]> git.lizzy.rs Git - rust.git/blob - src/librustc_span/symbol.rs
Rollup merge of #69047 - ehuss:rustfmt-vendor, r=Centril
[rust.git] / src / librustc_span / 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::stable_hasher::{HashStable, StableHasher, ToStableHashKey};
8 use rustc_macros::{symbols, HashStable_Generic};
9 use rustc_serialize::{Decodable, Decoder, Encodable, Encoder};
10 use rustc_serialize::{UseSpecializedDecodable, UseSpecializedEncodable};
11
12 use std::cmp::{Ord, PartialEq, PartialOrd};
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         Raw:                "raw",
101         Union:              "union",
102     }
103
104     // Symbols that can be referred to with rustc_span::sym::*. The symbol is
105     // the stringified identifier unless otherwise specified (e.g.
106     // `proc_dash_macro` represents "proc-macro").
107     //
108     // As well as the symbols listed, there are symbols for the the strings
109     // "0", "1", ..., "9", which are accessible via `sym::integer`.
110     Symbols {
111         aarch64_target_feature,
112         abi,
113         abi_amdgpu_kernel,
114         abi_efiapi,
115         abi_msp430_interrupt,
116         abi_ptx,
117         abi_sysv64,
118         abi_thiscall,
119         abi_unadjusted,
120         abi_vectorcall,
121         abi_x86_interrupt,
122         aborts,
123         address,
124         add_with_overflow,
125         advanced_slice_patterns,
126         adx_target_feature,
127         alias,
128         align,
129         alignstack,
130         all,
131         allocator,
132         allocator_internals,
133         alloc_error_handler,
134         allow,
135         allowed,
136         allow_fail,
137         allow_internal_unsafe,
138         allow_internal_unstable,
139         allow_internal_unstable_backcompat_hack,
140         always,
141         and,
142         any,
143         arbitrary_enum_discriminant,
144         arbitrary_self_types,
145         Arguments,
146         ArgumentV1,
147         arm_target_feature,
148         asm,
149         assert,
150         associated_consts,
151         associated_type_bounds,
152         associated_type_defaults,
153         associated_types,
154         assume_init,
155         async_await,
156         async_closure,
157         attr,
158         attributes,
159         attr_literals,
160         augmented_assignments,
161         automatically_derived,
162         avx512_target_feature,
163         await_macro,
164         begin_panic,
165         bench,
166         bin,
167         bind_by_move_pattern_guards,
168         bindings_after_at,
169         block,
170         bool,
171         borrowck_graphviz_format,
172         borrowck_graphviz_postflow,
173         borrowck_graphviz_preflow,
174         box_patterns,
175         box_syntax,
176         braced_empty_structs,
177         bswap,
178         bitreverse,
179         C,
180         caller_location,
181         cdylib,
182         cfg,
183         cfg_attr,
184         cfg_attr_multi,
185         cfg_doctest,
186         cfg_sanitize,
187         cfg_target_feature,
188         cfg_target_has_atomic,
189         cfg_target_thread_local,
190         cfg_target_vendor,
191         char,
192         clippy,
193         clone,
194         Clone,
195         clone_closures,
196         clone_from,
197         closure_to_fn_coercion,
198         cmp,
199         cmpxchg16b_target_feature,
200         cold,
201         column,
202         compile_error,
203         compiler_builtins,
204         concat,
205         concat_idents,
206         conservative_impl_trait,
207         console,
208         const_compare_raw_pointers,
209         const_constructor,
210         const_extern_fn,
211         const_fn,
212         const_fn_union,
213         const_generics,
214         const_if_match,
215         const_indexing,
216         const_in_array_repeat_expressions,
217         const_let,
218         const_loop,
219         const_mut_refs,
220         const_panic,
221         const_raw_ptr_deref,
222         const_raw_ptr_to_usize_cast,
223         const_transmute,
224         const_trait_bound_opt_out,
225         const_trait_impl,
226         contents,
227         context,
228         convert,
229         Copy,
230         copy_closures,
231         core,
232         core_intrinsics,
233         crate_id,
234         crate_in_paths,
235         crate_local,
236         crate_name,
237         crate_type,
