hash_tables.h revision eb525c5499e34cc9c4b825d6d9e75bb07cc06ace 
1// Copyright (c) 2011 The Chromium Authors. All rights reserved.
2// Use of this source code is governed by a BSD-style license that can be
3// found in the LICENSE file.
4//
5
6//
7// Deal with the differences between Microsoft and GNU implemenations
8// of hash_map. Allows all platforms to use |base::hash_map| and
9// |base::hash_set|.
10//  eg:
11//   base::hash_map<int> my_map;
12//   base::hash_set<int> my_set;
13//
14// NOTE: It is an explicit non-goal of this class to provide a generic hash
15// function for pointers.  If you want to hash a pointers to a particular class,
16// please define the template specialization elsewhere (for example, in its
17// header file) and keep it specific to just pointers to that class.  This is
18// because identity hashes are not desirable for all types that might show up
19// in containers as pointers.
20
21#ifndef BASE_CONTAINERS_HASH_TABLES_H_
22#define BASE_CONTAINERS_HASH_TABLES_H_
23
24#include <utility>
25
26#include "base/basictypes.h"
27#include "base/strings/string16.h"
28#include "build/build_config.h"
29
30#if defined(COMPILER_MSVC)
31#include <hash_map>
32#include <hash_set>
33
34#define BASE_HASH_NAMESPACE stdext
35
36#elif defined(COMPILER_GCC)
37#if defined(OS_ANDROID)
38#define BASE_HASH_NAMESPACE std
39#else
40#define BASE_HASH_NAMESPACE __gnu_cxx
41#endif
42
43// This is a hack to disable the gcc 4.4 warning about hash_map and hash_set
44// being deprecated.  We can get rid of this when we upgrade to VS2008 and we
45// can use <tr1/unordered_map> and <tr1/unordered_set>.
46#ifdef __DEPRECATED
47#define CHROME_OLD__DEPRECATED __DEPRECATED
48#undef __DEPRECATED
49#endif
50
51#if defined(OS_ANDROID)
52#include <hash_map>
53#include <hash_set>
54#else
55#include <ext/hash_map>
56#include <ext/hash_set>
57#endif
58
59#include <string>
60
61#ifdef CHROME_OLD__DEPRECATED
62#define __DEPRECATED CHROME_OLD__DEPRECATED
63#undef CHROME_OLD__DEPRECATED
64#endif
65
66namespace BASE_HASH_NAMESPACE {
67
68#if !defined(OS_ANDROID)
69// The GNU C++ library provides identity hash functions for many integral types,
70// but not for |long long|.  This hash function will truncate if |size_t| is
71// narrower than |long long|.  This is probably good enough for what we will
72// use it for.
73
74#define DEFINE_TRIVIAL_HASH(integral_type) \
75    template<> \
76    struct hash<integral_type> { \
77      std::size_t operator()(integral_type value) const { \
78        return static_cast<std::size_t>(value); \
79      } \
80    }
81
82DEFINE_TRIVIAL_HASH(long long);
83DEFINE_TRIVIAL_HASH(unsigned long long);
84
85#undef DEFINE_TRIVIAL_HASH
86#endif  // !defined(OS_ANDROID)
87
88// Implement string hash functions so that strings of various flavors can
89// be used as keys in STL maps and sets.  The hash algorithm comes from the
90// GNU C++ library, in <tr1/functional>.  It is duplicated here because GCC
91// versions prior to 4.3.2 are unable to compile <tr1/functional> when RTTI
92// is disabled, as it is in our build.
93
94#define DEFINE_STRING_HASH(string_type) \
95    template<> \
96    struct hash<string_type> { \
97      std::size_t operator()(const string_type& s) const { \
98        std::size_t result = 0; \
99        for (string_type::const_iterator i = s.begin(); i != s.end(); ++i) \
100          result = (result * 131) + *i; \
101        return result; \
102      } \
103    }
104
105DEFINE_STRING_HASH(std::string);
106DEFINE_STRING_HASH(string16);
107
108#undef DEFINE_STRING_HASH
109
110}  // namespace BASE_HASH_NAMESPACE
111
112#else  // COMPILER
113#error define BASE_HASH_NAMESPACE for your compiler
114#endif  // COMPILER
115
116namespace base {
117using BASE_HASH_NAMESPACE::hash_map;
118using BASE_HASH_NAMESPACE::hash_multimap;
119using BASE_HASH_NAMESPACE::hash_multiset;
120using BASE_HASH_NAMESPACE::hash_set;
121}
122
123
124// Implement methods for hashing a pair of integers, so they can be used as
125// keys in STL containers.
