src/libsyntax_pos/symbol.rs

   1 //! An "interner" is a data structure that associates values with usize tags and
   2 //! allows bidirectional lookup; i.e., given a value, one can easily find the
   3 //! type, and vice versa.
   4
   5 use arena::DroplessArena;
   6 use rustc_data_structures::fx::FxHashMap;
   7 use rustc_index::vec::Idx;
   8 use rustc_macros::symbols;
   9 use rustc_serialize::{Decodable, Decoder, Encodable, Encoder};
  10 use rustc_serialize::{UseSpecializedDecodable, UseSpecializedEncodable};
  11
  12 use std::cmp::{PartialEq, PartialOrd, Ord};
  13 use std::fmt;
  14 use std::hash::{Hash, Hasher};
  15 use std::str;
  16
  17 use crate::{Span, DUMMY_SP, GLOBALS};
  18
  19 #[cfg(test)]
  20 mod tests;
  21
  22 symbols! {
  23     // After modifying this list adjust `is_special`, `is_used_keyword`/`is_unused_keyword`,
  24     // this should be rarely necessary though if the keywords are kept in alphabetic order.
  25     Keywords {
  26         // Special reserved identifiers used internally for elided lifetimes,
  27         // unnamed method parameters, crate root module, error recovery etc.
  28         Invalid:            "",
  29         PathRoot:           "{{root}}",
  30         DollarCrate:        "$crate",
  31         Underscore:         "_",
  32
  33         // Keywords that are used in stable Rust.
  34         As:                 "as",
  35         Break:              "break",
  36         Const:              "const",
  37         Continue:           "continue",
  38         Crate:              "crate",
  39         Else:               "else",
  40         Enum:               "enum",
  41         Extern:             "extern",
  42         False:              "false",
  43         Fn:                 "fn",
  44         For:                "for",
  45         If:                 "if",
  46         Impl:               "impl",
  47         In:                 "in",
  48         Let:                "let",
  49         Loop:               "loop",
  50         Match:              "match",
  51         Mod:                "mod",
  52         Move:               "move",
  53         Mut:                "mut",
  54         Pub:                "pub",
  55         Ref:                "ref",
  56         Return:             "return",
  57         SelfLower:          "self",
  58         SelfUpper:          "Self",
  59         Static:             "static",
  60         Struct:             "struct",
  61         Super:              "super",
  62         Trait:              "trait",
  63         True:               "true",
  64         Type:               "type",
  65         Unsafe:             "unsafe",
  66         Use:                "use",
  67         Where:              "where",
  68         While:              "while",
  69
  70         // Keywords that are used in unstable Rust or reserved for future use.
  71         Abstract:           "abstract",
  72         Become:             "become",
  73         Box:                "box",
  74         Do:                 "do",
  75         Final:              "final",
  76         Macro:              "macro",
  77         Override:           "override",
  78         Priv:               "priv",
  79         Typeof:             "typeof",
  80         Unsized:            "unsized",
  81         Virtual:            "virtual",
  82         Yield:              "yield",
  83
  84         // Edition-specific keywords that are used in stable Rust.
  85         Async:              "async", // >= 2018 Edition only
  86         Await:              "await", // >= 2018 Edition only
  87         Dyn:                "dyn", // >= 2018 Edition only
  88
  89         // Edition-specific keywords that are used in unstable Rust or reserved for future use.
  90         Try:                "try", // >= 2018 Edition only
  91
  92         // Special lifetime names
  93         UnderscoreLifetime: "'_",
  94         StaticLifetime:     "'static",
  95
  96         // Weak keywords, have special meaning only in specific contexts.
  97         Auto:               "auto",
  98         Catch:              "catch",
  99         Default:            "default",
 100         Union:              "union",
 101     }
 102
 103     // Symbols that can be referred to with syntax_pos::sym::*. The symbol is
 104     // the stringified identifier unless otherwise specified (e.g.
 105     // `proc_dash_macro` represents "proc-macro").
 106     //
 107     // As well as the symbols listed, there are symbols for the the strings
 108     // "0", "1", ..., "9", which are accessible via `sym::integer`.
