Package it.unimi.dsi.bits

Class LongArrayBitVector

  • All Implemented Interfaces:
    BitVector, BigList<Boolean>, BooleanBigList, BooleanCollection, BooleanIterable, BooleanStack, Size64, Stack<Boolean>, Serializable, Cloneable, Comparable<BigList<? extends Boolean>>, Iterable<Boolean>, Collection<Boolean>, RandomAccess
    public class LongArrayBitVector
extends AbstractBitVector
implements Cloneable, Serializable
    A bit vector implementation based on arrays of longs.

    The main goal of this class is to be fast and flexible. It implements a lightweight, fast, open, optimized, reuse-oriented version of bit vectors. Instances of this class represent a bit vector an array of longs that is enlarged as needed when new entries are created (by dividing the current length by the golden ratio), but is never made smaller (even on a clear()). Use trim() for that purpose.

    Besides usual methods for setting and getting bits, this class provides views that make it possible to access comfortably the bit vector in different ways: for instance, asLongBigList(int) provide access as a list of longs, whereas AbstractBitVector.asLongSet() provides access in setwise form.

    When enlarging the underlying array (e.g., for append(long, int) operations or add operations on the big list view), or when invoking ensureCapacity(long), this class calls LongArrays.grow(long[], int, int), which could enlarge the array more than expected. On the contrary, length(long) (and the corresponding method in the big list view) sizes the underlying array in an exact manner.

    Bit numbering follows the right-to-left convention: bit k (counted from the right) of word w is bit 64w + k of the overall bit vector.

    If CHECKS is true at compile time, boundary checks for all bit operations will be compiled in. For maximum speed, you may want to recompile this class with CHECKS set to false. CHECKS is public, so you can check from your code whether you're being provided a version with checks or not.

    Warning: A few optional methods have still to be implemented (e.g., adding an element at an arbitrary position using the list view).

    Warning: In some cases, you might want to cache locally the result of bits() to speed up computations on immutable bit vectors (this is what happens, for instance, in static ranking structures). This class, however, does its own serialisation of the bit vector: as a result, all cached references to the result of bits() must be marked as transient and rebuilt at deserialisation time, or you will end up saving the bits twice.

    See Also:
    Serialized Form

    • Field Detail

      • CHECKS

        public static final boolean CHECKS
        Whether this class has been compiled with index checks or not.
        See Also:
        Constant Field Values

      • length

        protected long length
        The number of bits in this vector.

      • bits

        protected transient long[] bits
        The backing array of this vector. Bit 0 of the first element contains bit 0 of the bit vector, bit 1 of the second element contains bit BITS_PER_WORD of the bit vector and so on.

    • Constructor Detail

      • LongArrayBitVector

        protected LongArrayBitVector​(long capacity)

    • Method Detail

      • numWords

        protected static final int numWords​(long size)
        Returns the number of words that are necessary to hold the given number of bits.
        Parameters:
        size - a number of bits.
        Returns:
        the number of words that are necessary to hold the given number of bits.

      • word

        protected static final int word​(long index)
        Return the index of the word that holds a bit of specified index.
        Parameters:
        index - the index of a bit, or -1.
        Returns:
        the index of the word that holds the bit of given index, or -1 if index is -1.

      • bit

        protected static final int bit​(long index)
        Returns the inside-word index of the bit that would hold the bit of specified index.

        Note that bit 0 is positioned in word 0, index 0, bit 1 in word 0, index 1, …, bit BITS_PER_WORD in word 0, index 0, bit BITS_PER_WORD + 1 in word 1, index 1, and so on.

        Parameters:
        index - the index of a bit.
        Returns:
        the inside-word index of the bit that would hold the bit of specified index.

      • mask

        protected static final long mask​(long index)
        Returns a mask having a 1 exactly at the bit bit(index).
        Parameters:
        index - the index of a bit
        Returns:
        a mask having a 1 exactly at the bit bit(index).

