public void TestCardinality()
{
Console.WriteLine("testing EWAH GetCardinality");
var bitmap = new EwahCompressedBitArray();
bitmap.Set(int.MaxValue);
//Assert.AreEqual(true, false);
Console.WriteLine("Total Items %d\n", bitmap.GetCardinality());
Assert.AreEqual(bitmap.GetCardinality(), 1UL);
Console.WriteLine("testing EWAH GetCardinality:ok");
}
public void TestNot()
{
Console.WriteLine("testing not");
var bmp = new EwahCompressedBitArray();
for (int i = 0; i <= 184; i++)
{
bmp.Set(i);
}
Assert.AreEqual(185, bmp.GetCardinality());
bmp.Not();
Assert.AreEqual(0, bmp.GetCardinality());
Console.WriteLine("testing not:ok");
}
public void testsetSizeInBits()
{
Console.WriteLine("testing setSizeInBits");
for (int k = 0; k < 4096; ++k)
{
EwahCompressedBitArray ewah = new EwahCompressedBitArray();
ewah.SizeInBits = k;
Assert.AreEqual(ewah.SizeInBits, k);
Assert.AreEqual(ewah.GetCardinality(), 0);
EwahCompressedBitArray ewah2 = new EwahCompressedBitArray();
ewah2.SetSizeInBits(k, false);
Assert.AreEqual(ewah2.SizeInBits, k);
Assert.AreEqual(ewah2.GetCardinality(), 0);
EwahCompressedBitArray ewah3 = new EwahCompressedBitArray();
for (int i = 0; i < k; ++i)
{
ewah3.Set(i);
}
Assert.AreEqual(ewah3.SizeInBits, k);
Assert.AreEqual(ewah3.GetCardinality(), k);
EwahCompressedBitArray ewah4 = new EwahCompressedBitArray();
ewah4.SetSizeInBits(k, true);
Assert.AreEqual(ewah4.SizeInBits, k);
Assert.AreEqual(ewah4.GetCardinality(), k);
}
}
public void TestSizeInBits1()
{
Console.WriteLine("testing TestSizeInBits1");
EwahCompressedBitArray bitmap = new EwahCompressedBitArray();
bitmap.SetSizeInBits(1, false);
Assert.AreEqual(1, bitmap.SizeInBits);
bitmap.Not();
Assert.AreEqual(1, bitmap.GetCardinality());
}
/**
* Convenience function to assess equality between a compressed BitArray
* and an uncompressed BitArray
*
* @param x the compressed BitArray/bitmap
* @param y the uncompressed BitArray/bitmap
*/
private static void AreEqual(EwahCompressedBitArray x, BitArray y)
{
Assert.AreEqual(x.GetCardinality(), y.Cardinality());
var positions = new List <int>();
for (int ii = 0; ii < y.Count; ii++)
{
if (y[ii])
{
positions.Add(ii);
}
}
AreEqual(x.GetPositions(), positions);
}
/**
* Pseudo-non-deterministic test inspired by S.J.vanSchaik.
* (Yes, non-deterministic tests are bad, but the test is actually deterministic.)
*/
/**
* Pseudo-non-deterministic test inspired by Federico Fissore.
