• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Copyright (C) 2005, 2006, 2008, 2010, 2013 Apple Inc. All rights reserved.
3  * Copyright (C) 2010 Patrick Gansterer <paroga@paroga.com>
4  *
5  * This library is free software; you can redistribute it and/or
6  * modify it under the terms of the GNU Library General Public
7  * License as published by the Free Software Foundation; either
8  * version 2 of the License, or (at your option) any later version.
9  *
10  * This library is distributed in the hope that it will be useful,
11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
13  * Library General Public License for more details.
14  *
15  * You should have received a copy of the GNU Library General Public License
16  * along with this library; see the file COPYING.LIB.  If not, write to
17  * the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
18  * Boston, MA 02110-1301, USA.
19  *
20  */
21 
22 #ifndef WTF_StringHasher_h
23 #define WTF_StringHasher_h
24 
25 #include "wtf/unicode/Unicode.h"
26 
27 namespace WTF {
28 
29 // Paul Hsieh's SuperFastHash
30 // http://www.azillionmonkeys.com/qed/hash.html
31 
32 // LChar data is interpreted as Latin-1-encoded (zero extended to 16 bits).
33 
34 // NOTE: The hash computation here must stay in sync with the create_hash_table script in
35 // JavaScriptCore and the CodeGeneratorJS.pm script in WebCore.
36 
37 // Golden ratio. Arbitrary start value to avoid mapping all zeros to a hash value of zero.
38 static const unsigned stringHashingStartValue = 0x9E3779B9U;
39 
40 class StringHasher {
41 public:
42     static const unsigned flagCount = 8; // Save 8 bits for StringImpl to use as flags.
43 
StringHasher()44     StringHasher()
45         : m_hash(stringHashingStartValue)
46         , m_hasPendingCharacter(false)
47         , m_pendingCharacter(0)
48     {
49     }
50 
51     // The hasher hashes two characters at a time, and thus an "aligned" hasher is one
52     // where an even number of characters have been added. Callers that always add
53     // characters two at a time can use the "assuming aligned" functions.
addCharactersAssumingAligned(UChar a,UChar b)54     void addCharactersAssumingAligned(UChar a, UChar b)
55     {
56         ASSERT(!m_hasPendingCharacter);
57         m_hash += a;
58         m_hash = (m_hash << 16) ^ ((b << 11) ^ m_hash);
59         m_hash += m_hash >> 11;
60     }
61 
addCharacter(UChar character)62     void addCharacter(UChar character)
63     {
64         if (m_hasPendingCharacter) {
65             m_hasPendingCharacter = false;
66             addCharactersAssumingAligned(m_pendingCharacter, character);
67             return;
68         }
69 
70         m_pendingCharacter = character;
71         m_hasPendingCharacter = true;
72     }
73 
addCharacters(UChar a,UChar b)74     void addCharacters(UChar a, UChar b)
75     {
76         if (m_hasPendingCharacter) {
77 #if ENABLE(ASSERT)
78             m_hasPendingCharacter = false;
79 #endif
80             addCharactersAssumingAligned(m_pendingCharacter, a);
81             m_pendingCharacter = b;
82 #if ENABLE(ASSERT)
83             m_hasPendingCharacter = true;
84 #endif
85             return;
86         }
87 
88         addCharactersAssumingAligned(a, b);
89     }
90 
addCharactersAssumingAligned(const T * data,unsigned length)91     template<typename T, UChar Converter(T)> void addCharactersAssumingAligned(const T* data, unsigned length)
92     {
93         ASSERT(!m_hasPendingCharacter);
94 
95         bool remainder = length & 1;
96         length >>= 1;
97 
98         while (length--) {
99             addCharactersAssumingAligned(Converter(data[0]), Converter(data[1]));
100             data += 2;
101         }
102 
103         if (remainder)
104             addCharacter(Converter(*data));
105     }
106 
addCharactersAssumingAligned(const T * data,unsigned length)107     template<typename T> void addCharactersAssumingAligned(const T* data, unsigned length)
108     {
109         addCharactersAssumingAligned<T, defaultConverter>(data, length);
110     }
111 
addCharacters(const T * data,unsigned length)112     template<typename T, UChar Converter(T)> void addCharacters(const T* data, unsigned length)
113     {
114         if (m_hasPendingCharacter && length) {
115             m_hasPendingCharacter = false;
116             addCharactersAssumingAligned(m_pendingCharacter, Converter(*data++));
117             --length;
118         }
119         addCharactersAssumingAligned<T, Converter>(data, length);
120     }
121 
addCharacters(const T * data,unsigned length)122     template<typename T> void addCharacters(const T* data, unsigned length)
123     {
124         addCharacters<T, defaultConverter>(data, length);
125     }
126 
hashWithTop8BitsMasked()127     unsigned hashWithTop8BitsMasked() const
128     {
129         unsigned result = avalancheBits();
130 
131         // Reserving space from the high bits for flags preserves most of the hash's
132         // value, since hash lookup typically masks out the high bits anyway.