      • getInstance

        public static LongArrayBitVector getInstance​(long capacity)
        Creates a new empty bit vector of given capacity. The resulting vector will be able to contain capacity bits without reallocations of the backing array.

        Note that this constructor creates an empty bit vector. If you want a cleared bit vector of a specified size, please use the ofLength(long) factory method.

        Parameters:
        capacity - the capacity (in bits) of the new bit vector.
        Returns:
        a new bit vector of given capacity.

      • getInstance

        public static LongArrayBitVector getInstance()
        Creates a new empty bit vector. No allocation is actually performed.
        Returns:
        a new bit vector with no capacity.

      • ofLength

        public static LongArrayBitVector ofLength​(long length)
        Creates a new empty bit vector of given length.
        Parameters:
        length - the size (in bits) of the new bit vector.

      • of

        public static LongArrayBitVector of​(int... bit)
        Creates a new bit vector with given bits.
        Parameters:
        bit - a list of bits that will be set in the newly created bit vector.

      • bits

        public long[] bits()
        Description copied from interface: BitVector
        Returns the bits in this bit vector as an array of longs, not to be modified.
        Specified by:
        bits in interface BitVector
        Overrides:
        bits in class AbstractBitVector
        Returns:
        an array of longs whose first BitVector.length() bits contain the bits of this bit vector. The array cannot be modified.

      • length

        public long length()
        Description copied from interface: BitVector
        Returns the number of bits in this bit vector.

        If the number of bits in this vector is smaller than or equal to Integer.MAX_VALUE, this method is semantically equivalent to List.size(). In any case, this method is semantically equivalent to Size64.size64(), but it is prefererred.

        Specified by:
        length in interface BitVector
        Returns:
        the number of bits in this bit vector.

      • ensureCapacity

        public LongArrayBitVector ensureCapacity​(long numBits)
        Ensures that this bit vector can hold the specified number of bits.

        This method uses LongArrays.grow(long[], int, int) to ensure that there is enough space for the given number of bits. As a consequence, the actual length of the long array allocated might be larger than expected.

        Parameters:
        numBits - the number of bits that this vector must be able to contain.
        Returns:
        this bit vector.

      • length

        public LongArrayBitVector length​(long newLength)
        Description copied from interface: BitVector
        Sets the number of bits in this bit vector.

        It is expected that this method will try to allocate exactly the necessary space.

        If the argument fits an integer, this method has the same side effects of BooleanList.size(int). In any case, this method has the same side effects of BigList.size(long), but it is preferred, as it has the advantage of returning this bit vector, thus making it possible to chain methods.

        Specified by:
        length in interface BitVector
        Overrides:
        length in class AbstractBitVector
        Returns:
        this bit vector.

      • fill

        public void fill​(boolean value)
        Description copied from interface: BitVector
        Sets all bits this bit vector to the given boolean value (optional operation).
        Specified by:
        fill in interface BitVector
        Overrides:
        fill in class AbstractBitVector
        Parameters:
        value - the value (true or false).

      • fill

        public void fill​(long from,
                 long to,
                 boolean value)
        Description copied from interface: BitVector
        Fills a range of bits in this bit vector (optional operation).
        Specified by:
        fill in interface BitVector
        Overrides:
        fill in class AbstractBitVector
        Parameters:
        from - the first index (inclusive).
        to - the last index (not inclusive).
        value - the value (true or false).

      • flip

        public void flip()
        Description copied from interface: BitVector
        Flips all bits in this bit vector (optional operation).
        Specified by:
        flip in interface BitVector
        Overrides:
        flip in class AbstractBitVector

      • flip

        public void flip​(long from,
                 long to)
        Description copied from interface: BitVector
        Flips a range of bits in this bit vector (optional operation).
        Specified by:
        flip in interface BitVector
        Overrides:
        flip in class AbstractBitVector
        Parameters:
        from - the first index (inclusive).
        to - the last index (not inclusive).