*
* @param length the number of set bits in a bitmap
*/
private static void ShouldSetBits(int length)
{
Console.WriteLine("testing shouldSetBits " + length);
int[] bitsToSet = CreateSortedIntArrayOfBitsToSet(length, 434222);
var ewah = new EwahCompressedBitArray();
Console.WriteLine(" ... setting " + bitsToSet.Length + " values");
foreach (int i in bitsToSet)
{
ewah.Set(i);
}
Console.WriteLine(" ... verifying " + bitsToSet.Length + " values");
AreEqual(ewah, bitsToSet);
Console.WriteLine(" ... checking GetCardinality");
Assert.AreEqual(bitsToSet.Length, ewah.GetCardinality());
}
public void SsiYanKaiTest()
{
Console.WriteLine("testing SsiYanKaiTest");
EwahCompressedBitArray a = EwahCompressedBitArray.BitmapOf(39935, 39936, 39937, 39938, 39939, 39940, 39941, 39942, 39943, 39944, 39945, 39946, 39947, 39948, 39949, 39950, 39951, 39952, 39953, 39954, 39955, 39956, 39957, 39958, 39959, 39960, 39961, 39962, 39963, 39964, 39965, 39966, 39967, 39968, 39969, 39970, 39971, 39972, 39973, 39974, 39975, 39976, 39977, 39978, 39979, 39980, 39981, 39982, 39983, 39984, 39985, 39986, 39987, 39988, 39989, 39990, 39991, 39992, 39993, 39994, 39995, 39996, 39997, 39998, 39999, 40000, 40001, 40002, 40003, 40004, 40005, 40006, 40007, 40008, 40009, 40010, 40011, 40012, 40013, 40014, 40015, 40016, 40017, 40018, 40019, 40020, 40021, 40022, 40023, 40024, 40025, 40026, 40027, 40028, 40029, 40030, 40031, 40032, 40033, 40034, 40035, 40036, 40037, 40038, 40039, 40040, 40041, 40042, 40043, 40044, 40045, 40046, 40047, 40048, 40049, 40050, 40051, 40052, 40053, 40054, 40055, 40056, 40057, 40058, 40059, 40060, 40061, 40062, 40063, 40064, 40065, 40066, 40067, 40068, 40069, 40070, 40071, 40072, 40073, 40074, 40075, 40076, 40077, 40078, 40079, 40080, 40081, 40082, 40083, 40084, 40085, 40086, 40087, 40088, 40089, 40090, 40091, 40092, 40093, 40094, 40095, 40096, 40097, 40098, 40099, 40100);
EwahCompressedBitArray b = EwahCompressedBitArray.BitmapOf(39935, 39936, 39937, 39938, 39939, 39940, 39941, 39942, 39943, 39944, 39945, 39946, 39947, 39948, 39949, 39950, 39951, 39952, 39953, 39954, 39955, 39956, 39957, 39958, 39959, 39960, 39961, 39962, 39963, 39964, 39965, 39966, 39967, 39968, 39969, 39970, 39971, 39972, 39973, 39974, 39975, 39976, 39977, 39978, 39979, 39980, 39981, 39982, 39983, 39984, 39985, 39986, 39987, 39988, 39989, 39990, 39991, 39992, 39993, 39994, 39995, 39996, 39997, 39998, 39999, 270000);
HashSet <int> aPositions = new HashSet <int>(a.GetPositions());
int intersection = 0;
EwahCompressedBitArray inter = new EwahCompressedBitArray();
HashSet <int> bPositions = new HashSet <int>(b.GetPositions());
foreach (int integer in bPositions)
{
if (aPositions.Contains(integer))
{
inter.Set(integer);
++intersection;
}
}
EwahCompressedBitArray and2 = a.And(b);
List <int> l1 = inter.GetPositions();
List <int> l2 = and2.GetPositions();
var ok = true;
if (l1.Count != l2.Count)
{
Console.WriteLine("cardinality differs = " + l1.Count + " " + l2.Count);
ok = false;
}
for (int k = 0; k < l1.Count; ++k)
{
if (l1[k] != l2[k])
{
Console.WriteLine("differ at " + k + " = " + l1[k] + " " + l2[k]);
ok = false;
}
}
Assert.IsTrue(ok);
Assert.AreEqual(true, and2.Equals(inter));
Assert.AreEqual(inter.GetHashCode(), and2.GetHashCode());
Assert.AreEqual(intersection, and2.GetCardinality());
}
public void VanSchaikTest()
{
Console.WriteLine("testing vanSchaikTest (this takes some time)");
const int totalNumBits = 32768;
const double odds = 0.9;
var rand = new Random(323232323);
for (int t = 0; t < 100; t++)
{
int numBitsSet = 0;
var cBitMap = new EwahCompressedBitArray();
for (int i = 0; i < totalNumBits; i++)
{
if (rand.NextDouble() < odds)
{
cBitMap.Set(i);
numBitsSet++;
}
}
Assert.AreEqual(cBitMap.GetCardinality(), numBitsSet);
}
Console.WriteLine("testing vanSchaikTest:ok");
}
/**
* Assess equality between an uncompressed bitmap and a compressed one,
* part of a test contributed by Marc Polizzi
*
* @param clrBitArray the uncompressed bitmap
* @param ewahBitmap the compressed bitmap
*/
private static void assertCardinality(BitArray clrBitArray,
EwahCompressedBitArray ewahBitmap)
{
Assert.AreEqual(ewahBitmap.GetCardinality(), clrBitArray.Cardinality());
}
internal QueryResult Query(QueryInfo info, EwahCompressedBitArray exclusionBitMap = null, bool printLoggingMessages = false)
{
var bitMap = GetTagByQueryBitMapLookup(info.Type);
var questionLookup = GetTagByQueryLookup(info.Type)[TagServer.ALL_TAGS_KEY];
// Calculating the Cardinality can be (is?) expensive, we don't want to do it in Queries unless we really need to!?
