public System.Object Clone()
{
byte[] copyBits = new byte[bits.Length];
Array.Copy(bits, 0, copyBits, 0, bits.Length);
BitVector clone = new BitVector(copyBits, size);
clone.count = count;
return clone;
}
public override sealed int FindValues(BitVector bitset, int docId, int maxId)
{
while (true)
{
if (bitset.Get(_array[docId >> SHIFT_SIZE][docId & MASK])) return docId;
if (docId++ >= maxId) break;
}
return DocIdSetIterator.NO_MORE_DOCS;
}
private void DoTestGetSetVectorOfSize(int n)
{
BitVector bv = new BitVector(n);
for (int i = 0; i < bv.Size(); i++)
{
// ensure a set bit can be git'
Assert.IsFalse(bv.Get(i));
bv.Set(i);
Assert.IsTrue(bv.Get(i));
}
}
private void DoTestClearVectorOfSize(int n)
{
BitVector bv = new BitVector(n);
for (int i = 0; i < bv.Size(); i++)
{
// ensure a set bit is cleared
Assert.IsFalse(bv.Get(i));
bv.Set(i);
Assert.IsTrue(bv.Get(i));
bv.Clear(i);
Assert.IsFalse(bv.Get(i));
}
}
private BigSegmentedArray GetCollapsedCounts()
{
if (_collapsedCounts == null)
{
_collapsedCounts = new LazyBigIntArray(_bucketValues.Count);
FacetDataCache dataCache = _subCollector.DataCache;
ITermValueList subList = dataCache.ValArray;
BigSegmentedArray subcounts = _subCollector.Count;
BitVector indexSet = new BitVector(subcounts.Size());
int c = 0;
int i = 0;
foreach (string val in _bucketValues)
{
if (val.Length > 0)
{
string[] subVals = _predefinedBuckets.Get(val);
int count = 0;
foreach (string subVal in subVals)
{
int index = subList.IndexOf(subVal);
if (index > 0)
{
int subcount = subcounts.Get(index);
count += subcount;
if (!indexSet.Get(index))
{
indexSet.Set(index);
c += dataCache.Freqs[index];
}
}
}
_collapsedCounts.Add(i, count);
}
i++;
}
_collapsedCounts.Add(0, (_numdocs - c));
}
return _collapsedCounts;
}
private void DoTestConstructOfSize(int n)
{
BitVector bv = new BitVector(n);
Assert.AreEqual(n, bv.Size());
}
private void DoTestConstructOfSize(int n)
{
BitVector bv = new BitVector(n);
Assert.AreEqual(n, bv.Size());
}
private BitVector CreateSubsetTestVector()
{
BitVector bv = new BitVector(subsetPattern.Length);
for (int i = 0; i < subsetPattern.Length; i++)
{
if (subsetPattern[i] == 1)
{
bv.Set(i);
}
}
return bv;
}
/// <summary> Compare two BitVectors.
/// This should really be an equals method on the BitVector itself.
/// </summary>
/// <param name="bv">One bit vector
/// </param>
/// <param name="compare">The second to compare
/// </param>
private bool DoCompare(BitVector bv, BitVector compare)
{
bool equal = true;
for (int i = 0; i < bv.Size(); i++)
{
// bits must be equal
if (bv.Get(i) != compare.Get(i))
{
equal = false;
break;
}
}
return equal;
}
private void DoTestDgaps(int size, int count1, int count2)
{
Directory d = new RAMDirectory();
BitVector bv = new BitVector(size);
for (int i = 0; i < count1; i++)
{
bv.Set(i);
Assert.AreEqual(i + 1, bv.Count());
}
bv.Write(d, "TESTBV");
// gradually increase number of set bits
for (int i = count1; i < count2; i++)
{
BitVector bv2 = new BitVector(d, "TESTBV");
Assert.IsTrue(DoCompare(bv, bv2));
bv = bv2;
bv.Set(i);
Assert.AreEqual(i + 1, bv.Count());
bv.Write(d, "TESTBV");
}
// now start decreasing number of set bits
for (int i = count2 - 1; i >= count1; i--)
{
BitVector bv2 = new BitVector(d, "TESTBV");
Assert.IsTrue(DoCompare(bv, bv2));
bv = bv2;
bv.Clear(i);
Assert.AreEqual(i, bv.Count());
bv.Write(d, "TESTBV");
}
}
private void DoTestWriteRead(int n)
{
Directory d = new RAMDirectory();
BitVector bv = new BitVector(n);
// test count when incrementally setting bits
for (int i = 0; i < bv.Size(); i++)
{
Assert.IsFalse(bv.Get(i));
Assert.AreEqual(i, bv.Count());
bv.Set(i);
Assert.IsTrue(bv.Get(i));
Assert.AreEqual(i + 1, bv.Count());
bv.Write(d, "TESTBV");
BitVector compare = new BitVector(d, "TESTBV");
// compare bit vectors with bits set incrementally
