Mercurial > repos > rliterman > csp2

comparison CSP2/CSP2_env/env-d9b9114564458d9d-741b3de822f2aaca6c6caa4325c4afce/include/kj/hash.h @ 69:33d812a61356

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 2e9511a184a1ca667c7be0c6321a36dc4e3d116d
author jpayne
date Tue, 18 Mar 2025 17:55:14 -0400
parents
children
comparison
equal deleted inserted replaced
67:0e9998148a16 69:33d812a61356
1 // Copyright (c) 2018 Kenton Varda and contributors
2 // Licensed under the MIT License:
3 //
4 // Permission is hereby granted, free of charge, to any person obtaining a copy
5 // of this software and associated documentation files (the "Software"), to deal
6 // in the Software without restriction, including without limitation the rights
7 // to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
8 // copies of the Software, and to permit persons to whom the Software is
9 // furnished to do so, subject to the following conditions:
10 //
11 // The above copyright notice and this permission notice shall be included in
12 // all copies or substantial portions of the Software.
13 //
14 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
15 // IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16 // FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
17 // AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
18 // LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
19 // OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
20 // THE SOFTWARE.
21
22 #pragma once
23
24 #include "string.h"
25
26 KJ_BEGIN_HEADER
27
28 namespace kj {
29 namespace _ { // private
30
31 struct HashCoder {
32 // This is a dummy type with only one instance: HASHCODER (below). To make an arbitrary type
33 // hashable, define `operator*(HashCoder, T)` to return any other type that is already hashable.
34 // Be sure to declare the operator in the same namespace as `T` **or** in the global scope.
35 // You can use the KJ_HASHCODE() macro as syntax sugar for this.
36 //
37 // A more usual way to accomplish what we're doing here would be to require that you define
38 // a function like `hashCode(T)` and then rely on argument-dependent lookup. However, this has
39 // the problem that it pollutes other people's namespaces and even the global namespace. For
40 // example, some other project may already have functions called `hashCode` which do something
41 // different. Declaring `operator*` with `HashCoder` as the left operand cannot conflict with
42 // anything.
43
44 uint operator*(ArrayPtr<const byte> s) const;
45 inline uint operator*(ArrayPtr<byte> s) const { return operator*(s.asConst()); }
46
47 inline uint operator*(ArrayPtr<const char> s) const { return operator*(s.asBytes()); }
48 inline uint operator*(ArrayPtr<char> s) const { return operator*(s.asBytes()); }
49 inline uint operator*(const Array<const char>& s) const { return operator*(s.asBytes()); }
50 inline uint operator*(const Array<char>& s) const { return operator*(s.asBytes()); }
51 inline uint operator*(const String& s) const { return operator*(s.asBytes()); }
52 inline uint operator*(const StringPtr& s) const { return operator*(s.asBytes()); }
53 inline uint operator*(const ConstString& s) const { return operator*(s.asBytes()); }
54
55 inline uint operator*(decltype(nullptr)) const { return 0; }
56 inline uint operator*(bool b) const { return b; }
57 inline uint operator*(char i) const { return i; }
58 inline uint operator*(signed char i) const { return i; }
59 inline uint operator*(unsigned char i) const { return i; }
60 inline uint operator*(signed short i) const { return i; }
61 inline uint operator*(unsigned short i) const { return i; }
62 inline uint operator*(signed int i) const { return i; }
63 inline uint operator*(unsigned int i) const { return i; }
64
65 inline uint operator*(signed long i) const {
66 if (sizeof(i) == sizeof(uint)) {
67 return operator*(static_cast<uint>(i));
68 } else {
69 return operator*(static_cast<unsigned long long>(i));
70 }
71 }
72 inline uint operator*(unsigned long i) const {
73 if (sizeof(i) == sizeof(uint)) {
74 return operator*(static_cast<uint>(i));
75 } else {
76 return operator*(static_cast<unsigned long long>(i));
77 }
78 }
79 inline uint operator*(signed long long i) const {
80 return operator*(static_cast<unsigned long long>(i));
81 }
82 inline uint operator*(unsigned long long i) const {
83 // Mix 64 bits to 32 bits in such a way that if our input values differ primarily in the upper
84 // 32 bits, we still get good diffusion. (I.e. we cannot just truncate!)
85 //
86 // 49123 is an arbitrarily-chosen prime that is vaguely close to 2^16.
87 //
88 // TODO(perf): I just made this up. Is it OK?