238         crate_visibility_modifier,
239         ctpop,
240         cttz,
241         cttz_nonzero,
242         ctlz,
243         ctlz_nonzero,
244         custom_attribute,
245         custom_derive,
246         custom_inner_attributes,
247         custom_test_frameworks,
248         c_variadic,
249         debug_trait,
250         declare_lint_pass,
251         decl_macro,
252         Debug,
253         Decodable,
254         Default,
255         default_lib_allocator,
256         default_type_parameter_fallback,
257         default_type_params,
258         delay_span_bug_from_inside_query,
259         deny,
260         deprecated,
261         deref,
262         deref_mut,
263         derive,
264         diagnostic,
265         direct,
266         doc,
267         doc_alias,
268         doc_cfg,
269         doc_keyword,
270         doc_masked,
271         doc_spotlight,
272         doctest,
273         document_private_items,
274         dotdoteq_in_patterns,
275         dotdot_in_tuple_patterns,
276         double_braced_crate: "{{crate}}",
277         double_braced_impl: "{{impl}}",
278         double_braced_misc: "{{misc}}",
279         double_braced_closure: "{{closure}}",
280         double_braced_constructor: "{{constructor}}",
281         double_braced_constant: "{{constant}}",
282         double_braced_opaque: "{{opaque}}",
283         dropck_eyepatch,
284         dropck_parametricity,
285         drop_types_in_const,
286         dylib,
287         dyn_trait,
288         eh_personality,
289         eh_unwind_resume,
290         enable,
291         Encodable,
292         env,
293         eq,
294         err,
295         Err,
296         Eq,
297         Equal,
298         enclosing_scope,
299         except,
300         exclusive_range_pattern,
301         exhaustive_integer_patterns,
302         exhaustive_patterns,
303         existential_type,
304         expected,
305         export_name,
306         expr,
307         extern_absolute_paths,
308         external_doc,
309         extern_crate_item_prelude,
310         extern_crate_self,
311         extern_in_paths,
312         extern_prelude,
313         extern_types,
314         f16c_target_feature,
315         f32,
316         f64,
317         feature,
318         ffi_returns_twice,
319         field,
320         field_init_shorthand,
321         file,
322         fmt,
323         fmt_internals,
324         fn_must_use,
325         forbid,
326         format_args,
327         format_args_nl,
328         from,
329         From,
330         from_desugaring,
331         from_error,
332         from_generator,
333         from_method,
334         from_ok,
335         from_usize,
336         fundamental,
337         future,
338         Future,
339         FxHashSet,
340         FxHashMap,
341         gen_future,
342         gen_kill,
343         generators,
344         generic_associated_types,
345         generic_param_attrs,
346         global_allocator,
347         global_asm,
348         globs,
349         half_open_range_patterns,
350         hash,
351         Hash,
352         HashSet,
353         HashMap,
354         hexagon_target_feature,
355         hidden,
356         homogeneous_aggregate,
357         html_favicon_url,
358         html_logo_url,
359         html_no_source,
360         html_playground_url,
361         html_root_url,
362         i128,
363         i128_type,
364         i16,
365         i32,
366         i64,
367         i8,
368         ident,
369         if_let,
370         if_while_or_patterns,
371         ignore,
372         impl_header_lifetime_elision,
373         impl_lint_pass,
374         impl_trait_in_bindings,
375         import_shadowing,
376         index,
377         index_mut,
378         in_band_lifetimes,
379         include,
380         include_bytes,
381         include_str,
382         inclusive_range_syntax,
383         infer_outlives_requirements,
384         infer_static_outlives_requirements,
385         inline,
386         intel,
387         into_iter,
388         IntoIterator,
389         into_result,
390         intrinsics,
391         irrefutable_let_patterns,
392         isize,
393         issue,
394         issue_5723_bootstrap,
395         issue_tracker_base_url,
396         item,
397         item_context: "ItemContext",
398         item_like_imports,
399         iter,
400         Iterator,
401         keyword,
402         kind,
403         label,
404         label_break_value,