 109     Symbols {
 110         aarch64_target_feature,
 111         abi,
 112         abi_amdgpu_kernel,
 113         abi_efiapi,
 114         abi_msp430_interrupt,
 115         abi_ptx,
 116         abi_sysv64,
 117         abi_thiscall,
 118         abi_unadjusted,
 119         abi_vectorcall,
 120         abi_x86_interrupt,
 121         aborts,
 122         advanced_slice_patterns,
 123         adx_target_feature,
 124         alias,
 125         align,
 126         alignstack,
 127         all,
 128         allocator,
 129         allocator_internals,
 130         alloc_error_handler,
 131         allow,
 132         allowed,
 133         allow_fail,
 134         allow_internal_unsafe,
 135         allow_internal_unstable,
 136         allow_internal_unstable_backcompat_hack,
 137         always,
 138         and,
 139         any,
 140         arbitrary_enum_discriminant,
 141         arbitrary_self_types,
 142         Arguments,
 143         ArgumentV1,
 144         arm_target_feature,
 145         asm,
 146         assert,
 147         associated_consts,
 148         associated_type_bounds,
 149         associated_type_defaults,
 150         associated_types,
 151         async_await,
 152         async_closure,
 153         attr,
 154         attributes,
 155         attr_literals,
 156         augmented_assignments,
 157         automatically_derived,
 158         avx512_target_feature,
 159         await_macro,
 160         begin_panic,
 161         bench,
 162         bin,
 163         bind_by_move_pattern_guards,
 164         block,
 165         bool,
 166         borrowck_graphviz_postflow,
 167         borrowck_graphviz_preflow,
 168         box_patterns,
 169         box_syntax,
 170         braced_empty_structs,
 171         C,
 172         cdylib,
 173         cfg,
 174         cfg_attr,
 175         cfg_attr_multi,
 176         cfg_doctest,
 177         cfg_target_feature,
 178         cfg_target_has_atomic,
 179         cfg_target_thread_local,
 180         cfg_target_vendor,
 181         char,
 182         clippy,
 183         clone,
 184         Clone,
 185         clone_closures,
 186         clone_from,
 187         closure_to_fn_coercion,
 188         cmp,
 189         cmpxchg16b_target_feature,
 190         cold,
 191         column,
 192         compile_error,
 193         compiler_builtins,
 194         concat,
 195         concat_idents,
 196         conservative_impl_trait,
 197         console,
 198         const_compare_raw_pointers,
 199         const_constructor,
 200         const_extern_fn,
 201         const_fn,
 202         const_fn_union,
 203         const_generics,
 204         const_indexing,
 205         const_in_array_repeat_expressions,
 206         const_let,
 207         const_panic,
 208         const_raw_ptr_deref,
 209         const_raw_ptr_to_usize_cast,
 210         const_transmute,
 211         contents,
 212         context,
 213         convert,
 214         Copy,
 215         copy_closures,
 216         core,
 217         core_intrinsics,
 218         crate_id,
 219         crate_in_paths,
 220         crate_local,
 221         crate_name,
 222         crate_type,
 223         crate_visibility_modifier,
 224         custom_attribute,
 225         custom_derive,
 226         custom_inner_attributes,
 227         custom_test_frameworks,
 228         c_variadic,
 229         debug_trait,
 230         declare_lint_pass,
 231         decl_macro,
 232         Debug,
 233         Decodable,
 234         Default,
 235         default_lib_allocator,
 236         default_type_parameter_fallback,
 237         default_type_params,
 238         deny,
 239         deprecated,
 240         deref,
 241         deref_mut,
 242         derive,
 243         diagnostic,
 244         direct,
 245         doc,
 246         doc_alias,
 247         doc_cfg,
 248         doc_keyword,
 249         doc_masked,
 250         doc_spotlight,
 251         doctest,
 252         document_private_items,
 253         dotdoteq_in_patterns,
 254         dotdot_in_tuple_patterns,
 255         double_braced_crate: "{{crate}}",
 256         double_braced_impl: "{{impl}}",
 257         double_braced_misc: "{{misc}}",
 258         double_braced_closure: "{{closure}}",
 259         double_braced_constructor: "{{constructor}}",
 260         double_braced_constant: "{{constant}}",
 261         double_braced_opaque: "{{opaque}}",
 262         dropck_eyepatch,
 263         dropck_parametricity,
 264         drop_types_in_const,
 265         dylib,
 266         dyn_trait,
 267         eh_personality,
 268         eh_unwind_resume,
 269         enable,
 270         Encodable,
 271         env,
 272         eq,
 273         err,
