048             return Err(frag_name.to_string());
1049         }
1050     }
1051     Ok(())
1052 }
1053
1054 fn is_legal_fragment_specifier(_sess: &ParseSess,
1055                                _features: &Features,
1056                                _attrs: &[ast::Attribute],
1057                                frag_name: &str,
1058                                _frag_span: Span) -> bool {
1059     /*
1060      * If new fragment specifiers are invented in nightly, `_sess`,
1061      * `_features`, `_attrs`, and `_frag_span` will be useful here
1062      * for checking against feature gates. See past versions of
1063      * this function.
1064      */
1065     match frag_name {
1066         "item" | "block" | "stmt" | "expr" | "pat" | "lifetime" |
1067         "path" | "ty" | "ident" | "meta" | "tt" | "vis" | "literal" |
1068         "" => true,
1069         _ => false,
1070     }
1071 }
1072
1073 fn quoted_tt_to_string(tt: &quoted::TokenTree) -> String {
1074     match *tt {
1075         quoted::TokenTree::Token(_, ref tok) => ::print::pprust::token_to_string(tok),
1076         quoted::TokenTree::MetaVar(_, name) => format!("${}", name),
1077         quoted::TokenTree::MetaVarDecl(_, name, kind) => format!("${}:{}", name, kind),
1078         _ => panic!("unexpected quoted::TokenTree::{{Sequence or Delimited}} \
1079                      in follow set checker"),
1080     }
1081 }