405         lang,
406         lang_items,
407         let_chains,
408         lhs,
409         lib,
410         lifetime,
411         line,
412         link,
413         linkage,
414         link_args,
415         link_cfg,
416         link_llvm_intrinsics,
417         link_name,
418         link_ordinal,
419         link_section,
420         LintPass,
421         lint_reasons,
422         literal,
423         local_inner_macros,
424         log_syntax,
425         loop_break_value,
426         macro_at_most_once_rep,
427         macro_escape,
428         macro_export,
429         macro_lifetime_matcher,
430         macro_literal_matcher,
431         macro_reexport,
432         macro_rules,
433         macros_in_extern,
434         macro_use,
435         macro_vis_matcher,
436         main,
437         managed_boxes,
438         marker,
439         marker_trait_attr,
440         masked,
441         match_beginning_vert,
442         match_default_bindings,
443         may_dangle,
444         maybe_uninit_uninit,
445         maybe_uninit_zeroed,
446         mem_uninitialized,
447         mem_zeroed,
448         member_constraints,
449         memory,
450         message,
451         meta,
452         min_align_of,
453         min_const_fn,
454         min_const_unsafe_fn,
455         mips_target_feature,
456         mmx_target_feature,
457         module,
458         module_path,
459         more_struct_aliases,
460         move_ref_pattern,
461         move_val_init,
462         movbe_target_feature,
463         mul_with_overflow,
464         must_use,
465         naked,
466         naked_functions,
467         name,
468         needs_allocator,
469         needs_drop,
470         needs_panic_runtime,
471         negate_unsigned,
472         never,
473         never_type,
474         never_type_fallback,
475         new,
476         next,
477         __next,
478         nll,
479         no_builtins,
480         no_core,
481         no_crate_inject,
482         no_debug,
483         no_default_passes,
484         no_implicit_prelude,
485         no_inline,
486         no_link,
487         no_main,
488         no_mangle,
489         non_ascii_idents,
490         None,
491         non_exhaustive,
492         non_modrs_mods,
493         no_sanitize,
494         no_stack_check,
495         no_start,
496         no_std,
497         not,
498         note,
499         object_safe_for_dispatch,
500         Ok,
501         omit_gdb_pretty_printer_section,
502         on,
503         on_unimplemented,
504         oom,
505         ops,
506         optimize,
507         optimize_attribute,
508         optin_builtin_traits,
509         option,
510         Option,
511         option_env,
512         opt_out_copy,
513         or,
514         or_patterns,
515         Ord,
516         Ordering,
517         Output,
518         overlapping_marker_traits,
519         packed,
520         panic,
521         panic_handler,
522         panic_impl,
523         panic_implementation,
524         panic_runtime,
525         parent_trait,
526         partial_cmp,
527         param_attrs,
528         PartialEq,
529         PartialOrd,
530         passes,
531         pat,
532         path,
533         pattern_parentheses,
534         Pending,
535         pin,
536         Pin,
537         pinned,
538         platform_intrinsics,
539         plugin,
540         plugin_registrar,
541         plugins,
542         Poll,
543         poll_with_tls_context,
544         powerpc_target_feature,
545         precise_pointer_size_matching,
546         pref_align_of,
547         prelude,
548         prelude_import,
549         primitive,
550         proc_dash_macro: "proc-macro",
551         proc_macro,
552         proc_macro_attribute,
553         proc_macro_def_site,
554         proc_macro_derive,
555         proc_macro_expr,
556         proc_macro_gen,
557         proc_macro_hygiene,
558         proc_macro_internals,
559         proc_macro_mod,
560         proc_macro_non_items,
561         proc_macro_path_invoc,
562         profiler_runtime,
563         ptr_offset_from,
564         pub_restricted,
565         pushpop_unsafe,
566         quad_precision_float,
567         question_mark,
568         quote,
569         Range,
570         RangeFrom,
571         RangeFull,
572         RangeInclusive,
573         RangeTo,
574         RangeToInclusive,
575         raw_dylib,
576         raw_identifiers,
577         raw_ref_op,
578         Ready,
