]> git.lizzy.rs Git - rust.git/blob - src/librustc_span/symbol.rs
Rollup merge of #68958 - GuillaumeGomez:clean-up-e0277-e0282, r=Dylan-DPC
[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_val_init,
461         movbe_target_feature,
462         mul_with_overflow,
463         must_use,
464         naked,
465         naked_functions,
466         name,
467         needs_allocator,
468         needs_drop,
469         needs_panic_runtime,
470         negate_unsigned,
471         never,
472         never_type,
473         never_type_fallback,
474         new,
475         next,
476         __next,
477         nll,
478         no_builtins,
479         no_core,
480         no_crate_inject,
481         no_debug,
482         no_default_passes,
483         no_implicit_prelude,
484         no_inline,
485         no_link,
486         no_main,
487         no_mangle,
488         non_ascii_idents,
489         None,
490         non_exhaustive,
491         non_modrs_mods,
492         no_sanitize,
493         no_stack_check,
494         no_start,
495         no_std,
496         not,
497         note,
498         object_safe_for_dispatch,
499         Ok,
500         omit_gdb_pretty_printer_section,
501         on,
502         on_unimplemented,
503         oom,
504         ops,
505         optimize,
506         optimize_attribute,
507         optin_builtin_traits,
508         option,
509         Option,
510         option_env,
511         opt_out_copy,
512         or,
513         or_patterns,
514         Ord,
515         Ordering,
516         Output,
517         overlapping_marker_traits,
518         packed,
519         panic,
520         panic_handler,
521         panic_impl,
522         panic_implementation,
523         panic_runtime,
524         parent_trait,
525         partial_cmp,
526         param_attrs,
527         PartialEq,
528         PartialOrd,
529         passes,
530         pat,
531         path,
532         pattern_parentheses,
533         Pending,
534         pin,
535         Pin,
536         pinned,
537         platform_intrinsics,
538         plugin,
539         plugin_registrar,
540         plugins,
541         Poll,
542         poll_with_tls_context,
543         powerpc_target_feature,
544         precise_pointer_size_matching,
545         pref_align_of,
546         prelude,
547         prelude_import,
548         primitive,
549         proc_dash_macro: "proc-macro",
550         proc_macro,
551         proc_macro_attribute,
552         proc_macro_def_site,
553         proc_macro_derive,
554         proc_macro_expr,
555         proc_macro_gen,
556         proc_macro_hygiene,
557         proc_macro_internals,
558         proc_macro_mod,
559         proc_macro_non_items,
560         proc_macro_path_invoc,
561         profiler_runtime,
562         ptr_offset_from,
563         pub_restricted,
564         pushpop_unsafe,
565         quad_precision_float,
566         question_mark,
567         quote,
568         Range,
569         RangeFrom,
570         RangeFull,
571         RangeInclusive,
572         RangeTo,
573         RangeToInclusive,
574         raw_dylib,
575         raw_identifiers,
576         raw_ref_op,
577         Ready,
578         reason,
579         recursion_limit,
580         reexport_test_harness_main,
581         reflect,
582         register_attr,
583         register_tool,
584         relaxed_adts,
585         repr,
586         repr128,
587         repr_align,
588         repr_align_enum,
589         repr_packed,
590         repr_simd,
591         repr_transparent,
592         re_rebalance_coherence,
593         result,
594         Result,
595         Return,
596         rhs,
597         rlib,
598         rotate_left,
599         rotate_right,
600         rt,
601         rtm_target_feature,
602         rust,
603         rust_2015_preview,
604         rust_2018_preview,
605         rust_begin_unwind,
606         rustc,
607         RustcDecodable,
608         RustcEncodable,
609         rustc_allocator,
610         rustc_allocator_nounwind,
611         rustc_allow_const_fn_ptr,
612         rustc_args_required_const,
613         rustc_attrs,
614         rustc_builtin_macro,
615         rustc_clean,
616         rustc_const_unstable,
617         rustc_const_stable,
618         rustc_conversion_suggestion,
619         rustc_def_path,
620         rustc_deprecated,
621         rustc_diagnostic_item,
622         rustc_diagnostic_macros,
623         rustc_dirty,
624         rustc_dummy,
625         rustc_dump_env_program_clauses,
626         rustc_dump_program_clauses,
627         rustc_dump_user_substs,
628         rustc_error,
629         rustc_expected_cgu_reuse,
630         rustc_if_this_changed,
631         rustc_inherit_overflow_checks,
632         rustc_layout,
633         rustc_layout_scalar_valid_range_end,
634         rustc_layout_scalar_valid_range_start,
