Jeff Thompson | a28eed8 | 2013-08-22 16:21:10 -0700 | [diff] [blame] | 1 | |
| 2 | [/ Copyright 2005-2008 Daniel James. |
| 3 | / Distributed under the Boost Software License, Version 1.0. (See accompanying |
| 4 | / file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) ] |
| 5 | |
| 6 | [def __multi-index-short__ [@boost:/libs/multi_index/doc/index.html |
| 7 | Boost.MultiIndex]] |
| 8 | |
| 9 | [section:tutorial Tutorial] |
| 10 | |
| 11 | When using a hash index with __multi-index-short__, you don't need to do |
| 12 | anything to use [classref boost::hash] as it uses it by default. |
| 13 | To find out how to use a user-defined type, read the |
| 14 | [link hash.custom section on extending boost::hash for a custom data type]. |
| 15 | |
| 16 | If your standard library supplies its own implementation of the unordered |
| 17 | associative containers and you wish to use |
| 18 | [classref boost::hash], just use an extra template parameter: |
| 19 | |
| 20 | std::unordered_multiset<int, ``[classref boost::hash]``<int> > |
| 21 | set_of_ints; |
| 22 | |
| 23 | std::unordered_set<std::pair<int, int>, ``[classref boost::hash]``<std::pair<int, int> > |
| 24 | set_of_pairs; |
| 25 | |
| 26 | std::unordered_map<int, std::string, ``[classref boost::hash]``<int> > map_int_to_string; |
| 27 | |
| 28 | To use [classref boost::hash] directly, create an instance and call it as a function: |
| 29 | |
| 30 | #include <``[headerref boost/functional/hash.hpp]``> |
| 31 | |
| 32 | int main() |
| 33 | { |
| 34 | ``[classref boost::hash]``<std::string> string_hash; |
| 35 | |
| 36 | std::size_t h = string_hash("Hash me"); |
| 37 | } |
| 38 | |
| 39 | For an example of generic use, here is a function to generate a vector |
| 40 | containing the hashes of the elements of a container: |
| 41 | |
| 42 | template <class Container> |
| 43 | std::vector<std::size_t> get_hashes(Container const& x) |
| 44 | { |
| 45 | std::vector<std::size_t> hashes; |
| 46 | std::transform(x.begin(), x.end(), std::insert_iterator(hashes), |
| 47 | ``[classref boost::hash]``<typename Container::value_type>()); |
| 48 | |
| 49 | return hashes; |
| 50 | } |
| 51 | |
| 52 | [endsect] |
| 53 | |
| 54 | [section:custom Extending boost::hash for a custom data type] |
| 55 | |
| 56 | [classref boost::hash] is implemented by calling the function |
| 57 | [funcref boost::hash_value hash_value]. |
| 58 | The namespace isn't specified so that it can detect overloads via argument |
| 59 | dependant lookup. So if there is a free function `hash_value` in the same |
| 60 | namespace as a custom type, it will get called. |
| 61 | |
| 62 | If you have a structure `library::book`, where each `book` is uniquely |
| 63 | defined by it's member `id`: |
| 64 | |
| 65 | namespace library |
| 66 | { |
| 67 | struct book |
| 68 | { |
| 69 | int id; |
| 70 | std::string author; |
| 71 | std::string title; |
| 72 | |
| 73 | // .... |
| 74 | }; |
| 75 | |
| 76 | bool operator==(book const& a, book const& b) |
| 77 | { |
| 78 | return a.id == b.id; |
| 79 | } |
| 80 | } |
| 81 | |
| 82 | Then all you would need to do is write the function `library::hash_value`: |
| 83 | |
| 84 | namespace library |
| 85 | { |
| 86 | std::size_t hash_value(book const& b) |
| 87 | { |
| 88 | ``[classref boost::hash]``<int> hasher; |
| 89 | return hasher(b.id); |
| 90 | } |
| 91 | } |
| 92 | |
| 93 | And you can now use [classref boost::hash] with book: |
| 94 | |
| 95 | library::book knife(3458, "Zane Grey", "The Hash Knife Outfit"); |
| 96 | library::book dandelion(1354, "Paul J. Shanley", |
| 97 | "Hash & Dandelion Greens"); |
| 98 | |
| 99 | ``[classref boost::hash]``<library::book> book_hasher; |
| 100 | std::size_t knife_hash_value = book_hasher(knife); |
| 101 | |
| 102 | // If std::unordered_set is available: |
| 103 | std::unordered_set<library::book, ``[classref boost::hash]``<library::book> > books; |
| 104 | books.insert(knife); |
| 105 | books.insert(library::book(2443, "Lindgren, Torgny", "Hash")); |
| 106 | books.insert(library::book(1953, "Snyder, Bernadette M.", |
| 107 | "Heavenly Hash: A Tasty Mix of a Mother's Meditations")); |
| 108 | |
| 109 | assert(books.find(knife) != books.end()); |
| 110 | assert(books.find(dandelion) == books.end()); |
| 111 | |
| 112 | The full example can be found in: |