579         reason,
580         recursion_limit,
581         reexport_test_harness_main,
582         reflect,
583         register_attr,
584         register_tool,
585         relaxed_adts,
586         repr,
587         repr128,
588         repr_align,
589         repr_align_enum,
590         repr_packed,
591         repr_simd,
592         repr_transparent,
593         re_rebalance_coherence,
594         result,
595         Result,
596         Return,
597         rhs,
598         rlib,
599         rotate_left,
600         rotate_right,
601         rt,
602         rtm_target_feature,
603         rust,
604         rust_2015_preview,
605         rust_2018_preview,
606         rust_begin_unwind,
607         rustc,
608         RustcDecodable,
609         RustcEncodable,
610         rustc_allocator,
611         rustc_allocator_nounwind,
612         rustc_allow_const_fn_ptr,
613         rustc_args_required_const,
614         rustc_attrs,
615         rustc_builtin_macro,
616         rustc_clean,
617         rustc_const_unstable,
618         rustc_const_stable,
619         rustc_conversion_suggestion,
620         rustc_def_path,
621         rustc_deprecated,
622         rustc_diagnostic_item,
623         rustc_diagnostic_macros,
624         rustc_dirty,
625         rustc_dummy,
626         rustc_dump_env_program_clauses,
627         rustc_dump_program_clauses,
628         rustc_dump_user_substs,
629         rustc_error,
630         rustc_expected_cgu_reuse,
631         rustc_if_this_changed,
632         rustc_inherit_overflow_checks,
633         rustc_layout,
634         rustc_layout_scalar_valid_range_end,
635         rustc_layout_scalar_valid_range_start,
636         rustc_macro_transparency,
637         rustc_mir,
638         rustc_nonnull_optimization_guaranteed,
639         rustc_object_lifetime_default,
640         rustc_on_unimplemented,
641         rustc_outlives,
642         rustc_paren_sugar,
643         rustc_partition_codegened,
644         rustc_partition_reused,
645         rustc_peek,
646         rustc_peek_definite_init,
647         rustc_peek_maybe_init,
648         rustc_peek_maybe_uninit,
649         rustc_peek_indirectly_mutable,
650         rustc_private,
651         rustc_proc_macro_decls,
652         rustc_promotable,
653         rustc_regions,
654         rustc_stable,
655         rustc_std_internal_symbol,
656         rustc_symbol_name,
657         rustc_synthetic,
658         rustc_reservation_impl,
659         rustc_test_marker,
660         rustc_then_this_would_need,
661         rustc_variance,
662         rustfmt,
663         rust_eh_personality,
664         rust_eh_unwind_resume,
665         rust_oom,
666         rvalue_static_promotion,
667         sanitize,
668         sanitizer_runtime,
669         saturating_add,
670         saturating_sub,
671         _Self,
672         self_in_typedefs,
673         self_struct_ctor,
674         send_trait,
675         should_panic,
676         simd,
677         simd_extract,
678         simd_ffi,
679         simd_insert,
680         since,
681         size,
682         size_of,
683         slice_patterns,
684         slicing_syntax,
685         soft,
686         Some,
687         specialization,
688         speed,
689         spotlight,
690         sse4a_target_feature,
691         stable,
692         staged_api,
693         start,
694         static_in_const,
695         staticlib,
696         static_nobundle,
697         static_recursion,
698         std,
699         std_inject,
700         str,
701         stringify,
702         stmt,
703         stmt_expr_attributes,
704         stop_after_dataflow,
705         struct_field_attributes,
706         struct_inherit,
707         structural_match,
708         struct_variant,
709         sty,
710         sub_with_overflow,
711         suggestion,
712         sync_trait,
713         target_feature,
714         target_has_atomic,
715         target_has_atomic_load_store,
716         target_thread_local,
717         task,
718         tbm_target_feature,
719         termination_trait,
720         termination_trait_test,
721         test,
722         test_2018_feature,
723         test_accepted_feature,
724         test_case,
725         test_removed_feature,
726         test_runner,
727         then_with,
728         thread,
729         thread_local,
730         tool_attributes,
731         tool_lints,
732         trace_macros,
733         track_caller,
734         trait_alias,
