Class BetweenFilter<T,E extends Comparable<? super E>>

Type Parameters:
T - the type of the input argument to the filter
E - the type of the extracted attribute to use for comparison
All Implemented Interfaces:
ExternalizableLite, PortableObject, Filter, EntryFilter, IndexAwareFilter, QueryRecorderFilter, Serializable

public class BetweenFilter<T,E extends Comparable<? super E>> extends AndFilter
Filter which compares the result of a method invocation with a value for "Between" condition. We use the standard ISO/IEC 9075:1992 semantic, according to which "X between Y and Z" is equivalent to "X >= Y && X <= Z". In a case when either result of a method invocation or a value to compare are equal to null, the evaluate test yields false. This approach is equivalent to the way the NULL values are handled by SQL.
Author:
cp/gg 2002.10.29, jk 2014.05.20
See Also:
  • Field Details

  • Constructor Details

    • BetweenFilter

      public BetweenFilter()
      Default constructor (necessary for serialization).
    • BetweenFilter

      public BetweenFilter(ValueExtractor<? super T,? extends E> extractor, E from, E to)
      Construct a BetweenFilter for testing "Between" condition.
      Parameters:
      extractor - the ValueExtractor to use by this filter
      from - the object to compare the "Greater or Equals" boundary with
      to - the object to compare the "Less or Equals" boundary with
    • BetweenFilter

      public BetweenFilter(String sMethod, E from, E to)
      Construct a BetweenFilter for testing "Between" condition.
      Parameters:
      sMethod - the name of the method to invoke via reflection
      from - the object to compare the "Greater or Equals" boundary with
      to - the object to compare the "Less or Equals" boundary with
    • BetweenFilter

      public BetweenFilter(String sMethod, E lowerBound, E upperBound, boolean fIncludeLowerBound, boolean fIncludeUpperBound)
      Construct a BetweenFilter for testing "Between" condition.
      Parameters:
      sMethod - the name of the method to invoke via reflection
      lowerBound - the lower bound of the range
      upperBound - the upper bound of the range
      fIncludeLowerBound - a flag indicating whether values matching the lower bound evaluate to true
      fIncludeUpperBound - a flag indicating whether values matching the upper bound evaluate to true
    • BetweenFilter

      public BetweenFilter(ValueExtractor<? super T,? extends E> extractor, E lowerBound, E upperBound, boolean fIncludeLowerBound, boolean fIncludeUpperBound)
      Construct a BetweenFilter for testing "Between" condition.
      Parameters:
      extractor - the ValueExtractor to be used by this filter
      lowerBound - the lower bound of the range
      upperBound - the upper bound of the range
      fIncludeLowerBound - a flag indicating whether values matching the lower bound evaluate to true
      fIncludeUpperBound - a flag indicating whether values matching the upper bound evaluate to true
  • Method Details

    • getValueExtractor

      public ValueExtractor getValueExtractor()
      Obtain the ValueExtractor used by this filter.
      Returns:
      the ValueExtractor used by this filter
    • getLowerBound

      public E getLowerBound()
      Obtain the lower bound of the range being used to evaluate values by this BetweenFilter.
      Returns:
      the lower bound of the range being used to evaluate values by this BetweenFilter
    • getUpperBound

      public E getUpperBound()
      Obtain the upper bound of the range being used to evaluate values by this BetweenFilter.
      Returns:
      the upper bound of the range being used to evaluate values by this BetweenFilter
    • isLowerBoundInclusive

      public boolean isLowerBoundInclusive()
      Obtain the flag indicating whether values matching the lower bound of the range evaluate to true.
      Returns:
      the flag indicating whether values matching the lower bound of the range evaluate to true
    • isUpperBoundInclusive

      public boolean isUpperBoundInclusive()
      Obtain the flag indicating whether values matching the upper bound of the range evaluate to true.
      Returns:
      the flag indicating whether values matching the upper bound of the range evaluate to true
    • evaluate

      public boolean evaluate(Object oTarget)
      Apply the test to the input argument.
      Specified by:
      evaluate in interface Filter<T>
      Overrides:
      evaluate in class AllFilter
      Parameters:
      oTarget - the input argument to evaluate
      Returns:
      true if the input argument matches the filter, otherwise false
    • evaluateEntry

      public boolean evaluateEntry(Map.Entry entry)
      Apply the test to a Map Entry.
      Specified by:
      evaluateEntry in interface EntryFilter<T,E extends Comparable<? super E>>
      Overrides:
      evaluateEntry in class ArrayFilter
      Parameters:
      entry - the Map Entry to evaluate; never null
      Returns:
      true if the test passes, false otherwise
    • evaluateEntry

      protected boolean evaluateEntry(Map.Entry entry, QueryContext ctx, QueryRecord.PartialResult.TraceStep step)
      Check if the given entry passes the filter's evaluation. Record the actual cost of execution for each of the participating filters if a query context is provided.
      Overrides:
      evaluateEntry in class AllFilter
      Parameters:
      entry - a key value pair to filter
      ctx - the query ctx; may be null
      step - the step used to record the execution cost
      Returns:
      true if the entry passes the filter, false otherwise
    • applyIndex

      public Filter applyIndex(Map mapIndexes, Set setKeys)
      Filter remaining keys using a Map of available indexes.

