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_data_structures::indexed_vec::Idx;
   8 use rustc_data_structures::newtype_index;
   9 use rustc_macros::symbols;
  10 use serialize::{Decodable, Decoder, Encodable, Encoder};
  11
  12 use std::cmp::{PartialEq, Ordering, PartialOrd, Ord};
  13 use std::fmt;
  14 use std::hash::{Hash, Hasher};
  15 use std::str;
  16
  17 use crate::hygiene::SyntaxContext;
  18 use crate::{Span, DUMMY_SP, GLOBALS};
  19
  20 symbols! {
  21     // After modifying this list adjust `is_special`, `is_used_keyword`/`is_unused_keyword`,
  22     // this should be rarely necessary though if the keywords are kept in alphabetic order.
  23     Keywords {
  24         // Special reserved identifiers used internally for elided lifetimes,
  25         // unnamed method parameters, crate root module, error recovery etc.
  26         Invalid:            "",
  27         PathRoot:           "{{root}}",
  28         DollarCrate:        "$crate",
  29         Underscore:         "_",
  30
  31         // Keywords that are used in stable Rust.
  32         As:                 "as",
  33         Break:              "break",
  34         Const:              "const",
  35         Continue:           "continue",
  36         Crate:              "crate",
  37         Else:               "else",
  38         Enum:               "enum",
  39         Extern:             "extern",
  40         False:              "false",
  41         Fn:                 "fn",
  42         For:                "for",
  43         If:                 "if",
  44         Impl:               "impl",
  45         In:                 "in",
  46         Let:                "let",
  47         Loop:               "loop",
  48         Match:              "match",
  49         Mod:                "mod",
  50         Move:               "move",
  51         Mut:                "mut",
  52         Pub:                "pub",
  53         Ref:                "ref",
  54         Return:             "return",
  55         SelfLower:          "self",
  56         SelfUpper:          "Self",
  57         Static:             "static",
  58         Struct:             "struct",
  59         Super:              "super",
  60         Trait:              "trait",
  61         True:               "true",
  62         Type:               "type",
  63         Unsafe:             "unsafe",
  64         Use:                "use",
  65         Where:              "where",
  66         While:              "while",
  67
  68         // Keywords that are used in unstable Rust or reserved for future use.
  69         Abstract:           "abstract",
  70         Become:             "become",
  71         Box:                "box",
  72         Do:                 "do",
  73         Final:              "final",
  74         Macro:              "macro",
  75         Override:           "override",
  76         Priv:               "priv",
  77         Typeof:             "typeof",
  78         Unsized:            "unsized",
  79         Virtual:            "virtual",
  80         Yield:              "yield",
  81
  82         // Edition-specific keywords that are used in stable Rust.
  83         Dyn:                "dyn", // >= 2018 Edition only
  84
  85         // Edition-specific keywords that are used in unstable Rust or reserved for future use.
  86         Async:              "async", // >= 2018 Edition only
  87         Await:              "await", // >= 2018 Edition only
  88         Try:                "try", // >= 2018 Edition only
  89
  90         // Special lifetime names
  91         UnderscoreLifetime: "'_",
  92         StaticLifetime:     "'static",
  93
  94         // Weak keywords, have special meaning only in specific contexts.
  95         Auto:               "auto",
  96         Catch:              "catch",
  97         Default:            "default",
  98         Existential:        "existential",
  99         Union:              "union",
 100     }
 101
 102     // Symbols that can be referred to with syntax_pos::sym::*. The symbol is
 103     // the stringified identifier unless otherwise specified (e.g.
 104     // `proc_dash_macro` represents "proc-macro").
 105     //
 106     // As well as the symbols listed, there are symbols for the the strings
 107     // "0", "1", ..., "9", which are accessible via `sym::integer`.
 108     Symbols {
 109         aarch64_target_feature,
 110         abi,
 111         abi_amdgpu_kernel,
 112         abi_msp430_interrupt,
 113         abi_ptx,
 114         abi_sysv64,
 115         abi_thiscall,
 116         abi_unadjusted,
 117         abi_vectorcall,
 118         abi_x86_interrupt,
 119         aborts,
 120         advanced_slice_patterns,
 121         adx_target_feature,
 122         alias,
 123         align,
 124         alignstack,
 125         all,
 126         allocator,
 127         allocator_internals,
 128         alloc_error_handler,
 129         allow,
 130         allowed,
 131         allow_fail,
 132         allow_internal_unsafe,
 133         allow_internal_unstable,
 134         allow_internal_unstable_backcompat_hack,
 135         always,
 136         and,
 137         any,
 138         arbitrary_enum_discriminant,
 139         arbitrary_self_types,
 140         Arguments,
 141         ArgumentV1,
 142         arm_target_feature,
 143         asm,
 144         associated_consts,
 145         associated_type_bounds,
 146         associated_type_defaults,
 147         associated_types,
 148         async_await,
 149         attr,
 150         attributes,
 151         attr_literals,
 152         augmented_assignments,
 153         automatically_derived,
 154         avx512_target_feature,
 155         await_macro,
 156         begin_panic,
 157         bench,
 158         bin,
 159         bind_by_move_pattern_guards,
