1 | //===--- Ref.h ---------------------------------------------------*- C++-*-===// |
2 | // |
3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
4 | // See https://llvm.org/LICENSE.txt for license information. |
5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
6 | // |
7 | //===----------------------------------------------------------------------===// |
8 | |
9 | #ifndef LLVM_CLANG_TOOLS_EXTRA_CLANGD_INDEX_REF_H |
10 | #define |
11 | |
12 | #include "index/SymbolID.h" |
13 | #include "index/SymbolLocation.h" |
14 | #include "llvm/ADT/Hashing.h" |
15 | #include "llvm/Support/Allocator.h" |
16 | #include "llvm/Support/StringSaver.h" |
17 | #include "llvm/Support/raw_ostream.h" |
18 | #include <cstdint> |
19 | #include <set> |
20 | #include <utility> |
21 | |
22 | namespace clang { |
23 | namespace clangd { |
24 | |
25 | /// Describes the kind of a cross-reference. |
26 | /// |
27 | /// This is a bitfield which can be combined from different kinds. |
28 | enum class RefKind : uint8_t { |
29 | Unknown = 0, |
30 | // Points to symbol declaration. Example: |
31 | // |
32 | // class Foo; |
33 | // ^ Foo declaration |
34 | // Foo foo; |
35 | // ^ this does not reference Foo declaration |
36 | Declaration = 1 << 0, |
37 | // Points to symbol definition. Example: |
38 | // |
39 | // int foo(); |
40 | // ^ references foo declaration, but not foo definition |
41 | // int foo() { return 42; } |
42 | // ^ references foo definition, but not declaration |
43 | // bool bar() { return true; } |
44 | // ^ references both definition and declaration |
45 | Definition = 1 << 1, |
46 | // Points to symbol reference. Example: |
47 | // |
48 | // int Foo = 42; |
49 | // int Bar = Foo + 1; |
50 | // ^ this is a reference to Foo |
51 | Reference = 1 << 2, |
52 | // The reference explicitly spells out declaration's name. Such references can |
53 | // not come from macro expansions or implicit AST nodes. |
54 | // |
55 | // class Foo { public: Foo() {} }; |
56 | // ^ references declaration, definition and explicitly spells out name |
57 | // #define MACRO Foo |
58 | // v there is an implicit constructor call here which is not a spelled ref |
59 | // Foo foo; |
60 | // ^ this reference explicitly spells out Foo's name |
61 | // struct Bar { |
62 | // MACRO Internal; |
63 | // ^ this references Foo, but does not explicitly spell out its name |
64 | // }; |
65 | Spelled = 1 << 3, |
66 | All = Declaration | Definition | Reference | Spelled, |
67 | }; |
68 | |
69 | inline RefKind operator|(RefKind L, RefKind R) { |
70 | return static_cast<RefKind>(static_cast<uint8_t>(L) | |
71 | static_cast<uint8_t>(R)); |
72 | } |
73 | inline RefKind &operator|=(RefKind &L, RefKind R) { return L = L | R; } |
74 | inline RefKind operator&(RefKind A, RefKind B) { |
75 | return static_cast<RefKind>(static_cast<uint8_t>(A) & |
76 | static_cast<uint8_t>(B)); |
77 | } |
78 | |
79 | llvm::raw_ostream &operator<<(llvm::raw_ostream &, RefKind); |
80 | |
81 | /// Represents a symbol occurrence in the source file. |
82 | /// Despite the name, it could be a declaration/definition/reference. |
83 | /// |
84 | /// WARNING: Location does not own the underlying data - Copies are shallow. |
85 | struct Ref { |
86 | /// The source location where the symbol is named. |
87 | SymbolLocation Location; |
88 | RefKind Kind = RefKind::Unknown; |
89 | /// The ID of the symbol whose definition contains this reference. |
90 | /// For example, for a reference inside a function body, this would |
91 | /// be that function. For top-level definitions this isNull(). |
92 | SymbolID Container; |
93 | }; |
94 | |
95 | inline bool operator<(const Ref &L, const Ref &R) { |
96 | return std::tie(args: L.Location, args: L.Kind, args: L.Container) < |
97 | std::tie(args: R.Location, args: R.Kind, args: R.Container); |
98 | } |
99 | inline bool operator==(const Ref &L, const Ref &R) { |
