Package org.apache.datasketches.tuple

Class AnotB<S extends Summary>

  • Type Parameters:
    S - Type of Summary
    public final class AnotB<S extends Summary>
extends Object
    Computes a set difference, A-AND-NOT-B, of two generic tuple sketches. This class includes both stateful and stateless operations.

    The stateful operation is as follows:

    
 AnotB anotb = new AnotB();

 anotb.setA(Sketch skA); //The first argument.
 anotb.notB(Sketch skB); //The second (subtraction) argument.
 anotb.notB(Sketch skC); // ...any number of additional subtractions...
 anotb.getResult(false); //Get an interim result.
 anotb.notB(Sketch skD); //Additional subtractions.
 anotb.getResult(true);  //Final result and resets the AnotB operator.

    The stateless operation is as follows:

    
 AnotB anotb = new AnotB();

 CompactSketch csk = anotb.aNotB(Sketch skA, Sketch skB);

    Calling the setA operation a second time essentially clears the internal state and loads the new sketch.

    The stateless and stateful operations are independent of each other.

    Author:
    Lee Rhodes

    • Constructor Detail

      • AnotB

        public AnotB()

    • Method Detail

      • setA

        public void setA​(Sketch<S> skA)
        This is part of a multistep, stateful AnotB operation and sets the given Tuple sketch as the first argument A of A-AND-NOT-B. This overwrites the internal state of this AnotB operator with the contents of the given sketch. This sets the stage for multiple following notB steps.

        An input argument of null will throw an exception.

        Rationale: In mathematics a "null set" is a set with no members, which we call an empty set. That is distinctly different from the java null, which represents a nonexistent object. In most cases it is a programming error due to some object that was not properly initialized. With a null as the first argument, we cannot know what the user's intent is. Since it is very likely that a null is a programming error, we throw a an exception.

        An empty input argument will set the internal state to empty.

        Rationale: An empty set is a mathematically legal concept. Although it makes any subsequent, valid argument for B irrelevant, we must allow this and assume the user knows what they are doing.

        Performing getResult(boolean) just after this step will return a compact form of the given argument.

        Parameters:
        skA - The incoming sketch for the first argument, A.

      • notB

        public void notB​(Sketch<S> skB)
        This is part of a multistep, stateful AnotB operation and sets the given Tuple sketch as the second (or n+1th) argument B of A-AND-NOT-B. Performs an AND NOT operation with the existing internal state of this AnotB operator.

        An input argument of null or empty is ignored.

        Rationale: A null for the second or following arguments is more tolerable because A NOT null is still A even if we don't know exactly what the null represents. It clearly does not have any content that overlaps with A. Also, because this can be part of a multistep operation with multiple notB steps. Other following steps can still produce a valid result.

        Use getResult(boolean) to obtain the result.

        Parameters:
        skB - The incoming Tuple sketch for the second (or following) argument B.

      • notB

        public void notB​(Sketch skB)
        This is part of a multistep, stateful AnotB operation and sets the given Theta sketch as the second (or n+1th) argument B of A-AND-NOT-B. Performs an AND NOT operation with the existing internal state of this AnotB operator. Calls to this method can be intermingled with calls to notB(org.apache.datasketches.theta.Sketch).

        An input argument of null or empty is ignored.

        Rationale: A null for the second or following arguments is more tolerable because A NOT null is still A even if we don't know exactly what the null represents. It clearly does not have any content that overlaps with A. Also, because this can be part of a multistep operation with multiple notB steps. Other following steps can still produce a valid result.

        Use getResult(boolean) to obtain the result.

        Parameters:
        skB - The incoming Theta sketch for the second (or following) argument B.

      • getResult

        public CompactSketch<S> getResult​(boolean reset)
        Gets the result of the multistep, stateful operation AnotB that have been executed with calls to setA(Sketch) and (notB(Sketch) or notB(org.apache.datasketches.theta.Sketch)).
        Parameters:
        reset - If true, clears this operator to the empty state after this result is returned. Set this to false if you wish to obtain an intermediate result.
        Returns:
        the result of this operation as an unordered CompactSketch.

      • aNotB

        public static <S extends Summary> CompactSketch<S> aNotB​(Sketch<S> skA,
                                                         Sketch<S> skB)
        Returns the A-and-not-B set operation on the two given Tuple sketches.

        This a stateless operation and has no impact on the internal state of this operator. Thus, this is not an accumulating update and is independent of the setA(Sketch), notB(Sketch), notB(org.apache.datasketches.theta.Sketch), and getResult(boolean) methods.

        If either argument is null an exception is thrown.

        Rationale: In mathematics a "null set" is a set with no members, which we call an empty set. That is distinctly different from the java null, which represents a nonexistent object. In most cases it is a programming error due to some object that was not properly initialized. With a null as the first argument, we cannot know what the user's intent is. With a null as the second argument, we can't ignore it as we must return a result and there is no following possible viable arguments for the second argument. Since it is very likely that a null is a programming error, we throw an exception.

        Type Parameters:
        S - Type of Summary
        Parameters:
        skA - The incoming Tuple sketch for the first argument
        skB - The incoming Tuple sketch for the second argument
        Returns:
        the result as an unordered CompactSketch

      • aNotB

        public static <S extends Summary> CompactSketch<S> aNotB​(Sketch<S> skA,
                                                         Sketch skB)
        Returns the A-and-not-B set operation on a Tuple sketch and a Theta sketch.

        This a stateless operation and has no impact on the internal state of this operator. Thus, this is not an accumulating update and is independent of the setA(Sketch), notB(Sketch), notB(org.apache.datasketches.theta.Sketch), and getResult(boolean) methods.

        If either argument is null an exception is thrown.

        Rationale: In mathematics a "null set" is a set with no members, which we call an empty set. That is distinctly different from the java null, which represents a nonexistent object. In most cases it is a programming error due to some object that was not properly initialized. With a null as the first argument, we cannot know what the user's intent is. With a null as the second argument, we can't ignore it as we must return a result and there is no following possible viable arguments for the second argument. Since it is very likely that a null is a programming error for either argument we throw a an exception.

        Type Parameters:
        S - Type of Summary
        Parameters:
        skA - The incoming Tuple sketch for the first argument
        skB - The incoming Theta sketch for the second argument
        Returns:
        the result as an unordered CompactSketch

      • reset

        public void reset()
        Resets this operation back to the empty state.