735         transmute,
736         transparent,
737         transparent_enums,
738         transparent_unions,
739         trivial_bounds,
740         Try,
741         try_blocks,
742         try_trait,
743         tt,
744         tuple_indexing,
745         two_phase,
746         Ty,
747         ty,
748         type_alias_impl_trait,
749         type_id,
750         type_name,
751         TyCtxt,
752         TyKind,
753         type_alias_enum_variants,
754         type_ascription,
755         type_length_limit,
756         type_macros,
757         u128,
758         u16,
759         u32,
760         u64,
761         u8,
762         unboxed_closures,
763         unchecked_add,
764         unchecked_div,
765         unchecked_mul,
766         unchecked_rem,
767         unchecked_shl,
768         unchecked_shr,
769         unchecked_sub,
770         underscore_const_names,
771         underscore_imports,
772         underscore_lifetimes,
773         uniform_paths,
774         universal_impl_trait,
775         unmarked_api,
776         unreachable_code,
777         unrestricted_attribute_tokens,
778         unsafe_no_drop_flag,
779         unsized_locals,
780         unsized_tuple_coercion,
781         unstable,
782         untagged_unions,
783         unwind,
784         unwind_attributes,
785         unwrap_or,
786         used,
787         use_extern_macros,
788         use_nested_groups,
789         usize,
790         v1,
791         val,
792         var,
793         vec,
794         Vec,
795         vis,
796         visible_private_types,
797         volatile,
798         warn,
799         wasm_import_module,
800         wasm_target_feature,
801         while_let,
802         windows,
803         windows_subsystem,
804         wrapping_add,
805         wrapping_sub,
806         wrapping_mul,
807         Yield,
808     }
809 }
810
811 #[derive(Copy, Clone, Eq, HashStable_Generic)]
812 pub struct Ident {
813     pub name: Symbol,
814     pub span: Span,
815 }
816
817 impl Ident {
818     #[inline]
819     /// Constructs a new identifier from a symbol and a span.
820     pub const fn new(name: Symbol, span: Span) -> Ident {
821         Ident { name, span }
822     }
823
824     /// Constructs a new identifier with a dummy span.
825     #[inline]
826     pub const fn with_dummy_span(name: Symbol) -> Ident {
827         Ident::new(name, DUMMY_SP)
828     }
829
830     #[inline]
831     pub fn invalid() -> Ident {
832         Ident::with_dummy_span(kw::Invalid)
833     }
834
835     /// Maps a string to an identifier with a dummy span.
836     pub fn from_str(string: &str) -> Ident {
837         Ident::with_dummy_span(Symbol::intern(string))
838     }
839
840     /// Maps a string and a span to an identifier.
841     pub fn from_str_and_span(string: &str, span: Span) -> Ident {
842         Ident::new(Symbol::intern(string), span)
843     }
844
845     /// Replaces `lo` and `hi` with those from `span`, but keep hygiene context.
846     pub fn with_span_pos(self, span: Span) -> Ident {
847         Ident::new(self.name, span.with_ctxt(self.span.ctxt()))
848     }
849
850     pub fn without_first_quote(self) -> Ident {
851         Ident::new(Symbol::intern(self.as_str().trim_start_matches('\'')), self.span)
852     }
853
854     /// "Normalize" ident for use in comparisons using "item hygiene".
855     /// Identifiers with same string value become same if they came from the same "modern" macro
856     /// (e.g., `macro` item, but not `macro_rules` item) and stay different if they came from
857     /// different "modern" macros.
858     /// Technically, this operation strips all non-opaque marks from ident's syntactic context.
859     pub fn modern(self) -> Ident {
860         Ident::new(self.name, self.span.modern())
861     }
862
863     /// "Normalize" ident for use in comparisons using "local variable hygiene".
864     /// Identifiers with same string value become same if they came from the same non-transparent
865     /// macro (e.g., `macro` or `macro_rules!` items) and stay different if they came from different
866     /// non-transparent macros.
867     /// Technically, this operation strips all transparent marks from ident's syntactic context.
868     pub fn modern_and_legacy(self) -> Ident {
869         Ident::new(self.name, self.span.modern_and_legacy())
870     }
871
872     /// Convert the name to a `SymbolStr`. This is a slowish operation because
873     /// it requires locking the symbol interner.
