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};
  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(sess: &ParseSess, features: &Features, def: &ast::Item, edition: Edition)
 250                -> SyntaxExtension {
 251     let lhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("lhs"));
 252     let rhs_nm = ast::Ident::with_empty_ctxt(Symbol::gensym("rhs"));
 253
 254     // Parse the macro_rules! invocation
 255     let body = match def.node {
 256         ast::ItemKind::MacroDef(ref body) => body,
 257         _ => unreachable!(),
 258     };
 259
 260     // The pattern that macro_rules matches.
 261     // The grammar for macro_rules! is:
 262     // $( $lhs:tt => $rhs:tt );+
 263     // ...quasiquoting this would be nice.
 264     // These spans won't matter, anyways
 265     let argument_gram = vec![
 266         quoted::TokenTree::Sequence(DelimSpan::dummy(), Lrc::new(quoted::SequenceRepetition {
 267             tts: vec![
 268                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, lhs_nm, ast::Ident::from_str("tt")),
 269                 quoted::TokenTree::Token(DUMMY_SP, token::FatArrow),
 270                 quoted::TokenTree::MetaVarDecl(DUMMY_SP, rhs_nm, ast::Ident::from_str("tt")),
 271             ],
 272             separator: Some(if body.legacy { token::Semi } else { token::Comma }),
 273             op: quoted::KleeneOp::OneOrMore,
 274             num_captures: 2,
 275         })),
 276         // to phase into semicolon-termination instead of semicolon-separation
 277         quoted::TokenTree::Sequence(DelimSpan::dummy(), Lrc::new(quoted::SequenceRepetition {
 278             tts: vec![quoted::TokenTree::Token(DUMMY_SP, token::Semi)],
 279             separator: None,
 280             op: quoted::KleeneOp::ZeroOrMore,
 281             num_captures: 0
 282         })),
 283     ];
 284
 285     let argument_map = match parse(sess, body.stream(), &argument_gram, None, true) {
 286         Success(m) => m,
 287         Failure(sp, tok, t) => {
 288             let s = parse_failure_msg(tok);
 289             let sp = sp.substitute_dummy(def.span);
 290             let mut err = sess.span_diagnostic.struct_span_fatal(sp, &s);
 291             err.span_label(sp, t);
 292             err.emit();
 293             FatalError.raise();
 294         }
 295         Error(sp, s) => {
 296             sess.span_diagnostic.span_fatal(sp.substitute_dummy(def.span), &s).raise();
 297         }
 298     };
 299
 300     let mut valid = true;
 301
 302     // Extract the arguments:
 303     let lhses = match *argument_map[&lhs_nm] {
 304         MatchedSeq(ref s, _) => {
 305             s.iter().map(|m| {
 306                 if let MatchedNonterminal(ref nt) = *m {
 307                     if let NtTT(ref tt) = **nt {
 308                         let tt = quoted::parse(
 309                             tt.clone().into(),
 310                             true,
 311                             sess,
 312                             features,
 313                             &def.attrs,
 314                             edition,
 315                             def.id,
 316                         )
 317                         .pop()
 318                         .unwrap();
 319                         valid &= check_lhs_nt_follows(sess, features, &def.attrs, &tt);
 320                         return tt;
 321                     }
 322                 }
 323                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 324             }).collect::<Vec<quoted::TokenTree>>()
 325         }
 326         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 327     };
 328
 329     let rhses = match *argument_map[&rhs_nm] {
 330         MatchedSeq(ref s, _) => {
 331             s.iter().map(|m| {
 332                 if let MatchedNonterminal(ref nt) = *m {
 333                     if let NtTT(ref tt) = **nt {
 334                         return quoted::parse(
 335                             tt.clone().into(),
 336                             false,
 337                             sess,
 338                             features,
 339                             &def.attrs,
 340                             edition,
 341                             def.id,
 342                         ).pop()
 343                          .unwrap();
 344                     }
 345                 }
 346                 sess.span_diagnostic.span_bug(def.span, "wrong-structured lhs")
 347             }).collect::<Vec<quoted::TokenTree>>()
 348         }
 349         _ => sess.span_diagnostic.span_bug(def.span, "wrong-structured rhs")
 350     };
 351
 352     for rhs in &rhses {
 353         valid &= check_rhs(sess, rhs);
 354     }
 355
 356     // don't abort iteration early, so that errors for multiple lhses can be reported
 357     for lhs in &lhses {
 358         valid &= check_lhs_no_empty_seq(sess, &[lhs.clone()])
 359     }
 360
 361     let expander: Box<_> = Box::new(MacroRulesMacroExpander {
 362         name: def.ident,
