src/libsyntax/ext/tt/macro_rules.rs

   1 use crate::{ast, attr};
   2 use crate::edition::Edition;
   3 use crate::errors::FatalError;
   4 use crate::ext::base::{DummyResult, ExtCtxt, MacResult, SyntaxExtension};
   5 use crate::ext::base::{NormalTT, TTMacroExpander};
   6 use crate::ext::expand::{AstFragment, AstFragmentKind};
   7 use crate::ext::tt::macro_parser::{Success, Error, Failure};
   8 use crate::ext::tt::macro_parser::{MatchedSeq, MatchedNonterminal};
   9 use crate::ext::tt::macro_parser::{parse, parse_failure_msg};
  10 use crate::ext::tt::quoted;
  11 use crate::ext::tt::transcribe::transcribe;
  12 use crate::feature_gate::Features;
  13 use crate::parse::{Directory, ParseSess};
  14 use crate::parse::parser::Parser;
  15 use crate::parse::token::{self, NtTT};
  16 use crate::parse::token::Token::*;
  17 use crate::symbol::Symbol;
  18 use crate::tokenstream::{DelimSpan, TokenStream, TokenTree};
  19
  20 use syntax_pos::{Span, DUMMY_SP, symbol::Ident};
  21 use log::debug;
  22
  23 use rustc_data_structures::fx::{FxHashMap};
  24 use std::borrow::Cow;
  25 use std::collections::hash_map::Entry;
  26
  27 use rustc_data_structures::sync::Lrc;
  28 use crate::errors::Applicability;
  29
  30 const VALID_FRAGMENT_NAMES_MSG: &str = "valid fragment specifiers are \
  31     `ident`, `block`, `stmt`, `expr`, `pat`, `ty`, `lifetime`, `literal`, \
  32     `path`, `meta`, `tt`, `item` and `vis`";
  33
  34 pub struct ParserAnyMacro<'a> {
  35     parser: Parser<'a>,
  36
  37     /// Span of the expansion site of the macro this parser is for
  38     site_span: Span,
  39     /// The ident of the macro we're parsing
  40     macro_ident: ast::Ident,
  41     arm_span: Span,
  42 }
  43
  44 impl<'a> ParserAnyMacro<'a> {
  45     pub fn make(mut self: Box<ParserAnyMacro<'a>>, kind: AstFragmentKind) -> AstFragment {
  46         let ParserAnyMacro { site_span, macro_ident, ref mut parser, arm_span } = *self;
  47         let fragment = panictry!(parser.parse_ast_fragment(kind, true).map_err(|mut e| {
  48             if parser.token == token::Eof && e.message().ends_with(", found `<eof>`") {
  49                 if !e.span.is_dummy() {  // early end of macro arm (#52866)
  50                     e.replace_span_with(parser.sess.source_map().next_point(parser.span));
  51                 }
  52                 let msg = &e.message[0];
  53                 e.message[0] = (
  54                     format!(
  55                         "macro expansion ends with an incomplete expression: {}",
  56                         msg.0.replace(", found `<eof>`", ""),
  57                     ),
  58                     msg.1,
  59                 );
  60             }
  61             if e.span.is_dummy() {  // Get around lack of span in error (#30128)
  62                 e.replace_span_with(site_span);
  63                 if parser.sess.source_map().span_to_filename(arm_span).is_real() {
  64                     e.span_label(arm_span, "in this macro arm");
  65                 }
  66             } else if !parser.sess.source_map().span_to_filename(parser.span).is_real() {
  67                 e.span_label(site_span, "in this macro invocation");
  68             }
  69             e
  70         }));
  71
  72         // We allow semicolons at the end of expressions -- e.g., the semicolon in
  73         // `macro_rules! m { () => { panic!(); } }` isn't parsed by `.parse_expr()`,
  74         // but `m!()` is allowed in expression positions (cf. issue #34706).
  75         if kind == AstFragmentKind::Expr && parser.token == token::Semi {
  76             parser.bump();
  77         }
  78
  79         // Make sure we don't have any tokens left to parse so we don't silently drop anything.
  80         let path = ast::Path::from_ident(macro_ident.with_span_pos(site_span));
  81         parser.ensure_complete_parse(&path, kind.name(), site_span);
  82         fragment
  83     }
  84 }
  85
  86 struct MacroRulesMacroExpander {
  87     name: ast::Ident,
  88     lhses: Vec<quoted::TokenTree>,
  89     rhses: Vec<quoted::TokenTree>,
  90     valid: bool,
  91 }
  92
  93 impl TTMacroExpander for MacroRulesMacroExpander {
  94     fn expand<'cx>(
  95         &self,
  96         cx: &'cx mut ExtCtxt<'_>,
  97         sp: Span,
  98         input: TokenStream,
  99         def_span: Option<Span>,
 100     ) -> Box<dyn MacResult+'cx> {
 101         if !self.valid {
 102             return DummyResult::any(sp);
 103         }
 104         generic_extension(cx,
 105                           sp,
 106                           def_span,
 107                           self.name,
 108                           input,
 109                           &self.lhses,
 110                           &self.rhses)
 111     }
 112 }
 113
 114 fn trace_macros_note(cx: &mut ExtCtxt<'_>, sp: Span, message: String) {
 115     let sp = sp.macro_backtrace().last().map(|trace| trace.call_site).unwrap_or(sp);
 116     cx.expansions.entry(sp).or_default().push(message);
 117 }
 118
 119 /// Given `lhses` and `rhses`, this is the new macro we create
 120 fn generic_extension<'cx>(cx: &'cx mut ExtCtxt<'_>,
 121                           sp: Span,
 122                           def_span: Option<Span>,
 123                           name: ast::Ident,
 124                           arg: TokenStream,
 125                           lhses: &[quoted::TokenTree],
 126                           rhses: &[quoted::TokenTree])
 127                           -> Box<dyn MacResult+'cx> {
 128     if cx.trace_macros() {
 129         trace_macros_note(cx, sp, format!("expanding `{}! {{ {} }}`", name, arg));
 130     }
 131
 132     // Which arm's failure should we report? (the one furthest along)
 133     let mut best_fail_spot = DUMMY_SP;
 134     let mut best_fail_tok = None;
 135     let mut best_fail_text = None;
 136
 137     for (i, lhs) in lhses.iter().enumerate() { // try each arm's matchers
 138         let lhs_tt = match *lhs {
 139             quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 140             _ => cx.span_bug(sp, "malformed macro lhs")
 141         };
 142
 143         match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 144             Success(named_matches) => {
 145                 let rhs = match rhses[i] {
 146                     // ignore delimiters
 147                     quoted::TokenTree::Delimited(_, ref delimed) => delimed.tts.clone(),
 148                     _ => cx.span_bug(sp, "malformed macro rhs"),
 149                 };
 150                 let arm_span = rhses[i].span();
 151
 152                 let rhs_spans = rhs.iter().map(|t| t.span()).collect::<Vec<_>>();
 153                 // rhs has holes ( `$id` and `$(...)` that need filled)
 154                 let mut tts = transcribe(cx, Some(named_matches), rhs);
 155
 156                 // Replace all the tokens for the corresponding positions in the macro, to maintain
 157                 // proper positions in error reporting, while maintaining the macro_backtrace.