874     pub fn as_str(self) -> SymbolStr {
875         self.name.as_str()
876     }
877 }
878
879 impl PartialEq for Ident {
880     fn eq(&self, rhs: &Self) -> bool {
881         self.name == rhs.name && self.span.ctxt() == rhs.span.ctxt()
882     }
883 }
884
885 impl Hash for Ident {
886     fn hash<H: Hasher>(&self, state: &mut H) {
887         self.name.hash(state);
888         self.span.ctxt().hash(state);
889     }
890 }
891
892 impl fmt::Debug for Ident {
893     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
894         if self.is_raw_guess() {
895             write!(f, "r#")?;
896         }
897         write!(f, "{}{:?}", self.name, self.span.ctxt())
898     }
899 }
900
901 impl fmt::Display for Ident {
902     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
903         if self.is_raw_guess() {
904             write!(f, "r#")?;
905         }
906         fmt::Display::fmt(&self.name, f)
907     }
908 }
909
910 impl UseSpecializedEncodable for Ident {
911     fn default_encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
912         s.emit_struct("Ident", 2, |s| {
913             s.emit_struct_field("name", 0, |s| self.name.encode(s))?;
914             s.emit_struct_field("span", 1, |s| self.span.encode(s))
915         })
916     }
917 }
918
919 impl UseSpecializedDecodable for Ident {
920     fn default_decode<D: Decoder>(d: &mut D) -> Result<Self, D::Error> {
921         d.read_struct("Ident", 2, |d| {
922             Ok(Ident {
923                 name: d.read_struct_field("name", 0, Decodable::decode)?,
924                 span: d.read_struct_field("span", 1, Decodable::decode)?,
925             })
926         })
927     }
928 }
929
930 /// An interned string.
931 ///
932 /// Internally, a `Symbol` is implemented as an index, and all operations
933 /// (including hashing, equality, and ordering) operate on that index. The use
934 /// of `rustc_index::newtype_index!` means that `Option<Symbol>` only takes up 4 bytes,
935 /// because `rustc_index::newtype_index!` reserves the last 256 values for tagging purposes.
936 ///
937 /// Note that `Symbol` cannot directly be a `rustc_index::newtype_index!` because it
938 /// implements `fmt::Debug`, `Encodable`, and `Decodable` in special ways.
939 #[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
940 pub struct Symbol(SymbolIndex);
941
942 rustc_index::newtype_index! {
943     pub struct SymbolIndex { .. }
944 }
945
946 impl Symbol {
947     const fn new(n: u32) -> Self {
948         Symbol(SymbolIndex::from_u32_const(n))
949     }
950
951     /// Maps a string to its interned representation.
952     pub fn intern(string: &str) -> Self {
953         with_interner(|interner| interner.intern(string))
954     }
955
956     /// Access the symbol's chars. This is a slowish operation because it
957     /// requires locking the symbol interner.
958     pub fn with<F: FnOnce(&str) -> R, R>(self, f: F) -> R {
959         with_interner(|interner| f(interner.get(self)))
960     }
961
962     /// Convert to a `SymbolStr`. This is a slowish operation because it
963     /// requires locking the symbol interner.
964     pub fn as_str(self) -> SymbolStr {
965         with_interner(|interner| unsafe {
966             SymbolStr { string: std::mem::transmute::<&str, &str>(interner.get(self)) }
967         })
968     }
969
970     pub fn as_u32(self) -> u32 {
971         self.0.as_u32()
972     }
973 }
974
975 impl fmt::Debug for Symbol {
976     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
977         self.with(|str| fmt::Debug::fmt(&str, f))
978     }
979 }
980
981 impl fmt::Display for Symbol {
982     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
983         self.with(|str| fmt::Display::fmt(&str, f))
984     }
985 }
986
987 impl Encodable for Symbol {
988     fn encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
989         self.with(|string| s.emit_str(string))
990     }
991 }
992
993 impl Decodable for Symbol {
994     fn decode<D: Decoder>(d: &mut D) -> Result<Symbol, D::Error> {
995         Ok(Symbol::intern(&d.read_str()?))
996     }
997 }
998
999 impl<CTX> HashStable<CTX> for Symbol {
1000     #[inline]
1001     fn hash_stable(&self, hcx: &mut CTX, hasher: &mut StableHasher) {
1002         self.as_str().hash_stable(hcx, hasher);
1003     }
1004 }
1005
1006 impl<CTX> ToStableHashKey<CTX> for Symbol {
1007     type KeyType = SymbolStr;
1008
1009     #[inline]
1010     fn to_stable_hash_key(&self, _: &CTX) -> SymbolStr {
1011         self.as_str()
1012     }
1013 }
1014
1015 // The `&'static str`s in this type actually point into the arena.