 363         lhses,
 364         rhses,
 365         valid,
 366     });
 367
 368     if body.legacy {
 369         let allow_internal_unstable = attr::contains_name(&def.attrs, "allow_internal_unstable");
 370         let allow_internal_unsafe = attr::contains_name(&def.attrs, "allow_internal_unsafe");
 371         let mut local_inner_macros = false;
 372         if let Some(macro_export) = attr::find_by_name(&def.attrs, "macro_export") {
 373             if let Some(l) = macro_export.meta_item_list() {
 374                 local_inner_macros = attr::list_contains_name(&l, "local_inner_macros");
 375             }
 376         }
 377
 378         let unstable_feature = attr::find_stability(&sess,
 379                                                     &def.attrs, def.span).and_then(|stability| {
 380             if let attr::StabilityLevel::Unstable { issue, .. } = stability.level {
 381                 Some((stability.feature, issue))
 382             } else {
 383                 None
 384             }
 385         });
 386
 387         NormalTT {
 388             expander,
 389             def_info: Some((def.id, def.span)),
 390             allow_internal_unstable,
 391             allow_internal_unsafe,
 392             local_inner_macros,
 393             unstable_feature,
 394             edition,
 395         }
 396     } else {
 397         let is_transparent = attr::contains_name(&def.attrs, "rustc_transparent_macro");
 398
 399         SyntaxExtension::DeclMacro {
 400             expander,
 401             def_info: Some((def.id, def.span)),
 402             is_transparent,
 403             edition,
 404         }
 405     }
 406 }
 407
 408 fn check_lhs_nt_follows(sess: &ParseSess,
 409                         features: &Features,
 410                         attrs: &[ast::Attribute],
 411                         lhs: &quoted::TokenTree) -> bool {
 412     // lhs is going to be like TokenTree::Delimited(...), where the
 413     // entire lhs is those tts. Or, it can be a "bare sequence", not wrapped in parens.
 414     if let quoted::TokenTree::Delimited(_, ref tts) = *lhs {
 415         check_matcher(sess, features, attrs, &tts.tts)
 416     } else {
 417         let msg = "invalid macro matcher; matchers must be contained in balanced delimiters";
 418         sess.span_diagnostic.span_err(lhs.span(), msg);
 419         false
 420     }
 421     // we don't abort on errors on rejection, the driver will do that for us
 422     // after parsing/expansion. we can report every error in every macro this way.
 423 }
 424
 425 /// Check that the lhs contains no repetition which could match an empty token
 426 /// tree, because then the matcher would hang indefinitely.
 427 fn check_lhs_no_empty_seq(sess: &ParseSess, tts: &[quoted::TokenTree]) -> bool {
 428     use quoted::TokenTree;
 429     for tt in tts {
 430         match *tt {
 431             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => (),
 432             TokenTree::Delimited(_, ref del) => if !check_lhs_no_empty_seq(sess, &del.tts) {
 433                 return false;
 434             },
 435             TokenTree::Sequence(span, ref seq) => {
 436                 if seq.separator.is_none() && seq.tts.iter().all(|seq_tt| {
 437                     match *seq_tt {
 438                         TokenTree::MetaVarDecl(_, _, id) => id.name == "vis",
 439                         TokenTree::Sequence(_, ref sub_seq) =>
 440                             sub_seq.op == quoted::KleeneOp::ZeroOrMore
 441                             || sub_seq.op == quoted::KleeneOp::ZeroOrOne,
 442                         _ => false,
 443                     }
 444                 }) {
 445                     let sp = span.entire();
 446                     sess.span_diagnostic.span_err(sp, "repetition matches empty token tree");
 447                     return false;
 448                 }
 449                 if !check_lhs_no_empty_seq(sess, &seq.tts) {
 450                     return false;
 451                 }
 452             }
 453         }
 454     }
 455
 456     true
 457 }
 458
 459 fn check_rhs(sess: &ParseSess, rhs: &quoted::TokenTree) -> bool {
 460     match *rhs {
 461         quoted::TokenTree::Delimited(..) => return true,
 462         _ => sess.span_diagnostic.span_err(rhs.span(), "macro rhs must be delimited")
 463     }
 464     false
 465 }
 466
 467 fn check_matcher(sess: &ParseSess,
 468                  features: &Features,
 469                  attrs: &[ast::Attribute],
 470                  matcher: &[quoted::TokenTree]) -> bool {
 471     let first_sets = FirstSets::new(matcher);
 472     let empty_suffix = TokenSet::empty();
 473     let err = sess.span_diagnostic.err_count();
 474     check_matcher_core(sess, features, attrs, &first_sets, matcher, &empty_suffix);
 475     err == sess.span_diagnostic.err_count()
 476 }
 477
 478 // `The FirstSets` for a matcher is a mapping from subsequences in the
 479 // matcher to the FIRST set for that subsequence.