635         rustc_macro_transparency,
636         rustc_mir,
637         rustc_nonnull_optimization_guaranteed,
638         rustc_object_lifetime_default,
639         rustc_on_unimplemented,
640         rustc_outlives,
641         rustc_paren_sugar,
642         rustc_partition_codegened,
643         rustc_partition_reused,
644         rustc_peek,
645         rustc_peek_definite_init,
646         rustc_peek_maybe_init,
647         rustc_peek_maybe_uninit,
648         rustc_peek_indirectly_mutable,
649         rustc_private,
650         rustc_proc_macro_decls,
651         rustc_promotable,
652         rustc_regions,
653         rustc_stable,
654         rustc_std_internal_symbol,
655         rustc_symbol_name,
656         rustc_synthetic,
657         rustc_reservation_impl,
658         rustc_test_marker,
659         rustc_then_this_would_need,
660         rustc_variance,
661         rustfmt,
662         rust_eh_personality,
663         rust_eh_unwind_resume,
664         rust_oom,
665         rvalue_static_promotion,
666         sanitize,
667         sanitizer_runtime,
668         saturating_add,
669         saturating_sub,
670         _Self,
671         self_in_typedefs,
672         self_struct_ctor,
673         send_trait,
674         should_panic,
675         simd,
676         simd_extract,
677         simd_ffi,
678         simd_insert,
679         since,
680         size,
681         size_of,
682         slice_patterns,
683         slicing_syntax,
684         soft,
685         Some,
686         specialization,
687         speed,
688         spotlight,
689         sse4a_target_feature,
690         stable,
691         staged_api,
692         start,
693         static_in_const,
694         staticlib,
695         static_nobundle,
696         static_recursion,
697         std,
698         std_inject,
699         str,
700         stringify,
701         stmt,
702         stmt_expr_attributes,
703         stop_after_dataflow,
704         struct_field_attributes,
705         struct_inherit,
706         structural_match,
707         struct_variant,
708         sty,
709         sub_with_overflow,
710         suggestion,
711         sync_trait,
712         target_feature,
713         target_has_atomic,
714         target_has_atomic_load_store,
715         target_thread_local,
716         task,
717         tbm_target_feature,
718         termination_trait,
719         termination_trait_test,
720         test,
721         test_2018_feature,
722         test_accepted_feature,
723         test_case,
724         test_removed_feature,
725         test_runner,
726         then_with,
727         thread,
728         thread_local,
729         tool_attributes,
730         tool_lints,
731         trace_macros,
732         track_caller,
733         trait_alias,
734         transmute,
735         transparent,
736         transparent_enums,
737         transparent_unions,
738         trivial_bounds,
739         Try,
740         try_blocks,
741         try_trait,
742         tt,
743         tuple_indexing,
744         two_phase,
745         Ty,
746         ty,
747         type_alias_impl_trait,
748         type_id,
749         type_name,
750         TyCtxt,
751         TyKind,
752         type_alias_enum_variants,
753         type_ascription,
754         type_length_limit,
755         type_macros,
756         u128,
757         u16,
758         u32,
759         u64,
760         u8,
761         unboxed_closures,
762         unchecked_add,
763         unchecked_div,
764         unchecked_mul,
765         unchecked_rem,
766         unchecked_shl,
767         unchecked_shr,
768         unchecked_sub,
769         underscore_const_names,
770         underscore_imports,
771         underscore_lifetimes,
772         uniform_paths,
773         universal_impl_trait,
774         unmarked_api,
775         unreachable_code,
776         unrestricted_attribute_tokens,
777         unsafe_no_drop_flag,
778         unsized_locals,
779         unsized_tuple_coercion,
780         unstable,
781         untagged_unions,
782         unwind,
783         unwind_attributes,
784         unwrap_or,
785         used,
786         use_extern_macros,
787         use_nested_groups,
788         usize,
789         v1,
790         val,
791         var,
792         vec,
793         Vec,
794         vis,
795         visible_private_types,
796         volatile,
797         warn,
798         wasm_import_module,
799         wasm_target_feature,
800         while_let,
801         windows,
802         windows_subsystem,
803         wrapping_add,
804         wrapping_sub,
805         wrapping_mul,
806         Yield,
807     }
808 }
809
810 #[derive(Copy, Clone, Eq, HashStable_Generic)]
811 pub struct Ident {
812     pub name: Symbol,
813     pub span: Span,
814 }
815
816 impl Ident {
817     #[inline]
818     /// Constructs a new identifier from a symbol and a span.