1016 #[derive(Default)]
1017 pub struct Interner {
1018     arena: DroplessArena,
1019     names: FxHashMap<&'static str, Symbol>,
1020     strings: Vec<&'static str>,
1021 }
1022
1023 impl Interner {
1024     fn prefill(init: &[&'static str]) -> Self {
1025         Interner {
1026             strings: init.into(),
1027             names: init.iter().copied().zip((0..).map(Symbol::new)).collect(),
1028             ..Default::default()
1029         }
1030     }
1031
1032     pub fn intern(&mut self, string: &str) -> Symbol {
1033         if let Some(&name) = self.names.get(string) {
1034             return name;
1035         }
1036
1037         let name = Symbol::new(self.strings.len() as u32);
1038
1039         // `from_utf8_unchecked` is safe since we just allocated a `&str` which is known to be
1040         // UTF-8.
1041         let string: &str =
1042             unsafe { str::from_utf8_unchecked(self.arena.alloc_slice(string.as_bytes())) };
1043         // It is safe to extend the arena allocation to `'static` because we only access
1044         // these while the arena is still alive.
1045         let string: &'static str = unsafe { &*(string as *const str) };
1046         self.strings.push(string);
1047         self.names.insert(string, name);
1048         name
1049     }
1050
1051     // Get the symbol as a string. `Symbol::as_str()` should be used in
1052     // preference to this function.
1053     pub fn get(&self, symbol: Symbol) -> &str {
1054         self.strings[symbol.0.as_usize()]
1055     }
1056 }
1057
1058 // This module has a very short name because it's used a lot.
1059 pub mod kw {
1060     use super::Symbol;
1061     keywords!();
1062 }
1063
1064 // This module has a very short name because it's used a lot.
1065 #[allow(rustc::default_hash_types)]
1066 pub mod sym {
1067     use super::Symbol;
1068     use std::convert::TryInto;
1069
1070     symbols!();
1071
1072     // Get the symbol for an integer. The first few non-negative integers each
1073     // have a static symbol and therefore are fast.
1074     pub fn integer<N: TryInto<usize> + Copy + ToString>(n: N) -> Symbol {
1075         if let Result::Ok(idx) = n.try_into() {
1076             if let Option::Some(&sym) = digits_array.get(idx) {
1077                 return sym;
1078             }
1079         }
1080         Symbol::intern(&n.to_string())
1081     }
1082 }
1083
1084 impl Symbol {
1085     fn is_used_keyword_2018(self) -> bool {
1086         self >= kw::Async && self <= kw::Dyn
1087     }
1088
1089     fn is_unused_keyword_2018(self) -> bool {
1090         self == kw::Try
1091     }
1092
1093     /// Used for sanity checking rustdoc keyword sections.
1094     pub fn is_doc_keyword(self) -> bool {
1095         self <= kw::Union
1096     }
1097
1098     /// A keyword or reserved identifier that can be used as a path segment.
1099     pub fn is_path_segment_keyword(self) -> bool {
1100         self == kw::Super
1101             || self == kw::SelfLower
1102             || self == kw::SelfUpper
1103             || self == kw::Crate
1104             || self == kw::PathRoot
1105             || self == kw::DollarCrate
1106     }
1107
1108     /// Returns `true` if the symbol is `true` or `false`.
1109     pub fn is_bool_lit(self) -> bool {
1110         self == kw::True || self == kw::False
1111     }
1112
1113     /// This symbol can be a raw identifier.
1114     pub fn can_be_raw(self) -> bool {
1115         self != kw::Invalid && self != kw::Underscore && !self.is_path_segment_keyword()
1116     }
1117 }
1118
1119 impl Ident {
1120     // Returns `true` for reserved identifiers used internally for elided lifetimes,
1121     // unnamed method parameters, crate root module, error recovery etc.
1122     pub fn is_special(self) -> bool {
1123         self.name <= kw::Underscore
1124     }
1125
1126     /// Returns `true` if the token is a keyword used in the language.