126
127#if defined(COMPILER_MSVC)
128
129#define DEFINE_PAIR_HASH_FUNCTION_START(type1, type2) \
130    template<> \
131    inline std::size_t hash_value<std::pair<type1, type2> >( \
132        const std::pair<type1, type2>& value)
133
134#define DEFINE_PAIR_HASH_FUNCTION_END()
135
136#elif defined(COMPILER_GCC)
137
138#define DEFINE_PAIR_HASH_FUNCTION_START(type1, type2) \
139    template<> \
140    struct hash<std::pair<type1, type2> > { \
141      std::size_t operator()(std::pair<type1, type2> value) const
142
143#define DEFINE_PAIR_HASH_FUNCTION_END() \
144    };
145
146#else
147#error define DEFINE_PAIR_HASH_FUNCTION_START for your compiler
148#endif  // COMPILER
149
150namespace BASE_HASH_NAMESPACE {
151
152// Implement hashing for pairs of at-most 32 bit integer values.
153// When size_t is 32 bits, we turn the 64-bit hash code into 32 bits by using
154// multiply-add hashing. This algorithm, as described in
155// Theorem 4.3.3 of the thesis "Über die Komplexität der Multiplikation in
156// eingeschränkten Branchingprogrammmodellen" by Woelfel, is:
157//
158//   h32(x32, y32) = (h64(x32, y32) * rand_odd64 + rand16 * 2^16) % 2^64 / 2^32
159//
160// Contact danakj@chromium.org for any questions.
161#define DEFINE_32BIT_PAIR_HASH(type1, type2) \
162    DEFINE_PAIR_HASH_FUNCTION_START(type1, type2) { \
163      uint64 first = value.first; \
164      uint32 second = value.second; \
165      uint64 hash64 = (first << 32) | second; \
166      \
167      if (sizeof(std::size_t) >= sizeof(uint64)) \
168        return static_cast<std::size_t>(hash64); \
169      \
170      uint64 odd_random = 481046412LL << 32 | 1025306954LL; \
171      uint32 shift_random = 10121U << 16; \
172      \
173      hash64 = hash64 * odd_random + shift_random; \
174      std::size_t high_bits = static_cast<std::size_t>( \
175          hash64 >> (sizeof(uint64) - sizeof(std::size_t))); \
176      return high_bits; \
177    } \
178  DEFINE_PAIR_HASH_FUNCTION_END();
179
180DEFINE_32BIT_PAIR_HASH(int16, int16);
181DEFINE_32BIT_PAIR_HASH(int16, uint16);
182DEFINE_32BIT_PAIR_HASH(int16, int32);
183DEFINE_32BIT_PAIR_HASH(int16, uint32);
184DEFINE_32BIT_PAIR_HASH(uint16, int16);
185DEFINE_32BIT_PAIR_HASH(uint16, uint16);
186DEFINE_32BIT_PAIR_HASH(uint16, int32);
187DEFINE_32BIT_PAIR_HASH(uint16, uint32);
188DEFINE_32BIT_PAIR_HASH(int32, int16);
189DEFINE_32BIT_PAIR_HASH(int32, uint16);
190DEFINE_32BIT_PAIR_HASH(int32, int32);
191DEFINE_32BIT_PAIR_HASH(int32, uint32);
192DEFINE_32BIT_PAIR_HASH(uint32, int16);
193DEFINE_32BIT_PAIR_HASH(uint32, uint16);
194DEFINE_32BIT_PAIR_HASH(uint32, int32);
195DEFINE_32BIT_PAIR_HASH(uint32, uint32);
196
197#undef DEFINE_32BIT_PAIR_HASH
198
199// Implement hashing for pairs of up-to 64-bit integer values.