 274         Err,
 275         Eq,
 276         Equal,
 277         except,
 278         exclusive_range_pattern,
 279         exhaustive_integer_patterns,
 280         exhaustive_patterns,
 281         existential_type,
 282         expected,
 283         export_name,
 284         expr,
 285         extern_absolute_paths,
 286         external_doc,
 287         extern_crate_item_prelude,
 288         extern_crate_self,
 289         extern_in_paths,
 290         extern_prelude,
 291         extern_types,
 292         f16c_target_feature,
 293         f32,
 294         f64,
 295         feature,
 296         ffi_returns_twice,
 297         field,
 298         field_init_shorthand,
 299         file,
 300         fmt,
 301         fmt_internals,
 302         fn_must_use,
 303         forbid,
 304         format_args,
 305         format_args_nl,
 306         from,
 307         From,
 308         from_desugaring,
 309         from_error,
 310         from_generator,
 311         from_method,
 312         from_ok,
 313         from_usize,
 314         fundamental,
 315         future,
 316         Future,
 317         FxHashSet,
 318         FxHashMap,
 319         gen_future,
 320         generators,
 321         generic_associated_types,
 322         generic_param_attrs,
 323         global_allocator,
 324         global_asm,
 325         globs,
 326         hash,
 327         Hash,
 328         HashSet,
 329         HashMap,
 330         hexagon_target_feature,
 331         hidden,
 332         homogeneous_aggregate,
 333         html_favicon_url,
 334         html_logo_url,
 335         html_no_source,
 336         html_playground_url,
 337         html_root_url,
 338         i128,
 339         i128_type,
 340         i16,
 341         i32,
 342         i64,
 343         i8,
 344         ident,
 345         if_let,
 346         if_while_or_patterns,
 347         ignore,
 348         impl_header_lifetime_elision,
 349         impl_lint_pass,
 350         impl_trait_in_bindings,
 351         import_shadowing,
 352         index,
 353         index_mut,
 354         in_band_lifetimes,
 355         include,
 356         include_bytes,
 357         include_str,
 358         inclusive_range_syntax,
 359         infer_outlives_requirements,
 360         infer_static_outlives_requirements,
 361         inline,
 362         intel,
 363         into_iter,
 364         IntoIterator,
 365         into_result,
 366         intrinsics,
 367         irrefutable_let_patterns,
 368         isize,
 369         issue,
 370         issue_5723_bootstrap,
 371         issue_tracker_base_url,
 372         item,
 373         item_context: "ItemContext",
 374         item_like_imports,
 375         iter,
 376         Iterator,
 377         keyword,
 378         kind,
 379         label,
 380         label_break_value,
 381         lang,
 382         lang_items,
 383         let_chains,
 384         lhs,
 385         lib,
 386         lifetime,
 387         line,
 388         link,
 389         linkage,
 390         link_args,
 391         link_cfg,
 392         link_llvm_intrinsics,
 393         link_name,
 394         link_ordinal,
 395         link_section,
 396         LintPass,
 397         lint_reasons,
 398         literal,
 399         local_inner_macros,
 400         log_syntax,
 401         loop_break_value,
 402         macro_at_most_once_rep,
 403         macro_escape,
 404         macro_export,
 405         macro_lifetime_matcher,
 406         macro_literal_matcher,
 407         macro_reexport,
 408         macro_rules,
 409         macros_in_extern,
 410         macro_use,
 411         macro_vis_matcher,
 412         main,
 413         managed_boxes,
 414         marker,
 415         marker_trait_attr,
 416         masked,
 417         match_beginning_vert,
 418         match_default_bindings,
 419         may_dangle,
 420         mem,
 421         member_constraints,
 422         message,
 423         meta,
 424         min_const_fn,
 425         min_const_unsafe_fn,
 426         mips_target_feature,
 427         mmx_target_feature,
 428         module,
 429         module_path,
 430         more_struct_aliases,
 431         movbe_target_feature,
 432         must_use,
 433         naked,
 434         naked_functions,
 435         name,
 436         needs_allocator,
 437         needs_panic_runtime,
 438         negate_unsigned,
 439         never,
 440         never_type,
 441         new,
 442         next,
 443         __next,
 444         nll,
 445         no_builtins,
 446         no_core,
 447         no_crate_inject,
 448         no_debug,
 449         no_default_passes,
 450         no_implicit_prelude,
 451         no_inline,