819     pub const fn new(name: Symbol, span: Span) -> Ident {
820         Ident { name, span }
821     }
822
823     /// Constructs a new identifier with a dummy span.
824     #[inline]
825     pub const fn with_dummy_span(name: Symbol) -> Ident {
826         Ident::new(name, DUMMY_SP)
827     }
828
829     #[inline]
830     pub fn invalid() -> Ident {
831         Ident::with_dummy_span(kw::Invalid)
832     }
833
834     /// Maps a string to an identifier with a dummy span.
835     pub fn from_str(string: &str) -> Ident {
836         Ident::with_dummy_span(Symbol::intern(string))
837     }
838
839     /// Maps a string and a span to an identifier.
840     pub fn from_str_and_span(string: &str, span: Span) -> Ident {
841         Ident::new(Symbol::intern(string), span)
842     }
843
844     /// Replaces `lo` and `hi` with those from `span`, but keep hygiene context.
845     pub fn with_span_pos(self, span: Span) -> Ident {
846         Ident::new(self.name, span.with_ctxt(self.span.ctxt()))
847     }
848
849     pub fn without_first_quote(self) -> Ident {
850         Ident::new(Symbol::intern(self.as_str().trim_start_matches('\'')), self.span)
851     }
852
853     /// "Normalize" ident for use in comparisons using "item hygiene".
854     /// Identifiers with same string value become same if they came from the same "modern" macro
855     /// (e.g., `macro` item, but not `macro_rules` item) and stay different if they came from
856     /// different "modern" macros.
857     /// Technically, this operation strips all non-opaque marks from ident's syntactic context.
858     pub fn modern(self) -> Ident {
859         Ident::new(self.name, self.span.modern())
860     }
861
862     /// "Normalize" ident for use in comparisons using "local variable hygiene".
863     /// Identifiers with same string value become same if they came from the same non-transparent
864     /// macro (e.g., `macro` or `macro_rules!` items) and stay different if they came from different
865     /// non-transparent macros.
866     /// Technically, this operation strips all transparent marks from ident's syntactic context.
867     pub fn modern_and_legacy(self) -> Ident {
868         Ident::new(self.name, self.span.modern_and_legacy())
869     }
870
871     /// Convert the name to a `SymbolStr`. This is a slowish operation because
872     /// it requires locking the symbol interner.