200// We use the compound integer hash method to produce a 64-bit hash code, by
201// breaking the two 64-bit inputs into 4 32-bit values:
202// http://opendatastructures.org/versions/edition-0.1d/ods-java/node33.html#SECTION00832000000000000000
203// Then we reduce our result to 32 bits if required, similar to above.
204#define DEFINE_64BIT_PAIR_HASH(type1, type2) \
205    DEFINE_PAIR_HASH_FUNCTION_START(type1, type2) { \
206      uint32 short_random1 = 842304669U; \
207      uint32 short_random2 = 619063811U; \
208      uint32 short_random3 = 937041849U; \
209      uint32 short_random4 = 3309708029U; \
210      \
211      uint64 value1 = value.first; \
212      uint64 value2 = value.second; \
213      uint32 value1a = static_cast<uint32>(value1 & 0xffffffff); \
214      uint32 value1b = static_cast<uint32>((value1 >> 32) & 0xffffffff); \
215      uint32 value2a = static_cast<uint32>(value2 & 0xffffffff); \
216      uint32 value2b = static_cast<uint32>((value2 >> 32) & 0xffffffff); \
217      \
218      uint64 product1 = static_cast<uint64>(value1a) * short_random1; \
219      uint64 product2 = static_cast<uint64>(value1b) * short_random2; \
220      uint64 product3 = static_cast<uint64>(value2a) * short_random3; \
221      uint64 product4 = static_cast<uint64>(value2b) * short_random4; \
222      \
223      uint64 hash64 = product1 + product2 + product3 + product4; \
224      \
225      if (sizeof(std::size_t) >= sizeof(uint64)) \
226        return static_cast<std::size_t>(hash64); \
227      \
228      uint64 odd_random = 1578233944LL << 32 | 194370989LL; \
229      uint32 shift_random = 20591U << 16; \
230      \
231      hash64 = hash64 * odd_random + shift_random; \
232      std::size_t high_bits = static_cast<std::size_t>( \
233          hash64 >> (sizeof(uint64) - sizeof(std::size_t))); \
234      return high_bits; \
235    } \
236  DEFINE_PAIR_HASH_FUNCTION_END();
237
238DEFINE_64BIT_PAIR_HASH(int16, int64);
239DEFINE_64BIT_PAIR_HASH(int16, uint64);
240DEFINE_64BIT_PAIR_HASH(uint16, int64);
241DEFINE_64BIT_PAIR_HASH(uint16, uint64);
242DEFINE_64BIT_PAIR_HASH(int32, int64);
243DEFINE_64BIT_PAIR_HASH(int32, uint64);
244DEFINE_64BIT_PAIR_HASH(uint32, int64);
245DEFINE_64BIT_PAIR_HASH(uint32, uint64);
246DEFINE_64BIT_PAIR_HASH(int64, int16);
247DEFINE_64BIT_PAIR_HASH(int64, uint16);
248DEFINE_64BIT_PAIR_HASH(int64, int32);
249DEFINE_64BIT_PAIR_HASH(int64, uint32);
250DEFINE_64BIT_PAIR_HASH(int64, int64);
251DEFINE_64BIT_PAIR_HASH(int64, uint64);
252DEFINE_64BIT_PAIR_HASH(uint64, int16);
253DEFINE_64BIT_PAIR_HASH(uint64, uint16);
254DEFINE_64BIT_PAIR_HASH(uint64, int32);
255DEFINE_64BIT_PAIR_HASH(uint64, uint32);
256DEFINE_64BIT_PAIR_HASH(uint64, int64);
257DEFINE_64BIT_PAIR_HASH(uint64, uint64);
258
259#undef DEFINE_64BIT_PAIR_HASH
260}
261
262#undef DEFINE_PAIR_HASH_FUNCTION_START
263#undef DEFINE_PAIR_HASH_FUNCTION_END
264
265#endif  // BASE_CONTAINERS_HASH_TABLES_H_
266