[Note: this document is formatted similarly to the SGI STL implementation documentation pages, and refers to concepts and classes defined there. However, neither this document nor the code it describes is associated with SGI, nor is it necessary to have SGI's STL implementation installed in order to use this class.]

sparse_hash_set<Key, HashFcn, EqualKey, Alloc>

sparse_hash_set is a Hashed Associative Container that stores objects of type Key. sparse_hash_set is a Simple Associative Container, meaning that its value type, as well as its key type, is key. It is also a Unique Associative Container, meaning that no two elements have keys that compare equal using EqualKey.

Looking up an element in a sparse_hash_set by its key is efficient, so sparse_hash_set is useful for "dictionaries" where the order of elements is irrelevant. If it is important for the elements to be in a particular order, however, then map is more appropriate.

sparse_hash_set is distinguished from other hash-set implementations by its stingy use of memory and by the ability to save and restore contents to disk. On the other hand, this hash-set implementation, while still efficient, is slower than other hash-set implementations, and it also has requirements -- for instance, for a distinguished "deleted key" -- that may not be easy for all applications to satisfy.

This class is appropriate for applications that need to store large "dictionaries" in memory, or for applications that need these dictionaries to be persistent.

Example

(Note: this example uses SGI semantics for hash<> -- the kind used by gcc and most Unix compiler suites -- and not Dinkumware semantics -- the kind used by Microsoft Visual Studio. If you are using MSVC, this example will not compile as-is: you'll need to change hash to hash_compare, and you won't use eqstr at all. See the MSVC documentation for hash_map and hash_compare, for more details.)
#include <iostream>
#include <sparsehash/sparse_hash_set>

using google::sparse_hash_set;      // namespace where class lives by default
using std::cout;
using std::endl;
using ext::hash;  // or __gnu_cxx::hash, or maybe tr1::hash, depending on your OS

struct eqstr
{
  bool operator()(const char* s1, const char* s2) const
  {
    return (s1 == s2) || (s1 && s2 && strcmp(s1, s2) == 0);
  }
};

void lookup(const hash_set<const char*, hash<const char*>, eqstr>& Set,
            const char* word)
{
  sparse_hash_set<const char*, hash<const char*>, eqstr>::const_iterator it
    = Set.find(word);
  cout << word << ": "
       << (it != Set.end() ? "present" : "not present")
       << endl;
}

int main()
{
  sparse_hash_set<const char*, hash<const char*>, eqstr> Set;
  Set.insert("kiwi");
  Set.insert("plum");
  Set.insert("apple");
  Set.insert("mango");
  Set.insert("apricot");
  Set.insert("banana");

  lookup(Set, "mango");
  lookup(Set, "apple");
  lookup(Set, "durian");
}

Definition

Defined in the header sparse_hash_set. This class is not part of the C++ standard, though it is mostly compatible with the tr1 class unordered_set.

Template parameters

ParameterDescriptionDefault
Key The hash_set's key and value type. This is also defined as sparse_hash_set::key_type and sparse_hash_set::value_type.  
HashFcn The hash function used by the hash_set. This is also defined as sparse_hash_set::hasher.
Note: Hashtable performance depends heavily on the choice of hash function. See the performance page for more information.
hash<Key>
EqualKey The hash_set key equality function: a binary predicate that determines whether two keys are equal. This is also defined as sparse_hash_set::key_equal. equal_to<Key>
Alloc The STL allocator to use. By default, uses the provided allocator libc_allocator_with_realloc, which likely gives better performance than other STL allocators due to its built-in support for realloc, which this container takes advantage of. If you use an allocator other than the default, note that this container imposes an additional requirement on the STL allocator type beyond those in [lib.allocator.requirements]: it does not support allocators that define alternate memory models. That is, it assumes that pointer, const_pointer, size_type, and difference_type are just T*, const T*, size_t, and ptrdiff_t, respectively. This is also defined as sparse_hash_set::allocator_type.

Model of

Unique Hashed Associative Container, Simple Associative Container

Type requirements

Public base classes

None.