 158                 if rhs_spans.len() == tts.len() {
 159                     tts = tts.map_enumerated(|i, mut tt| {
 160                         let mut sp = rhs_spans[i];
 161                         sp = sp.with_ctxt(tt.span().ctxt());
 162                         tt.set_span(sp);
 163                         tt
 164                     });
 165                 }
 166
 167                 if cx.trace_macros() {
 168                     trace_macros_note(cx, sp, format!("to `{}`", tts));
 169                 }
 170
 171                 let directory = Directory {
 172                     path: Cow::from(cx.current_expansion.module.directory.as_path()),
 173                     ownership: cx.current_expansion.directory_ownership,
 174                 };
 175                 let mut p = Parser::new(cx.parse_sess(), tts, Some(directory), true, false);
 176                 p.root_module_name = cx.current_expansion.module.mod_path.last()
 177                     .map(|id| id.as_str().to_string());
 178
 179                 p.process_potential_macro_variable();
 180                 // Let the context choose how to interpret the result.
 181                 // Weird, but useful for X-macros.
 182                 return Box::new(ParserAnyMacro {
 183                     parser: p,
 184
 185                     // Pass along the original expansion site and the name of the macro
 186                     // so we can print a useful error message if the parse of the expanded
 187                     // macro leaves unparsed tokens.
 188                     site_span: sp,
 189                     macro_ident: name,
 190                     arm_span,
 191                 })
 192             }
 193             Failure(sp, tok, t) => if sp.lo() >= best_fail_spot.lo() {
 194                 best_fail_spot = sp;
 195                 best_fail_tok = Some(tok);
 196                 best_fail_text = Some(t);
 197             },
 198             Error(err_sp, ref msg) => {
 199                 cx.span_fatal(err_sp.substitute_dummy(sp), &msg[..])
 200             }
 201         }
 202     }
 203
 204     let best_fail_msg = parse_failure_msg(best_fail_tok.expect("ran no matchers"));
 205     let span = best_fail_spot.substitute_dummy(sp);
 206     let mut err = cx.struct_span_err(span, &best_fail_msg);
 207     err.span_label(span, best_fail_text.unwrap_or(&best_fail_msg));
 208     if let Some(sp) = def_span {
 209         if cx.source_map().span_to_filename(sp).is_real() && !sp.is_dummy() {
 210             err.span_label(cx.source_map().def_span(sp), "when calling this macro");
 211         }
 212     }
 213
 214     // Check whether there's a missing comma in this macro call, like `println!("{}" a);`
 215     if let Some((arg, comma_span)) = arg.add_comma() {
 216         for lhs in lhses { // try each arm's matchers
 217             let lhs_tt = match *lhs {
 218                 quoted::TokenTree::Delimited(_, ref delim) => &delim.tts[..],
 219                 _ => continue,
 220             };
 221             match TokenTree::parse(cx, lhs_tt, arg.clone()) {
 222                 Success(_) => {
 223                     if comma_span.is_dummy() {
 224                         err.note("you might be missing a comma");
 225                     } else {
 226                         err.span_suggestion_short(
 227                             comma_span,
 228                             "missing comma here",
 229                             ", ".to_string(),
 230                             Applicability::MachineApplicable,
 231                         );
 232                     }
 233                 }
 234                 _ => {}
 235             }
 236         }
 237     }
 238     err.emit();
 239     cx.trace_macros_diag();
 240     DummyResult::any(sp)
 241 }
 242
 243 // Note that macro-by-example's input is also matched against a token tree:
 244 //                   $( $lhs:tt => $rhs:tt );+
 245 //
 246 // Holy self-referential!
 247
 248 /// Converts a `macro_rules!` invocation into a syntax extension.
 249 pub fn compile(
 250     sess: &ParseSess,
 251     features: &Features,
 252     def: &ast::Item,
 253     edition: Edition
 254 ) -> SyntaxExtension {
 255     let lhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("lhs"));
 256     let rhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("rhs"));
 257
 258     // Parse the macro_rules! invocation
 259     let body = match def.node {
 260         ast::ItemKind::MacroDef(ref body) => body,
 261         _ => unreachable!(),
 262     };
 263
 264     // The pattern that macro_rules matches.
 265     // The grammar for macro_rules! is:
 266     // $( $lhs:tt => $rhs:tt );+
 267     // ...quasiquoting this would be nice.