 452         no_link,
 453         no_main,
 454         no_mangle,
 455         non_ascii_idents,
 456         None,
 457         non_exhaustive,
 458         non_modrs_mods,
 459         no_stack_check,
 460         no_start,
 461         no_std,
 462         not,
 463         note,
 464         object_safe_for_dispatch,
 465         Ok,
 466         omit_gdb_pretty_printer_section,
 467         on,
 468         on_unimplemented,
 469         oom,
 470         ops,
 471         optimize,
 472         optimize_attribute,
 473         optin_builtin_traits,
 474         option,
 475         Option,
 476         option_env,
 477         opt_out_copy,
 478         or,
 479         or_patterns,
 480         Ord,
 481         Ordering,
 482         Output,
 483         overlapping_marker_traits,
 484         packed,
 485         panic,
 486         panic_handler,
 487         panic_impl,
 488         panic_implementation,
 489         panic_runtime,
 490         parent_trait,
 491         partial_cmp,
 492         param_attrs,
 493         PartialEq,
 494         PartialOrd,
 495         passes,
 496         pat,
 497         path,
 498         pattern_parentheses,
 499         Pending,
 500         pin,
 501         Pin,
 502         pinned,
 503         platform_intrinsics,
 504         plugin,
 505         plugin_registrar,
 506         plugins,
 507         Poll,
 508         poll_with_tls_context,
 509         powerpc_target_feature,
 510         precise_pointer_size_matching,
 511         prelude,
 512         prelude_import,
 513         primitive,
 514         proc_dash_macro: "proc-macro",
 515         proc_macro,
 516         proc_macro_attribute,
 517         proc_macro_def_site,
 518         proc_macro_derive,
 519         proc_macro_expr,
 520         proc_macro_gen,
 521         proc_macro_hygiene,
 522         proc_macro_internals,
 523         proc_macro_mod,
 524         proc_macro_non_items,
 525         proc_macro_path_invoc,
 526         profiler_runtime,
 527         pub_restricted,
 528         pushpop_unsafe,
 529         quad_precision_float,
 530         question_mark,
 531         quote,
 532         Range,
 533         RangeFrom,
 534         RangeFull,
 535         RangeInclusive,
 536         RangeTo,
 537         RangeToInclusive,
 538         raw_dylib,
 539         raw_identifiers,
 540         Ready,
 541         reason,
 542         recursion_limit,
 543         reexport_test_harness_main,
 544         reflect,
 545         relaxed_adts,
 546         repr,
 547         repr128,
 548         repr_align,
 549         repr_align_enum,
 550         repr_packed,
 551         repr_simd,
 552         repr_transparent,
 553         re_rebalance_coherence,
 554         result,
 555         Result,
 556         Return,
 557         rhs,
 558         rlib,
 559         rt,
 560         rtm_target_feature,
 561         rust,
 562         rust_2015_preview,
 563         rust_2018_preview,
 564         rust_begin_unwind,
 565         rustc,
 566         Rust,
 567         RustcDecodable,
 568         RustcEncodable,
 569         rustc_allocator,
 570         rustc_allocator_nounwind,
 571         rustc_allow_const_fn_ptr,
 572         rustc_args_required_const,
 573         rustc_attrs,
 574         rustc_builtin_macro,
 575         rustc_clean,
 576         rustc_const_unstable,
 577         rustc_conversion_suggestion,
 578         rustc_def_path,
 579         rustc_deprecated,
 580         rustc_diagnostic_item,
 581         rustc_diagnostic_macros,
 582         rustc_dirty,
 583         rustc_dummy,
 584         rustc_dump_env_program_clauses,
 585         rustc_dump_program_clauses,
 586         rustc_dump_user_substs,
 587         rustc_error,
 588         rustc_expected_cgu_reuse,
 589         rustc_if_this_changed,
 590         rustc_inherit_overflow_checks,
 591         rustc_layout,
 592         rustc_layout_scalar_valid_range_end,
 593         rustc_layout_scalar_valid_range_start,
 594         rustc_macro_transparency,
 595         rustc_mir,
 596         rustc_nonnull_optimization_guaranteed,
 597         rustc_object_lifetime_default,
 598         rustc_on_unimplemented,
 599         rustc_outlives,
 600         rustc_paren_sugar,
 601         rustc_partition_codegened,
 602         rustc_partition_reused,
 603         rustc_peek,
 604         rustc_peek_definite_init,
 605         rustc_peek_maybe_init,
 606         rustc_peek_maybe_uninit,
 607         rustc_peek_indirectly_mutable,
 608         rustc_private,
 609         rustc_proc_macro_decls,
 610         rustc_promotable,
 611         rustc_regions,
 612         rustc_stable,
 613         rustc_std_internal_symbol,
 614         rustc_symbol_name,