Members

MemberWhere definedDescription
value_type Container The type of object, T, stored in the hash_set.
key_type Associative Container The key type associated with value_type.
hasher Hashed Associative Container The sparse_hash_set's hash function.
key_equal Hashed Associative Container Function object that compares keys for equality.
allocator_type Unordered Associative Container (tr1) The type of the Allocator given as a template parameter.
pointer Container Pointer to T.
reference Container Reference to T
const_reference Container Const reference to T
size_type Container An unsigned integral type.
difference_type Container A signed integral type.
iterator Container Iterator used to iterate through a sparse_hash_set.
const_iterator Container Const iterator used to iterate through a sparse_hash_set. (iterator and const_iterator are the same type.)
local_iterator Unordered Associative Container (tr1) Iterator used to iterate through a subset of sparse_hash_set.
const_local_iterator Unordered Associative Container (tr1) Const iterator used to iterate through a subset of sparse_hash_set.
iterator begin() const Container Returns an iterator pointing to the beginning of the sparse_hash_set.
iterator end() const Container Returns an iterator pointing to the end of the sparse_hash_set.
local_iterator begin(size_type i) Unordered Associative Container (tr1) Returns a local_iterator pointing to the beginning of bucket i in the sparse_hash_set.
local_iterator end(size_type i) Unordered Associative Container (tr1) Returns a local_iterator pointing to the end of bucket i in the sparse_hash_set. For sparse_hash_set, each bucket contains either 0 or 1 item.
const_local_iterator begin(size_type i) const Unordered Associative Container (tr1) Returns a const_local_iterator pointing to the beginning of bucket i in the sparse_hash_set.
const_local_iterator end(size_type i) const Unordered Associative Container (tr1) Returns a const_local_iterator pointing to the end of bucket i in the sparse_hash_set. For sparse_hash_set, each bucket contains either 0 or 1 item.
size_type size() const Container Returns the size of the sparse_hash_set.
size_type max_size() const Container Returns the largest possible size of the sparse_hash_set.
bool empty() const Container true if the sparse_hash_set's size is 0.
size_type bucket_count() const Hashed Associative Container Returns the number of buckets used by the sparse_hash_set.
size_type max_bucket_count() const Hashed Associative Container Returns the largest possible number of buckets used by the sparse_hash_set.
size_type bucket_size(size_type i) const Unordered Associative Container (tr1) Returns the number of elements in bucket i. For sparse_hash_set, this will be either 0 or 1.
size_type bucket(const key_type& key) const Unordered Associative Container (tr1) If the key exists in the set, returns the index of the bucket containing the given key, otherwise, return the bucket the key would be inserted into. This value may be passed to begin(size_type) and end(size_type).
float load_factor() const Unordered Associative Container (tr1) The number of elements in the sparse_hash_set divided by the number of buckets.
float max_load_factor() const Unordered Associative Container (tr1) The maximum load factor before increasing the number of buckets in the sparse_hash_set.
void max_load_factor(float new_grow) Unordered Associative Container (tr1) Sets the maximum load factor before increasing the number of buckets in the sparse_hash_set.
float min_load_factor() const sparse_hash_set The minimum load factor before decreasing the number of buckets in the sparse_hash_set.
void min_load_factor(float new_grow) sparse_hash_set Sets the minimum load factor before decreasing the number of buckets in the sparse_hash_set.
void set_resizing_parameters(float shrink, float grow) sparse_hash_set DEPRECATED. See below.
void resize(size_type n) Hashed Associative Container Increases the bucket count to hold at least n items. [2] [3]
void rehash(size_type n) Unordered Associative Container (tr1) Increases the bucket count to hold at least n items. This is identical to resize. [2] [3]
hasher hash_funct() const Hashed Associative Container Returns the hasher object used by the sparse_hash_set.
hasher hash_function() const Unordered Associative Container (tr1) Returns the hasher object used by the sparse_hash_set. This is idential to hash_funct.
key_equal key_eq() const Hashed Associative Container Returns the key_equal object used by the sparse_hash_set.