      The filter is responsible for removing all keys from the passed set of keys that the applicable indexes can prove should be filtered. If the filter does not fully evaluate the remaining keys using just the index information, it must return a filter (which may be an EntryFilter) that can complete the task using an iterating implementation. If, on the other hand, the filter does fully evaluate the remaining keys using just the index information, then it should return null to indicate that no further filtering is necessary.

      Specified by:
      applyIndex in interface IndexAwareFilter<T,E extends Comparable<? super E>>
      Overrides:
      applyIndex in class ArrayFilter
      Parameters:
      mapIndexes - the available MapIndex objects keyed by the related ValueExtractor; read-only
      setKeys - the mutable set of keys that remain to be filtered
      Returns:
      a Filter object (which may be an EntryFilter) that can be used to process the remaining keys, or null if no additional filter processing is necessary
    • calculateEffectiveness

      public int calculateEffectiveness(Map mapIndexes, Set setKeys)
      Given a Map of available indexes, determine if this IndexAwareFilter can use any of the indexes to assist in its processing, and if so, determine how effective the use of that index would be.

      The returned value is an effectiveness estimate of how well this filter can use the specified indexes to filter the specified keys. An operation that requires no more than a single access to the index content (i.e. Equals, NotEquals) has an effectiveness of one. Evaluation of a single entry is assumed to have an effectiveness that depends on the index implementation and is usually measured as a constant number of the single operations. This number is referred to as evaluation cost.

      If the effectiveness of a filter evaluates to a number larger than the keySet.size() * <evaluation cost> then a user could avoid using the index and iterate through the keySet calling evaluate rather then applyIndex.

      Specified by:
      calculateEffectiveness in interface IndexAwareFilter<T,E extends Comparable<? super E>>
      Overrides:
      calculateEffectiveness in class AllFilter
      Parameters:
      mapIndexes - the available MapIndex objects keyed by the related ValueExtractor; read-only
      setKeys - the set of keys that will be filtered; read-only
      Returns:
      an effectiveness estimate of how well this filter can use the specified indexes to filter the specified keys
    • explain

      public void explain(QueryContext ctx, QueryRecord.PartialResult.ExplainStep step, Set setKeys)
      Record the projected query execution cost by this filter.

      This method is expected to record the order of execution and estimated cost of applying corresponding indexes in the given step without actually applying any indexes or evaluating entries.

      Specified by:
      explain in interface QueryRecorderFilter<T>
      Overrides:
      explain in class ArrayFilter
      Parameters:
      ctx - the query context
      step - the step used to record the estimated execution cost
      setKeys - the set of keys that would be filtered
    • trace

      public Filter trace(QueryContext ctx, QueryRecord.PartialResult.TraceStep step, Set setKeys)
      Filter the given keys using available indexes and record the cost of execution on the given step of the QueryRecord.

      This method should record the size of the given key set before and after applying corresponding indexes using QueryRecord.PartialResult.RecordableStep.recordPreFilterKeys(int) and QueryRecord.PartialResult.TraceStep.recordPostFilterKeys(int) as well as the corresponding execution time using the QueryRecord.PartialResult.TraceStep.recordDuration(long) method.

      This method is only called if the filter is an IndexAwareFilter and its implementations should explicitly call applyIndex() to actually perform the query. Additionally, this method should return the filter object (if any) returned by the applyIndex() call.

      Specified by:
      trace in interface QueryRecorderFilter<T>
      Overrides:
      trace in class ArrayFilter
      Parameters:
      ctx - the query context
      step - the step used to record the execution cost
      setKeys - the mutable set of keys that remain to be filtered
      Returns:
      the filter returned from IndexAwareFilter.applyIndex(Map, Set)
    • trace

      public boolean trace(QueryContext ctx, QueryRecord.PartialResult.TraceStep step, Map.Entry entry)
      Evaluate the specified entry against this filter and record the evaluation cost on the given step of the QueryRecord.

      This method should record the corresponding latencies using QueryRecord.PartialResult.TraceStep.recordDuration(long).

      Implementations are responsible for explicitly calling evaluateEntry() method to perform the actual entry evaluation. Additionally, this method should return the result of the evaluateEntry call.

      Specified by:
      trace in interface QueryRecorderFilter<T>
      Overrides:
      trace in class ArrayFilter
      Parameters:
      ctx - the context
      step - the step used to record the evaluation cost
      entry - the entry to evaluate
      Returns:
      the result returned from EntryFilter.evaluateEntry(Map.Entry)
    • toString

      public String toString()
      Return a human-readable description for this Filter.
      Overrides:
      toString in class ArrayFilter
      Returns:
      a String description of the Filter
    • evaluateExtracted

      protected boolean evaluateExtracted(Object oExtracted)
      Evaluate the specified extracted value.
      Parameters:
      oExtracted - an extracted value to evaluate
      Returns:
      true if the test passes
    • applySortedIndex

      protected void applySortedIndex(Set setKeys, SortedMap<Object,Set> mapInverted)
      Called by the applyIndex(java.util.Map, java.util.Set) method if the index corresponding to this filter's value extractor is a sorted index.
      Parameters:
      setKeys - the set of keys of the entries being filtered
      mapInverted - the index to apply