 480 //
 481 // This mapping is partially precomputed via a backwards scan over the
 482 // token trees of the matcher, which provides a mapping from each
 483 // repetition sequence to its *first* set.
 484 //
 485 // (Hypothetically, sequences should be uniquely identifiable via their
 486 // spans, though perhaps that is false, e.g., for macro-generated macros
 487 // that do not try to inject artificial span information. My plan is
 488 // to try to catch such cases ahead of time and not include them in
 489 // the precomputed mapping.)
 490 struct FirstSets {
 491     // this maps each TokenTree::Sequence `$(tt ...) SEP OP` that is uniquely identified by its
 492     // span in the original matcher to the First set for the inner sequence `tt ...`.
 493     //
 494     // If two sequences have the same span in a matcher, then map that
 495     // span to None (invalidating the mapping here and forcing the code to
 496     // use a slow path).
 497     first: FxHashMap<Span, Option<TokenSet>>,
 498 }
 499
 500 impl FirstSets {
 501     fn new(tts: &[quoted::TokenTree]) -> FirstSets {
 502         use quoted::TokenTree;
 503
 504         let mut sets = FirstSets { first: FxHashMap::default() };
 505         build_recur(&mut sets, tts);
 506         return sets;
 507
 508         // walks backward over `tts`, returning the FIRST for `tts`
 509         // and updating `sets` at the same time for all sequence
 510         // substructure we find within `tts`.
 511         fn build_recur(sets: &mut FirstSets, tts: &[TokenTree]) -> TokenSet {
 512             let mut first = TokenSet::empty();
 513             for tt in tts.iter().rev() {
 514                 match *tt {
 515                     TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 516                         first.replace_with(tt.clone());
 517                     }
 518                     TokenTree::Delimited(span, ref delimited) => {
 519                         build_recur(sets, &delimited.tts[..]);
 520                         first.replace_with(delimited.open_tt(span.open));
 521                     }
 522                     TokenTree::Sequence(sp, ref seq_rep) => {
 523                         let subfirst = build_recur(sets, &seq_rep.tts[..]);
 524
 525                         match sets.first.entry(sp.entire()) {
 526                             Entry::Vacant(vac) => {
 527                                 vac.insert(Some(subfirst.clone()));
 528                             }
 529                             Entry::Occupied(mut occ) => {
 530                                 // if there is already an entry, then a span must have collided.
 531                                 // This should not happen with typical macro_rules macros,
 532                                 // but syntax extensions need not maintain distinct spans,
 533                                 // so distinct syntax trees can be assigned the same span.
 534                                 // In such a case, the map cannot be trusted; so mark this
 535                                 // entry as unusable.
 536                                 occ.insert(None);
 537                             }
 538                         }
 539
 540                         // If the sequence contents can be empty, then the first
 541                         // token could be the separator token itself.
 542
 543                         if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 544                                                         subfirst.maybe_empty) {
 545                             first.add_one_maybe(TokenTree::Token(sp.entire(), sep.clone()));
 546                         }
 547
 548                         // Reverse scan: Sequence comes before `first`.
 549                         if subfirst.maybe_empty
 550                            || seq_rep.op == quoted::KleeneOp::ZeroOrMore
 551                            || seq_rep.op == quoted::KleeneOp::ZeroOrOne
 552                         {
 553                             // If sequence is potentially empty, then
 554                             // union them (preserving first emptiness).