      • trim

        public boolean trim()
        Reduces as must as possible the size of the backing array.
        Returns:
        true if some trimming was actually necessary.

      • clear

        public void clear()
        Sets the size of this bit vector to 0.

        Note that this method does not try to reallocate that backing array. If you want to force that behaviour, call trim() afterwards.

        Specified by:
        clear in interface Collection<Boolean>
        Overrides:
        clear in class AbstractBitVector

      • copy

        public LongArrayBitVector copy​(long from,
                               long to)
        Description copied from interface: BitVector
        Returns a copy of a part of this bit vector.
        Specified by:
        copy in interface BitVector
        Overrides:
        copy in class AbstractBitVector
        Parameters:
        from - the starting bit, inclusive.
        to - the ending bit, not inclusive.
        Returns:
        a copy of the part of this bit vector going from bit from (inclusive) to bit to (not inclusive)

      • set

        public void set​(long index)
        Description copied from interface: BitVector
        Sets a bit in this bit vector (optional operation).
        Specified by:
        set in interface BitVector
        Overrides:
        set in class AbstractBitVector
        Parameters:
        index - the index of a bit.

      • clear

        public void clear​(long index)
        Description copied from interface: BitVector
        Clears a bit in this bit vector (optional operation).
        Specified by:
        clear in interface BitVector
        Overrides:
        clear in class AbstractBitVector
        Parameters:
        index - the index of a bit.

      • append

        public LongArrayBitVector append​(long value,
                                 int width)
        Description copied from interface: BitVector
        Appends the less significant bits of a long integer to this bit vector.
        Specified by:
        append in interface BitVector
        Overrides:
        append in class AbstractBitVector
        Parameters:
        value - a value to be appended
        width - the number of less significant bits to be added to this bit vector.
        Returns:
        this bit vector.

      • getLong

        public long getLong​(long from,
                    long to)
        Description copied from interface: BitVector
        Returns the specified bit range as a long.

        Note that bit 0 of the returned long will be bit from of this bit vector.

        Implementations are invited to provide high-speed implementations for the case in which from is a multiple of Long.SIZE and to is from + Long.SIZE (or less, in case the vector length is exceeded). This behaviour make it possible to implement high-speed hashing, copies, etc.

        Specified by:
        getLong in interface BitVector
        Overrides:
        getLong in class AbstractBitVector
        Parameters:
        from - the starting bit (inclusive).
        to - the ending bit (exclusive).
        Returns:
        the long value contained in the specified bits.

      • count

        public long count()
        Description copied from interface: BitVector
        Counts the number of bits set to true in this bit vector.
        Specified by:
        count in interface BitVector
        Overrides:
        count in class AbstractBitVector
        Returns:
        the number of bits set to true in this bit vector.

      • nextOne

        public long nextOne​(long index)
        Description copied from interface: BitVector
        Returns the position of the first bit set at of after the given position.
        Specified by:
        nextOne in interface BitVector
        Overrides:
        nextOne in class AbstractBitVector
        Returns:
        the position of the first bit set at or after position index, or -1 if no such bit exists.

      • previousOne

        public long previousOne​(long index)
        Description copied from interface: BitVector
        Returns the position of the first bit set strictly before the given position.
        Specified by:
        previousOne in interface BitVector
        Overrides:
        previousOne in class AbstractBitVector
        Returns:
        the position of the first bit set strictly before position index, or -1 if no such bit exists.

      • nextZero

        public long nextZero​(long index)
        Description copied from interface: BitVector
        Returns the position of the first bit unset after the given position.
        Specified by:
        nextZero in interface BitVector
        Overrides:
        nextZero in class AbstractBitVector
        Returns:
        the first bit unset after position index (inclusive), or -1 if no such bit exists.