bool calculateCardinality = true; // false
if (printLoggingMessages)
{
Logger.Log("Tag \"{0}\" is in {1:N0} Questions, Tag \"{2}\" is in {3:N0} Questions",
info.Tag, allTags[info.Tag], info.OtherTag, allTags[info.OtherTag]);
}
//PrintResults(Enumerable.Range(0, questionLookup.Length), qu => questionLookup[qu], TagServer.ALL_TAGS_KEY, info.Type);
//PrintResults(bitMap[info.Tag], qu => questionLookup[qu], info.Tag, info.Type);
//PrintResults(bitMap[info.OtherTag], qu => questionLookup[qu], info.OtherTag, info.Type);
var timer = Stopwatch.StartNew();
var tag1BitMap = bitMap[info.Tag];
var tag2BitMap = bitMap[info.OtherTag];
EwahCompressedBitArray bitMapResult = new EwahCompressedBitArray();
switch (info.Operator)
{
case "AND":
bitMapResult = tag1BitMap.And(tag2BitMap);
break;
case "AND-NOT":
bitMapResult = tag1BitMap.AndNot(tag2BitMap);
break;
case "OR":
bitMapResult = tag1BitMap.Or(tag2BitMap);
break;
case "OR-NOT": //"i.e. .net+or+jquery-"
bitMapResult = tag1BitMap.OrNot(tag2BitMap);
break;
// TODO Work out what a "NOT" query really means, the LINQ version was "result = tag1Query.Except(tag2Query)" (which is the same as AND-NOT?!)
//case "NOT":
// var bitMapResult = (EwahCompressedBitArray)tag2BitMap.Clone();
// bitMapResult.Not();
// break;
default:
throw new InvalidOperationException(string.Format("Invalid operator specified: {0}", info.Operator ?? "<NULL>"));
}
if (exclusionBitMap != null)
{
ulong cardinalityBeforeExclusions = 0;
if (printLoggingMessages)
{
cardinalityBeforeExclusions = bitMapResult.GetCardinality();
}
// The Exclusiong BitMap is Set (i.e. 1) in places where you CAN use the question, i.e. it's NOT excluded
// That way we can efficiently apply the exclusions by ANDing this BitMap to the previous results
var exclusionTimer = Stopwatch.StartNew();
bitMapResult = bitMapResult.AndNot(exclusionBitMap);
exclusionTimer.Stop();
if (printLoggingMessages)
{
if (calculateCardinality)
{
Logger.Log("Took {0,5:N2} ms to apply exclusion BitMap (Cardinality={1:N0}), Results Cardinality: Before={2:N0}, After={3:N0}",
exclusionTimer.Elapsed.TotalMilliseconds, exclusionBitMap.GetCardinality(), cardinalityBeforeExclusions, bitMapResult.GetCardinality());
}
else
{
Logger.Log("Took {0,5:N2} ms to apply exclusion BitMap", exclusionTimer.Elapsed.TotalMilliseconds);
}
}
}
var resultCollectionTimer = Stopwatch.StartNew();
var result = bitMapResult.Skip(info.Skip)
.Take(info.PageSize)
.Select(i => questions[questionLookup[i]])
.ToList();
resultCollectionTimer.Stop();
if (printLoggingMessages)
{
Logger.Log("Took {0,5:N2} ms to collect the results", resultCollectionTimer.Elapsed.TotalMilliseconds);
}
timer.Stop();
Results.AddData(timer.Elapsed.TotalMilliseconds.ToString("#.##"));
if (printLoggingMessages)
{
using (Utils.SetConsoleColour(ConsoleColor.DarkYellow))
{
if (calculateCardinality)
{
Logger.Log("Took {0,5:N2} ms in TOTAL to calculate \"{1} {2} {3}\", Got {4} results, (Result Cardinality={5:N0})",
timer.Elapsed.TotalMilliseconds, info.Tag, info.Operator, info.OtherTag, result.Count, bitMapResult.GetCardinality());
}
else
{