 615         rustc_synthetic,
 616         rustc_reservation_impl,
 617         rustc_test_marker,
 618         rustc_then_this_would_need,
 619         rustc_variance,
 620         rustdoc,
 621         rustfmt,
 622         rust_eh_personality,
 623         rust_eh_unwind_resume,
 624         rust_oom,
 625         rvalue_static_promotion,
 626         sanitizer_runtime,
 627         _Self,
 628         self_in_typedefs,
 629         self_struct_ctor,
 630         should_panic,
 631         simd,
 632         simd_ffi,
 633         since,
 634         size,
 635         slice_patterns,
 636         slicing_syntax,
 637         soft,
 638         Some,
 639         specialization,
 640         speed,
 641         spotlight,
 642         sse4a_target_feature,
 643         stable,
 644         staged_api,
 645         start,
 646         static_in_const,
 647         staticlib,
 648         static_nobundle,
 649         static_recursion,
 650         std,
 651         std_inject,
 652         str,
 653         stringify,
 654         stmt,
 655         stmt_expr_attributes,
 656         stop_after_dataflow,
 657         struct_field_attributes,
 658         struct_inherit,
 659         structural_match,
 660         struct_variant,
 661         sty,
 662         suggestion,
 663         target_feature,
 664         target_has_atomic,
 665         target_has_atomic_load_store,
 666         target_thread_local,
 667         task,
 668         tbm_target_feature,
 669         termination_trait,
 670         termination_trait_test,
 671         test,
 672         test_2018_feature,
 673         test_accepted_feature,
 674         test_case,
 675         test_removed_feature,
 676         test_runner,
 677         then_with,
 678         thread_local,
 679         tool_attributes,
 680         tool_lints,
 681         trace_macros,
 682         track_caller,
 683         trait_alias,
 684         transmute,
 685         transparent,
 686         transparent_enums,
 687         transparent_unions,
 688         trivial_bounds,
 689         Try,
 690         try_blocks,
 691         try_trait,
 692         tt,
 693         tuple_indexing,
 694         Ty,
 695         ty,
 696         type_alias_impl_trait,
 697         TyCtxt,
 698         TyKind,
 699         type_alias_enum_variants,
 700         type_ascription,
 701         type_length_limit,
 702         type_macros,
 703         u128,
 704         u16,
 705         u32,
 706         u64,
 707         u8,
 708         unboxed_closures,
 709         underscore_const_names,
 710         underscore_imports,
 711         underscore_lifetimes,
 712         uniform_paths,
 713         uninitialized,
 714         universal_impl_trait,
 715         unmarked_api,
 716         unreachable_code,
 717         unrestricted_attribute_tokens,
 718         unsafe_no_drop_flag,
 719         unsized_locals,
 720         unsized_tuple_coercion,
 721         unstable,
 722         untagged_unions,
 723         unwind,
 724         unwind_attributes,
 725         unwrap_or,
 726         used,
 727         use_extern_macros,
 728         use_nested_groups,
 729         usize,
 730         v1,
 731         val,
 732         vec,
 733         Vec,
 734         vis,
 735         visible_private_types,
 736         volatile,
 737         warn,
 738         warn_directory_ownership,
 739         wasm_import_module,
 740         wasm_target_feature,
 741         while_let,
 742         windows,
 743         windows_subsystem,
 744         Yield,
 745         zeroed,
 746     }
 747 }
 748
 749 #[derive(Copy, Clone, Eq)]
 750 pub struct Ident {
 751     pub name: Symbol,
 752     pub span: Span,
 753 }
 754
 755 impl Ident {
 756     #[inline]
 757     /// Constructs a new identifier from a symbol and a span.
 758     pub const fn new(name: Symbol, span: Span) -> Ident {
 759         Ident { name, span }
 760     }
 761
 762     /// Constructs a new identifier with a dummy span.
 763     #[inline]
 764     pub const fn with_dummy_span(name: Symbol) -> Ident {
 765         Ident::new(name, DUMMY_SP)
 766     }
 767
 768     #[inline]
 769     pub fn invalid() -> Ident {
 770         Ident::with_dummy_span(kw::Invalid)
 771     }
 772
 773     /// Maps a string to an identifier with a dummy span.
 774     pub fn from_str(string: &str) -> Ident {
 775         Ident::with_dummy_span(Symbol::intern(string))
 776     }
 777
 778     /// Maps a string and a span to an identifier.
 779     pub fn from_str_and_span(string: &str, span: Span) -> Ident {
 780         Ident::new(Symbol::intern(string), span)
 781     }
 782
 783     /// Replaces `lo` and `hi` with those from `span`, but keep hygiene context.