 160         block,
 161         bool,
 162         borrowck_graphviz_postflow,
 163         borrowck_graphviz_preflow,
 164         box_patterns,
 165         box_syntax,
 166         braced_empty_structs,
 167         C,
 168         cdylib,
 169         cfg,
 170         cfg_attr,
 171         cfg_attr_multi,
 172         cfg_target_feature,
 173         cfg_target_has_atomic,
 174         cfg_target_thread_local,
 175         cfg_target_vendor,
 176         char,
 177         clone,
 178         Clone,
 179         clone_closures,
 180         clone_from,
 181         closure_to_fn_coercion,
 182         cmp,
 183         cmpxchg16b_target_feature,
 184         cold,
 185         compile_error,
 186         compiler_builtins,
 187         concat_idents,
 188         conservative_impl_trait,
 189         console,
 190         const_compare_raw_pointers,
 191         const_constructor,
 192         const_fn,
 193         const_fn_union,
 194         const_generics,
 195         const_indexing,
 196         const_let,
 197         const_panic,
 198         const_raw_ptr_deref,
 199         const_raw_ptr_to_usize_cast,
 200         const_transmute,
 201         contents,
 202         context,
 203         convert,
 204         copy_closures,
 205         core,
 206         core_intrinsics,
 207         crate_id,
 208         crate_in_paths,
 209         crate_local,
 210         crate_name,
 211         crate_type,
 212         crate_visibility_modifier,
 213         custom_attribute,
 214         custom_derive,
 215         custom_inner_attributes,
 216         custom_test_frameworks,
 217         c_variadic,
 218         decl_macro,
 219         Default,
 220         default_lib_allocator,
 221         default_type_parameter_fallback,
 222         default_type_params,
 223         deny,
 224         deprecated,
 225         deref,
 226         deref_mut,
 227         derive,
 228         direct,
 229         doc,
 230         doc_alias,
 231         doc_cfg,
 232         doc_keyword,
 233         doc_masked,
 234         doc_spotlight,
 235         document_private_items,
 236         dotdoteq_in_patterns,
 237         dotdot_in_tuple_patterns,
 238         double_braced_crate: "{{crate}}",
 239         double_braced_impl: "{{impl}}",
 240         double_braced_misc: "{{misc}}",
 241         double_braced_closure: "{{closure}}",
 242         double_braced_constructor: "{{constructor}}",
 243         double_braced_constant: "{{constant}}",
 244         double_braced_opaque: "{{opaque}}",
 245         dropck_eyepatch,
 246         dropck_parametricity,
 247         drop_types_in_const,
 248         dylib,
 249         dyn_trait,
 250         eh_personality,
 251         eh_unwind_resume,
 252         enable,
 253         eq,
 254         err,
 255         Err,
 256         Equal,
 257         except,
 258         exclusive_range_pattern,
 259         exhaustive_integer_patterns,
 260         exhaustive_patterns,
 261         existential_type,
 262         expected,
 263         export_name,
 264         expr,
 265         extern_absolute_paths,
 266         external_doc,
 267         extern_crate_item_prelude,
 268         extern_crate_self,
 269         extern_in_paths,
 270         extern_prelude,
 271         extern_types,
 272         f16c_target_feature,
 273         f32,
 274         f64,
 275         feature,
 276         ffi_returns_twice,
 277         field,
 278         field_init_shorthand,
 279         file,
 280         fmt,
 281         fmt_internals,
 282         fn_must_use,
 283         forbid,
 284         format_args_nl,
 285         from,
 286         From,
 287         from_desugaring,
 288         from_error,
 289         from_generator,
 290         from_method,
 291         from_ok,
 292         from_usize,
 293         fundamental,
 294         future,
 295         Future,
 296         FxHashSet,
 297         FxHashMap,
 298         gen_future,
 299         generators,
 300         generic_associated_types,
 301         generic_param_attrs,
 302         global_allocator,
 303         global_asm,
 304         globs,
 305         hash,
 306         Hash,
 307         HashSet,
 308         HashMap,
 309         hexagon_target_feature,
 310         hidden,
 311         homogeneous_aggregate,
 312         html_favicon_url,
 313         html_logo_url,
 314         html_no_source,
 315         html_playground_url,
 316         html_root_url,
 317         i128,
 318         i128_type,
 319         i16,
 320         i32,
 321         i64,
 322         i8,
 323         ident,
 324         if_let,
 325         if_while_or_patterns,
 326         ignore,
 327         impl_header_lifetime_elision,
 328         impl_trait_in_bindings,
 329         import_shadowing,
 330         index,
 331         index_mut,
 332         in_band_lifetimes,
 333         include,
 334         inclusive_range_syntax,
 335         infer_outlives_requirements,
 336         infer_static_outlives_requirements,
 337         inline,
 338         intel,
 339         into_iter,
 340         IntoIterator,
 341         into_result,
 342         intrinsics,
 343         irrefutable_let_patterns,
 344         isize,
 345         issue,
 346         issue_5723_bootstrap,
 347         issue_tracker_base_url,
 348         item,
 349         item_like_imports,
 350         iter,
 351         Iterator,
 352         keyword,
 353         kind,
 354         label,
 355         label_break_value,
 356         lang,
 357         lang_items,
 358         let_chains,
 359         lhs,
 360         lib,
 361         lifetime,
 362         link,
 363         linkage,
 364         link_args,
 365         link_cfg,
 366         link_llvm_intrinsics,
 367         link_name,
 368         link_section,
 369         lint_reasons,
 370         literal,
 371         local_inner_macros,