 268     // These spans won't matter, anyways
 269     let argument_gram = vec![
 270         quoted::TokenTree::Sequence(DelimSpan::dummy(), Lrc::new(quoted::SequenceRepetition {
 271             tts: vec![
 272                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, lhs_nm, ast::Ident::from_str("tt")),
 273                 quoted::TokenTree::Token(DUMMY_SP, token::FatArrow),
 274                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, rhs_nm, ast::Ident::from_str("tt")),
 275             ],
 276             separator: Some(if body.legacy { token::Semi } else { token::Comma }),
 277             op: quoted::KleeneOp::OneOrMore,
 278             num_captures: 2,
 279         })),
 280         // to phase into semicolon-termination instead of semicolon-separation
 281         quoted::TokenTree::Sequence(DelimSpan::dummy(), Lrc::new(quoted::SequenceRepetition {
 282             tts: vec![quoted::TokenTree::Token(DUMMY_SP, token::Semi)],
 283             separator: None,
 284             op: quoted::KleeneOp::ZeroOrMore,
 285             num_captures: 0
 286         })),
 287     ];
 288
 289     let argument_map = match parse(sess, body.stream(), &argument_gram, None, true) {
 290         Success(m) => m,
 291         Failure(sp, tok, t) => {
 292             let s = parse_failure_msg(tok);
 293             let sp = sp.substitute_dummy(def.span);
 294             let mut err = sess.span_diagnostic.struct_span_fatal(sp, &s);
 295             err.span_label(sp, t);
 296             err.emit();
 297             FatalError.raise();
 298         }
 299         Error(sp, s) => {
 300             sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s).raise();
 301         }
 302     };
 303
 304     let mut valid = true;
 305
 306     // Extract the arguments:
 307     let lhses = match *argument_map[&lhs_nm] {
 308         MatchedSeq(ref s, _) => {
 309             s.iter().map(|m| {
 310                 if let MatchedNonterminal(ref nt) = *m {
 311                     if let NtTT(ref tt) = **nt {
 312                         let tt = quoted::parse(
 313                             tt.clone().into(),
 314                             true,
 315                             sess,
 316                             features,
 317                             &def.attrs,
 318                             edition,
 319                             def.id,
 320                         )
 321                         .pop()
 322                         .unwrap();
 323                         valid &= check_lhs_nt_follows(sess, features, &def.attrs, &tt);
 324                         return tt;
 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 lhs")
 331     };
 332
 333     let rhses = match *argument_map[&rhs_nm] {
 334         MatchedSeq(ref s, _) => {
 335             s.iter().map(|m| {
 336                 if let MatchedNonterminal(ref nt) = *m {
 337                     if let NtTT(ref tt) = **nt {
 338                         return quoted::parse(
 339                             tt.clone().into(),
 340                             false,
 341                             sess,
 342                             features,
 343                             &def.attrs,
 344                             edition,
 345                             def.id,
 346                         ).pop()
 347                          .unwrap();
 348                     }
 349                 }
 350                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 351             }).collect::<Vec<quoted::TokenTree>>()
 352         }
 353         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured rhs")
 354     };
 355
 356     for rhs in &rhses {
 357         valid &= check_rhs(sess, rhs);
 358     }
 359
 360     // don't abort iteration early, so that errors for multiple lhses can be reported
 361     for lhs in &lhses {
 362         valid &= check_lhs_no_empty_seq(sess, &[lhs.clone()]);
 363         valid &= check_lhs_duplicate_matcher_bindings(
 364             sess,
 365             &[lhs.clone()],
 366             &mut FxHashMap::default(),
 367             def.id
 368         );
 369     }
 370
 371     let expander: Box<_> = Box::new(MacroRulesMacroExpander {
 372         name: def.ident,
 373         lhses,
 374         rhses,
 375         valid,
 376     });
 377
 378     if body.legacy {
 379         let allow_internal_unstable = attr::find_by_name(&def.attrs, "allow_internal_unstable")
 380             .map_or(Vec::new(), |attr| attr
 381                 .meta_item_list()
 382                 .unwrap_or_else(|| sess.span_diagnostic.span_bug(
 383                     attr.span, "allow_internal_unstable expects list of feature names",
 384                 ))
 385                 .iter()
 386                 .map(|it| it.name().unwrap_or_else(|| sess.span_diagnostic.span_bug(
 387                     it.span, "allow internal unstable expects feature names",
 388                 )))
 389                 .collect()
 390             );
 391         let allow_internal_unsafe = attr::contains_name(&def.attrs, "allow_internal_unsafe");
 392         let mut local_inner_macros = false;
 393         if let Some(macro_export) = attr::find_by_name(&def.attrs, "macro_export") {
 394             if let Some(l) = macro_export.meta_item_list() {
 395                 local_inner_macros = attr::list_contains_name(&l, "local_inner_macros");
 396             }
 397         }
 398
 399         let unstable_feature = attr::find_stability(&sess,
 400                                                     &def.attrs, def.span).and_then(|stability| {
 401             if let attr::StabilityLevel::Unstable { issue, .. } = stability.level {
 402                 Some((stability.feature, issue))
 403             } else {
 404                 None
 405             }
 406         });
 407
 408         NormalTT {
 409             expander,
 410             def_info: Some((def.id, def.span)),
 411             allow_internal_unstable,
 412             allow_internal_unsafe,
 413             local_inner_macros,
 414             unstable_feature,
 415             edition,
 416         }
 417     } else {
 418         let is_transparent = attr::contains_name(&def.attrs, "rustc_transparent_macro");
 419
 420         SyntaxExtension::DeclMacro {
 421             expander,
 422             def_info: Some((def.id, def.span)),
 423             is_transparent,
 424             edition,
 425         }
 426     }
 427 }
 428
 429 fn check_lhs_nt_follows(sess: &ParseSess,
 430                         features: &Features,
 431                         attrs: &[ast::Attribute],
 432                         lhs: &quoted::TokenTree) -> bool {
 433     // lhs is going to be like TokenTree::Delimited(...), where the
 434     // entire lhs is those tts. Or, it can be a "bare sequence", not wrapped in parens.