 784     pub fn with_span_pos(self, span: Span) -> Ident {
 785         Ident::new(self.name, span.with_ctxt(self.span.ctxt()))
 786     }
 787
 788     pub fn without_first_quote(self) -> Ident {
 789         Ident::new(Symbol::intern(self.as_str().trim_start_matches('\'')), self.span)
 790     }
 791
 792     /// "Normalize" ident for use in comparisons using "item hygiene".
 793     /// Identifiers with same string value become same if they came from the same "modern" macro
 794     /// (e.g., `macro` item, but not `macro_rules` item) and stay different if they came from
 795     /// different "modern" macros.
 796     /// Technically, this operation strips all non-opaque marks from ident's syntactic context.
 797     pub fn modern(self) -> Ident {
 798         Ident::new(self.name, self.span.modern())
 799     }
 800
 801     /// "Normalize" ident for use in comparisons using "local variable hygiene".
 802     /// Identifiers with same string value become same if they came from the same non-transparent
 803     /// macro (e.g., `macro` or `macro_rules!` items) and stay different if they came from different
 804     /// non-transparent macros.
 805     /// Technically, this operation strips all transparent marks from ident's syntactic context.
 806     pub fn modern_and_legacy(self) -> Ident {
 807         Ident::new(self.name, self.span.modern_and_legacy())
 808     }
 809
 810     /// Convert the name to a `SymbolStr`. This is a slowish operation because
 811     /// it requires locking the symbol interner.
 812     pub fn as_str(self) -> SymbolStr {
 813         self.name.as_str()
 814     }
 815 }
 816
 817 impl PartialEq for Ident {
 818     fn eq(&self, rhs: &Self) -> bool {
 819         self.name == rhs.name && self.span.ctxt() == rhs.span.ctxt()
 820     }
 821 }
 822
 823 impl Hash for Ident {
 824     fn hash<H: Hasher>(&self, state: &mut H) {
 825         self.name.hash(state);
 826         self.span.ctxt().hash(state);
 827     }
 828 }
 829
 830 impl fmt::Debug for Ident {
 831     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
 832         write!(f, "{}{:?}", self.name, self.span.ctxt())
 833     }
 834 }
 835
 836 impl fmt::Display for Ident {
 837     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
 838         fmt::Display::fmt(&self.name, f)
 839     }
 840 }
 841
 842 impl UseSpecializedEncodable for Ident {
 843     fn default_encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
 844         s.emit_struct("Ident", 2, |s| {
 845             s.emit_struct_field("name", 0, |s| {
 846                 self.name.encode(s)
 847             })?;
 848             s.emit_struct_field("span", 1, |s| {
 849                 self.span.encode(s)
 850             })
 851         })
 852     }
 853 }
 854
 855 impl UseSpecializedDecodable for Ident {
 856     fn default_decode<D: Decoder>(d: &mut D) -> Result<Self, D::Error> {
 857         d.read_struct("Ident", 2, |d| {
 858             Ok(Ident {
 859                 name: d.read_struct_field("name", 0, Decodable::decode)?,
 860                 span: d.read_struct_field("span", 1, Decodable::decode)?,
 861             })
 862         })
 863     }
 864 }
 865
 866 /// An interned string.
 867 ///
 868 /// Internally, a `Symbol` is implemented as an index, and all operations
 869 /// (including hashing, equality, and ordering) operate on that index. The use
 870 /// of `rustc_index::newtype_index!` means that `Option<Symbol>` only takes up 4 bytes,
 871 /// because `rustc_index::newtype_index!` reserves the last 256 values for tagging purposes.
 872 ///
 873 /// Note that `Symbol` cannot directly be a `rustc_index::newtype_index!` because it
 874 /// implements `fmt::Debug`, `Encodable`, and `Decodable` in special ways.
 875 #[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
 876 pub struct Symbol(SymbolIndex);
 877
 878 rustc_index::newtype_index! {
 879     pub struct SymbolIndex { .. }
 880 }
 881
 882 impl Symbol {
 883     const fn new(n: u32) -> Self {
 884         Symbol(SymbolIndex::from_u32_const(n))
 885     }
 886
 887     /// Maps a string to its interned representation.
 888     pub fn intern(string: &str) -> Self {
 889         with_interner(|interner| interner.intern(string))
 890     }
 891
 892     /// Access the symbol's chars. This is a slowish operation because it
 893     /// requires locking the symbol interner.
