pylucene 3.5.0-3
[pylucene.git] / lucene-java-3.5.0 / lucene / src / java / org / apache / lucene / search / FieldCacheRangeFilter.java
diff --git a/lucene-java-3.5.0/lucene/src/java/org/apache/lucene/search/FieldCacheRangeFilter.java b/lucene-java-3.5.0/lucene/src/java/org/apache/lucene/search/FieldCacheRangeFilter.java
new file mode 100644 (file)
index 0000000..067f881
--- /dev/null
@@ -0,0 +1,602 @@
+package org.apache.lucene.search;
+/**
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+import java.io.IOException;
+
+import org.apache.lucene.index.IndexReader;
+import org.apache.lucene.index.TermDocs;
+import org.apache.lucene.util.NumericUtils;
+import org.apache.lucene.document.NumericField; // for javadocs
+
+/**
+ * A range filter built on top of a cached single term field (in {@link FieldCache}).
+ * 
+ * <p>{@code FieldCacheRangeFilter} builds a single cache for the field the first time it is used.
+ * Each subsequent {@code FieldCacheRangeFilter} on the same field then reuses this cache,
+ * even if the range itself changes. 
+ * 
+ * <p>This means that {@code FieldCacheRangeFilter} is much faster (sometimes more than 100x as fast) 
+ * as building a {@link TermRangeFilter}, if using a {@link #newStringRange}.
+ * However, if the range never changes it is slower (around 2x as slow) than building
+ * a CachingWrapperFilter on top of a single {@link TermRangeFilter}.
+ *
+ * For numeric data types, this filter may be significantly faster than {@link NumericRangeFilter}.
+ * Furthermore, it does not need the numeric values encoded by {@link NumericField}. But
+ * it has the problem that it only works with exact one value/document (see below).
+ *
+ * <p>As with all {@link FieldCache} based functionality, {@code FieldCacheRangeFilter} is only valid for 
+ * fields which exact one term for each document (except for {@link #newStringRange}
+ * where 0 terms are also allowed). Due to a restriction of {@link FieldCache}, for numeric ranges
+ * all terms that do not have a numeric value, 0 is assumed.
+ *
+ * <p>Thus it works on dates, prices and other single value fields but will not work on
+ * regular text fields. It is preferable to use a <code>NOT_ANALYZED</code> field to ensure that
+ * there is only a single term. 
+ *
+ * <p>This class does not have an constructor, use one of the static factory methods available,
+ * that create a correct instance for different data types supported by {@link FieldCache}.
+ */
+
+public abstract class FieldCacheRangeFilter<T> extends Filter {
+  final String field;
+  final FieldCache.Parser parser;
+  final T lowerVal;
+  final T upperVal;
+  final boolean includeLower;
+  final boolean includeUpper;
+  
+  private FieldCacheRangeFilter(String field, FieldCache.Parser parser, T lowerVal, T upperVal, boolean includeLower, boolean includeUpper) {
+    this.field = field;
+    this.parser = parser;
+    this.lowerVal = lowerVal;
+    this.upperVal = upperVal;
+    this.includeLower = includeLower;
+    this.includeUpper = includeUpper;
+  }
+  
+  /** This method is implemented for each data type */
+  @Override
+  public abstract DocIdSet getDocIdSet(IndexReader reader) throws IOException;
+
+  /**
+   * Creates a string range filter using {@link FieldCache#getStringIndex}. This works with all
+   * fields containing zero or one term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<String> newStringRange(String field, String lowerVal, String upperVal, boolean includeLower, boolean includeUpper) {
+    return new FieldCacheRangeFilter<String>(field, null, lowerVal, upperVal, includeLower, includeUpper) {
+      @Override
+      public DocIdSet getDocIdSet(IndexReader reader) throws IOException {
+        final FieldCache.StringIndex fcsi = FieldCache.DEFAULT.getStringIndex(reader, field);
+        final int lowerPoint = fcsi.binarySearchLookup(lowerVal);
+        final int upperPoint = fcsi.binarySearchLookup(upperVal);
+        
+        final int inclusiveLowerPoint, inclusiveUpperPoint;
+
+        // Hints:
+        // * binarySearchLookup returns 0, if value was null.
