base/containers/hash_tables.h

   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 <unordered_map>
  32 #include <unordered_set>
  33
  34 #define BASE_HASH_NAMESPACE std
  35
  36 #elif defined(COMPILER_GCC)
  37
  38 #define BASE_HASH_NAMESPACE base_hash
  39
  40 // This is a hack to disable the gcc 4.4 warning about hash_map and hash_set
  41 // being deprecated.  We can get rid of this when we upgrade to VS2008 and we
  42 // can use <tr1/unordered_map> and <tr1/unordered_set>.
  43 #ifdef __DEPRECATED
  44 #define CHROME_OLD__DEPRECATED __DEPRECATED
  45 #undef __DEPRECATED
  46 #endif
  47
  48 #if defined(OS_ANDROID)
  49 #include <hash_map>
  50 #include <hash_set>
  51 #define BASE_HASH_IMPL_NAMESPACE std
  52 #else
  53 #include <ext/hash_map>
  54 #include <ext/hash_set>
  55 #define BASE_HASH_IMPL_NAMESPACE __gnu_cxx
  56 #endif
  57
  58 #include <string>
  59
  60 #ifdef CHROME_OLD__DEPRECATED
  61 #define __DEPRECATED CHROME_OLD__DEPRECATED
  62 #undef CHROME_OLD__DEPRECATED
  63 #endif
  64
  65 namespace BASE_HASH_NAMESPACE {
  66
  67 // The pre-standard hash behaves like C++11's std::hash, except around pointers.
  68 // const char* is specialized to hash the C string and hash functions for
  69 // general T* are missing. Define a BASE_HASH_NAMESPACE::hash which aligns with
  70 // the C++11 behavior.
  71
  72 template<typename T>
  73 struct hash {
  74   std::size_t operator()(const T& value) const {
  75     return BASE_HASH_IMPL_NAMESPACE::hash<T>()(value);
  76   }
  77 };
  78
  79 template<typename T>
  80 struct hash<T*> {
  81   std::size_t operator()(T* value) const {
  82     return BASE_HASH_IMPL_NAMESPACE::hash<uintptr_t>()(
  83         reinterpret_cast<uintptr_t>(value));
  84   }
  85 };
  86
  87 #if !defined(OS_ANDROID)
  88 // The GNU C++ library provides identity hash functions for many integral types,
  89 // but not for |long long|.  This hash function will truncate if |size_t| is
  90 // narrower than |long long|.  This is probably good enough for what we will
  91 // use it for.
  92
  93 #define DEFINE_TRIVIAL_HASH(integral_type) \
  94     template<> \
  95     struct hash<integral_type> { \
  96       std::size_t operator()(integral_type value) const { \
  97         return static_cast<std::size_t>(value); \
  98       } \
  99     }
 100
 101 DEFINE_TRIVIAL_HASH(long long);
 102 DEFINE_TRIVIAL_HASH(unsigned long long);
 103
 104 #undef DEFINE_TRIVIAL_HASH
 105 #endif  // !defined(OS_ANDROID)
 106
 107 // Implement string hash functions so that strings of various flavors can
 108 // be used as keys in STL maps and sets.  The hash algorithm comes from the
 109 // GNU C++ library, in <tr1/functional>.  It is duplicated here because GCC
 110 // versions prior to 4.3.2 are unable to compile <tr1/functional> when RTTI
 111 // is disabled, as it is in our build.
 112
 113 #define DEFINE_STRING_HASH(string_type) \
 114     template<> \
 115     struct hash<string_type> { \
 116       std::size_t operator()(const string_type& s) const { \
 117         std::size_t result = 0; \
 118         for (string_type::const_iterator i = s.begin(); i != s.end(); ++i) \
 119           result = (result * 131) + *i; \
 120         return result; \
 121       } \
 122     }
 123
 124 DEFINE_STRING_HASH(std::string);
 125 DEFINE_STRING_HASH(base::string16);
 126
 127 #undef DEFINE_STRING_HASH
 128
 129 }  // namespace BASE_HASH_NAMESPACE
 130
 131 #else  // COMPILER
 132 #error define BASE_HASH_NAMESPACE for your compiler
 133 #endif  // COMPILER
 134
 135 namespace base {
 136
 137 // On MSVC, use the C++11 containers.