 555                             first.add_all(&TokenSet { maybe_empty: true, ..subfirst });
 556                         } else {
 557                             // Otherwise, sequence guaranteed
 558                             // non-empty; replace first.
 559                             first = subfirst;
 560                         }
 561                     }
 562                 }
 563             }
 564
 565             first
 566         }
 567     }
 568
 569     // walks forward over `tts` until all potential FIRST tokens are
 570     // identified.
 571     fn first(&self, tts: &[quoted::TokenTree]) -> TokenSet {
 572         use quoted::TokenTree;
 573
 574         let mut first = TokenSet::empty();
 575         for tt in tts.iter() {
 576             assert!(first.maybe_empty);
 577             match *tt {
 578                 TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 579                     first.add_one(tt.clone());
 580                     return first;
 581                 }
 582                 TokenTree::Delimited(span, ref delimited) => {
 583                     first.add_one(delimited.open_tt(span.open));
 584                     return first;
 585                 }
 586                 TokenTree::Sequence(sp, ref seq_rep) => {
 587                     match self.first.get(&sp.entire()) {
 588                         Some(&Some(ref subfirst)) => {
 589
 590                             // If the sequence contents can be empty, then the first
 591                             // token could be the separator token itself.
 592
 593                             if let (Some(ref sep), true) = (seq_rep.separator.clone(),
 594                                                             subfirst.maybe_empty) {
 595                                 first.add_one_maybe(TokenTree::Token(sp.entire(), sep.clone()));
 596                             }
 597
 598                             assert!(first.maybe_empty);
 599                             first.add_all(subfirst);
 600                             if subfirst.maybe_empty
 601                                || seq_rep.op == quoted::KleeneOp::ZeroOrMore
 602                                || seq_rep.op == quoted::KleeneOp::ZeroOrOne
 603                             {
 604                                 // continue scanning for more first
 605                                 // tokens, but also make sure we
 606                                 // restore empty-tracking state
 607                                 first.maybe_empty = true;
 608                                 continue;
 609                             } else {
 610                                 return first;
 611                             }
 612                         }
 613
 614                         Some(&None) => {
 615                             panic!("assume all sequences have (unique) spans for now");
 616                         }
 617
 618                         None => {
 619                             panic!("We missed a sequence during FirstSets construction");
 620                         }
 621                     }
 622                 }
 623             }
 624         }
 625
 626         // we only exit the loop if `tts` was empty or if every
 627         // element of `tts` matches the empty sequence.
 628         assert!(first.maybe_empty);
 629         first
 630     }
 631 }
 632
 633 // A set of `quoted::TokenTree`s, which may include `TokenTree::Match`s
 634 // (for macro-by-example syntactic variables). It also carries the
 635 // `maybe_empty` flag; that is true if and only if the matcher can
 636 // match an empty token sequence.
 637 //
 638 // The First set is computed on submatchers like `$($a:expr b),* $(c)* d`,
 639 // which has corresponding FIRST = {$a:expr, c, d}.
 640 // Likewise, `$($a:expr b),* $(c)+ d` has FIRST = {$a:expr, c}.
 641 //
 642 // (Notably, we must allow for *-op to occur zero times.)
 643 #[derive(Clone, Debug)]
 644 struct TokenSet {
 645     tokens: Vec<quoted::TokenTree>,
 646     maybe_empty: bool,
 647 }
 648
 649 impl TokenSet {
 650     // Returns a set for the empty sequence.
 651     fn empty() -> Self { TokenSet { tokens: Vec::new(), maybe_empty: true } }
 652
 653     // Returns the set `{ tok }` for the single-token (and thus
 654     // non-empty) sequence [tok].
 655     fn singleton(tok: quoted::TokenTree) -> Self {
 656         TokenSet { tokens: vec![tok], maybe_empty: false }
 657     }
 658
 659     // Changes self to be the set `{ tok }`.
 660     // Since `tok` is always present, marks self as non-empty.
 661     fn replace_with(&mut self, tok: quoted::TokenTree) {
 662         self.tokens.clear();
 663         self.tokens.push(tok);
 664         self.maybe_empty = false;
 665     }
 666
 667     // Changes self to be the empty set `{}`; meant for use when
 668     // the particular token does not matter, but we want to
 669     // record that it occurs.
 670     fn replace_with_irrelevant(&mut self) {
 671         self.tokens.clear();
 672         self.maybe_empty = false;
 673     }
 674
 675     // Adds `tok` to the set for `self`, marking sequence as non-empy.