+        // * the value is <0 if no exact hit was found, the returned value
+        //   is (-(insertion point) - 1)
+        if (lowerPoint == 0) {
+          assert lowerVal == null;
+          inclusiveLowerPoint = 1;
+        } else if (includeLower && lowerPoint > 0) {
+          inclusiveLowerPoint = lowerPoint;
+        } else if (lowerPoint > 0) {
+          inclusiveLowerPoint = lowerPoint + 1;
+        } else {
+          inclusiveLowerPoint = Math.max(1, -lowerPoint - 1);
+        }
+        
+        if (upperPoint == 0) {
+          assert upperVal == null;
+          inclusiveUpperPoint = Integer.MAX_VALUE;  
+        } else if (includeUpper && upperPoint > 0) {
+          inclusiveUpperPoint = upperPoint;
+        } else if (upperPoint > 0) {
+          inclusiveUpperPoint = upperPoint - 1;
+        } else {
+          inclusiveUpperPoint = -upperPoint - 2;
+        }      
+
+        if (inclusiveUpperPoint <= 0 || inclusiveLowerPoint > inclusiveUpperPoint)
+          return DocIdSet.EMPTY_DOCIDSET;
+        
+        assert inclusiveLowerPoint > 0 && inclusiveUpperPoint > 0;
+        
+        // for this DocIdSet, we never need to use TermDocs,
+        // because deleted docs have an order of 0 (null entry in StringIndex)
+        return new FieldCacheDocIdSet(reader, false) {
+          @Override
+          final boolean matchDoc(int doc) {
+            return fcsi.order[doc] >= inclusiveLowerPoint && fcsi.order[doc] <= inclusiveUpperPoint;
+          }
+        };
+      }
+    };
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getBytes(IndexReader,String)}. This works with all
+   * byte fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Byte> newByteRange(String field, Byte lowerVal, Byte upperVal, boolean includeLower, boolean includeUpper) {
+    return newByteRange(field, null, lowerVal, upperVal, includeLower, includeUpper);
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getBytes(IndexReader,String,FieldCache.ByteParser)}. This works with all
+   * byte fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Byte> newByteRange(String field, FieldCache.ByteParser parser, Byte lowerVal, Byte upperVal, boolean includeLower, boolean includeUpper) {
+    return new FieldCacheRangeFilter<Byte>(field, parser, lowerVal, upperVal, includeLower, includeUpper) {
+      @Override
+      public DocIdSet getDocIdSet(IndexReader reader) throws IOException {
+        final byte inclusiveLowerPoint, inclusiveUpperPoint;
+        if (lowerVal != null) {
+          final byte i = lowerVal.byteValue();
+          if (!includeLower && i == Byte.MAX_VALUE)
+            return DocIdSet.EMPTY_DOCIDSET;
+          inclusiveLowerPoint = (byte) (includeLower ?  i : (i + 1));
+        } else {
+          inclusiveLowerPoint = Byte.MIN_VALUE;
+        }
+        if (upperVal != null) {
+          final byte i = upperVal.byteValue();
+          if (!includeUpper && i == Byte.MIN_VALUE)
+            return DocIdSet.EMPTY_DOCIDSET;
+          inclusiveUpperPoint = (byte) (includeUpper ? i : (i - 1));
+        } else {
+          inclusiveUpperPoint = Byte.MAX_VALUE;
+        }
+        
+        if (inclusiveLowerPoint > inclusiveUpperPoint)
+          return DocIdSet.EMPTY_DOCIDSET;
+        
+        final byte[] values = FieldCache.DEFAULT.getBytes(reader, field, (FieldCache.ByteParser) parser);
+        // we only request the usage of termDocs, if the range contains 0
+        return new FieldCacheDocIdSet(reader, (inclusiveLowerPoint <= 0 && inclusiveUpperPoint >= 0)) {
+          @Override
+          boolean matchDoc(int doc) {
+            return values[doc] >= inclusiveLowerPoint && values[doc] <= inclusiveUpperPoint;
+          }
+        };
+      }
+    };
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getShorts(IndexReader,String)}. This works with all
+   * short fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Short> newShortRange(String field, Short lowerVal, Short upperVal, boolean includeLower, boolean includeUpper) {
+    return newShortRange(field, null, lowerVal, upperVal, includeLower, includeUpper);
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getShorts(IndexReader,String,FieldCache.ShortParser)}. This works with all
+   * short fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Short> newShortRange(String field, FieldCache.ShortParser parser, Short lowerVal, Short upperVal, boolean includeLower, boolean includeUpper) {
+    return new FieldCacheRangeFilter<Short>(field, parser, lowerVal, upperVal, includeLower, includeUpper) {