133         result &= (1U << (sizeof(result) * 8 - flagCount)) - 1;
134 
135         // This avoids ever returning a hash code of 0, since that is used to
136         // signal "hash not computed yet". Setting the high bit maintains
137         // reasonable fidelity to a hash code of 0 because it is likely to yield
138         // exactly 0 when hash lookup masks out the high bits.
139         if (!result)
140             result = 0x80000000 >> flagCount;
141 
142         return result;
143     }
144 
hash()145     unsigned hash() const
146     {
147         unsigned result = avalancheBits();
148 
149         // This avoids ever returning a hash code of 0, since that is used to
150         // signal "hash not computed yet". Setting the high bit maintains
151         // reasonable fidelity to a hash code of 0 because it is likely to yield
152         // exactly 0 when hash lookup masks out the high bits.
153         if (!result)
154             result = 0x80000000;
155 
156         return result;
157     }
158 
computeHashAndMaskTop8Bits(const T * data,unsigned length)159     template<typename T, UChar Converter(T)> static unsigned computeHashAndMaskTop8Bits(const T* data, unsigned length)
160     {
161         StringHasher hasher;
162         hasher.addCharactersAssumingAligned<T, Converter>(data, length);
163         return hasher.hashWithTop8BitsMasked();
164     }
165 
computeHashAndMaskTop8Bits(const T * data,unsigned length)166     template<typename T> static unsigned computeHashAndMaskTop8Bits(const T* data, unsigned length)
167     {
168         return computeHashAndMaskTop8Bits<T, defaultConverter>(data, length);
169     }
170 
computeHash(const T * data,unsigned length)171     template<typename T, UChar Converter(T)> static unsigned computeHash(const T* data, unsigned length)
172     {
173         StringHasher hasher;
174         hasher.addCharactersAssumingAligned<T, Converter>(data, length);
175         return hasher.hash();
176     }
177 
computeHash(const T * data,unsigned length)178     template<typename T> static unsigned computeHash(const T* data, unsigned length)
179     {
180         return computeHash<T, defaultConverter>(data, length);
181     }
182 
hashMemory(const void * data,unsigned length)183     static unsigned hashMemory(const void* data, unsigned length)
184     {
185         // FIXME: Why does this function use the version of the hash that drops the top 8 bits?
186         // We want that for all string hashing so we can use those bits in StringImpl and hash
187         // strings consistently, but I don't see why we'd want that for general memory hashing.
188         ASSERT(!(length % 2));
189         return computeHashAndMaskTop8Bits<UChar>(static_cast<const UChar*>(data), length / sizeof(UChar));
190     }
191 
hashMemory(const void * data)192     template<size_t length> static unsigned hashMemory(const void* data)
193     {
194         COMPILE_ASSERT(!(length % 2), length_must_be_a_multiple_of_two);
195         return hashMemory(data, length);
196     }
197 
198 private:
defaultConverter(UChar character)199     static UChar defaultConverter(UChar character)
200     {
201         return character;
202     }
203 
defaultConverter(LChar character)204     static UChar defaultConverter(LChar character)
205     {
206         return character;
207     }
208 
avalancheBits()209     unsigned avalancheBits() const
210     {
211         unsigned result = m_hash;
212 
213         // Handle end case.
214         if (m_hasPendingCharacter) {
215             result += m_pendingCharacter;
216             result ^= result << 11;
217             result += result >> 17;
218         }
219 
220         // Force "avalanching" of final 31 bits.
221         result ^= result << 3;
222         result += result >> 5;
223         result ^= result << 2;
224         result += result >> 15;
225         result ^= result << 10;
226 
227         return result;
228     }
229 
230     unsigned m_hash;
231     bool m_hasPendingCharacter;
232     UChar m_pendingCharacter;
233 };
234 
235 } // namespace WTF
236 
237 using WTF::StringHasher;
238 
239 #endif // WTF_StringHasher_h
240