 372         log_syntax,
 373         loop_break_value,
 374         macro_at_most_once_rep,
 375         macro_escape,
 376         macro_export,
 377         macro_lifetime_matcher,
 378         macro_literal_matcher,
 379         macro_reexport,
 380         macro_rules,
 381         macros_in_extern,
 382         macro_use,
 383         macro_vis_matcher,
 384         main,
 385         managed_boxes,
 386         marker,
 387         marker_trait_attr,
 388         masked,
 389         match_beginning_vert,
 390         match_default_bindings,
 391         may_dangle,
 392         member_constraints,
 393         message,
 394         meta,
 395         min_const_fn,
 396         min_const_unsafe_fn,
 397         mips_target_feature,
 398         mmx_target_feature,
 399         module,
 400         more_struct_aliases,
 401         movbe_target_feature,
 402         must_use,
 403         naked,
 404         naked_functions,
 405         name,
 406         needs_allocator,
 407         needs_panic_runtime,
 408         negate_unsigned,
 409         never,
 410         never_type,
 411         new,
 412         next,
 413         __next,
 414         nll,
 415         no_builtins,
 416         no_core,
 417         no_crate_inject,
 418         no_debug,
 419         no_default_passes,
 420         no_implicit_prelude,
 421         no_inline,
 422         no_link,
 423         no_main,
 424         no_mangle,
 425         non_ascii_idents,
 426         None,
 427         non_exhaustive,
 428         non_modrs_mods,
 429         no_stack_check,
 430         no_start,
 431         no_std,
 432         not,
 433         note,
 434         Ok,
 435         omit_gdb_pretty_printer_section,
 436         on,
 437         on_unimplemented,
 438         oom,
 439         ops,
 440         optimize,
 441         optimize_attribute,
 442         optin_builtin_traits,
 443         option,
 444         Option,
 445         opt_out_copy,
 446         or,
 447         Ord,
 448         Ordering,
 449         Output,
 450         overlapping_marker_traits,
 451         packed,
 452         panic,
 453         panic_handler,
 454         panic_impl,
 455         panic_implementation,
 456         panic_runtime,
 457         parent_trait,
 458         partial_cmp,
 459         param_attrs,
 460         PartialOrd,
 461         passes,
 462         pat,
 463         path,
 464         pattern_parentheses,
 465         Pending,
 466         pin,
 467         Pin,
 468         pinned,
 469         platform_intrinsics,
 470         plugin,
 471         plugin_registrar,
 472         plugins,
 473         Poll,
 474         poll_with_tls_context,
 475         powerpc_target_feature,
 476         precise_pointer_size_matching,
 477         prelude,
 478         prelude_import,
 479         primitive,
 480         proc_dash_macro: "proc-macro",
 481         proc_macro,
 482         proc_macro_attribute,
 483         proc_macro_def_site,
 484         proc_macro_derive,
 485         proc_macro_expr,
 486         proc_macro_gen,
 487         proc_macro_hygiene,
 488         proc_macro_mod,
 489         proc_macro_non_items,
 490         proc_macro_path_invoc,
 491         profiler_runtime,
 492         pub_restricted,
 493         pushpop_unsafe,
 494         quad_precision_float,
 495         question_mark,
 496         quote,
 497         Range,
 498         RangeFrom,
 499         RangeFull,
 500         RangeInclusive,
 501         RangeTo,
 502         RangeToInclusive,
 503         raw_identifiers,
 504         Ready,
 505         reason,
 506         recursion_limit,
 507         reexport_test_harness_main,
 508         reflect,
 509         relaxed_adts,
 510         repr,
 511         repr128,
 512         repr_align,
 513         repr_align_enum,
 514         repr_packed,
 515         repr_simd,
 516         repr_transparent,
 517         re_rebalance_coherence,
 518         result,
 519         Result,
 520         Return,
 521         rhs,
 522         rlib,
 523         rt,
 524         rtm_target_feature,
 525         rust,
 526         rust_2015_preview,
 527         rust_2018_preview,
 528         rust_begin_unwind,
 529         rustc,
 530         rustc_allocator,
 531         rustc_allocator_nounwind,
 532         rustc_allow_const_fn_ptr,
 533         rustc_args_required_const,
 534         rustc_attrs,
 535         rustc_clean,
 536         rustc_const_unstable,
 537         rustc_conversion_suggestion,
 538         rustc_copy_clone_marker,
 539         rustc_def_path,
 540         rustc_deprecated,
 541         rustc_diagnostic_macros,
 542         rustc_dirty,
 543         rustc_doc_only_macro,
 544         rustc_dummy,
 545         rustc_dump_env_program_clauses,
 546         rustc_dump_program_clauses,
 547         rustc_dump_user_substs,
 548         rustc_error,
 549         rustc_expected_cgu_reuse,
 550         rustc_if_this_changed,
 551         rustc_inherit_overflow_checks,
 552         rustc_layout,
 553         rustc_layout_scalar_valid_range_end,
 554         rustc_layout_scalar_valid_range_start,
 555         rustc_mir,
 556         rustc_nonnull_optimization_guaranteed,
 557         rustc_object_lifetime_default,
 558         rustc_on_unimplemented,
 559         rustc_outlives,
 560         rustc_paren_sugar,
 561         rustc_partition_codegened,
 562         rustc_partition_reused,
 563         rustc_peek,
 564         rustc_peek_definite_init,
 565         rustc_peek_maybe_init,
 566         rustc_peek_maybe_uninit,
 567         rustc_private,
 568         rustc_proc_macro_decls,
 569         rustc_promotable,
 570         rustc_regions,
 571         rustc_stable,
 572         rustc_std_internal_symbol,
 573         rustc_symbol_name,
 574         rustc_synthetic,
 575         rustc_test_marker,
 576         rustc_then_this_would_need,