+      @Override
+      public DocIdSet getDocIdSet(IndexReader reader) throws IOException {
+        final short inclusiveLowerPoint, inclusiveUpperPoint;
+        if (lowerVal != null) {
+          short i = lowerVal.shortValue();
+          if (!includeLower && i == Short.MAX_VALUE)
+            return DocIdSet.EMPTY_DOCIDSET;
+          inclusiveLowerPoint = (short) (includeLower ? i : (i + 1));
+        } else {
+          inclusiveLowerPoint = Short.MIN_VALUE;
+        }
+        if (upperVal != null) {
+          short i = upperVal.shortValue();
+          if (!includeUpper && i == Short.MIN_VALUE)
+            return DocIdSet.EMPTY_DOCIDSET;
+          inclusiveUpperPoint = (short) (includeUpper ? i : (i - 1));
+        } else {
+          inclusiveUpperPoint = Short.MAX_VALUE;
+        }
+        
+        if (inclusiveLowerPoint > inclusiveUpperPoint)
+          return DocIdSet.EMPTY_DOCIDSET;
+        
+        final short[] values = FieldCache.DEFAULT.getShorts(reader, field, (FieldCache.ShortParser) parser);
+        // we only request the usage of termDocs, if the range contains 0
+        return new FieldCacheDocIdSet(reader, (inclusiveLowerPoint <= 0 && inclusiveUpperPoint >= 0)) {
+          @Override
+          boolean matchDoc(int doc) {
+            return values[doc] >= inclusiveLowerPoint && values[doc] <= inclusiveUpperPoint;
+          }
+        };
+      }
+    };
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getInts(IndexReader,String)}. This works with all
+   * int fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Integer> newIntRange(String field, Integer lowerVal, Integer upperVal, boolean includeLower, boolean includeUpper) {
+    return newIntRange(field, null, lowerVal, upperVal, includeLower, includeUpper);
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getInts(IndexReader,String,FieldCache.IntParser)}. This works with all
+   * int fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Integer> newIntRange(String field, FieldCache.IntParser parser, Integer lowerVal, Integer upperVal, boolean includeLower, boolean includeUpper) {
+    return new FieldCacheRangeFilter<Integer>(field, parser, lowerVal, upperVal, includeLower, includeUpper) {
+      @Override
+      public DocIdSet getDocIdSet(IndexReader reader) throws IOException {
+        final int inclusiveLowerPoint, inclusiveUpperPoint;
+        if (lowerVal != null) {
+          int i = lowerVal.intValue();
+          if (!includeLower && i == Integer.MAX_VALUE)
+            return DocIdSet.EMPTY_DOCIDSET;
+          inclusiveLowerPoint = includeLower ? i : (i + 1);
+        } else {
+          inclusiveLowerPoint = Integer.MIN_VALUE;
+        }
+        if (upperVal != null) {
+          int i = upperVal.intValue();
+          if (!includeUpper && i == Integer.MIN_VALUE)
+            return DocIdSet.EMPTY_DOCIDSET;
+          inclusiveUpperPoint = includeUpper ? i : (i - 1);
+        } else {
+          inclusiveUpperPoint = Integer.MAX_VALUE;
+        }
+        
+        if (inclusiveLowerPoint > inclusiveUpperPoint)
+          return DocIdSet.EMPTY_DOCIDSET;
+        
+        final int[] values = FieldCache.DEFAULT.getInts(reader, field, (FieldCache.IntParser) parser);
+        // we only request the usage of termDocs, if the range contains 0
+        return new FieldCacheDocIdSet(reader, (inclusiveLowerPoint <= 0 && inclusiveUpperPoint >= 0)) {
+          @Override
+          boolean matchDoc(int doc) {
+            return values[doc] >= inclusiveLowerPoint && values[doc] <= inclusiveUpperPoint;
+          }
+        };
+      }
+    };
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getLongs(IndexReader,String)}. This works with all
+   * long fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Long> newLongRange(String field, Long lowerVal, Long upperVal, boolean includeLower, boolean includeUpper) {
+    return newLongRange(field, null, lowerVal, upperVal, includeLower, includeUpper);
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getLongs(IndexReader,String,FieldCache.LongParser)}. This works with all
+   * long fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Long> newLongRange(String field, FieldCache.LongParser parser, Long lowerVal, Long upperVal, boolean includeLower, boolean includeUpper) {
+    return new FieldCacheRangeFilter<Long>(field, parser, lowerVal, upperVal, includeLower, includeUpper) {
+      @Override
+      public DocIdSet getDocIdSet(IndexReader reader) throws IOException {