100 | return std::tie(args: L.Location, args: L.Kind, args: L.Container) == |
101 | std::tie(args: R.Location, args: R.Kind, args: R.Container); |
102 | } |
103 | |
104 | llvm::raw_ostream &operator<<(llvm::raw_ostream &, const Ref &); |
105 | |
106 | /// An efficient structure of storing large set of symbol references in memory. |
107 | /// Filenames are deduplicated. |
108 | class RefSlab { |
109 | public: |
110 | // Refs are stored in order. |
111 | using value_type = std::pair<SymbolID, llvm::ArrayRef<Ref>>; |
112 | using const_iterator = std::vector<value_type>::const_iterator; |
113 | using iterator = const_iterator; |
114 | |
115 | RefSlab() = default; |
116 | RefSlab(RefSlab &&Slab) = default; |
117 | RefSlab &operator=(RefSlab &&RHS) = default; |
118 | |
119 | const_iterator begin() const { return Refs.begin(); } |
120 | const_iterator end() const { return Refs.end(); } |
121 | /// Gets the number of symbols. |
122 | size_t size() const { return Refs.size(); } |
123 | size_t numRefs() const { return NumRefs; } |
124 | bool empty() const { return Refs.empty(); } |
125 | |
126 | size_t bytes() const { |
127 | return sizeof(*this) + Arena.getTotalMemory() + |
128 | sizeof(value_type) * Refs.capacity(); |
129 | } |
130 | |
131 | /// RefSlab::Builder is a mutable container that can 'freeze' to RefSlab. |
132 | class Builder { |
133 | public: |
134 | Builder() : UniqueStrings(Arena) {} |
135 | /// Adds a ref to the slab. Deep copy: Strings will be owned by the slab. |
136 | void insert(const SymbolID &ID, const Ref &S); |
137 | /// Consumes the builder to finalize the slab. |
138 | RefSlab build() &&; |
139 | |
140 | private: |
141 | // A ref we're storing with its symbol to consume with build(). |
142 | // All strings are interned, so DenseMapInfo can use pointer comparisons. |
143 | struct Entry { |
144 | SymbolID Symbol; |
145 | Ref Reference; |
146 | }; |
147 | friend struct llvm::DenseMapInfo<Entry>; |
148 | |
149 | llvm::BumpPtrAllocator Arena; |
150 | llvm::UniqueStringSaver UniqueStrings; // Contents on the arena. |
151 | llvm::DenseSet<Entry> Entries; |
152 | }; |
153 | |
154 | private: |
155 | RefSlab(std::vector<value_type> Refs, llvm::BumpPtrAllocator Arena, |
156 | size_t NumRefs) |
157 | : Arena(std::move(Arena)), Refs(std::move(Refs)), NumRefs(NumRefs) {} |
158 | |
159 | llvm::BumpPtrAllocator Arena; |
160 | std::vector<value_type> Refs; |
161 | /// Number of all references. |
162 | size_t NumRefs = 0; |
163 | }; |
164 | |
165 | } // namespace clangd |
166 | } // namespace clang |
167 | |
168 | namespace llvm { |
169 | template <> struct DenseMapInfo<clang::clangd::RefSlab::Builder::Entry> { |
170 | using Entry = clang::clangd::RefSlab::Builder::Entry; |
171 | static inline Entry getEmptyKey() { |
172 | static Entry E{.Symbol: clang::clangd::SymbolID("" ), .Reference: {}}; |
173 | return E; |
174 | } |
175 | static inline Entry getTombstoneKey() { |
176 | static Entry E{.Symbol: clang::clangd::SymbolID("TOMBSTONE" ), .Reference: {}}; |
177 | return E; |
178 | } |
179 | static unsigned getHashValue(const Entry &Val) { |
180 | return llvm::hash_combine( |
181 | args: Val.Symbol, args: reinterpret_cast<uintptr_t>(Val.Reference.Location.FileURI), |
182 | args: Val.Reference.Location.Start.rep(), args: Val.Reference.Location.End.rep()); |
183 | } |
184 | static bool isEqual(const Entry &LHS, const Entry &RHS) { |
185 | return std::tie(args: LHS.Symbol, args: LHS.Reference.Location.FileURI, |
186 | args: LHS.Reference.Kind) == |
187 | std::tie(args: RHS.Symbol, args: RHS.Reference.Location.FileURI, |
188 | args: RHS.Reference.Kind) && |
189 | LHS.Reference.Location.Start == RHS.Reference.Location.Start && |
190 | LHS.Reference.Location.End == RHS.Reference.Location.End; |
191 | } |
192 | }; |
193 | } // namespace llvm |
194 | |
195 | #endif // LLVM_CLANG_TOOLS_EXTRA_CLANGD_INDEX_REF_H |
196 | |