 577         rustc_transparent_macro,
 578         rustc_variance,
 579         rustdoc,
 580         rust_eh_personality,
 581         rust_eh_unwind_resume,
 582         rust_oom,
 583         __rust_unstable_column,
 584         rvalue_static_promotion,
 585         sanitizer_runtime,
 586         _Self,
 587         self_in_typedefs,
 588         self_struct_ctor,
 589         Send,
 590         should_panic,
 591         simd,
 592         simd_ffi,
 593         since,
 594         size,
 595         slice_patterns,
 596         slicing_syntax,
 597         Some,
 598         specialization,
 599         speed,
 600         spotlight,
 601         sse4a_target_feature,
 602         stable,
 603         staged_api,
 604         start,
 605         static_in_const,
 606         staticlib,
 607         static_nobundle,
 608         static_recursion,
 609         std,
 610         str,
 611         stmt,
 612         stmt_expr_attributes,
 613         stop_after_dataflow,
 614         struct_field_attributes,
 615         struct_inherit,
 616         structural_match,
 617         struct_variant,
 618         sty,
 619         suggestion,
 620         target_feature,
 621         target_has_atomic,
 622         target_thread_local,
 623         task,
 624         tbm_target_feature,
 625         termination_trait,
 626         termination_trait_test,
 627         test,
 628         test_2018_feature,
 629         test_accepted_feature,
 630         test_case,
 631         test_removed_feature,
 632         test_runner,
 633         then_with,
 634         thread_local,
 635         tool_attributes,
 636         tool_lints,
 637         trace_macros,
 638         trait_alias,
 639         transmute,
 640         transparent,
 641         transparent_enums,
 642         transparent_unions,
 643         trivial_bounds,
 644         Try,
 645         try_blocks,
 646         try_trait,
 647         tt,
 648         tuple_indexing,
 649         Ty,
 650         ty,
 651         TyCtxt,
 652         TyKind,
 653         type_alias_enum_variants,
 654         type_ascription,
 655         type_length_limit,
 656         type_macros,
 657         u128,
 658         u16,
 659         u32,
 660         u64,
 661         u8,
 662         unboxed_closures,
 663         underscore_const_names,
 664         underscore_imports,
 665         underscore_lifetimes,
 666         uniform_paths,
 667         universal_impl_trait,
 668         unmarked_api,
 669         unreachable_code,
 670         unrestricted_attribute_tokens,
 671         unsafe_destructor_blind_to_params,
 672         unsafe_no_drop_flag,
 673         unsized_locals,
 674         unsized_tuple_coercion,
 675         unstable,
 676         untagged_unions,
 677         unwind,
 678         unwind_attributes,
 679         unwrap_or,
 680         used,
 681         use_extern_macros,
 682         use_nested_groups,
 683         usize,
 684         v1,
 685         val,
 686         vec,
 687         Vec,
 688         vis,
 689         visible_private_types,
 690         volatile,
 691         warn,
 692         warn_directory_ownership,
 693         wasm_import_module,
 694         wasm_target_feature,
 695         while_let,
 696         windows,
 697         windows_subsystem,
 698         Yield,
 699     }
 700 }
 701
 702 #[derive(Copy, Clone, Eq)]
 703 pub struct Ident {
 704     pub name: Symbol,
 705     pub span: Span,
 706 }
 707
 708 impl Ident {
 709     #[inline]
 710     /// Constructs a new identifier from a symbol and a span.
 711     pub const fn new(name: Symbol, span: Span) -> Ident {
 712         Ident { name, span }
 713     }
 714
 715     /// Constructs a new identifier with an empty syntax context.
 716     #[inline]
 717     pub const fn with_empty_ctxt(name: Symbol) -> Ident {
 718         Ident::new(name, DUMMY_SP)
 719     }
 720
 721     #[inline]
 722     pub fn invalid() -> Ident {
 723         Ident::with_empty_ctxt(kw::Invalid)
 724     }
 725
 726     /// Maps an interned string to an identifier with an empty syntax context.
 727     pub fn from_interned_str(string: InternedString) -> Ident {
 728         Ident::with_empty_ctxt(string.as_symbol())
 729     }
 730
 731     /// Maps a string to an identifier with an empty span.
 732     pub fn from_str(string: &str) -> Ident {
 733         Ident::with_empty_ctxt(Symbol::intern(string))
 734     }
 735
 736     /// Maps a string and a span to an identifier.
 737     pub fn from_str_and_span(string: &str, span: Span) -> Ident {
 738         Ident::new(Symbol::intern(string), span)
 739     }
 740
 741     /// Replaces `lo` and `hi` with those from `span`, but keep hygiene context.
 742     pub fn with_span_pos(self, span: Span) -> Ident {
 743         Ident::new(self.name, span.with_ctxt(self.span.ctxt()))
 744     }
 745
 746     pub fn without_first_quote(self) -> Ident {
 747         Ident::new(Symbol::intern(self.as_str().trim_start_matches('\'')), self.span)
 748     }
 749
 750     /// "Normalize" ident for use in comparisons using "item hygiene".
 751     /// Identifiers with same string value become same if they came from the same "modern" macro
 752     /// (e.g., `macro` item, but not `macro_rules` item) and stay different if they came from
 753     /// different "modern" macros.
 754     /// Technically, this operation strips all non-opaque marks from ident's syntactic context.
 755     pub fn modern(self) -> Ident {
 756         Ident::new(self.name, self.span.modern())
 757     }
 758
 759     /// "Normalize" ident for use in comparisons using "local variable hygiene".
 760     /// Identifiers with same string value become same if they came from the same non-transparent
 761     /// macro (e.g., `macro` or `macro_rules!` items) and stay different if they came from different
 762     /// non-transparent macros.