+        final long inclusiveLowerPoint, inclusiveUpperPoint;
+        if (lowerVal != null) {
+          long i = lowerVal.longValue();
+          if (!includeLower && i == Long.MAX_VALUE)
+            return DocIdSet.EMPTY_DOCIDSET;
+          inclusiveLowerPoint = includeLower ? i : (i + 1L);
+        } else {
+          inclusiveLowerPoint = Long.MIN_VALUE;
+        }
+        if (upperVal != null) {
+          long i = upperVal.longValue();
+          if (!includeUpper && i == Long.MIN_VALUE)
+            return DocIdSet.EMPTY_DOCIDSET;
+          inclusiveUpperPoint = includeUpper ? i : (i - 1L);
+        } else {
+          inclusiveUpperPoint = Long.MAX_VALUE;
+        }
+        
+        if (inclusiveLowerPoint > inclusiveUpperPoint)
+          return DocIdSet.EMPTY_DOCIDSET;
+        
+        final long[] values = FieldCache.DEFAULT.getLongs(reader, field, (FieldCache.LongParser) parser);
+        // we only request the usage of termDocs, if the range contains 0
+        return new FieldCacheDocIdSet(reader, (inclusiveLowerPoint <= 0L && inclusiveUpperPoint >= 0L)) {
+          @Override
+          boolean matchDoc(int doc) {
+            return values[doc] >= inclusiveLowerPoint && values[doc] <= inclusiveUpperPoint;
+          }
+        };
+      }
+    };
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getFloats(IndexReader,String)}. This works with all
+   * float fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Float> newFloatRange(String field, Float lowerVal, Float upperVal, boolean includeLower, boolean includeUpper) {
+    return newFloatRange(field, null, lowerVal, upperVal, includeLower, includeUpper);
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getFloats(IndexReader,String,FieldCache.FloatParser)}. This works with all
+   * float fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Float> newFloatRange(String field, FieldCache.FloatParser parser, Float lowerVal, Float upperVal, boolean includeLower, boolean includeUpper) {
+    return new FieldCacheRangeFilter<Float>(field, parser, lowerVal, upperVal, includeLower, includeUpper) {
+      @Override
+      public DocIdSet getDocIdSet(IndexReader reader) throws IOException {
+        // we transform the floating point numbers to sortable integers
+        // using NumericUtils to easier find the next bigger/lower value
+        final float inclusiveLowerPoint, inclusiveUpperPoint;
+        if (lowerVal != null) {
+          float f = lowerVal.floatValue();
+          if (!includeUpper && f > 0.0f && Float.isInfinite(f))
+            return DocIdSet.EMPTY_DOCIDSET;
+          int i = NumericUtils.floatToSortableInt(f);
+          inclusiveLowerPoint = NumericUtils.sortableIntToFloat( includeLower ?  i : (i + 1) );
+        } else {
+          inclusiveLowerPoint = Float.NEGATIVE_INFINITY;
+        }
+        if (upperVal != null) {
+          float f = upperVal.floatValue();
+          if (!includeUpper && f < 0.0f && Float.isInfinite(f))
+            return DocIdSet.EMPTY_DOCIDSET;
+          int i = NumericUtils.floatToSortableInt(f);
+          inclusiveUpperPoint = NumericUtils.sortableIntToFloat( includeUpper ? i : (i - 1) );
+        } else {
+          inclusiveUpperPoint = Float.POSITIVE_INFINITY;
+        }
+        
+        if (inclusiveLowerPoint > inclusiveUpperPoint)
+          return DocIdSet.EMPTY_DOCIDSET;
+        
+        final float[] values = FieldCache.DEFAULT.getFloats(reader, field, (FieldCache.FloatParser) parser);
+        // we only request the usage of termDocs, if the range contains 0
+        return new FieldCacheDocIdSet(reader, (inclusiveLowerPoint <= 0.0f && inclusiveUpperPoint >= 0.0f)) {
+          @Override
+          boolean matchDoc(int doc) {
+            return values[doc] >= inclusiveLowerPoint && values[doc] <= inclusiveUpperPoint;
+          }
+        };
+      }
+    };
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getDoubles(IndexReader,String)}. This works with all
+   * double fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Double> newDoubleRange(String field, Double lowerVal, Double upperVal, boolean includeLower, boolean includeUpper) {
+    return newDoubleRange(field, null, lowerVal, upperVal, includeLower, includeUpper);
+  }
+  
+  /**
+   * Creates a numeric range filter using {@link FieldCache#getDoubles(IndexReader,String,FieldCache.DoubleParser)}. This works with all
+   * double fields containing exactly one numeric term in the field. The range can be half-open by setting one
+   * of the values to <code>null</code>.