 676     fn add_one(&mut self, tok: quoted::TokenTree) {
 677         if !self.tokens.contains(&tok) {
 678             self.tokens.push(tok);
 679         }
 680         self.maybe_empty = false;
 681     }
 682
 683     // Adds `tok` to the set for `self`. (Leaves `maybe_empty` flag alone.)
 684     fn add_one_maybe(&mut self, tok: quoted::TokenTree) {
 685         if !self.tokens.contains(&tok) {
 686             self.tokens.push(tok);
 687         }
 688     }
 689
 690     // Adds all elements of `other` to this.
 691     //
 692     // (Since this is a set, we filter out duplicates.)
 693     //
 694     // If `other` is potentially empty, then preserves the previous
 695     // setting of the empty flag of `self`. If `other` is guaranteed
 696     // non-empty, then `self` is marked non-empty.
 697     fn add_all(&mut self, other: &Self) {
 698         for tok in &other.tokens {
 699             if !self.tokens.contains(tok) {
 700                 self.tokens.push(tok.clone());
 701             }
 702         }
 703         if !other.maybe_empty {
 704             self.maybe_empty = false;
 705         }
 706     }
 707 }
 708
 709 // Checks that `matcher` is internally consistent and that it
 710 // can legally by followed by a token N, for all N in `follow`.
 711 // (If `follow` is empty, then it imposes no constraint on
 712 // the `matcher`.)
 713 //
 714 // Returns the set of NT tokens that could possibly come last in
 715 // `matcher`. (If `matcher` matches the empty sequence, then
 716 // `maybe_empty` will be set to true.)
 717 //
 718 // Requires that `first_sets` is pre-computed for `matcher`;
 719 // see `FirstSets::new`.
 720 fn check_matcher_core(sess: &ParseSess,
 721                       features: &Features,
 722                       attrs: &[ast::Attribute],
 723                       first_sets: &FirstSets,
 724                       matcher: &[quoted::TokenTree],
 725                       follow: &TokenSet) -> TokenSet {
 726     use quoted::TokenTree;
 727
 728     let mut last = TokenSet::empty();
 729
 730     // 2. For each token and suffix  [T, SUFFIX] in M:
 731     // ensure that T can be followed by SUFFIX, and if SUFFIX may be empty,
 732     // then ensure T can also be followed by any element of FOLLOW.
 733     'each_token: for i in 0..matcher.len() {
 734         let token = &matcher[i];
 735         let suffix = &matcher[i+1..];
 736
 737         let build_suffix_first = || {
 738             let mut s = first_sets.first(suffix);
 739             if s.maybe_empty { s.add_all(follow); }
 740             s
 741         };
 742
 743         // (we build `suffix_first` on demand below; you can tell
 744         // which cases are supposed to fall through by looking for the
 745         // initialization of this variable.)
 746         let suffix_first;
 747
 748         // First, update `last` so that it corresponds to the set
 749         // of NT tokens that might end the sequence `... token`.
 750         match *token {
 751             TokenTree::Token(..) | TokenTree::MetaVar(..) | TokenTree::MetaVarDecl(..) => {
 752                 let can_be_followed_by_any;
 753                 if let Err(bad_frag) = has_legal_fragment_specifier(sess, features, attrs, token) {
 754                     let msg = format!("invalid fragment specifier `{}`", bad_frag);
 755                     sess.span_diagnostic.struct_span_err(token.span(), &msg)
 756                         .help(VALID_FRAGMENT_NAMES_MSG)
 757                         .emit();
 758                     // (This eliminates false positives and duplicates
 759                     // from error messages.)
 760                     can_be_followed_by_any = true;
 761                 } else {
 762                     can_be_followed_by_any = token_can_be_followed_by_any(token);
 763                 }
 764
 765                 if can_be_followed_by_any {
 766                     // don't need to track tokens that work with any,
 767                     last.replace_with_irrelevant();
 768                     // ... and don't need to check tokens that can be
 769                     // followed by anything against SUFFIX.