 763     /// Technically, this operation strips all transparent marks from ident's syntactic context.
 764     pub fn modern_and_legacy(self) -> Ident {
 765         Ident::new(self.name, self.span.modern_and_legacy())
 766     }
 767
 768     /// Transforms an identifier into one with the same name, but gensymed.
 769     pub fn gensym(self) -> Ident {
 770         let name = with_interner(|interner| interner.gensymed(self.name));
 771         Ident::new(name, self.span)
 772     }
 773
 774     /// Transforms an underscore identifier into one with the same name, but
 775     /// gensymed. Leaves non-underscore identifiers unchanged.
 776     pub fn gensym_if_underscore(self) -> Ident {
 777         if self.name == kw::Underscore { self.gensym() } else { self }
 778     }
 779
 780     // WARNING: this function is deprecated and will be removed in the future.
 781     pub fn is_gensymed(self) -> bool {
 782         with_interner(|interner| interner.is_gensymed(self.name))
 783     }
 784
 785     pub fn as_str(self) -> LocalInternedString {
 786         self.name.as_str()
 787     }
 788
 789     pub fn as_interned_str(self) -> InternedString {
 790         self.name.as_interned_str()
 791     }
 792 }
 793
 794 impl PartialEq for Ident {
 795     fn eq(&self, rhs: &Self) -> bool {
 796         self.name == rhs.name && self.span.ctxt() == rhs.span.ctxt()
 797     }
 798 }
 799
 800 impl Hash for Ident {
 801     fn hash<H: Hasher>(&self, state: &mut H) {
 802         self.name.hash(state);
 803         self.span.ctxt().hash(state);
 804     }
 805 }
 806
 807 impl fmt::Debug for Ident {
 808     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
 809         write!(f, "{}{:?}", self.name, self.span.ctxt())
 810     }
 811 }
 812
 813 impl fmt::Display for Ident {
 814     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
 815         fmt::Display::fmt(&self.name, f)
 816     }
 817 }
 818
 819 impl Encodable for Ident {
 820     fn encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
 821         if self.span.ctxt().modern() == SyntaxContext::empty() {
 822             s.emit_str(&self.as_str())
 823         } else { // FIXME(jseyfried): intercrate hygiene
 824             let mut string = "#".to_owned();
 825             string.push_str(&self.as_str());
 826             s.emit_str(&string)
 827         }
 828     }
 829 }
 830
 831 impl Decodable for Ident {
 832     fn decode<D: Decoder>(d: &mut D) -> Result<Ident, D::Error> {
 833         let string = d.read_str()?;
 834         Ok(if !string.starts_with('#') {
 835             Ident::from_str(&string)
 836         } else { // FIXME(jseyfried): intercrate hygiene
 837             Ident::from_str(&string[1..]).gensym()
 838         })
 839     }
 840 }
 841
 842 /// A symbol is an interned or gensymed string. A gensym is a symbol that is
 843 /// never equal to any other symbol.
 844 ///
 845 /// Conceptually, a gensym can be thought of as a normal symbol with an
 846 /// invisible unique suffix. Gensyms are useful when creating new identifiers
 847 /// that must not match any existing identifiers, e.g. during macro expansion
 848 /// and syntax desugaring. Because gensyms should always be identifiers, all
 849 /// gensym operations are on `Ident` rather than `Symbol`. (Indeed, in the
 850 /// future the gensym-ness may be moved from `Symbol` to hygiene data.)
 851 ///
 852 /// Examples:
 853 /// ```
 854 /// assert_eq!(Ident::from_str("x"), Ident::from_str("x"))
 855 /// assert_ne!(Ident::from_str("x").gensym(), Ident::from_str("x"))
 856 /// assert_ne!(Ident::from_str("x").gensym(), Ident::from_str("x").gensym())
 857 /// ```
 858 /// Internally, a symbol is implemented as an index, and all operations
 859 /// (including hashing, equality, and ordering) operate on that index. The use
 860 /// of `newtype_index!` means that `Option<Symbol>` only takes up 4 bytes,
 861 /// because `newtype_index!` reserves the last 256 values for tagging purposes.
 862 ///
 863 /// Note that `Symbol` cannot directly be a `newtype_index!` because it
 864 /// implements `fmt::Debug`, `Encodable`, and `Decodable` in special ways.
 865 #[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
 866 pub struct Symbol(SymbolIndex);
 867
 868 newtype_index! {
 869     pub struct SymbolIndex { .. }
 870 }
 871
 872 impl Symbol {
 873     const fn new(n: u32) -> Self {
 874         Symbol(SymbolIndex::from_u32_const(n))
 875     }
 876
 877     /// Maps a string to its interned representation.
 878     pub fn intern(string: &str) -> Self {
 879         with_interner(|interner| interner.intern(string))
 880     }
 881
 882     pub fn as_str(self) -> LocalInternedString {
 883         with_interner(|interner| unsafe {
 884             LocalInternedString {
 885                 string: std::mem::transmute::<&str, &str>(interner.get(self))
 886             }
 887         })
 888     }
 889
 890     pub fn as_interned_str(self) -> InternedString {
 891         with_interner(|interner| InternedString {
 892             symbol: interner.interned(self)
 893         })
 894     }
 895
 896     pub fn as_u32(self) -> u32 {
 897         self.0.as_u32()
 898     }
 899 }
 900
 901 impl fmt::Debug for Symbol {
 902     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
 903         let is_gensymed = with_interner(|interner| interner.is_gensymed(*self));
 904         if is_gensymed {
 905             write!(f, "{}({:?})", self, self.0)
 906         } else {
 907             write!(f, "{}", self)
 908         }
 909     }
 910 }
 911
 912 impl fmt::Display for Symbol {
 913     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
 914         fmt::Display::fmt(&self.as_str(), f)
 915     }
 916 }
 917
 918 impl Encodable for Symbol {
 919     fn encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
 920         s.emit_str(&self.as_str())
 921     }
 922 }
 923
 924 impl Decodable for Symbol {
 925     fn decode<D: Decoder>(d: &mut D) -> Result<Symbol, D::Error> {
 926         Ok(Symbol::intern(&d.read_str()?))
 927     }
 928 }
 929
 930 // The `&'static str`s in this type actually point into the arena.