      • previousZero

        public long previousZero​(long index)
        Description copied from interface: BitVector
        Returns the position of the first bit unset before or at the given position.
        Specified by:
        previousZero in interface BitVector
        Overrides:
        previousZero in class AbstractBitVector
        Returns:
        the first bit unset before or at the given position, or -1 if no such bit exists.

      • longestCommonPrefixLength

        public long longestCommonPrefixLength​(LongArrayBitVector v)

      • and

        public BitVector and​(BitVector v)
        Description copied from interface: BitVector
        Performs a logical and between this bit vector and another one, leaving the result in this vector.
        Specified by:
        and in interface BitVector
        Overrides:
        and in class AbstractBitVector
        Parameters:
        v - a bit vector.
        Returns:
        this bit vector.

      • or

        public BitVector or​(BitVector v)
        Description copied from interface: BitVector
        Performs a logical or between this bit vector and another one, leaving the result in this vector.
        Specified by:
        or in interface BitVector
        Overrides:
        or in class AbstractBitVector
        Parameters:
        v - a bit vector.
        Returns:
        this bit vector.

      • xor

        public BitVector xor​(BitVector v)
        Description copied from interface: BitVector
        Performs a logical xor between this bit vector and another one, leaving the result in this vector.
        Specified by:
        xor in interface BitVector
        Overrides:
        xor in class AbstractBitVector
        Parameters:
        v - a bit vector.
        Returns:
        this bit vector.

      • wrap

        public static LongArrayBitVector wrap​(long[] array,
                                      long size)
        Wraps the given array of longs in a bit vector for the given number of bits.

        Note that all bits in array beyond that of index size must be unset, or an exception will be thrown.

        Parameters:
        array - an array of longs.
        size - the number of bits of the newly created bit vector.
        Returns:
        a bit vector of size size using array as backing array.

      • wrap

        public static LongArrayBitVector wrap​(long[] array)
        Wraps the given array of longs in a bit vector.
        Parameters:
        array - an array of longs.
        Returns:
        a bit vector of size array.length * Long.SIZE using array as backing array.

      • hashCode

        public int hashCode()
        Description copied from interface: BitVector
        Returns a hash code for this bit vector.

        Hash codes for bit vectors are defined as follows: 

         final long length = length();
 long fullLength = length - length % Long.SIZE;
 long h = 0x9e3779b97f4a7c13L ^ length;
 for(long i = 0; i < fullLength; i += Long.SIZE) h ^= (h << 5) + getLong(i, i + Long.SIZE) + (h >>> 2);
 if (length != fullLength) h ^= (h << 5) + getLong(fullLength, length) + (h >>> 2);
 (int)((h >>> 32) ^ h);

        The last value is the hash code of the bit vector. This hashing is based on shift-add-xor hashing (M.V. Ramakrishna and Justin Zobel, “Performance in practice of string hashing functions”, Proc. of the Fifth International Conference on Database Systems for Advanced Applications, 1997, pages 215−223).

        The returned value is not a high-quality hash such as Jenkins's, but it can be computed very quickly; in any case, 32 bits are too few for a high-quality hash to be used in large-scale applications.

        Important: all bit vector implementations are required to return the value defined here. The simplest way to obtain this result is to subclass AbstractBitVector.

        Specified by:
        hashCode in interface BitVector
        Specified by:
        hashCode in interface Collection<Boolean>
        Overrides:
        hashCode in class AbstractBitVector
        Returns:
        a hash code for this bit vector.

      • asLongBigList

        public LongBigList asLongBigList​(int width)
        Description copied from interface: BitVector
        Returns a view of this bit vector as a list of nonnegative integers of specified width.

        More formally, getLong(p) will return the nonnegative integer defined by the bits starting at p * width (bit 0, inclusive) and ending at (p + 1) * width (bit width − 1, exclusive).

        Specified by:
        asLongBigList in interface BitVector
        Overrides:
        asLongBigList in class AbstractBitVector