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