allocator_type get_allocator() const Unordered Associative Container (tr1) Returns the allocator_type object used by the sparse_hash_set: either the one passed in to the constructor, or a default Alloc instance.
sparse_hash_set() Container Creates an empty sparse_hash_set.
sparse_hash_set(size_type n) Hashed Associative Container Creates an empty sparse_hash_set that's optimized for holding up to n items. [3]
sparse_hash_set(size_type n, const hasher& h) Hashed Associative Container Creates an empty sparse_hash_set that's optimized for up to n items, using h as the hash function.
sparse_hash_set(size_type n, const hasher& h, const key_equal& k) Hashed Associative Container Creates an empty sparse_hash_set that's optimized for up to n items, using h as the hash function and k as the key equal function.
sparse_hash_set(size_type n, const hasher& h, const key_equal& k, const allocator_type& a) Unordered Associative Container (tr1) Creates an empty sparse_hash_set that's optimized for up to n items, using h as the hash function, k as the key equal function, and a as the allocator object.
template <class InputIterator>
sparse_hash_set(InputIterator f, InputIterator l) 
[2]
Unique Hashed Associative Container Creates a sparse_hash_set with a copy of a range.
template <class InputIterator>
sparse_hash_set(InputIterator f, InputIterator l, size_type n) 
[2]
Unique Hashed Associative Container Creates a hash_set with a copy of a range that's optimized to hold up to n items.
template <class InputIterator>
sparse_hash_set(InputIterator f, InputIterator l, size_type n, const
hasher& h) 
[2]
Unique Hashed Associative Container Creates a hash_set with a copy of a range that's optimized to hold up to n items, using h as the hash function.
template <class InputIterator>
sparse_hash_set(InputIterator f, InputIterator l, size_type n, const
hasher& h, const key_equal& k) 
[2]
Unique Hashed Associative Container Creates a hash_set with a copy of a range that's optimized for holding up to n items, using h as the hash function and k as the key equal function.
template <class InputIterator>
sparse_hash_set(InputIterator f, InputIterator l, size_type n, const
hasher& h, const key_equal& k, const allocator_type& a) 
[2]
Unordered Associative Container (tr1) Creates a hash_set with a copy of a range that's optimized for holding up to n items, using h as the hash function, k as the key equal function, and a as the allocator object.
sparse_hash_set(const hash_set&) Container The copy constructor.
sparse_hash_set& operator=(const hash_set&) Container The assignment operator
void swap(hash_set&) Container Swaps the contents of two hash_sets.
pair<iterator, bool> insert(const value_type& x)
Unique Associative Container Inserts x into the sparse_hash_set.
template <class InputIterator>
void insert(InputIterator f, InputIterator l) 
[2]
Unique Associative Container Inserts a range into the sparse_hash_set.
void set_deleted_key(const key_type& key) [4] sparse_hash_set See below.
void clear_deleted_key() [4] sparse_hash_set See below.
void erase(iterator pos) Associative Container Erases the element pointed to by pos. [4]
size_type erase(const key_type& k) Associative Container Erases the element whose key is k. [4]
void erase(iterator first, iterator last) Associative Container Erases all elements in a range. [4]
void clear() Associative Container Erases all of the elements.
iterator find(const key_type& k) const Associative Container Finds an element whose key is k.
size_type count(const key_type& k) const Unique Associative Container Counts the number of elements whose key is k.
pair<iterator, iterator> equal_range(const
key_type& k) const
Associative Container Finds a range containing all elements whose key is k.
template <ValueSerializer, OUTPUT> bool serialize(ValueSerializer serializer, OUTPUT *fp) sparse_hash_set See below.
template <ValueSerializer, INPUT> bool unserialize(ValueSerializer serializer, INPUT *fp) sparse_hash_set See below.
NopointerSerializer sparse_hash_set See below.
bool write_metadata(FILE *fp) sparse_hash_set DEPRECATED. See below.
bool read_metadata(FILE *fp) sparse_hash_set DEPRECATED. See below.
bool write_nopointer_data(FILE *fp) sparse_hash_set DEPRECATED. See below.
bool read_nopointer_data(FILE *fp) sparse_hash_set DEPRECATED. See below.
bool operator==(const hash_set&, const hash_set&)
Hashed Associative Container Tests two hash_sets for equality. This is a global function, not a member function.