 931 //
 932 // Note that normal symbols are indexed upward from 0, and gensyms are indexed
 933 // downward from SymbolIndex::MAX_AS_U32.
 934 #[derive(Default)]
 935 pub struct Interner {
 936     arena: DroplessArena,
 937     names: FxHashMap<&'static str, Symbol>,
 938     strings: Vec<&'static str>,
 939     gensyms: Vec<Symbol>,
 940 }
 941
 942 impl Interner {
 943     fn prefill(init: &[&'static str]) -> Self {
 944         Interner {
 945             strings: init.into(),
 946             names: init.iter().copied().zip((0..).map(Symbol::new)).collect(),
 947             ..Default::default()
 948         }
 949     }
 950
 951     pub fn intern(&mut self, string: &str) -> Symbol {
 952         if let Some(&name) = self.names.get(string) {
 953             return name;
 954         }
 955
 956         let name = Symbol::new(self.strings.len() as u32);
 957
 958         // `from_utf8_unchecked` is safe since we just allocated a `&str` which is known to be
 959         // UTF-8.
 960         let string: &str = unsafe {
 961             str::from_utf8_unchecked(self.arena.alloc_slice(string.as_bytes()))
 962         };
 963         // It is safe to extend the arena allocation to `'static` because we only access
 964         // these while the arena is still alive.
 965         let string: &'static str =  unsafe {
 966             &*(string as *const str)
 967         };
 968         self.strings.push(string);
 969         self.names.insert(string, name);
 970         name
 971     }
 972
 973     fn interned(&self, symbol: Symbol) -> Symbol {
 974         if (symbol.0.as_usize()) < self.strings.len() {
 975             symbol
 976         } else {
 977             self.gensyms[(SymbolIndex::MAX_AS_U32 - symbol.0.as_u32()) as usize]
 978         }
 979     }
 980
 981     fn gensymed(&mut self, symbol: Symbol) -> Symbol {
 982         self.gensyms.push(symbol);
 983         Symbol::new(SymbolIndex::MAX_AS_U32 - self.gensyms.len() as u32 + 1)
 984     }
 985
 986     fn is_gensymed(&mut self, symbol: Symbol) -> bool {
 987         symbol.0.as_usize() >= self.strings.len()
 988     }
 989
 990     // Get the symbol as a string. `Symbol::as_str()` should be used in
 991     // preference to this function.
 992     pub fn get(&self, symbol: Symbol) -> &str {
 993         match self.strings.get(symbol.0.as_usize()) {
 994             Some(string) => string,
 995             None => {
 996                 let symbol = self.gensyms[(SymbolIndex::MAX_AS_U32 - symbol.0.as_u32()) as usize];
 997                 self.strings[symbol.0.as_usize()]
 998             }
 999         }
1000     }
1001 }
1002
1003 // This module has a very short name because it's used a lot.
1004 pub mod kw {
1005     use super::Symbol;
1006     keywords!();
1007 }
1008
1009 // This module has a very short name because it's used a lot.
1010 pub mod sym {
1011     use std::convert::TryInto;
1012     use super::Symbol;
1013
1014     symbols!();
1015
1016     // Get the symbol for an integer. The first few non-negative integers each
1017     // have a static symbol and therefore are fast.
1018     pub fn integer<N: TryInto<usize> + Copy + ToString>(n: N) -> Symbol {
1019         if let Result::Ok(idx) = n.try_into() {
1020             if let Option::Some(&sym) = digits_array.get(idx) {
1021                 return sym;
1022             }
1023         }
1024         Symbol::intern(&n.to_string())
1025     }
1026 }
1027
1028 impl Symbol {
1029     fn is_used_keyword_2018(self) -> bool {
1030         self == kw::Dyn
1031     }
1032
1033     fn is_unused_keyword_2018(self) -> bool {
1034         self >= kw::Async && self <= kw::Try
1035     }
1036
1037     /// Used for sanity checking rustdoc keyword sections.
1038     pub fn is_doc_keyword(self) -> bool {
1039         self <= kw::Union
1040     }
1041
1042     /// A keyword or reserved identifier that can be used as a path segment.
1043     pub fn is_path_segment_keyword(self) -> bool {
1044         self == kw::Super ||
1045         self == kw::SelfLower ||
1046         self == kw::SelfUpper ||
1047         self == kw::Crate ||
1048         self == kw::PathRoot ||
1049         self == kw::DollarCrate
1050     }
1051
1052     /// This symbol can be a raw identifier.
1053     pub fn can_be_raw(self) -> bool {
1054         self != kw::Invalid && self != kw::Underscore && !self.is_path_segment_keyword()
1055     }
1056 }
1057
1058 impl Ident {
1059     // Returns `true` for reserved identifiers used internally for elided lifetimes,
1060     // unnamed method parameters, crate root module, error recovery etc.
1061     pub fn is_special(self) -> bool {
1062         self.name <= kw::Underscore
1063     }
1064
1065     /// Returns `true` if the token is a keyword used in the language.