 138 #if defined(COMPILER_MSVC)
 139
 140 template<class Key, class T,
 141          class Hash = std::hash<Key>,
 142          class Pred = std::equal_to<Key>,
 143          class Alloc = std::allocator<std::pair<const Key, T>>>
 144 using hash_map = std::unordered_map<Key, T, Hash, Pred, Alloc>;
 145
 146 template<class Key, class T,
 147          class Hash = std::hash<Key>,
 148          class Pred = std::equal_to<Key>,
 149          class Alloc = std::allocator<std::pair<const Key, T>>>
 150 using hash_multimap = std::unordered_multimap<Key, T, Hash, Pred, Alloc>;
 151
 152 template<class Key,
 153          class Hash = std::hash<Key>,
 154          class Pred = std::equal_to<Key>,
 155          class Alloc = std::allocator<Key>>
 156 using hash_multiset = std::unordered_multiset<Key, Hash, Pred, Alloc>;
 157
 158 template<class Key,
 159          class Hash = std::hash<Key>,
 160          class Pred = std::equal_to<Key>,
 161          class Alloc = std::allocator<Key>>
 162 using hash_set = std::unordered_set<Key, Hash, Pred, Alloc>;
 163
 164 #else  // !COMPILER_MSVC
 165
 166 // Otherwise, use the pre-standard ones, but override the default hash to match
 167 // C++11.
 168 template<class Key, class T,
 169          class Hash = BASE_HASH_NAMESPACE::hash<Key>,
 170          class Pred = std::equal_to<Key>,
 171          class Alloc = std::allocator<std::pair<const Key, T>>>
 172 using hash_map = BASE_HASH_IMPL_NAMESPACE::hash_map<Key, T, Hash, Pred, Alloc>;
 173
 174 template<class Key, class T,
 175          class Hash = BASE_HASH_NAMESPACE::hash<Key>,
 176          class Pred = std::equal_to<Key>,
 177          class Alloc = std::allocator<std::pair<const Key, T>>>
 178 using hash_multimap =
 179     BASE_HASH_IMPL_NAMESPACE::hash_multimap<Key, T, Hash, Pred, Alloc>;
 180
 181 template<class Key,
 182          class Hash = BASE_HASH_NAMESPACE::hash<Key>,
 183          class Pred = std::equal_to<Key>,
 184          class Alloc = std::allocator<Key>>
 185 using hash_multiset =
 186     BASE_HASH_IMPL_NAMESPACE::hash_multiset<Key, Hash, Pred, Alloc>;
 187
 188 template<class Key,
 189          class Hash = BASE_HASH_NAMESPACE::hash<Key>,
 190          class Pred = std::equal_to<Key>,
 191          class Alloc = std::allocator<Key>>
 192 using hash_set = BASE_HASH_IMPL_NAMESPACE::hash_set<Key, Hash, Pred, Alloc>;
 193
 194 #undef BASE_HASH_IMPL_NAMESPACE
 195
 196 #endif  // COMPILER_MSVC
 197
 198 // Implement hashing for pairs of at-most 32 bit integer values.
 199 // When size_t is 32 bits, we turn the 64-bit hash code into 32 bits by using
 200 // multiply-add hashing. This algorithm, as described in
 201 // Theorem 4.3.3 of the thesis "Über die Komplexität der Multiplikation in
 202 // eingeschränkten Branchingprogrammmodellen" by Woelfel, is:
 203 //
 204 //   h32(x32, y32) = (h64(x32, y32) * rand_odd64 + rand16 * 2^16) % 2^64 / 2^32
 205 //
 206 // Contact danakj@chromium.org for any questions.
 207 inline std::size_t HashInts32(uint32 value1, uint32 value2) {
 208   uint64 value1_64 = value1;
 209   uint64 hash64 = (value1_64 << 32) | value2;
 210
 211   if (sizeof(std::size_t) >= sizeof(uint64))
 212     return static_cast<std::size_t>(hash64);
 213
 214   uint64 odd_random = 481046412LL << 32 | 1025306955LL;
 215   uint32 shift_random = 10121U << 16;
 216
 217   hash64 = hash64 * odd_random + shift_random;
 218   std::size_t high_bits = static_cast<std::size_t>(
 219       hash64 >> (8 * (sizeof(uint64) - sizeof(std::size_t))));
 220   return high_bits;
 221 }
 222
 223 // Implement hashing for pairs of up-to 64-bit integer values.
 224 // We use the compound integer hash method to produce a 64-bit hash code, by
 225 // breaking the two 64-bit inputs into 4 32-bit values:
 226 // http://opendatastructures.org/versions/edition-0.1d/ods-java/node33.html#SECTION00832000000000000000
 227 // Then we reduce our result to 32 bits if required, similar to above.