New members

These members are not defined in the Unique Hashed Associative Container, Simple Associative Container, or tr1's Unordered Associative Container requirements, but are specific to sparse_hash_set.
MemberDescription
void set_deleted_key(const key_type& key) Sets the distinguished "deleted" key to key. This must be called before any calls to erase(). [4]
void clear_deleted_key() Clears the distinguished "deleted" key. After this is called, calls to erase() are not valid on this object. [4]
void set_resizing_parameters(float shrink, float grow) This function is DEPRECATED. It is equivalent to calling min_load_factor(shrink); max_load_factor(grow).
template <ValueSerializer, OUTPUT> bool serialize(ValueSerializer serializer, OUTPUT *fp) Emit a serialization of the hash_set to a stream. See below.
template <ValueSerializer, INPUT> bool unserialize(ValueSerializer serializer, INPUT *fp) Read in a serialization of a hash_set from a stream, replacing the existing hash_set contents with the serialized contents. See below.
bool write_metadata(FILE *fp) This function is DEPRECATED. See below.
bool read_metadata(FILE *fp) This function is DEPRECATED. See below.
bool write_nopointer_data(FILE *fp) This function is DEPRECATED. See below.
bool read_nopointer_data(FILE *fp) This function is DEPRECATED. See below.

Notes

[1] This member function relies on member template functions, which may not be supported by all compilers. If your compiler supports member templates, you can call this function with any type of input iterator. If your compiler does not yet support member templates, though, then the arguments must either be of type const value_type* or of type sparse_hash_set::const_iterator.

[2] In order to preserve iterators, erasing hashtable elements does not cause a hashtable to resize. This means that after a string of erase() calls, the hashtable will use more space than is required. At a cost of invalidating all current iterators, you can call resize() to manually compact the hashtable. The hashtable promotes too-small resize() arguments to the smallest legal value, so to compact a hashtable, it's sufficient to call resize(0).

[3] Unlike some other hashtable implementations, the optional n in the calls to the constructor, resize, and rehash indicates not the desired number of buckets that should be allocated, but instead the expected number of items to be inserted. The class then sizes the hash-set appropriately for the number of items specified. It's not an error to actually insert more or fewer items into the hashtable, but the implementation is most efficient -- does the fewest hashtable resizes -- if the number of inserted items is n or slightly less.

[4] sparse_hash_set requires you call set_deleted_key() before calling erase(). (This is the largest difference between the sparse_hash_set API and other hash-set APIs. See implementation.html for why this is necessary.) The argument to set_deleted_key() should be a key-value that is never used for legitimate hash-set entries. It is an error to call erase() without first calling set_deleted_key(), and it is also an error to call insert() with an item whose key is the "deleted key."

There is no need to call set_deleted_key if you do not wish to call erase() on the hash-set.

It is acceptable to change the deleted-key at any time by calling set_deleted_key() with a new argument. You can also call clear_deleted_key(), at which point all keys become valid for insertion but no hashtable entries can be deleted until set_deleted_key() is called again.

Input/Output

It is possible to save and restore sparse_hash_set objects to an arbitrary stream (such as a disk file) using the serialize() and unserialize() methods.

Each of these methods takes two arguments: a serializer, which says how to write hashtable items to disk, and a stream, which can be a C++ stream (istream or its subclasses for input, ostream or its subclasses for output), a FILE*, or a user-defined type (as described below).