89 return static_cast<uint>(i) + static_cast<uint>(i >> 32) * 49123;
90 }
91
92 template <typename T>
93 uint operator*(T* ptr) const {
94 static_assert(!isSameType<Decay<T>, char>(), "Wrap in StringPtr if you want to hash string "
95 "contents. If you want to hash the pointer, cast to void*");
96 if (sizeof(ptr) == sizeof(uint)) {
97 // TODO(cleanup): In C++17, make the if() above be `if constexpr ()`, then change this to
98 // reinterpret_cast<uint>(ptr).
99 return reinterpret_cast<unsigned long long>(ptr);
100 } else {
101 return operator*(reinterpret_cast<unsigned long long>(ptr));
102 }
103 }
104
105 template <typename T, typename = decltype(instance<const HashCoder&>() * instance<const T&>())>
106 uint operator*(ArrayPtr<T> arr) const;
107 template <typename T, typename = decltype(instance<const HashCoder&>() * instance<const T&>())>
108 uint operator*(const Array<T>& arr) const;
109 template <typename T, typename = EnableIf<__is_enum(T)>>
110 inline uint operator*(T e) const;
111
112 template <typename T, typename Result = decltype(instance<T>().hashCode())>
113 inline Result operator*(T&& value) const { return kj::fwd<T>(value).hashCode(); }
114 };
115 static KJ_CONSTEXPR(const) HashCoder HASHCODER = HashCoder();
116
117 } // namespace _ (private)
118
119 #define KJ_HASHCODE(...) operator*(::kj::_::HashCoder, __VA_ARGS__)
120 // Defines a hash function for a custom type. Example:
121 //
122 // class Foo {...};
123 // inline uint KJ_HASHCODE(const Foo& foo) { return kj::hashCode(foo.x, foo.y); }
124 //
125 // This allows Foo to be passed to hashCode().
126 //
127 // The function should be declared either in the same namespace as the target type or in the global
128 // namespace. It can return any type which itself is hashable -- that value will be hashed in turn
129 // until a `uint` comes out.
130
131 inline uint hashCode(uint value) { return value; }
132 template <typename T>
133 inline uint hashCode(T&& value) { return hashCode(_::HASHCODER * kj::fwd<T>(value)); }
134 template <typename T, size_t N>
135 inline uint hashCode(T (&arr)[N]) {
136 static_assert(!isSameType<Decay<T>, char>(), "Wrap in StringPtr if you want to hash string "
137 "contents. If you want to hash the pointer, cast to void*");
138 static_assert(isSameType<Decay<T>, char>(), "Wrap in ArrayPtr if you want to hash a C array. "
139 "If you want to hash the pointer, cast to void*");
140 return 0;
141 }
142 template <typename... T>
143 inline uint hashCode(T&&... values) {
144 uint hashes[] = { hashCode(kj::fwd<T>(values))... };
145 return hashCode(kj::ArrayPtr<uint>(hashes).asBytes());
146 }
147 // kj::hashCode() is a universal hashing function, like kj::str() is a universal stringification
148 // function. Throw stuff in, get a hash code.
149 //
150 // Hash codes may differ between different processes, even running exactly the same code.
151 //
152 // NOT SUITABLE FOR CRYPTOGRAPHY. This is for hash tables, not crypto.
153
154 // =======================================================================================
155 // inline implementation details
156
157 namespace _ { // private
158
159 template <typename T, typename>
160 inline uint HashCoder::operator*(ArrayPtr<T> arr) const {
161 // Hash each array element to create a string of hashes, then murmur2 over those.
162 //
163 // TODO(perf): Choose a more-modern hash. (See hash.c++.)
164
165 constexpr uint m = 0x5bd1e995;
166 constexpr uint r = 24;
167 uint h = arr.size() * sizeof(uint);
168
169 for (auto& e: arr) {
170 uint k = kj::hashCode(e);
171 k *= m;
172 k ^= k >> r;
173 k *= m;
174 h *= m;
175 h ^= k;
176 }
177
178 h ^= h >> 13;
179 h *= m;
180 h ^= h >> 15;
181 return h;
182 }
183 template <typename T, typename>
184 inline uint HashCoder::operator*(const Array<T>& arr) const {
185 return operator*(arr.asPtr());
186 }
187
188 template <typename T, typename>
189 inline uint HashCoder::operator*(T e) const {
190 return operator*(static_cast<__underlying_type(T)>(e));
191 }
192
193 } // namespace _ (private)
194 } // namespace kj
195
196 KJ_END_HEADER