 770                     continue 'each_token;
 771                 } else {
 772                     last.replace_with(token.clone());
 773                     suffix_first = build_suffix_first();
 774                 }
 775             }
 776             TokenTree::Delimited(span, ref d) => {
 777                 let my_suffix = TokenSet::singleton(d.close_tt(span.close));
 778                 check_matcher_core(sess, features, attrs, first_sets, &d.tts, &my_suffix);
 779                 // don't track non NT tokens
 780                 last.replace_with_irrelevant();
 781
 782                 // also, we don't need to check delimited sequences
 783                 // against SUFFIX
 784                 continue 'each_token;
 785             }
 786             TokenTree::Sequence(sp, ref seq_rep) => {
 787                 suffix_first = build_suffix_first();
 788                 // The trick here: when we check the interior, we want
 789                 // to include the separator (if any) as a potential
 790                 // (but not guaranteed) element of FOLLOW. So in that
 791                 // case, we make a temp copy of suffix and stuff
 792                 // delimiter in there.
 793                 //
 794                 // FIXME: Should I first scan suffix_first to see if
 795                 // delimiter is already in it before I go through the
 796                 // work of cloning it? But then again, this way I may
 797                 // get a "tighter" span?
 798                 let mut new;
 799                 let my_suffix = if let Some(ref u) = seq_rep.separator {
 800                     new = suffix_first.clone();
 801                     new.add_one_maybe(TokenTree::Token(sp.entire(), u.clone()));
 802                     &new
 803                 } else {
 804                     &suffix_first
 805                 };
 806
 807                 // At this point, `suffix_first` is built, and
 808                 // `my_suffix` is some TokenSet that we can use
 809                 // for checking the interior of `seq_rep`.
 810                 let next = check_matcher_core(sess,
 811                                               features,
 812                                               attrs,
 813                                               first_sets,
 814                                               &seq_rep.tts,
 815                                               my_suffix);
 816                 if next.maybe_empty {
 817                     last.add_all(&next);
 818                 } else {
 819                     last = next;
 820                 }
 821
 822                 // the recursive call to check_matcher_core already ran the 'each_last
 823                 // check below, so we can just keep going forward here.
 824                 continue 'each_token;
 825             }
 826         }
 827
 828         // (`suffix_first` guaranteed initialized once reaching here.)
 829
 830         // Now `last` holds the complete set of NT tokens that could
 831         // end the sequence before SUFFIX. Check that every one works with `suffix`.
 832         'each_last: for token in &last.tokens {
 833             if let TokenTree::MetaVarDecl(_, ref name, ref frag_spec) = *token {
 834                 for next_token in &suffix_first.tokens {
 835                     match is_in_follow(next_token, &frag_spec.as_str()) {
 836                         IsInFollow::Invalid(msg, help) => {
 837                             sess.span_diagnostic.struct_span_err(next_token.span(), &msg)
 838                                 .help(help).emit();
 839                             // don't bother reporting every source of
 840                             // conflict for a particular element of `last`.
 841                             continue 'each_last;
 842                         }
 843                         IsInFollow::Yes => {}
 844                         IsInFollow::No(ref possible) => {
 845                             let may_be = if last.tokens.len() == 1 &&
 846                                 suffix_first.tokens.len() == 1
 847                             {
 848                                 "is"
 849                             } else {
 850                                 "may be"
 851                             };
 852
 853                             let sp = next_token.span();
 854                             let mut err = sess.span_diagnostic.struct_span_err(
 855                                 sp,
 856                                 &format!("`${name}:{frag}` {may_be} followed by `{next}`, which \
 857                                           is not allowed for `{frag}` fragments",
 858                                          name=name,
 859                                          frag=frag_spec,
 860                                          next=quoted_tt_to_string(next_token),
 861                                          may_be=may_be),
 862                             );
 863                             err.span_label(
 864                                 sp,
 865                                 format!("not allowed after `{}` fragments", frag_spec),
 866                             );
 867                             let msg = "allowed there are: ";
 868                             match &possible[..] {
 869                                 &[] => {}
 870                                 &[t] => {
 871                                     err.note(&format!(
 872                                         "only {} is allowed after `{}` fragments",
 873                                         t,
 874                                         frag_spec,
 875                                     ));
 876                                 }
 877                                 ts => {
 878                                     err.note(&format!(
 879                                         "{}{} or {}",
 880                                         msg,
 881                                         ts[..ts.len() - 1].iter().map(|s| *s)
 882                                             .collect::<Vec<_>>().join(", "),
 883                                         ts[ts.len() - 1],
 884                                     ));
 885                                 }
 886                             }
 887                             err.emit();
 888                         }
 889                     }
 890                 }
 891             }
 892         }
 893     }
 894     last
 895 }
 896
 897 fn token_can_be_followed_by_any(tok: &quoted::TokenTree) -> bool {
 898     if let quoted::TokenTree::MetaVarDecl(_, _, frag_spec) = *tok {
 899         frag_can_be_followed_by_any(&frag_spec.as_str())
 900     } else {
 901         // (Non NT's can always be followed by anthing in matchers.)