The serializer is a functor that takes a stream and a single hashtable element (a value_type) and copies the hashtable element to the stream (for serialize()) or fills the hashtable element contents from the stream (for unserialize()), and returns true on success or false on error. The copy-in and copy-out functions can be provided in a single functor. Here is a sample serializer that read/writes a hashtable element for a string hash_set to a FILE*:

struct StringSerializer {
  bool operator()(FILE* fp, const std::string& value) const {
    assert(value.length() <= 255);   // we only support writing small strings
    const unsigned char size = value.length();
    if (fwrite(&size, 1, 1, fp) != 1)
      return false;
    if (fwrite(value.data(), size, 1, fp) != 1)
      return false;
    return true;
  }
  bool operator()(FILE* fp, std::string* value) const {
    unsigned char size;    // all strings are <= 255 chars long
    if (fread(&size, 1, 1, fp) != 1)
      return false;
    char* buf = new char[size];
    if (fread(buf, size, 1, fp) != 1) {
      delete[] buf;
      return false;
    }
    new(value) string(buf, size);
    delete[] buf;
    return true;
  }
};

Here is the functor being used in code (error checking omitted):

   sparse_hash_set<string> myset = CreateSet();
   FILE* fp = fopen("hashtable.data", "w");
   myset.serialize(StringSerializer(), fp);
   fclose(fp);

   sparse_hash_set<string> myset2;
   FILE* fp_in = fopen("hashtable.data", "r");
   myset2.unserialize(StringSerializer(), fp_in);
   fclose(fp_in);
   assert(myset == myset2);

Important note: the code above uses placement-new to instantiate the string. This is required for any non-POD type. The value_type passed in to the unserializer points to garbage memory, so it is not safe to assign to it directly if doing so causes a destructor to be called.

Also note that this example serializer can only serialize to a FILE*. If you want to also be able to use this serializer with C++ streams, you will need to write two more overloads of operator()'s, one that reads from an istream, and one that writes to an ostream. Likewise if you want to support serializing to a custom class.

If the key is "simple" enough, you can use the pre-supplied functor NopointerSerializer. This copies the hashtable data using the equivalent of a memcpy<>. Native C data types can be serialized this way, as can structs of native C data types. Pointers and STL objects cannot.

Note that NopointerSerializer() does not do any endian conversion. Thus, it is only appropriate when you intend to read the data on the same endian architecture as you write the data.

If you wish to serialize to your own stream type, you can do so by creating an object which supports two methods:

   bool Write(const void* data, size_t length);
   bool Read(void* data, size_t length);

Write() writes length bytes of data to a stream (presumably a stream owned by the object), while Read() reads data bytes from the stream into data. Both return true on success or false on error.

To unserialize a hashtable from a stream, you wil typically create a new sparse_hash_set object, then call unserialize() on it. unserialize() destroys the old contents of the object. You must pass in the appropriate ValueSerializer for the data being read in.

Both serialize() and unserialize() return true on success, or false if there was an error streaming the data.

Note that serialize() is not a const method, since it purges deleted elements before serializing. It is not safe to serialize from two threads at once, without synchronization.

NOTE: older versions of sparse_hash_set provided a different API, consisting of read_metadata(), read_nopointer_data(), write_metadata(), write_nopointer_data(). Writing to disk consisted of a call to write_metadata() followed by write_nopointer_data() (if the hash data was POD) or a custom loop over the hashtable buckets to write the data (otherwise). Reading from disk was similar. Prefer the new API for new code.

Validity of Iterators

erase() is guaranteed not to invalidate any iterators -- except for any iterators pointing to the item being erased, of course. insert() invalidates all iterators, as does resize().

This is implemented by making erase() not resize the hashtable. If you desire maximum space efficiency, you can call resize(0) after a string of erase() calls, to force the hashtable to resize to the smallest possible size.

In addition to invalidating iterators, insert() and resize() invalidate all pointers into the hashtable. If you want to store a pointer to an object held in a sparse_hash_set, either do so after finishing hashtable inserts, or store the object on the heap and a pointer to it in the sparse_hash_set.

See also

The following are SGI STL, and some Google STL, concepts and classes related to sparse_hash_set.

hash_set, Associative Container, Hashed Associative Container, Simple Associative Container, Unique Hashed Associative Container, set, map multiset, multimap, hash_map, hash_multiset, hash_multimap, sparsetable, sparse_hash_map, dense_hash_set, dense_hash_map