1127     pub fn is_used_keyword(self) -> bool {
1128         // Note: `span.edition()` is relatively expensive, don't call it unless necessary.
1129         self.name >= kw::As && self.name <= kw::While
1130             || self.name.is_used_keyword_2018() && self.span.rust_2018()
1131     }
1132
1133     /// Returns `true` if the token is a keyword reserved for possible future use.
1134     pub fn is_unused_keyword(self) -> bool {
1135         // Note: `span.edition()` is relatively expensive, don't call it unless necessary.
1136         self.name >= kw::Abstract && self.name <= kw::Yield
1137             || self.name.is_unused_keyword_2018() && self.span.rust_2018()
1138     }
1139
1140     /// Returns `true` if the token is either a special identifier or a keyword.
1141     pub fn is_reserved(self) -> bool {
1142         self.is_special() || self.is_used_keyword() || self.is_unused_keyword()
1143     }
1144
1145     /// A keyword or reserved identifier that can be used as a path segment.
1146     pub fn is_path_segment_keyword(self) -> bool {
1147         self.name.is_path_segment_keyword()
1148     }
1149
1150     /// We see this identifier in a normal identifier position, like variable name or a type.
1151     /// How was it written originally? Did it use the raw form? Let's try to guess.
1152     pub fn is_raw_guess(self) -> bool {
1153         self.name.can_be_raw() && self.is_reserved()
1154     }
1155 }
1156
1157 #[inline]
1158 fn with_interner<T, F: FnOnce(&mut Interner) -> T>(f: F) -> T {
1159     GLOBALS.with(|globals| f(&mut *globals.symbol_interner.lock()))
1160 }
1161
1162 /// An alternative to `Symbol`, useful when the chars within the symbol need to
1163 /// be accessed. It deliberately has limited functionality and should only be
1164 /// used for temporary values.
1165 ///
1166 /// Because the interner outlives any thread which uses this type, we can
1167 /// safely treat `string` which points to interner data, as an immortal string,
1168 /// as long as this type never crosses between threads.
1169 //
1170 // FIXME: ensure that the interner outlives any thread which uses `SymbolStr`,
1171 // by creating a new thread right after constructing the interner.
1172 #[derive(Clone, Eq, PartialOrd, Ord)]
1173 pub struct SymbolStr {
1174     string: &'static str,
1175 }
1176
1177 // This impl allows a `SymbolStr` to be directly equated with a `String` or
1178 // `&str`.
1179 impl<T: std::ops::Deref<Target = str>> std::cmp::PartialEq<T> for SymbolStr {
1180     fn eq(&self, other: &T) -> bool {
1181         self.string == other.deref()
1182     }
1183 }
1184
1185 impl !Send for SymbolStr {}
1186 impl !Sync for SymbolStr {}
1187
1188 /// This impl means that if `ss` is a `SymbolStr`:
1189 /// - `*ss` is a `str`;
1190 /// - `&*ss` is a `&str`;
1191 /// - `&ss as &str` is a `&str`, which means that `&ss` can be passed to a
1192 ///   function expecting a `&str`.
1193 impl std::ops::Deref for SymbolStr {
1194     type Target = str;
1195     #[inline]
1196     fn deref(&self) -> &str {
1197         self.string
1198     }
1199 }
1200
1201 impl fmt::Debug for SymbolStr {
1202     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1203         fmt::Debug::fmt(self.string, f)
1204     }
1205 }
1206
1207 impl fmt::Display for SymbolStr {
1208     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1209         fmt::Display::fmt(self.string, f)
1210     }
1211 }
1212
1213 impl<CTX> HashStable<CTX> for SymbolStr {
1214     #[inline]
1215     fn hash_stable(&self, hcx: &mut CTX, hasher: &mut StableHasher) {
1216         self.string.hash_stable(hcx, hasher)
1217     }
1218 }
1219
1220 impl<CTX> ToStableHashKey<CTX> for SymbolStr {
1221     type KeyType = SymbolStr;
1222
1223     #[inline]
1224     fn to_stable_hash_key(&self, _: &CTX) -> SymbolStr {
1225         self.clone()
1226     }
1227 }