1066     pub fn is_used_keyword(self) -> bool {
1067         // Note: `span.edition()` is relatively expensive, don't call it unless necessary.
1068         self.name >= kw::As && self.name <= kw::While ||
1069         self.name.is_used_keyword_2018() && self.span.rust_2018()
1070     }
1071
1072     /// Returns `true` if the token is a keyword reserved for possible future use.
1073     pub fn is_unused_keyword(self) -> bool {
1074         // Note: `span.edition()` is relatively expensive, don't call it unless necessary.
1075         self.name >= kw::Abstract && self.name <= kw::Yield ||
1076         self.name.is_unused_keyword_2018() && self.span.rust_2018()
1077     }
1078
1079     /// Returns `true` if the token is either a special identifier or a keyword.
1080     pub fn is_reserved(self) -> bool {
1081         self.is_special() || self.is_used_keyword() || self.is_unused_keyword()
1082     }
1083
1084     /// A keyword or reserved identifier that can be used as a path segment.
1085     pub fn is_path_segment_keyword(self) -> bool {
1086         self.name.is_path_segment_keyword()
1087     }
1088
1089     /// We see this identifier in a normal identifier position, like variable name or a type.
1090     /// How was it written originally? Did it use the raw form? Let's try to guess.
1091     pub fn is_raw_guess(self) -> bool {
1092         self.name.can_be_raw() && self.is_reserved()
1093     }
1094 }
1095
1096 // If an interner exists, return it. Otherwise, prepare a fresh one.
1097 #[inline]
1098 fn with_interner<T, F: FnOnce(&mut Interner) -> T>(f: F) -> T {
1099     GLOBALS.with(|globals| f(&mut *globals.symbol_interner.lock()))
1100 }
1101
1102 /// An alternative to `Symbol` and `InternedString`, useful when the chars
1103 /// within the symbol need to be accessed. It is best used for temporary
1104 /// values.
1105 ///
1106 /// Because the interner outlives any thread which uses this type, we can
1107 /// safely treat `string` which points to interner data, as an immortal string,
1108 /// as long as this type never crosses between threads.
1109 //
1110 // FIXME: ensure that the interner outlives any thread which uses
1111 // `LocalInternedString`, by creating a new thread right after constructing the
1112 // interner.
1113 #[derive(Clone, Copy, Hash, PartialOrd, Eq, Ord)]
1114 pub struct LocalInternedString {
1115     string: &'static str,
1116 }
1117
1118 impl LocalInternedString {
1119     /// Maps a string to its interned representation.
1120     pub fn intern(string: &str) -> Self {
1121         let string = with_interner(|interner| {
1122             let symbol = interner.intern(string);
1123             interner.strings[symbol.0.as_usize()]
1124         });
1125         LocalInternedString {
1126             string: unsafe { std::mem::transmute::<&str, &str>(string) }
1127         }
1128     }
1129
1130     pub fn as_interned_str(self) -> InternedString {
1131         InternedString {
1132             symbol: Symbol::intern(self.string)
1133         }
1134     }
1135
1136     #[inline]
1137     pub fn get(&self) -> &str {
1138         // This returns a valid string since we ensure that `self` outlives the interner
1139         // by creating the interner on a thread which outlives threads which can access it.
1140         // This type cannot move to a thread which outlives the interner since it does
1141         // not implement Send.
1142         self.string
1143     }
1144 }
1145
1146 impl<U: ?Sized> std::convert::AsRef<U> for LocalInternedString
1147 where
1148     str: std::convert::AsRef<U>
1149 {
1150     #[inline]
1151     fn as_ref(&self) -> &U {
1152         self.string.as_ref()
1153     }
1154 }
1155
1156 impl<T: std::ops::Deref<Target = str>> std::cmp::PartialEq<T> for LocalInternedString {
1157     fn eq(&self, other: &T) -> bool {
1158         self.string == other.deref()
1159     }
1160 }
1161
1162 impl std::cmp::PartialEq<LocalInternedString> for str {
1163     fn eq(&self, other: &LocalInternedString) -> bool {
1164         self == other.string
1165     }
1166 }
1167
1168 impl<'a> std::cmp::PartialEq<LocalInternedString> for &'a str {
1169     fn eq(&self, other: &LocalInternedString) -> bool {
1170         *self == other.string
1171     }
1172 }
1173
1174 impl std::cmp::PartialEq<LocalInternedString> for String {
1175     fn eq(&self, other: &LocalInternedString) -> bool {
1176         self == other.string
1177     }
1178 }
1179
1180 impl<'a> std::cmp::PartialEq<LocalInternedString> for &'a String {
1181     fn eq(&self, other: &LocalInternedString) -> bool {
1182         *self == other.string
1183     }
1184 }
1185
1186 impl !Send for LocalInternedString {}
1187 impl !Sync for LocalInternedString {}
1188
1189 impl std::ops::Deref for LocalInternedString {
1190     type Target = str;
1191     #[inline]
1192     fn deref(&self) -> &str { self.string }
1193 }
1194
1195 impl fmt::Debug for LocalInternedString {
1196     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1197         fmt::Debug::fmt(self.string, f)
1198     }
1199 }
1200
1201 impl fmt::Display for LocalInternedString {
1202     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1203         fmt::Display::fmt(self.string, f)
1204     }
1205 }
1206
1207 impl Decodable for LocalInternedString {
1208     fn decode<D: Decoder>(d: &mut D) -> Result<LocalInternedString, D::Error> {
1209         Ok(LocalInternedString::intern(&d.read_str()?))
1210     }
1211 }
1212
1213 impl Encodable for LocalInternedString {
1214     fn encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
1215         s.emit_str(self.string)
1216     }
1217 }
1218
1219 /// An alternative to `Symbol` that is focused on string contents. It has two
1220 /// main differences to `Symbol`.