 435     if let quoted::TokenTree::Delimited(_, ref tts) = *lhs {
 436         check_matcher(sess, features, attrs, &tts.tts)
 437     } else {
 438         let msg = "invalid macro matcher; matchers must be contained in balanced delimiters";
 439         sess.span_diagnostic.span_err(lhs.span(), msg);
 440         false
 441     }
 442     // we don't abort on errors on rejection, the driver will do that for us
 443     // after parsing/expansion. we can report every error in every macro this way.
 444 }
 445
 446 /// Check that the lhs contains no repetition which could match an empty token
 447 /// tree, because then the matcher would hang indefinitely.
 448 fn check_lhs_no_empty_seq(sess: &ParseSess, tts: &[quoted::TokenTree]) -> bool {
 449     use quoted::TokenTree;
 450     for tt in tts {
 451         match *tt {
 452             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => (),
 453             TokenTree::Delimited(_, ref del) => if !check_lhs_no_empty_seq(sess, &del.tts) {
 454                 return false;
 455             },
 456             TokenTree::Sequence(span, ref seq) => {
 457                 if seq.separator.is_none() && seq.tts.iter().all(|seq_tt| {
 458                     match *seq_tt {
 459                         TokenTree::MetaVarDecl(_, _, id) => id.name == "vis",
 460                         TokenTree::Sequence(_, ref sub_seq) =>
 461                             sub_seq.op == quoted::KleeneOp::ZeroOrMore
 462                             || sub_seq.op == quoted::KleeneOp::ZeroOrOne,
 463                         _ => false,
 464                     }
 465                 }) {
 466                     let sp = span.entire();
 467                     sess.span_diagnostic.span_err(sp, "repetition matches empty token tree");
 468                     return false;
 469                 }
 470                 if !check_lhs_no_empty_seq(sess, &seq.tts) {
 471                     return false;
 472                 }
 473             }
 474         }
 475     }
 476
 477     true
 478 }
 479
 480 /// Check that the LHS contains no duplicate matcher bindings. e.g. `$a:expr, $a:expr` would be
 481 /// illegal, since it would be ambiguous which `$a` to use if we ever needed to.
 482 fn check_lhs_duplicate_matcher_bindings(
 483     sess: &ParseSess,
 484     tts: &[quoted::TokenTree],
 485     metavar_names: &mut FxHashMap<Ident, Span>,
 486     node_id: ast::NodeId,
 487 ) -> bool {
 488     use self::quoted::TokenTree;
 489     use crate::early_buffered_lints::BufferedEarlyLintId;
 490     for tt in tts {
 491         match *tt {
 492             TokenTree::MetaVarDecl(span, name, _kind) => {
 493                 if let Some(&prev_span) = metavar_names.get(&name) {
 494                     // FIXME(mark-i-m): in a few cycles, make this a hard error.
 495                     // sess.span_diagnostic
 496                     //     .struct_span_err(span, "duplicate matcher binding")
 497                     //     .span_note(prev_span, "previous declaration was here")
 498                     //     .emit();
 499                     sess.buffer_lint(
 500                         BufferedEarlyLintId::DuplicateMacroMatcherBindingName,
 501                         crate::source_map::MultiSpan::from(vec![prev_span, span]),
 502                         node_id,
 503                         "duplicate matcher binding"
 504                     );
 505                     return false;
 506                 } else {
 507                     metavar_names.insert(name, span);
 508                 }
 509             }
 510             TokenTree::Delimited(_, ref del) => {
 511                 if !check_lhs_duplicate_matcher_bindings(sess, &del.tts, metavar_names, node_id) {
 512                     return false;
 513                 }
 514             },
 515             TokenTree::Sequence(_, ref seq) => {
 516                 if !check_lhs_duplicate_matcher_bindings(sess, &seq.tts, metavar_names, node_id) {
 517                     return false;
 518                 }
 519             }
 520             _ => {}
 521         }
 522     }
 523
 524     true
 525 }
 526
 527 fn check_rhs(sess: &ParseSess, rhs: &quoted::TokenTree) -> bool {
 528     match *rhs {
 529         quoted::TokenTree::Delimited(..) => return true,
 530         _ => sess.span_diagnostic.span_err(rhs.span(), "macro rhs must be delimited")
 531     }
 532     false
 533 }
 534
 535 fn check_matcher(sess: &ParseSess,
 536                  features: &Features,
 537                  attrs: &[ast::Attribute],
 538                  matcher: &[quoted::TokenTree]) -> bool {
 539     let first_sets = FirstSets::new(matcher);
 540     let empty_suffix = TokenSet::empty();
 541     let err = sess.span_diagnostic.err_count();
 542     check_matcher_core(sess, features, attrs, &first_sets, matcher, &empty_suffix);
 543     err == sess.span_diagnostic.err_count()
 544 }
 545
 546 // `The FirstSets` for a matcher is a mapping from subsequences in the
 547 // matcher to the FIRST set for that subsequence.
 548 //
 549 // This mapping is partially precomputed via a backwards scan over the
 550 // token trees of the matcher, which provides a mapping from each
 551 // repetition sequence to its *first* set.
 552 //
 553 // (Hypothetically, sequences should be uniquely identifiable via their
 554 // spans, though perhaps that is false, e.g., for macro-generated macros
 555 // that do not try to inject artificial span information. My plan is
 556 // to try to catch such cases ahead of time and not include them in
 557 // the precomputed mapping.)
 558 struct FirstSets {
 559     // this maps each TokenTree::Sequence `$(tt ...) SEP OP` that is uniquely identified by its
 560     // span in the original matcher to the First set for the inner sequence `tt ...`.
 561     //
 562     // If two sequences have the same span in a matcher, then map that
 563     // span to None (invalidating the mapping here and forcing the code to
 564     // use a slow path).