873     pub fn as_str(self) -> SymbolStr {
874         self.name.as_str()
875     }
876 }
877
878 impl PartialEq for Ident {
879     fn eq(&self, rhs: &Self) -> bool {
880         self.name == rhs.name && self.span.ctxt() == rhs.span.ctxt()
881     }
882 }
883
884 impl Hash for Ident {
885     fn hash<H: Hasher>(&self, state: &mut H) {
886         self.name.hash(state);
887         self.span.ctxt().hash(state);
888     }
889 }
890
891 impl fmt::Debug for Ident {
892     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
893         if self.is_raw_guess() {
894             write!(f, "r#")?;
895         }
896         write!(f, "{}{:?}", self.name, self.span.ctxt())
897     }
898 }
899
900 impl fmt::Display for Ident {
901     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
902         if self.is_raw_guess() {
903             write!(f, "r#")?;
904         }
905         fmt::Display::fmt(&self.name, f)
906     }
907 }
908
909 impl UseSpecializedEncodable for Ident {
910     fn default_encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
911         s.emit_struct("Ident", 2, |s| {
912             s.emit_struct_field("name", 0, |s| self.name.encode(s))?;
913             s.emit_struct_field("span", 1, |s| self.span.encode(s))
914         })
915     }
916 }
917
918 impl UseSpecializedDecodable for Ident {
919     fn default_decode<D: Decoder>(d: &mut D) -> Result<Self, D::Error> {
920         d.read_struct("Ident", 2, |d| {
921             Ok(Ident {
922                 name: d.read_struct_field("name", 0, Decodable::decode)?,
923                 span: d.read_struct_field("span", 1, Decodable::decode)?,
924             })
925         })
926     }
927 }
928
929 /// An interned string.
930 ///
931 /// Internally, a `Symbol` is implemented as an index, and all operations
932 /// (including hashing, equality, and ordering) operate on that index. The use
933 /// of `rustc_index::newtype_index!` means that `Option<Symbol>` only takes up 4 bytes,
934 /// because `rustc_index::newtype_index!` reserves the last 256 values for tagging purposes.
935 ///
936 /// Note that `Symbol` cannot directly be a `rustc_index::newtype_index!` because it
937 /// implements `fmt::Debug`, `Encodable`, and `Decodable` in special ways.
938 #[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
939 pub struct Symbol(SymbolIndex);
940
941 rustc_index::newtype_index! {
942     pub struct SymbolIndex { .. }
943 }
944
945 impl Symbol {
946     const fn new(n: u32) -> Self {
947         Symbol(SymbolIndex::from_u32_const(n))
948     }
949
950     /// Maps a string to its interned representation.
951     pub fn intern(string: &str) -> Self {
952         with_interner(|interner| interner.intern(string))
953     }
954
955     /// Access the symbol's chars. This is a slowish operation because it
956     /// requires locking the symbol interner.
957     pub fn with<F: FnOnce(&str) -> R, R>(self, f: F) -> R {
958         with_interner(|interner| f(interner.get(self)))
959     }
960
961     /// Convert to a `SymbolStr`. This is a slowish operation because it
962     /// requires locking the symbol interner.
963     pub fn as_str(self) -> SymbolStr {
964         with_interner(|interner| unsafe {
965             SymbolStr { string: std::mem::transmute::<&str, &str>(interner.get(self)) }
966         })
967     }
968
969     pub fn as_u32(self) -> u32 {
970         self.0.as_u32()
971     }
972 }
973
974 impl fmt::Debug for Symbol {
975     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
976         self.with(|str| fmt::Debug::fmt(&str, f))
977     }
978 }
979
980 impl fmt::Display for Symbol {
981     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
982         self.with(|str| fmt::Display::fmt(&str, f))
983     }
984 }
985
986 impl Encodable for Symbol {
987     fn encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
988         self.with(|string| s.emit_str(string))
989     }
990 }
991
992 impl Decodable for Symbol {
993     fn decode<D: Decoder>(d: &mut D) -> Result<Symbol, D::Error> {
994         Ok(Symbol::intern(&d.read_str()?))
995     }
996 }
997
998 impl<CTX> HashStable<CTX> for Symbol {
999     #[inline]
1000     fn hash_stable(&self, hcx: &mut CTX, hasher: &mut StableHasher) {
1001         self.as_str().hash_stable(hcx, hasher);
1002     }
1003 }
1004
1005 impl<CTX> ToStableHashKey<CTX> for Symbol {
1006     type KeyType = SymbolStr;
1007
1008     #[inline]
1009     fn to_stable_hash_key(&self, _: &CTX) -> SymbolStr {
1010         self.as_str()
1011     }
1012 }
1013
1014 // The `&'static str`s in this type actually point into the arena.