1221 ///
1222 /// First, its implementations of `Hash`, `PartialOrd` and `Ord` work with the
1223 /// string chars rather than the symbol integer. This is useful when hash
1224 /// stability is required across compile sessions, or a guaranteed sort
1225 /// ordering is required.
1226 ///
1227 /// Second, gensym-ness is irrelevant. E.g.:
1228 /// ```
1229 /// assert_ne!(Symbol::gensym("x"), Symbol::gensym("x"))
1230 /// assert_eq!(Symbol::gensym("x").as_interned_str(), Symbol::gensym("x").as_interned_str())
1231 /// ```
1232 #[derive(Clone, Copy, PartialEq, Eq)]
1233 pub struct InternedString {
1234     symbol: Symbol,
1235 }
1236
1237 impl InternedString {
1238     /// Maps a string to its interned representation.
1239     pub fn intern(string: &str) -> Self {
1240         InternedString {
1241             symbol: Symbol::intern(string)
1242         }
1243     }
1244
1245     pub fn with<F: FnOnce(&str) -> R, R>(self, f: F) -> R {
1246         let str = with_interner(|interner| {
1247             interner.get(self.symbol) as *const str
1248         });
1249         // This is safe because the interner keeps string alive until it is dropped.
1250         // We can access it because we know the interner is still alive since we use a
1251         // scoped thread local to access it, and it was alive at the beginning of this scope
1252         unsafe { f(&*str) }
1253     }
1254
1255     fn with2<F: FnOnce(&str, &str) -> R, R>(self, other: &InternedString, f: F) -> R {
1256         let (self_str, other_str) = with_interner(|interner| {
1257             (interner.get(self.symbol) as *const str,
1258              interner.get(other.symbol) as *const str)
1259         });
1260         // This is safe for the same reason that `with` is safe.
1261         unsafe { f(&*self_str, &*other_str) }
1262     }
1263
1264     pub fn as_symbol(self) -> Symbol {
1265         self.symbol
1266     }
1267
1268     pub fn as_str(self) -> LocalInternedString {
1269         self.symbol.as_str()
1270     }
1271 }
1272
1273 impl Hash for InternedString {
1274     fn hash<H: Hasher>(&self, state: &mut H) {
1275         self.with(|str| str.hash(state))
1276     }
1277 }
1278
1279 impl PartialOrd<InternedString> for InternedString {
1280     fn partial_cmp(&self, other: &InternedString) -> Option<Ordering> {
1281         if self.symbol == other.symbol {
1282             return Some(Ordering::Equal);
1283         }
1284         self.with2(other, |self_str, other_str| self_str.partial_cmp(other_str))
1285     }
1286 }
1287
1288 impl Ord for InternedString {
1289     fn cmp(&self, other: &InternedString) -> Ordering {
1290         if self.symbol == other.symbol {
1291             return Ordering::Equal;
1292         }
1293         self.with2(other, |self_str, other_str| self_str.cmp(other_str))
1294     }
1295 }
1296
1297 impl fmt::Debug for InternedString {
1298     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1299         self.with(|str| fmt::Debug::fmt(&str, f))
1300     }
1301 }
1302
1303 impl fmt::Display for InternedString {
1304     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1305         self.with(|str| fmt::Display::fmt(&str, f))
1306     }
1307 }
1308
1309 impl Decodable for InternedString {
1310     fn decode<D: Decoder>(d: &mut D) -> Result<InternedString, D::Error> {
1311         Ok(InternedString::intern(&d.read_str()?))
1312     }
1313 }
1314
1315 impl Encodable for InternedString {
1316     fn encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
1317         self.with(|string| s.emit_str(string))
1318     }
1319 }
1320
1321 #[cfg(test)]
1322 mod tests {
1323     use super::*;
1324     use crate::Globals;
1325     use crate::edition;
1326
1327     #[test]
1328     fn interner_tests() {
1329         let mut i: Interner = Interner::default();
1330         // first one is zero:
1331         assert_eq!(i.intern("dog"), Symbol::new(0));
1332         // re-use gets the same entry:
1333         assert_eq!(i.intern("dog"), Symbol::new(0));
1334         // different string gets a different #:
1335         assert_eq!(i.intern("cat"), Symbol::new(1));
1336         assert_eq!(i.intern("cat"), Symbol::new(1));
1337         // dog is still at zero
1338         assert_eq!(i.intern("dog"), Symbol::new(0));
1339         let z = i.intern("zebra");
1340         assert_eq!(i.gensymed(z), Symbol::new(SymbolIndex::MAX_AS_U32));
1341         // gensym of same string gets new number:
1342         assert_eq!(i.gensymed(z), Symbol::new(SymbolIndex::MAX_AS_U32 - 1));
1343         // gensym of *existing* string gets new number:
1344         let d = i.intern("dog");
1345         assert_eq!(i.gensymed(d), Symbol::new(SymbolIndex::MAX_AS_U32 - 2));
1346     }
1347
1348     #[test]
1349     fn without_first_quote_test() {
1350         GLOBALS.set(&Globals::new(edition::DEFAULT_EDITION), || {
1351             let i = Ident::from_str("'break");
1352             assert_eq!(i.without_first_quote().name, kw::Break);
1353         });
1354     }
1355 }