+   */
+  public static FieldCacheRangeFilter<Double> newDoubleRange(String field, FieldCache.DoubleParser parser, Double lowerVal, Double upperVal, boolean includeLower, boolean includeUpper) {
+    return new FieldCacheRangeFilter<Double>(field, parser, lowerVal, upperVal, includeLower, includeUpper) {
+      @Override
+      public DocIdSet getDocIdSet(IndexReader reader) throws IOException {
+        // we transform the floating point numbers to sortable integers
+        // using NumericUtils to easier find the next bigger/lower value
+        final double inclusiveLowerPoint, inclusiveUpperPoint;
+        if (lowerVal != null) {
+          double f = lowerVal.doubleValue();
+          if (!includeUpper && f > 0.0 && Double.isInfinite(f))
+            return DocIdSet.EMPTY_DOCIDSET;
+          long i = NumericUtils.doubleToSortableLong(f);
+          inclusiveLowerPoint = NumericUtils.sortableLongToDouble( includeLower ?  i : (i + 1L) );
+        } else {
+          inclusiveLowerPoint = Double.NEGATIVE_INFINITY;
+        }
+        if (upperVal != null) {
+          double f = upperVal.doubleValue();
+          if (!includeUpper && f < 0.0 && Double.isInfinite(f))
+            return DocIdSet.EMPTY_DOCIDSET;
+          long i = NumericUtils.doubleToSortableLong(f);
+          inclusiveUpperPoint = NumericUtils.sortableLongToDouble( includeUpper ? i : (i - 1L) );
+        } else {
+          inclusiveUpperPoint = Double.POSITIVE_INFINITY;
+        }
+        
+        if (inclusiveLowerPoint > inclusiveUpperPoint)
+          return DocIdSet.EMPTY_DOCIDSET;
+        
+        final double[] values = FieldCache.DEFAULT.getDoubles(reader, field, (FieldCache.DoubleParser) parser);
+        // we only request the usage of termDocs, if the range contains 0
+        return new FieldCacheDocIdSet(reader, (inclusiveLowerPoint <= 0.0 && inclusiveUpperPoint >= 0.0)) {
+          @Override
+          boolean matchDoc(int doc) {
+            return values[doc] >= inclusiveLowerPoint && values[doc] <= inclusiveUpperPoint;
+          }
+        };
+      }
+    };
+  }
+  
+  @Override
+  public final String toString() {
+    final StringBuilder sb = new StringBuilder(field).append(":");
+    return sb.append(includeLower ? '[' : '{')
+      .append((lowerVal == null) ? "*" : lowerVal.toString())
+      .append(" TO ")
+      .append((upperVal == null) ? "*" : upperVal.toString())
+      .append(includeUpper ? ']' : '}')
+      .toString();
+  }
+
+  @Override
+  public final boolean equals(Object o) {
+    if (this == o) return true;
+    if (!(o instanceof FieldCacheRangeFilter)) return false;
+    FieldCacheRangeFilter other = (FieldCacheRangeFilter) o;
+
+    if (!this.field.equals(other.field)
+        || this.includeLower != other.includeLower
+        || this.includeUpper != other.includeUpper
+    ) { return false; }
+    if (this.lowerVal != null ? !this.lowerVal.equals(other.lowerVal) : other.lowerVal != null) return false;
+    if (this.upperVal != null ? !this.upperVal.equals(other.upperVal) : other.upperVal != null) return false;
+    if (this.parser != null ? !this.parser.equals(other.parser) : other.parser != null) return false;
+    return true;
+  }
+  
+  @Override
+  public final int hashCode() {
+    int h = field.hashCode();
+    h ^= (lowerVal != null) ? lowerVal.hashCode() : 550356204;
+    h = (h << 1) | (h >>> 31);  // rotate to distinguish lower from upper
+    h ^= (upperVal != null) ? upperVal.hashCode() : -1674416163;
+    h ^= (parser != null) ? parser.hashCode() : -1572457324;
+    h ^= (includeLower ? 1549299360 : -365038026) ^ (includeUpper ? 1721088258 : 1948649653);
+    return h;
+  }
+
+  /** Returns the field name for this filter */
+  public String getField() { return field; }
+