 228 inline std::size_t HashInts64(uint64 value1, uint64 value2) {
 229   uint32 short_random1 = 842304669U;
 230   uint32 short_random2 = 619063811U;
 231   uint32 short_random3 = 937041849U;
 232   uint32 short_random4 = 3309708029U;
 233
 234   uint32 value1a = static_cast<uint32>(value1 & 0xffffffff);
 235   uint32 value1b = static_cast<uint32>((value1 >> 32) & 0xffffffff);
 236   uint32 value2a = static_cast<uint32>(value2 & 0xffffffff);
 237   uint32 value2b = static_cast<uint32>((value2 >> 32) & 0xffffffff);
 238
 239   uint64 product1 = static_cast<uint64>(value1a) * short_random1;
 240   uint64 product2 = static_cast<uint64>(value1b) * short_random2;
 241   uint64 product3 = static_cast<uint64>(value2a) * short_random3;
 242   uint64 product4 = static_cast<uint64>(value2b) * short_random4;
 243
 244   uint64 hash64 = product1 + product2 + product3 + product4;
 245
 246   if (sizeof(std::size_t) >= sizeof(uint64))
 247     return static_cast<std::size_t>(hash64);
 248
 249   uint64 odd_random = 1578233944LL << 32 | 194370989LL;
 250   uint32 shift_random = 20591U << 16;
 251
 252   hash64 = hash64 * odd_random + shift_random;
 253   std::size_t high_bits = static_cast<std::size_t>(
 254       hash64 >> (8 * (sizeof(uint64) - sizeof(std::size_t))));
 255   return high_bits;
 256 }
 257
 258 #define DEFINE_32BIT_PAIR_HASH(Type1, Type2) \
 259 inline std::size_t HashPair(Type1 value1, Type2 value2) { \
 260   return HashInts32(value1, value2); \
 261 }
 262
 263 DEFINE_32BIT_PAIR_HASH(int16, int16);
 264 DEFINE_32BIT_PAIR_HASH(int16, uint16);
 265 DEFINE_32BIT_PAIR_HASH(int16, int32);
 266 DEFINE_32BIT_PAIR_HASH(int16, uint32);
 267 DEFINE_32BIT_PAIR_HASH(uint16, int16);
 268 DEFINE_32BIT_PAIR_HASH(uint16, uint16);
 269 DEFINE_32BIT_PAIR_HASH(uint16, int32);
 270 DEFINE_32BIT_PAIR_HASH(uint16, uint32);
 271 DEFINE_32BIT_PAIR_HASH(int32, int16);
 272 DEFINE_32BIT_PAIR_HASH(int32, uint16);
 273 DEFINE_32BIT_PAIR_HASH(int32, int32);
 274 DEFINE_32BIT_PAIR_HASH(int32, uint32);
 275 DEFINE_32BIT_PAIR_HASH(uint32, int16);
 276 DEFINE_32BIT_PAIR_HASH(uint32, uint16);
 277 DEFINE_32BIT_PAIR_HASH(uint32, int32);
 278 DEFINE_32BIT_PAIR_HASH(uint32, uint32);
 279
 280 #undef DEFINE_32BIT_PAIR_HASH
 281
 282 #define DEFINE_64BIT_PAIR_HASH(Type1, Type2) \
 283 inline std::size_t HashPair(Type1 value1, Type2 value2) { \
 284   return HashInts64(value1, value2); \
 285 }
 286
 287 DEFINE_64BIT_PAIR_HASH(int16, int64);
 288 DEFINE_64BIT_PAIR_HASH(int16, uint64);
 289 DEFINE_64BIT_PAIR_HASH(uint16, int64);
 290 DEFINE_64BIT_PAIR_HASH(uint16, uint64);
 291 DEFINE_64BIT_PAIR_HASH(int32, int64);
 292 DEFINE_64BIT_PAIR_HASH(int32, uint64);
 293 DEFINE_64BIT_PAIR_HASH(uint32, int64);
 294 DEFINE_64BIT_PAIR_HASH(uint32, uint64);
 295 DEFINE_64BIT_PAIR_HASH(int64, int16);
 296 DEFINE_64BIT_PAIR_HASH(int64, uint16);
 297 DEFINE_64BIT_PAIR_HASH(int64, int32);
 298 DEFINE_64BIT_PAIR_HASH(int64, uint32);
 299 DEFINE_64BIT_PAIR_HASH(int64, int64);
 300 DEFINE_64BIT_PAIR_HASH(int64, uint64);
 301 DEFINE_64BIT_PAIR_HASH(uint64, int16);
 302 DEFINE_64BIT_PAIR_HASH(uint64, uint16);
 303 DEFINE_64BIT_PAIR_HASH(uint64, int32);
 304 DEFINE_64BIT_PAIR_HASH(uint64, uint32);
 305 DEFINE_64BIT_PAIR_HASH(uint64, int64);
 306 DEFINE_64BIT_PAIR_HASH(uint64, uint64);
 307
 308 #undef DEFINE_64BIT_PAIR_HASH
 309 }  // namespace base
 310
 311 namespace BASE_HASH_NAMESPACE {
 312
 313 // Implement methods for hashing a pair of integers, so they can be used as
 314 // keys in STL containers.
 315
 316 template<typename Type1, typename Type2>
 317 struct hash<std::pair<Type1, Type2> > {
 318   std::size_t operator()(std::pair<Type1, Type2> value) const {
 319     return base::HashPair(value.first, value.second);
 320   }
 321 };
 322
 323 }
 324
 325 #undef DEFINE_PAIR_HASH_FUNCTION_START
 326 #undef DEFINE_PAIR_HASH_FUNCTION_END
 327
 328 #endif  // BASE_CONTAINERS_HASH_TABLES_H_