Logger.Log("Took {0,5:N2} ms in TOTAL to calculate \"{1} {2} {3}\", Got {4} results",
timer.Elapsed.TotalMilliseconds, info.Tag, info.Operator, info.OtherTag, result.Count);
}
}
//PrintResults(bitMapResult, qu => questionLookup[qu], string.Format("{0} {1} {2}", info.Tag, info.Operator, info.OtherTag), info.Type);
Logger.Log();
}
return(new QueryResult
{
Questions = result,
// TODO see if we can get meaningful numbers here, WITHOUT calling GetCardinality() (because it's expensive)
//Counters = new Dictionary<string, int>
//{
// { "TagCounter", tagCounter },
// { "OtherTagCounter", otherTagCounter },
// { "ExclusionCounter", exclusionCounter.Counter }
//}
});
}
internal EwahCompressedBitArray CreateBitMapIndexForExcludedTags(CLR.HashSet <string> tagsToExclude, QueryType queryType, bool printLoggingMessages = false)
{
var bitMapTimer = Stopwatch.StartNew();
var tagLookupForQueryType = GetTagByQueryLookup(queryType);
var collectIdsTimer = Stopwatch.StartNew();
var excludedQuestionIds = cache.Value.GetCachedHashSet();
foreach (var tag in tagsToExclude)
{
foreach (var id in tagLookupForQueryType[tag])
{
excludedQuestionIds.Add(id);
}
}
collectIdsTimer.Stop();
// At the end we need to have the BitMap Set (i.e. 1) in places where you CAN use the question, i.e. it's NOT excluded
// That way we can efficiently apply the exclusions by ANDing this BitMap to the previous results
var allQuestions = tagLookupForQueryType[TagServer.ALL_TAGS_KEY];
var setBitsTimer = Stopwatch.StartNew();
var bitMap = new EwahCompressedBitArray();
for (int index = 0; index < allQuestions.Length; index++)
{
if (excludedQuestionIds.Contains(allQuestions[index]))
{
var wasSet = bitMap.SetOptimised(index); // Set a bit where you CAN'T use a question
if (wasSet == false)
{
Logger.LogStartupMessage("Error, unable to set bit {0:N0} (SizeInBits = {1:N0})", index, bitMap.SizeInBits);
}
}
}
setBitsTimer.Stop();
var tidyUpTimer = Stopwatch.StartNew();
bitMap.SetSizeInBits(questions.Count, defaultvalue: false);
bitMap.Shrink();
tidyUpTimer.Stop();
bitMapTimer.Stop();
if (printLoggingMessages)
{
Logger.LogStartupMessage("Took {0} ({1,6:N0} ms) to collect {2:N0} Question Ids from {3:N0} Tags",
collectIdsTimer.Elapsed, collectIdsTimer.ElapsedMilliseconds, excludedQuestionIds.Count, tagsToExclude.Count);
Logger.LogStartupMessage("Took {0} ({1,6:N0} ms) to set {2:N0} bits",
setBitsTimer.Elapsed, setBitsTimer.ElapsedMilliseconds, bitMap.GetCardinality());
Logger.LogStartupMessage("Took {0} ({1,6:N0} ms) to tidy-up the Bit Map (SetSizeInBits(..) and Shrink()), Size={2:N0} bytes ({3:N2} MB)",
tidyUpTimer.Elapsed, tidyUpTimer.ElapsedMilliseconds, bitMap.SizeInBytes, bitMap.SizeInBytes / 1024.0 / 1024.0);
using (Utils.SetConsoleColour(ConsoleColor.DarkYellow))
{
Logger.LogStartupMessage("Took {0} ({1,6:N0} ms) in TOTAL, made BitMap from {2:N0} Tags ({3:N0} Qu Ids), Cardinality={4:N0} ({5:N0})\n",
bitMapTimer.Elapsed, bitMapTimer.ElapsedMilliseconds,
tagsToExclude.Count,
excludedQuestionIds.Count,
bitMap.GetCardinality(),
(ulong)questions.Count - bitMap.GetCardinality());
}
}
return(bitMap);
}