| 113 | [@boost:/libs/functional/hash/examples/books.hpp /libs/functional/hash/examples/books.hpp] |
| 114 | and |
| 115 | [@boost:/libs/functional/hash/examples/books.cpp /libs/functional/hash/examples/books.cpp]. |
| 116 | |
| 117 | [tip |
| 118 | When writing a hash function, first look at how the equality function works. |
| 119 | Objects that are equal must generate the same hash value. |
| 120 | When objects are not equal they should generate different hash values. |
| 121 | In this object equality was based just on the id so the hash function |
| 122 | only hashes the id. If it was based on the object's name and author |
| 123 | then the hash function should take them into account |
| 124 | (how to do this is discussed in the next section). |
| 125 | ] |
| 126 | |
| 127 | [endsect] |
| 128 | |
| 129 | [section:combine Combining hash values] |
| 130 | |
| 131 | Say you have a point class, representing a two dimensional location: |
| 132 | |
| 133 | class point |
| 134 | { |
| 135 | int x; |
| 136 | int y; |
| 137 | public: |
| 138 | point() : x(0), y(0) {} |
| 139 | point(int x, int y) : x(x), y(y) {} |
| 140 | |
| 141 | bool operator==(point const& other) const |
| 142 | { |
| 143 | return x == other.x && y == other.y; |
| 144 | } |
| 145 | }; |
| 146 | |
| 147 | and you wish to use it as the key for an `unordered_map`. You need to |
| 148 | customise the hash for this structure. To do this we need to combine |
| 149 | the hash values for `x` and `y`. The function |
| 150 | [funcref boost::hash_combine] is supplied for this purpose: |
| 151 | |
| 152 | class point |
| 153 | { |
| 154 | ... |
| 155 | |
| 156 | friend std::size_t hash_value(point const& p) |
| 157 | { |
| 158 | std::size_t seed = 0; |
| 159 | ``[funcref boost::hash_combine]``(seed, p.x); |
| 160 | ``[funcref boost::hash_combine]``(seed, p.y); |
| 161 | |
| 162 | return seed; |
| 163 | } |
| 164 | |
| 165 | ... |
| 166 | }; |
| 167 | |
| 168 | Calls to hash_combine incrementally build the hash from the different members |
| 169 | of point, it can be repeatedly called for any number of elements. It calls |
| 170 | [funcref boost::hash_value hash_value] on the supplied element, and combines it with the seed. |
| 171 | |
| 172 | Full code for this example is at |
| 173 | [@boost:/libs/functional/hash/examples/point.cpp /libs/functional/hash/examples/point.cpp]. |
| 174 | |
| 175 | [note |
| 176 | When using [funcref boost::hash_combine] the order of the |
| 177 | calls matters. |
| 178 | ''' |
| 179 | <programlisting> |
| 180 | std::size_t seed = 0; |
| 181 | boost::hash_combine(seed, 1); |
| 182 | boost::hash_combine(seed, 2); |
| 183 | </programlisting> |
| 184 | results in a different seed to: |
| 185 | <programlisting> |
| 186 | std::size_t seed = 0; |
| 187 | boost::hash_combine(seed, 2); |
| 188 | boost::hash_combine(seed, 1); |
| 189 | </programlisting> |
| 190 | ''' |
| 191 | If you are calculating a hash value for data where the order of the data |
| 192 | doesn't matter in comparisons (e.g. a set) you will have to ensure that the |
| 193 | data is always supplied in the same order. |
| 194 | ] |
| 195 | |
| 196 | To calculate the hash of an iterator range you can use [funcref boost::hash_range]: |
| 197 | |
| 198 | std::vector<std::string> some_strings; |
| 199 | std::size_t hash = ``[funcref boost::hash_range]``(some_strings.begin(), some_strings.end()); |
| 200 | |
| 201 | Note that when writing template classes, you might not want to include the main |
| 202 | hash header as it's quite an expensive include that brings in a lot of other |
| 203 | headers, so instead you can include the `<boost/functional/hash_fwd.hpp>` |
| 204 | header which forward declares [classref boost::hash], |
| 205 | [funcref boost::hash_range] and [funcref boost::hash_combine]. You'll need to |
| 206 | include the main header before instantiating [classref boost::hash]. When using |
| 207 | a container that uses [classref boost::hash] it should do that for you, so your |
| 208 | type will work fine with the boost hash containers. There's an example of this |
| 209 | in [@boost:/libs/functional/hash/examples/template.hpp template.hpp] and |
| 210 | [@boost:/libs/functional/hash/examples/template.cpp template.cpp]. |
| 211 | |
| 212 | [endsect] |