1015 #[derive(Default)]
1016 pub struct Interner {
1017     arena: DroplessArena,
1018     names: FxHashMap<&'static str, Symbol>,
1019     strings: Vec<&'static str>,
1020 }
1021
1022 impl Interner {
1023     fn prefill(init: &[&'static str]) -> Self {
1024         Interner {
1025             strings: init.into(),
1026             names: init.iter().copied().zip((0..).map(Symbol::new)).collect(),
1027             ..Default::default()
1028         }
1029     }
1030
1031     pub fn intern(&mut self, string: &str) -> Symbol {
1032         if let Some(&name) = self.names.get(string) {
1033             return name;
1034         }
1035
1036         let name = Symbol::new(self.strings.len() as u32);
1037
1038         // `from_utf8_unchecked` is safe since we just allocated a `&str` which is known to be
1039         // UTF-8.
1040         let string: &str =
1041             unsafe { str::from_utf8_unchecked(self.arena.alloc_slice(string.as_bytes())) };
1042         // It is safe to extend the arena allocation to `'static` because we only access
1043         // these while the arena is still alive.
1044         let string: &'static str = unsafe { &*(string as *const str) };
1045         self.strings.push(string);
1046         self.names.insert(string, name);
1047         name
1048     }
1049
1050     // Get the symbol as a string. `Symbol::as_str()` should be used in
1051     // preference to this function.
1052     pub fn get(&self, symbol: Symbol) -> &str {
1053         self.strings[symbol.0.as_usize()]
1054     }
1055 }
1056
1057 // This module has a very short name because it's used a lot.
1058 pub mod kw {
1059     use super::Symbol;
1060     keywords!();
1061 }
1062
1063 // This module has a very short name because it's used a lot.
1064 #[allow(rustc::default_hash_types)]
1065 pub mod sym {
1066     use super::Symbol;
1067     use std::convert::TryInto;
1068
1069     symbols!();
1070
1071     // Get the symbol for an integer. The first few non-negative integers each
1072     // have a static symbol and therefore are fast.
1073     pub fn integer<N: TryInto<usize> + Copy + ToString>(n: N) -> Symbol {
1074         if let Result::Ok(idx) = n.try_into() {
1075             if let Option::Some(&sym) = digits_array.get(idx) {
1076                 return sym;
1077             }
1078         }
1079         Symbol::intern(&n.to_string())
1080     }
1081 }
1082
1083 impl Symbol {
1084     fn is_used_keyword_2018(self) -> bool {
1085         self >= kw::Async && self <= kw::Dyn
1086     }
1087
1088     fn is_unused_keyword_2018(self) -> bool {
1089         self == kw::Try
1090     }
1091
1092     /// Used for sanity checking rustdoc keyword sections.
1093     pub fn is_doc_keyword(self) -> bool {
1094         self <= kw::Union
1095     }
1096
1097     /// A keyword or reserved identifier that can be used as a path segment.
1098     pub fn is_path_segment_keyword(self) -> bool {
1099         self == kw::Super
1100             || self == kw::SelfLower
1101             || self == kw::SelfUpper
1102             || self == kw::Crate
1103             || self == kw::PathRoot
1104             || self == kw::DollarCrate
1105     }
1106
1107     /// Returns `true` if the symbol is `true` or `false`.
1108     pub fn is_bool_lit(self) -> bool {
1109         self == kw::True || self == kw::False
1110     }
1111
1112     /// This symbol can be a raw identifier.
1113     pub fn can_be_raw(self) -> bool {
1114         self != kw::Invalid && self != kw::Underscore && !self.is_path_segment_keyword()
1115     }
1116 }
1117
1118 impl Ident {
1119     // Returns `true` for reserved identifiers used internally for elided lifetimes,
1120     // unnamed method parameters, crate root module, error recovery etc.
1121     pub fn is_special(self) -> bool {
1122         self.name <= kw::Underscore
1123     }
1124
1125     /// Returns `true` if the token is a keyword used in the language.
1126     pub fn is_used_keyword(self) -> bool {
1127         // Note: `span.edition()` is relatively expensive, don't call it unless necessary.