 894     pub fn with<F: FnOnce(&str) -> R, R>(self, f: F) -> R {
 895         with_interner(|interner| {
 896             f(interner.get(self))
 897         })
 898     }
 899
 900     /// Convert to a `SymbolStr`. This is a slowish operation because it
 901     /// requires locking the symbol interner.
 902     pub fn as_str(self) -> SymbolStr {
 903         with_interner(|interner| unsafe {
 904             SymbolStr {
 905                 string: std::mem::transmute::<&str, &str>(interner.get(self))
 906             }
 907         })
 908     }
 909
 910     pub fn as_u32(self) -> u32 {
 911         self.0.as_u32()
 912     }
 913 }
 914
 915 impl fmt::Debug for Symbol {
 916     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
 917         self.with(|str| fmt::Debug::fmt(&str, f))
 918     }
 919 }
 920
 921 impl fmt::Display for Symbol {
 922     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
 923         self.with(|str| fmt::Display::fmt(&str, f))
 924     }
 925 }
 926
 927 impl Encodable for Symbol {
 928     fn encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
 929         self.with(|string| s.emit_str(string))
 930     }
 931 }
 932
 933 impl Decodable for Symbol {
 934     fn decode<D: Decoder>(d: &mut D) -> Result<Symbol, D::Error> {
 935         Ok(Symbol::intern(&d.read_str()?))
 936     }
 937 }
 938
 939 // The `&'static str`s in this type actually point into the arena.
 940 #[derive(Default)]
 941 pub struct Interner {
 942     arena: DroplessArena,
 943     names: FxHashMap<&'static str, Symbol>,
 944     strings: Vec<&'static str>,
 945 }
 946
 947 impl Interner {
 948     fn prefill(init: &[&'static str]) -> Self {
 949         Interner {
 950             strings: init.into(),
 951             names: init.iter().copied().zip((0..).map(Symbol::new)).collect(),
 952             ..Default::default()
 953         }
 954     }
 955
 956     pub fn intern(&mut self, string: &str) -> Symbol {
 957         if let Some(&name) = self.names.get(string) {
 958             return name;
 959         }
 960
 961         let name = Symbol::new(self.strings.len() as u32);
 962
 963         // `from_utf8_unchecked` is safe since we just allocated a `&str` which is known to be
 964         // UTF-8.
 965         let string: &str = unsafe {
 966             str::from_utf8_unchecked(self.arena.alloc_slice(string.as_bytes()))
 967         };
 968         // It is safe to extend the arena allocation to `'static` because we only access
 969         // these while the arena is still alive.
 970         let string: &'static str =  unsafe {
 971             &*(string as *const str)
 972         };
 973         self.strings.push(string);
 974         self.names.insert(string, name);
 975         name
 976     }
 977
 978     // Get the symbol as a string. `Symbol::as_str()` should be used in
 979     // preference to this function.
 980     pub fn get(&self, symbol: Symbol) -> &str {
 981         self.strings[symbol.0.as_usize()]
 982     }
 983 }
 984
 985 // This module has a very short name because it's used a lot.
 986 pub mod kw {
 987     use super::Symbol;
 988     keywords!();
 989 }
 990
 991 // This module has a very short name because it's used a lot.
 992 pub mod sym {
 993     use std::convert::TryInto;
 994     use super::Symbol;
 995
 996     symbols!();
 997
 998     // Get the symbol for an integer. The first few non-negative integers each
 999     // have a static symbol and therefore are fast.
1000     pub fn integer<N: TryInto<usize> + Copy + ToString>(n: N) -> Symbol {
1001         if let Result::Ok(idx) = n.try_into() {
1002             if let Option::Some(&sym) = digits_array.get(idx) {
1003                 return sym;
1004             }
1005         }
1006         Symbol::intern(&n.to_string())
1007     }
1008 }
1009
1010 impl Symbol {
1011     fn is_used_keyword_2018(self) -> bool {
1012         self >= kw::Async && self <= kw::Dyn
1013     }
1014
1015     fn is_unused_keyword_2018(self) -> bool {
1016         self == kw::Try
1017     }
1018
1019     /// Used for sanity checking rustdoc keyword sections.
1020     pub fn is_doc_keyword(self) -> bool {
1021         self <= kw::Union
1022     }
1023
1024     /// A keyword or reserved identifier that can be used as a path segment.