 902         true
 903     }
 904 }
 905
 906 /// True if a fragment of type `frag` can be followed by any sort of
 907 /// token.  We use this (among other things) as a useful approximation
 908 /// for when `frag` can be followed by a repetition like `$(...)*` or
 909 /// `$(...)+`. In general, these can be a bit tricky to reason about,
 910 /// so we adopt a conservative position that says that any fragment
 911 /// specifier which consumes at most one token tree can be followed by
 912 /// a fragment specifier (indeed, these fragments can be followed by
 913 /// ANYTHING without fear of future compatibility hazards).
 914 fn frag_can_be_followed_by_any(frag: &str) -> bool {
 915     match frag {
 916         "item"     | // always terminated by `}` or `;`
 917         "block"    | // exactly one token tree
 918         "ident"    | // exactly one token tree
 919         "literal"  | // exactly one token tree
 920         "meta"     | // exactly one token tree
 921         "lifetime" | // exactly one token tree
 922         "tt" =>   // exactly one token tree
 923             true,
 924
 925         _ =>
 926             false,
 927     }
 928 }
 929
 930 enum IsInFollow {
 931     Yes,
 932     No(Vec<&'static str>),
 933     Invalid(String, &'static str),
 934 }
 935
 936 /// True if `frag` can legally be followed by the token `tok`. For
 937 /// fragments that can consume an unbounded number of tokens, `tok`
 938 /// must be within a well-defined follow set. This is intended to
 939 /// guarantee future compatibility: for example, without this rule, if
 940 /// we expanded `expr` to include a new binary operator, we might
 941 /// break macros that were relying on that binary operator as a
 942 /// separator.
 943 // when changing this do not forget to update doc/book/macros.md!
 944 fn is_in_follow(tok: &quoted::TokenTree, frag: &str) -> IsInFollow {
 945     use quoted::TokenTree;
 946
 947     if let TokenTree::Token(_, token::CloseDelim(_)) = *tok {
 948         // closing a token tree can never be matched by any fragment;
 949         // iow, we always require that `(` and `)` match, etc.
 950         IsInFollow::Yes
 951     } else {
 952         match frag {
 953             "item" => {
 954                 // since items *must* be followed by either a `;` or a `}`, we can
 955                 // accept anything after them
 956                 IsInFollow::Yes
 957             },
 958             "block" => {
 959                 // anything can follow block, the braces provide an easy boundary to
 960                 // maintain
 961                 IsInFollow::Yes
 962             },
 963             "stmt" | "expr"  => {
 964                 let tokens = vec!["`=>`", "`,`", "`;`"];
 965                 match *tok {
 966                     TokenTree::Token(_, ref tok) => match *tok {
 967                         FatArrow | Comma | Semi => IsInFollow::Yes,
 968                         _ => IsInFollow::No(tokens),
 969                     },
 970                     _ => IsInFollow::No(tokens),
 971                 }
 972             },
 973             "pat" => {
 974                 let tokens = vec!["`=>`", "`,`", "`=`", "`|`", "`if`", "`in`"];
 975                 match *tok {
 976                     TokenTree::Token(_, ref tok) => match *tok {
 977                         FatArrow | Comma | Eq | BinOp(token::Or) => IsInFollow::Yes,
 978                         Ident(i, false) if i.name == "if" || i.name == "in" => IsInFollow::Yes,
 979                         _ => IsInFollow::No(tokens),
 980                     },
 981                     _ => IsInFollow::No(tokens),
 982                 }
 983             },
 984             "path" | "ty" => {
 985                 let tokens = vec![
 986                     "`{`", "`[`", "`=>`", "`,`", "`>`","`=`", "`:`", "`;`", "`|`", "`as`",
 987                     "`where`",
 988                 ];
 989                 match *tok {
 990                     TokenTree::Token(_, ref tok) => match *tok {
 991                         OpenDelim(token::DelimToken::Brace) |
 992                         OpenDelim(token::DelimToken::Bracket) |
 993                         Comma | FatArrow | Colon | Eq | Gt | BinOp(token::Shr) | Semi |
 994                         BinOp(token::Or) => IsInFollow::Yes,
 995                         Ident(i, false) if i.name == "as" || i.name == "where" => IsInFollow::Yes,
 996                         _ => IsInFollow::No(tokens),
 997                     },
 998                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "block" => IsInFollow::Yes,
 999                     _ => IsInFollow::No(tokens),
1000                 }
1001             },
1002             "ident" | "lifetime" => {
1003                 // being a single token, idents and lifetimes are harmless
1004                 IsInFollow::Yes
1005             },
1006             "literal" => {
1007                 // literals may be of a single token, or two tokens (negative numbers)
1008                 IsInFollow::Yes
1009             },
1010             "meta" | "tt" => {
1011                 // being either a single token or a delimited sequence, tt is
1012                 // harmless
1013                 IsInFollow::Yes
1014             },
1015             "vis" => {
1016                 // Explicitly disallow `priv`, on the off chance it comes back.