Assert.IsTrue(DoCompare(bv, compare));
}
}
private void DoTestCountVectorOfSize(int n)
{
BitVector bv = new BitVector(n);
// test count when incrementally setting bits
for (int i = 0; i < bv.Size(); i++)
{
Assert.IsFalse(bv.Get(i));
Assert.AreEqual(i, bv.Count());
bv.Set(i);
Assert.IsTrue(bv.Get(i));
Assert.AreEqual(i + 1, bv.Count());
}
bv = new BitVector(n);
// test count when setting then clearing bits
for (int i = 0; i < bv.Size(); i++)
{
Assert.IsFalse(bv.Get(i));
Assert.AreEqual(0, bv.Count());
bv.Set(i);
Assert.IsTrue(bv.Get(i));
Assert.AreEqual(1, bv.Count());
bv.Clear(i);
Assert.IsFalse(bv.Get(i));
Assert.AreEqual(0, bv.Count());
}
}
/// <summary>
/// (non-Javadoc)
/// see com.browseengine.bobo.util.BigSegmentedArray#findValues(org.apache.lucene.util.BitVector, int, int)
/// </summary>
/// <param name="bitset"></param>
/// <param name="id"></param>
/// <param name="maxId"></param>
/// <returns></returns>
public override int FindValues(BitVector bitset, int id, int maxId)
{
while (id <= maxId)
{
int i = id >> SHIFT_SIZE;
if (_array[i] == null)
{
if (bitset.Get(_fillValue))
return id;
else
id = (i + 1) << SHIFT_SIZE; // jump to next segment
}
else
{
if (bitset.Get(_array[i][id & MASK]))
return id;
else
id++;
}
}
return DocIdSetIterator.NO_MORE_DOCS;
}
public abstract int FindValues(BitVector bitset, int id, int maxId);
public int FindValues(BitVector values, int id, int maxID, bool withMissing)
{
// NOTE: Added Get() extension method call because
// the default .NET behavior throws an exception if the
// index is out of bounds, rather than returning null.
int[] page = _list.Get(id >> PAGEID_SHIFT);
if (page == null) page = MISSING_PAGE;
while (true)
{
int val = page[id & SLOTID_MASK];
if (val >= 0)
{
if (values.Get(val)) return id;
}
else if (val != MISSING)
{
int idx = -(val >> VALIDX_SHIFT);// signed shift, remember this is a negative number
int end = idx + (val & COUNT_MASK);
while (idx < end)
{
if (values.Get(page[idx++])) return id;
}
}
else if (withMissing)
{
if (values.Get(0)) return id;
}
if (id >= maxID) break;
if ((++id & SLOTID_MASK) == 0)
{
// NOTE: Added Get() extension method call because
// the default .NET behavior throws an exception if the
// index is out of bounds, rather than returning null.
page = _list.Get(id >> PAGEID_SHIFT);
if (page == null) page = MISSING_PAGE;
}
}
return DocIdSetIterator.NO_MORE_DOCS;
}
public int FindValues(BitVector values, int id, int maxID)
{
return FindValues(values, id, maxID, false);
}
public bool Contains(int id, BitVector values)
{
// NOTE: Added Get() extension method call because
// the default .NET behavior throws an exception if the
// index is out of bounds, rather than returning null.
int[] page = _list.Get(id >> PAGEID_SHIFT);
if (page == null) return false;
int val = page[id & SLOTID_MASK];
if (val >= 0)
{
return (values.Get(val));
}
else if (val != MISSING)
{
int idx = -(val >> VALIDX_SHIFT);// signed shift, remember this is a negative number
int end = idx + (val & COUNT_MASK);
while (idx < end)
{
if (values.Get(page[idx++])) return true;
}
}
return false;
}
public void CountNoReturnWithFilter(int id, int[] count, BitVector filter)
{
// NOTE: Added Get() extension method call because
// the default .NET behavior throws an exception if the
// index is out of bounds, rather than returning null.
int[] page = _list.Get(id >> PAGEID_SHIFT);
if (page == null)
{
count[0]++;
return;
}
int val = page[id & SLOTID_MASK];
if (val >= 0)
{
if (filter.Get(val))
{
count[val]++;
}
return;
}
else if (val != MISSING)
{
int idx = -(val >> VALIDX_SHIFT); // signed shift, remember val is a negative number
int cnt = (val & COUNT_MASK);
int end = idx + cnt;
while (idx < end)
{
int value = page[idx++];
if (filter.Get(value))
{
count[value]++;
}
}
return;
}
count[0]++;
return;
}