+  /** Returns <code>true</code> if the lower endpoint is inclusive */
+  public boolean includesLower() { return includeLower; }
+  
+  /** Returns <code>true</code> if the upper endpoint is inclusive */
+  public boolean includesUpper() { return includeUpper; }
+
+  /** Returns the lower value of this range filter */
+  public T getLowerVal() { return lowerVal; }
+
+  /** Returns the upper value of this range filter */
+  public T getUpperVal() { return upperVal; }
+  
+  /** Returns the current numeric parser ({@code null} for {@code T} is {@code String}} */
+  public FieldCache.Parser getParser() { return parser; }
+  
+  static abstract class FieldCacheDocIdSet extends DocIdSet {
+    private final IndexReader reader;
+    private boolean mayUseTermDocs;
+  
+    FieldCacheDocIdSet(IndexReader reader, boolean mayUseTermDocs) {
+      this.reader = reader;
+      this.mayUseTermDocs = mayUseTermDocs;
+    }
+  
+    /** this method checks, if a doc is a hit, should throw AIOBE, when position invalid */
+    abstract boolean matchDoc(int doc) throws ArrayIndexOutOfBoundsException;
+    
+    /** this DocIdSet is cacheable, if it works solely with FieldCache and no TermDocs */
+    @Override
+    public boolean isCacheable() {
+      return !(mayUseTermDocs && reader.hasDeletions());
+    }
+
+    @Override
+    public DocIdSetIterator iterator() throws IOException {
+      // Synchronization needed because deleted docs BitVector
+      // can change after call to hasDeletions until TermDocs creation.
+      // We only use an iterator with termDocs, when this was requested (e.g. range contains 0)
+      // and the index has deletions
+      final TermDocs termDocs;
+      synchronized(reader) {
+        termDocs = isCacheable() ? null : reader.termDocs(null);
+      }
+      if (termDocs != null) {
+        // a DocIdSetIterator using TermDocs to iterate valid docIds
+        return new DocIdSetIterator() {
+          private int doc = -1;
+          
+          @Override
+          public int docID() {
+            return doc;
+          }
+          
+          @Override
+          public int nextDoc() throws IOException {
+            do {
+              if (!termDocs.next())
+                return doc = NO_MORE_DOCS;
+            } while (!matchDoc(doc = termDocs.doc()));
+            return doc;
+          }
+          
+          @Override
+          public int advance(int target) throws IOException {
+            if (!termDocs.skipTo(target))
+              return doc = NO_MORE_DOCS;
+            while (!matchDoc(doc = termDocs.doc())) { 
+              if (!termDocs.next())
+                return doc = NO_MORE_DOCS;
+            }
+            return doc;
+          }
+        };
+      } else {
+        // a DocIdSetIterator generating docIds by incrementing a variable -
+        // this one can be used if there are no deletions are on the index
+        return new DocIdSetIterator() {
+          private int doc = -1;
+          
+          @Override
+          public int docID() {
+            return doc;
+          }
+          
+          @Override
+          public int nextDoc() {
+            try {
+              do {
+                doc++;
+              } while (!matchDoc(doc));
+              return doc;
+            } catch (ArrayIndexOutOfBoundsException e) {
+              return doc = NO_MORE_DOCS;
+            }
+          }
+          
+          @Override
+          public int advance(int target) {
+            try {
+              doc = target;
+              while (!matchDoc(doc)) { 
+                doc++;
+              }
+              return doc;
+            } catch (ArrayIndexOutOfBoundsException e) {
+              return doc = NO_MORE_DOCS;
+            }
+          }
+        };
+      }
+    }
+  }
+
+}