1128         self.name >= kw::As && self.name <= kw::While
1129             || self.name.is_used_keyword_2018() && self.span.rust_2018()
1130     }
1131
1132     /// Returns `true` if the token is a keyword reserved for possible future use.
1133     pub fn is_unused_keyword(self) -> bool {
1134         // Note: `span.edition()` is relatively expensive, don't call it unless necessary.
1135         self.name >= kw::Abstract && self.name <= kw::Yield
1136             || self.name.is_unused_keyword_2018() && self.span.rust_2018()
1137     }
1138
1139     /// Returns `true` if the token is either a special identifier or a keyword.
1140     pub fn is_reserved(self) -> bool {
1141         self.is_special() || self.is_used_keyword() || self.is_unused_keyword()
1142     }
1143
1144     /// A keyword or reserved identifier that can be used as a path segment.
1145     pub fn is_path_segment_keyword(self) -> bool {
1146         self.name.is_path_segment_keyword()
1147     }
1148
1149     /// We see this identifier in a normal identifier position, like variable name or a type.
1150     /// How was it written originally? Did it use the raw form? Let's try to guess.
1151     pub fn is_raw_guess(self) -> bool {
1152         self.name.can_be_raw() && self.is_reserved()
1153     }
1154 }
1155
1156 #[inline]
1157 fn with_interner<T, F: FnOnce(&mut Interner) -> T>(f: F) -> T {
1158     GLOBALS.with(|globals| f(&mut *globals.symbol_interner.lock()))
1159 }
1160
1161 /// An alternative to `Symbol`, useful when the chars within the symbol need to
1162 /// be accessed. It deliberately has limited functionality and should only be
1163 /// used for temporary values.
1164 ///
1165 /// Because the interner outlives any thread which uses this type, we can
1166 /// safely treat `string` which points to interner data, as an immortal string,
1167 /// as long as this type never crosses between threads.
1168 //
1169 // FIXME: ensure that the interner outlives any thread which uses `SymbolStr`,
1170 // by creating a new thread right after constructing the interner.
1171 #[derive(Clone, Eq, PartialOrd, Ord)]
1172 pub struct SymbolStr {
1173     string: &'static str,
1174 }
1175
1176 // This impl allows a `SymbolStr` to be directly equated with a `String` or
1177 // `&str`.
1178 impl<T: std::ops::Deref<Target = str>> std::cmp::PartialEq<T> for SymbolStr {
1179     fn eq(&self, other: &T) -> bool {
1180         self.string == other.deref()
1181     }
1182 }
1183
1184 impl !Send for SymbolStr {}
1185 impl !Sync for SymbolStr {}
1186
1187 /// This impl means that if `ss` is a `SymbolStr`:
1188 /// - `*ss` is a `str`;
1189 /// - `&*ss` is a `&str`;
1190 /// - `&ss as &str` is a `&str`, which means that `&ss` can be passed to a
1191 ///   function expecting a `&str`.
1192 impl std::ops::Deref for SymbolStr {
1193     type Target = str;
1194     #[inline]
1195     fn deref(&self) -> &str {
1196         self.string
1197     }
1198 }
1199
1200 impl fmt::Debug for SymbolStr {
1201     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1202         fmt::Debug::fmt(self.string, f)
1203     }
1204 }
1205
1206 impl fmt::Display for SymbolStr {
1207     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1208         fmt::Display::fmt(self.string, f)
1209     }
1210 }
1211
1212 impl<CTX> HashStable<CTX> for SymbolStr {
1213     #[inline]
1214     fn hash_stable(&self, hcx: &mut CTX, hasher: &mut StableHasher) {
1215         self.string.hash_stable(hcx, hasher)
1216     }
1217 }
1218
1219 impl<CTX> ToStableHashKey<CTX> for SymbolStr {
1220     type KeyType = SymbolStr;
1221
1222     #[inline]
1223     fn to_stable_hash_key(&self, _: &CTX) -> SymbolStr {
1224         self.clone()
1225     }
1226 }