 565     first: FxHashMap<Span, Option<TokenSet>>,
 566 }
 567
 568 impl FirstSets {
 569     fn new(tts: &[quoted::TokenTree]) -> FirstSets {
 570         use quoted::TokenTree;
 571
 572         let mut sets = FirstSets { first: FxHashMap::default() };
 573         build_recur(&mut sets, tts);
 574         return sets;
 575
 576         // walks backward over `tts`, returning the FIRST for `tts`
 577         // and updating `sets` at the same time for all sequence
 578         // substructure we find within `tts`.
 579         fn build_recur(sets: &mut FirstSets, tts: &[TokenTree]) -> TokenSet {
 580             let mut first = TokenSet::empty();
 581             for tt in tts.iter().rev() {
 582                 match *tt {
 583                     TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 584                         first.replace_with(tt.clone());
 585                     }
 586                     TokenTree::Delimited(span, ref delimited) => {
 587                         build_recur(sets, &delimited.tts[..]);
 588                         first.replace_with(delimited.open_tt(span.open));
 589                     }
 590                     TokenTree::Sequence(sp, ref seq_rep) => {
 591                         let subfirst = build_recur(sets, &seq_rep.tts[..]);
 592
 593                         match sets.first.entry(sp.entire()) {
 594                             Entry::Vacant(vac) => {
 595                                 vac.insert(Some(subfirst.clone()));
 596                             }
 597                             Entry::Occupied(mut occ) => {
 598                                 // if there is already an entry, then a span must have collided.
 599                                 // This should not happen with typical macro_rules macros,
 600                                 // but syntax extensions need not maintain distinct spans,
 601                                 // so distinct syntax trees can be assigned the same span.
 602                                 // In such a case, the map cannot be trusted; so mark this
 603                                 // entry as unusable.
 604                                 occ.insert(None);
 605                             }
 606                         }
 607
 608                         // If the sequence contents can be empty, then the first
 609                         // token could be the separator token itself.
 610
 611                         if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 612                                                         subfirst.maybe_empty) {
 613                             first.add_one_maybe(TokenTree::Token(sp.entire(), sep.clone()));
 614                         }
 615
 616                         // Reverse scan: Sequence comes before `first`.
 617                         if subfirst.maybe_empty
 618                            || seq_rep.op == quoted::KleeneOp::ZeroOrMore
 619                            || seq_rep.op == quoted::KleeneOp::ZeroOrOne
 620                         {
 621                             // If sequence is potentially empty, then
 622                             // union them (preserving first emptiness).
 623                             first.add_all(&TokenSet { maybe_empty: true, ..subfirst });
 624                         } else {
 625                             // Otherwise, sequence guaranteed
 626                             // non-empty; replace first.
 627                             first = subfirst;
 628                         }
 629                     }
 630                 }
 631             }
 632
 633             first
 634         }
 635     }
 636
 637     // walks forward over `tts` until all potential FIRST tokens are
 638     // identified.
 639     fn first(&self, tts: &[quoted::TokenTree]) -> TokenSet {
 640         use quoted::TokenTree;
 641
 642         let mut first = TokenSet::empty();
 643         for tt in tts.iter() {
 644             assert!(first.maybe_empty);
 645             match *tt {
 646                 TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 647                     first.add_one(tt.clone());
 648                     return first;
 649                 }
 650                 TokenTree::Delimited(span, ref delimited) => {
 651                     first.add_one(delimited.open_tt(span.open));
 652                     return first;
 653                 }
 654                 TokenTree::Sequence(sp, ref seq_rep) => {
 655                     match self.first.get(&sp.entire()) {
 656                         Some(&Some(ref subfirst)) => {
 657
 658                             // If the sequence contents can be empty, then the first
 659                             // token could be the separator token itself.
 660
 661                             if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 662                                                             subfirst.maybe_empty) {
 663                                 first.add_one_maybe(TokenTree::Token(sp.entire(), sep.clone()));
 664                             }
 665
 666                             assert!(first.maybe_empty);
 667                             first.add_all(subfirst);
 668                             if subfirst.maybe_empty
 669                                || seq_rep.op == quoted::KleeneOp::ZeroOrMore
 670                                || seq_rep.op == quoted::KleeneOp::ZeroOrOne
 671                             {
 672                                 // continue scanning for more first
 673                                 // tokens, but also make sure we
 674                                 // restore empty-tracking state
 675                                 first.maybe_empty = true;
 676                                 continue;
 677                             } else {
 678                                 return first;
 679                             }
 680                         }
 681
 682                         Some(&None) => {
 683                             panic!("assume all sequences have (unique) spans for now");
 684                         }
 685
 686                         None => {
 687                             panic!("We missed a sequence during FirstSets construction");
 688                         }
 689                     }
 690                 }
 691             }
 692         }
 693
 694         // we only exit the loop if `tts` was empty or if every
 695         // element of `tts` matches the empty sequence.
 696         assert!(first.maybe_empty);
 697         first
 698     }
 699 }
 700
 701 // A set of `quoted::TokenTree`s, which may include `TokenTree::Match`s
 702 // (for macro-by-example syntactic variables). It also carries the
 703 // `maybe_empty` flag; that is true if and only if the matcher can
 704 // match an empty token sequence.
 705 //
 706 // The First set is computed on submatchers like `$($a:expr b),* $(c)* d`,
 707 // which has corresponding FIRST = {$a:expr, c, d}.
 708 // Likewise, `$($a:expr b),* $(c)+ d` has FIRST = {$a:expr, c}.
 709 //
 710 // (Notably, we must allow for *-op to occur zero times.)
 711 #[derive(Clone, Debug)]
 712 struct TokenSet {
 713     tokens: Vec<quoted::TokenTree>,
 714     maybe_empty: bool,
 715 }
 716
 717 impl TokenSet {
 718     // Returns a set for the empty sequence.
 719     fn empty() -> Self { TokenSet { tokens: Vec::new(), maybe_empty: true } }
 720
 721     // Returns the set `{ tok }` for the single-token (and thus
 722     // non-empty) sequence [tok].