1025     pub fn is_path_segment_keyword(self) -> bool {
1026         self == kw::Super ||
1027         self == kw::SelfLower ||
1028         self == kw::SelfUpper ||
1029         self == kw::Crate ||
1030         self == kw::PathRoot ||
1031         self == kw::DollarCrate
1032     }
1033
1034     /// Returns `true` if the symbol is `true` or `false`.
1035     pub fn is_bool_lit(self) -> bool {
1036         self == kw::True || self == kw::False
1037     }
1038
1039     /// This symbol can be a raw identifier.
1040     pub fn can_be_raw(self) -> bool {
1041         self != kw::Invalid && self != kw::Underscore && !self.is_path_segment_keyword()
1042     }
1043 }
1044
1045 impl Ident {
1046     // Returns `true` for reserved identifiers used internally for elided lifetimes,
1047     // unnamed method parameters, crate root module, error recovery etc.
1048     pub fn is_special(self) -> bool {
1049         self.name <= kw::Underscore
1050     }
1051
1052     /// Returns `true` if the token is a keyword used in the language.
1053     pub fn is_used_keyword(self) -> bool {
1054         // Note: `span.edition()` is relatively expensive, don't call it unless necessary.
1055         self.name >= kw::As && self.name <= kw::While ||
1056         self.name.is_used_keyword_2018() && self.span.rust_2018()
1057     }
1058
1059     /// Returns `true` if the token is a keyword reserved for possible future use.
1060     pub fn is_unused_keyword(self) -> bool {
1061         // Note: `span.edition()` is relatively expensive, don't call it unless necessary.
1062         self.name >= kw::Abstract && self.name <= kw::Yield ||
1063         self.name.is_unused_keyword_2018() && self.span.rust_2018()
1064     }
1065
1066     /// Returns `true` if the token is either a special identifier or a keyword.
1067     pub fn is_reserved(self) -> bool {
1068         self.is_special() || self.is_used_keyword() || self.is_unused_keyword()
1069     }
1070
1071     /// A keyword or reserved identifier that can be used as a path segment.
1072     pub fn is_path_segment_keyword(self) -> bool {
1073         self.name.is_path_segment_keyword()
1074     }
1075
1076     /// We see this identifier in a normal identifier position, like variable name or a type.
1077     /// How was it written originally? Did it use the raw form? Let's try to guess.
1078     pub fn is_raw_guess(self) -> bool {
1079         self.name.can_be_raw() && self.is_reserved()
1080     }
1081 }
1082
1083 #[inline]
1084 fn with_interner<T, F: FnOnce(&mut Interner) -> T>(f: F) -> T {
1085     GLOBALS.with(|globals| f(&mut *globals.symbol_interner.lock()))
1086 }
1087
1088 /// An alternative to `Symbol`, useful when the chars within the symbol need to
1089 /// be accessed. It deliberately has limited functionality and should only be
1090 /// used for temporary values.
1091 ///
1092 /// Because the interner outlives any thread which uses this type, we can
1093 /// safely treat `string` which points to interner data, as an immortal string,
1094 /// as long as this type never crosses between threads.
1095 //
1096 // FIXME: ensure that the interner outlives any thread which uses `SymbolStr`,
1097 // by creating a new thread right after constructing the interner.
1098 #[derive(Clone, Eq, PartialOrd, Ord)]
1099 pub struct SymbolStr {
1100     string: &'static str,
1101 }
1102
1103 // This impl allows a `SymbolStr` to be directly equated with a `String` or
1104 // `&str`.
1105 impl<T: std::ops::Deref<Target = str>> std::cmp::PartialEq<T> for SymbolStr {
1106     fn eq(&self, other: &T) -> bool {
1107         self.string == other.deref()
1108     }
1109 }
1110
1111 impl !Send for SymbolStr {}
1112 impl !Sync for SymbolStr {}
1113
1114 /// This impl means that if `ss` is a `SymbolStr`:
1115 /// - `*ss` is a `str`;
1116 /// - `&*ss` is a `&str`;
1117 /// - `&ss as &str` is a `&str`, which means that `&ss` can be passed to a
1118 ///   function expecting a `&str`.
1119 impl std::ops::Deref for SymbolStr {
1120     type Target = str;
1121     #[inline]
1122     fn deref(&self) -> &str { self.string }
1123 }
1124
1125 impl fmt::Debug for SymbolStr {
1126     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1127         fmt::Debug::fmt(self.string, f)
1128     }
1129 }
1130
1131 impl fmt::Display for SymbolStr {
1132     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1133         fmt::Display::fmt(self.string, f)
1134     }
1135 }