1017                 let tokens = vec!["`,`", "an ident", "a type"];
1018                 match *tok {
1019                     TokenTree::Token(_, ref tok) => match *tok {
1020                         Comma => IsInFollow::Yes,
1021                         Ident(i, is_raw) if is_raw || i.name != "priv" => IsInFollow::Yes,
1022                         ref tok => if tok.can_begin_type() {
1023                             IsInFollow::Yes
1024                         } else {
1025                             IsInFollow::No(tokens)
1026                         }
1027                     },
1028                     TokenTree::MetaVarDecl(_, _, frag) if frag.name == "ident"
1029                                                        || frag.name == "ty"
1030                                                        || frag.name == "path" => IsInFollow::Yes,
1031                     _ => IsInFollow::No(tokens),
1032                 }
1033             },
1034             "" => IsInFollow::Yes, // keywords::Invalid
1035             _ => IsInFollow::Invalid(format!("invalid fragment specifier `{}`", frag),
1036                                      VALID_FRAGMENT_NAMES_MSG),
1037         }
1038     }
1039 }
1040
1041 fn has_legal_fragment_specifier(sess: &ParseSess,
1042                                 features: &Features,
1043                                 attrs: &[ast::Attribute],
1044                                 tok: &quoted::TokenTree) -> Result<(), String> {
1045     debug!("has_legal_fragment_specifier({:?})", tok);
1046     if let quoted::TokenTree::MetaVarDecl(_, _, ref frag_spec) = *tok {
1047         let frag_name = frag_spec.as_str();
1048         let frag_span = tok.span();
1049         if !is_legal_fragment_specifier(sess, features, attrs, &frag_name, frag_span) {
1050             return Err(frag_name.to_string());
1051         }
1052     }
1053     Ok(())
1054 }
1055
1056 fn is_legal_fragment_specifier(_sess: &ParseSess,
1057                                _features: &Features,
1058                                _attrs: &[ast::Attribute],
1059                                frag_name: &str,
1060                                _frag_span: Span) -> bool {
1061     /*
1062      * If new fragment specifiers are invented in nightly, `_sess`,
1063      * `_features`, `_attrs`, and `_frag_span` will be useful here
1064      * for checking against feature gates. See past versions of
1065      * this function.
1066      */
1067     match frag_name {
1068         "item" | "block" | "stmt" | "expr" | "pat" | "lifetime" |
1069         "path" | "ty" | "ident" | "meta" | "tt" | "vis" | "literal" |
1070         "" => true,
1071         _ => false,
1072     }
1073 }
1074
1075 fn quoted_tt_to_string(tt: &quoted::TokenTree) -> String {
1076     match *tt {
1077         quoted::TokenTree::Token(_, ref tok) => crate::print::pprust::token_to_string(tok),
1078         quoted::TokenTree::MetaVar(_, name) => format!("${}", name),
1079         quoted::TokenTree::MetaVarDecl(_, name, kind) => format!("${}:{}", name, kind),
1080         _ => panic!("unexpected quoted::TokenTree::{{Sequence or Delimited}} \
1081                      in follow set checker"),
1082     }
1083 }