 723     fn singleton(tok: quoted::TokenTree) -> Self {
 724         TokenSet { tokens: vec![tok], maybe_empty: false }
 725     }
 726
 727     // Changes self to be the set `{ tok }`.
 728     // Since `tok` is always present, marks self as non-empty.
 729     fn replace_with(&mut self, tok: quoted::TokenTree) {
 730         self.tokens.clear();
 731         self.tokens.push(tok);
 732         self.maybe_empty = false;
 733     }
 734
 735     // Changes self to be the empty set `{}`; meant for use when
 736     // the particular token does not matter, but we want to
 737     // record that it occurs.
 738     fn replace_with_irrelevant(&mut self) {
 739         self.tokens.clear();
 740         self.maybe_empty = false;
 741     }
 742
 743     // Adds `tok` to the set for `self`, marking sequence as non-empy.
 744     fn add_one(&mut self, tok: quoted::TokenTree) {
 745         if !self.tokens.contains(&tok) {
 746             self.tokens.push(tok);
 747         }
 748         self.maybe_empty = false;
 749     }
 750
 751     // Adds `tok` to the set for `self`. (Leaves `maybe_empty` flag alone.)
 752     fn add_one_maybe(&mut self, tok: quoted::TokenTree) {
 753         if !self.tokens.contains(&tok) {
 754             self.tokens.push(tok);
 755         }
 756     }
 757
 758     // Adds all elements of `other` to this.
 759     //
 760     // (Since this is a set, we filter out duplicates.)
 761     //
 762     // If `other` is potentially empty, then preserves the previous
 763     // setting of the empty flag of `self`. If `other` is guaranteed
 764     // non-empty, then `self` is marked non-empty.
 765     fn add_all(&mut self, other: &Self) {
 766         for tok in &other.tokens {
 767             if !self.tokens.contains(tok) {
 768                 self.tokens.push(tok.clone());
 769             }
 770         }
 771         if !other.maybe_empty {
 772             self.maybe_empty = false;
 773         }
 774     }
 775 }
 776
 777 // Checks that `matcher` is internally consistent and that it
 778 // can legally by followed by a token N, for all N in `follow`.
 779 // (If `follow` is empty, then it imposes no constraint on
 780 // the `matcher`.)
 781 //
 782 // Returns the set of NT tokens that could possibly come last in
 783 // `matcher`. (If `matcher` matches the empty sequence, then
 784 // `maybe_empty` will be set to true.)
 785 //
 786 // Requires that `first_sets` is pre-computed for `matcher`;
 787 // see `FirstSets::new`.
 788 fn check_matcher_core(sess: &ParseSess,
 789                       features: &Features,
 790                       attrs: &[ast::Attribute],
 791                       first_sets: &FirstSets,
 792                       matcher: &[quoted::TokenTree],
 793                       follow: &TokenSet) -> TokenSet {
 794     use quoted::TokenTree;
 795
 796     let mut last = TokenSet::empty();
 797
 798     // 2. For each token and suffix  [T, SUFFIX] in M:
 799     // ensure that T can be followed by SUFFIX, and if SUFFIX may be empty,
 800     // then ensure T can also be followed by any element of FOLLOW.
 801     'each_token: for i in 0..matcher.len() {
 802         let token = &matcher[i];
 803         let suffix = &matcher[i+1..];
 804
 805         let build_suffix_first = || {
 806             let mut s = first_sets.first(suffix);
 807             if s.maybe_empty { s.add_all(follow); }
 808             s
 809         };
 810
 811         // (we build `suffix_first` on demand below; you can tell
 812         // which cases are supposed to fall through by looking for the
 813         // initialization of this variable.)
 814         let suffix_first;
 815
 816         // First, update `last` so that it corresponds to the set
 817         // of NT tokens that might end the sequence `... token`.
 818         match *token {
 819             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 820                 let can_be_followed_by_any;
 821                 if let Err(bad_frag) = has_legal_fragment_specifier(sess, features, attrs, token) {
 822                     let msg = format!("invalid fragment specifier `{}`", bad_frag);
 823                     sess.span_diagnostic.struct_span_err(token.span(), &msg)
 824                         .help(VALID_FRAGMENT_NAMES_MSG)
 825                         .emit();
 826                     // (This eliminates false positives and duplicates
 827                     // from error messages.)
 828                     can_be_followed_by_any = true;
 829                 } else {
 830                     can_be_followed_by_any = token_can_be_followed_by_any(token);
 831                 }
 832
 833                 if can_be_followed_by_any {
 834                     // don't need to track tokens that work with any,
 835                     last.replace_with_irrelevant();
 836                     // ... and don't need to check tokens that can be
 837                     // followed by anything against SUFFIX.
 838                     continue 'each_token;
 839                 } else {
 840                     last.replace_with(token.clone());
 841                     suffix_first = build_suffix_first();
 842                 }
 843             }
 844             TokenTree::Delimited(span, ref d) => {
 845                 let my_suffix = TokenSet::singleton(d.close_tt(span.close));
 846                 check_matcher_core(sess, features, attrs, first_sets, &d.tts, &my_suffix);
 847                 // don't track non NT tokens
 848                 last.replace_with_irrelevant();
 849
 850                 // also, we don't need to check delimited sequences
 851                 // against SUFFIX
 852                 continue 'each_token;
 853             }
 854             TokenTree::Sequence(sp, ref seq_rep) => {
 855                 suffix_first = build_suffix_first();
 856                 // The trick here: when we check the interior, we want
 857                 // to include the separator (if any) as a potential
 858                 // (but not guaranteed) element of FOLLOW. So in that
 859                 // case, we make a temp copy of suffix and stuff
 860                 // delimiter in there.
 861                 //
 862                 // FIXME: Should I first scan suffix_first to see if
 863                 // delimiter is already in it before I go through the
 864                 // work of cloning it? But then again, this way I may
 865                 // get a "tighter" span?
 866                 let mut new;
 867                 let my_suffix = if let Some(ref u) = seq_rep.separator {
 868                     new = suffix_first.clone();
 869                     new.add_one_maybe(TokenTree::Token(sp.entire(), u.clone()));
 870                     &new
 871                 } else {
 872                     &suffix_first
 873                 };
 874
 875                 // At this point, `suffix_first` is built, and
 876                 // `my_suffix` is some TokenSet that we can use
 877                 // for checking the interior of `seq_rep`.
 878                 let next = check_matcher_core(sess,
 879                                               features,
 880                                               attrs,
 881                                               first_sets,
 882                                               &seq_rep.tts,
 883                                               my_suffix);
 884                 if next.maybe_empty {
 885                     last.add_all(&next);
 886                 } else {
 887                     last = next;
 888                 }
 889
 890                 // the recursive call to check_matcher_core already ran the 'each_last
 891                 // check below, so we can just keep going forward here.
 892                 continue 'each_token;
 893             }
 894         }
 895
 896         // (`suffix_first` guaranteed initialized once reaching here.)
 897
 898         // Now `last` holds the complete set of NT tokens that could
 899         // end the sequence before SUFFIX. Check that every one works with `suffix`.
 900         'each_last: for token in &last.tokens {
 901             if let TokenTree::MetaVarDecl(_, ref name, ref frag_spec) = *token {
 902                 for next_token in &suffix_first.tokens {
 903                     match is_in_follow(next_token, &frag_spec.as_str()) {
 904                         IsInFollow::Invalid(msg, help) => {
 905                             sess.span_diagnostic.struct_span_err(next_token.span(), &msg)
 906                                 .help(help).emit();
 907                             // don't bother reporting every source of
 908                             // conflict for a particular element of `last`.
 909                             continue 'each_last;
 910                         }
 911                         IsInFollow::Yes => {}
 912                         IsInFollow::No(ref possible) => {
 913                             let may_be = if last.tokens.len() == 1 &&
 914                                 suffix_first.tokens.len() == 1
 915                             {
 916                                 "is"
 917                             } else {
 918                                 "may be"
 919                             };
 920
 921                             let sp = next_token.span();
 922                             let mut err = sess.span_diagnostic.struct_span_err(
 923                                 sp,
 924                                 &format!("`${name}:{frag}` {may_be} followed by `{next}`, which \
 925                                           is not allowed for `{frag}` fragments",
 926                                          name=name,
 927                                          frag=frag_spec,
 928                                          next=quoted_tt_to_string(next_token),
 929                                          may_be=may_be),
 930                             );
 931                             err.span_label(
 932                                 sp,
 933                                 format!("not allowed after `{}` fragments", frag_spec),
 934                             );
 935                             let msg = "allowed there are: ";
 936                             match &possible[..] {
 937                                 &[] => {}
 938                                 &[t] => {
 939                                     err.note(&format!(
 940                                         "only {} is allowed after `{}` fragments",
 941                                         t,
 942                                         frag_spec,
 943                                     ));
 944                                 }
 945                                 ts => {
 946                                     err.note(&format!(
 947                                         "{}{} or {}",
 948                                         msg,
 949                                         ts[..ts.len() - 1].iter().map(|s| *s)
 950                                             .collect::<Vec<_>>().join(", "),
 951                                         ts[ts.len() - 1],
 952                                     ));
 953                                 }
 954                             }
 955                             err.emit();
 956                         }
 957                     }
 958                 }
 959             }
 960         }
 961     }
 962     last
 963 }
 964
 965 fn token_can_be_followed_by_any(tok: &quoted::TokenTree) -> bool {
 966     if let quoted::TokenTree::MetaVarDecl(_, _, frag_spec) = *tok {
 967         frag_can_be_followed_by_any(&frag_spec.as_str())
 968     } else {
 969         // (Non NT's can always be followed by anthing in matchers.)
 970         true
 971     }
 972 }
 973
 974 /// True if a fragment of type `frag` can be followed by any sort of
 975 /// token.  We use this (among other things) as a useful approximation
 976 /// for when `frag` can be followed by a repetition like `$(...)*` or
 977 /// `$(...)+`. In general, these can be a bit tricky to reason about,
 978 /// so we adopt a conservative position that says that any fragment
 979 /// specifier which consumes at most one token tree can be followed by
 980 /// a fragment specifier (indeed, these fragments can be followed by
 981 /// ANYTHING without fear of future compatibility hazards).
 982 fn frag_can_be_followed_by_any(frag: &str) -> bool {
 983     match frag {
 984         "item"     | // always terminated by `}` or `;`
 985         "block"    | // exactly one token tree
 986         "ident"    | // exactly one token tree
 987         "literal"  | // exactly one token tree
 988         "meta"     | // exactly one token tree
 989         "lifetime" | // exactly one token tree
 990         "tt" =>   // exactly one token tree
 991             true,
 992
 993         _ =>
 994             false,
 995     }
 996 }
 997
 998 enum IsInFollow {
 999     Yes,
1000     No(Vec<&'static str>),
1001     Invalid(String, &'static str),
1002 }
1003
1004 /// True if `frag` can legally be followed by the token `tok`. For
1005 /// fragments that can consume an unbounded number of tokens, `tok`
1006 /// must be within a well-defined follow set. This is intended to
1007 /// guarantee future compatibility: for example, without this rule, if
1008 /// we expanded `expr` to include a new binary operator, we might
1009 /// break macros that were relying on that binary operator as a
1010 /// separator.
1011 // when changing this do not forget to update doc/book/macros.md!
1012 fn is_in_follow(tok: &quoted::TokenTree, frag: &str) -> IsInFollow {
1013     use quoted::TokenTree;
1014
1015     if let TokenTree::Token(_, token::CloseDelim(_)) = *tok {
1016         // closing a token tree can never be matched by any fragment;
1017         // iow, we always require that `(` and `)` match, etc.
1018         IsInFollow::Yes
1019     } else {
1020         match frag {
1021             "item" => {
1022                 // since items *must* be followed by either a `;` or a `}`, we can
1023                 // accept anything after them
1024                 IsInFollow::Yes
1025             },
1026             "block" => {
1027                 // anything can follow block, the braces provide an easy boundary to
1028                 // maintain
1029                 IsInFollow::Yes
1030             },
1031             "stmt" | "expr"  => {
1032                 let tokens = vec!["`=>`", "`,`", "`;`"];
1033                 match *tok {
1034                     TokenTree::Token(_, ref tok) => match *tok {
1035                         FatArrow | Comma | Semi => IsInFollow::Yes,
1036                         _ => IsInFollow::No(tokens),
1037                     },
1038                     _ => IsInFollow::No(tokens),
1039                 }
1040             },
1041             "pat" => {
1042                 let tokens = vec!["`=>`", "`,`", "`=`", "`|`", "`if`", "`in`"];
1043                 match *tok {
1044                     TokenTree::Token(_, ref tok) => match *tok {
1045                         FatArrow | Comma | Eq | BinOp(token::Or) => IsInFollow::Yes,
1046                         Ident(i, false) if i.name == "if" || i.name == "in" => IsInFollow::Yes,
1047                         _ => IsInFollow::No(tokens),
1048                     },
1049                     _ => IsInFollow::No(tokens),
1050                 }
1051             },
1052             "path" | "ty" => {
1053                 let tokens = vec![
1054                     "`{`", "`[`", "`=>`", "`,`", "`>`","`=`", "`:`", "`;`", "`|`", "`as`",
1055                     "`where`",
1056                 ];
1057                 match *tok {
1058                     TokenTree::Token(_, ref tok) => match *tok {
1059                         OpenDelim(token::DelimToken::Brace) |
1060                         OpenDelim(token::DelimToken::Bracket) |
1061                         Comma | FatArrow | Colon | Eq | Gt | BinOp(token::Shr) | Semi |
1062                         BinOp(token::Or) => IsInFollow::Yes,
1063                         Ident(i, false) if i.name == "as" || i.name == "where" => IsInFollow::Yes,
1064                         _ => IsInFollow::No(tokens),
1065                     },
1066                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "block" => IsInFollow::Yes,
1067                     _ => IsInFollow::No(tokens),
1068                 }
1069             },
1070             "ident" | "lifetime" => {
1071                 // being a single token, idents and lifetimes are harmless
1072                 IsInFollow::Yes
1073             },
1074             "literal" => {
1075                 // literals may be of a single token, or two tokens (negative numbers)
1076                 IsInFollow::Yes
1077             },
1078             "meta" | "tt" => {
1079                 // being either a single token or a delimited sequence, tt is
1080                 // harmless
1081                 IsInFollow::Yes
1082             },
1083             "vis" => {
1084                 // Explicitly disallow `priv`, on the off chance it comes back.
1085                 let tokens = vec!["`,`", "an ident", "a type"];
1086                 match *tok {
1087                     TokenTree::Token(_, ref tok) => match *tok {
1088                         Comma => IsInFollow::Yes,
1089                         Ident(i, is_raw) if is_raw || i.name != "priv" => IsInFollow::Yes,
1090                         ref tok => if tok.can_begin_type() {
1091                             IsInFollow::Yes
1092                         } else {
1093                             IsInFollow::No(tokens)
1094                         }
1095                     },
1096                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "ident"
1097                                                        || frag.name == "ty"
1098                                                        || frag.name == "path" => IsInFollow::Yes,
1099                     _ => IsInFollow::No(tokens),
1100                 }
1101             },
1102             "" => IsInFollow::Yes, // keywords::Invalid
1103             _ => IsInFollow::Invalid(format!("invalid fragment specifier `{}`", frag),
1104                                      VALID_FRAGMENT_NAMES_MSG),
1105         }
1106     }
1107 }
1108
1109 fn has_legal_fragment_specifier(sess: &ParseSess,
1110                                 features: &Features,
1111                                 attrs: &[ast::Attribute],
1112                                 tok: &quoted::TokenTree) -> Result<(), String> {
1113     debug!("has_legal_fragment_specifier({:?})", tok);
1114     if let quoted::TokenTree::MetaVarDecl(_, _, ref frag_spec) = *tok {
1115         let frag_name = frag_spec.as_str();
1116         let frag_span = tok.span();
1117         if !is_legal_fragment_specifier(sess, features, attrs, &frag_name, frag_span) {
1118             return Err(frag_name.to_string());
1119         }
1120     }
1121     Ok(())
1122 }
1123
1124 fn is_legal_fragment_specifier(_sess: &ParseSess,
1125                                _features: &Features,
1126                                _attrs: &[ast::Attribute],
1127                                frag_name: &str,
1128                                _frag_span: Span) -> bool {
1129     /*
1130      * If new fragment specifiers are invented in nightly, `_sess`,
1131      * `_features`, `_attrs`, and `_frag_span` will be useful here
1132      * for checking against feature gates. See past versions of
1133      * this function.
1134      */
1135     match frag_name {
1136         "item" | "block" | "stmt" | "expr" | "pat" | "lifetime" |
1137         "path" | "ty" | "ident" | "meta" | "tt" | "vis" | "literal" |
1138         "" => true,
1139         _ => false,
1140     }
1141 }
1142
1143 fn quoted_tt_to_string(tt: &quoted::TokenTree) -> String {
1144     match *tt {
1145         quoted::TokenTree::Token(_, ref tok) => crate::print::pprust::token_to_string(tok),
1146         quoted::TokenTree::MetaVar(_, name) => format!("${}", name),
1147         quoted::TokenTree::MetaVarDecl(_, name, kind) => format!("${}:{}", name, kind),
1148         _ => panic!("unexpected quoted::TokenTree::{{Sequence or Delimited}} \
1149                      in follow set checker"),
1150     }
1151 }