private static void PlainSearch(string src, string sub)
{
RkSearch rksearch = new RkSearch();
SimpleSearch ssearch = new SimpleSearch();
DumbSearch dsearch = new DumbSearch();
DateTime start, end;
int idx = 0;
Console.WriteLine("Searching");
Console.WriteLine("Dumb Search ... ");
start = DateTime.Now;
idx = dsearch.Search(src, sub);
end = DateTime.Now;
Console.WriteLine("Dumb Search Time: {0}ms", (end - start).TotalMilliseconds);
Console.WriteLine("Index: {0}", idx);
Console.WriteLine("Simple Search ... ");
start = DateTime.Now;
idx = ssearch.Search(src, sub);
end = DateTime.Now;
Console.WriteLine("Simple Search Time: {0}ms", (end - start).TotalMilliseconds);
Console.WriteLine("Index: {0}", idx);
Console.WriteLine("Rabin Karp Search ... ");
start = DateTime.Now;
idx = rksearch.Search(src, sub);
end = DateTime.Now;
Console.WriteLine("RK Search Time: {0}ms", (end - start).TotalMilliseconds);
Console.WriteLine("Index: {0}", idx);
}
public void RkSimpleSearchCompareTest()
{
RkSearch rksearch = new RkSearch();
SimpleSearch ssearch = new SimpleSearch();
SeqUtils gen = new SeqUtils();
string src = gen.GenerateSequence(1000);
string sub = src.Substring(850, 15);
int idx = 0;
Console.WriteLine(src.Length);
DateTime start, end;
start = DateTime.Now;
idx = ssearch.Search(src, sub);
end = DateTime.Now;
Console.WriteLine("Simple Search Time: {0}ms", (end - start).TotalMilliseconds);
Console.WriteLine("Index: {0}", idx);
start = DateTime.Now;
idx = rksearch.Search(src, sub);
end = DateTime.Now;
Console.WriteLine("RK Search Time: {0}ms", (end - start).TotalMilliseconds);
Console.WriteLine("Index: {0}", idx);
}
public void RkSearchTestset()
{
RkSearch search = new RkSearch();
ulong x = 25223;
int c = search.CountSetPairs2(x);
Assert.AreEqual(c, 6);
x = 10578;
c = search.CountSetPairs2(x);
Assert.AreEqual(c, 6);
}
public void RkSearchTestSetMany()
{
RkSearch search = new RkSearch();
for (ulong i = ulong.MaxValue - 20000000; i > ulong.MaxValue - 4000; i--)
{
Console.WriteLine(i);
int c1 = search.CountSetPairs(i);
int c2 = search.CountSetPairs2(i);
Console.WriteLine("C1: {0} C2: {1}", c1, c2);
Assert.AreEqual(c1, c2);
}
}
public void RkSearchRandomSequence()
{
SeqUtils gen = new SeqUtils();
string src = gen.GenerateSequence(1000);
// 13 is the max limit due to size of long
string sub = src.Substring(900, 23);
RkSearch search = new RkSearch();
int idx = search.Search(src, sub);
Assert.IsTrue(idx > 0);
Assert.IsTrue(idx == 900);
}
private static void BasicDemo()
{
RkSearch rksearch = new RkSearch();
SimpleSearch ssearch = new SimpleSearch();
DumbSearch dsearch = new DumbSearch();
SeqUtils gen = new SeqUtils();
int seqLen = 200000000;
Console.Write("Generating sequence of length {0} ....", seqLen);
string src = gen.GenerateSequence(200000000);
Console.WriteLine("Done.");
int subPos = 85000001;
int subLen = 25;
Console.WriteLine("Substring of length {0} and at position {1}", subLen, subPos);
string sub = src.Substring(85000001, 25);
PlainSearch(src, sub);
int diff = 4;
sub = gen.InsertDiff(sub, diff);
DiffSearch(src, sub, diff);
}
public void RkSearchTest()
{
string src = "acgtaaactgggacct";
string sub = "ctg";
RkSearch search = new RkSearch();
int idx = search.Search(src, sub);
Assert.IsTrue(idx > 0);
Assert.IsTrue(idx == 7);
Console.WriteLine("Found at {0}", idx);
sub = "a";
idx = search.Search(src, sub);
Assert.IsTrue(idx >= 0);
Assert.IsTrue(idx == 0);
sub = "acct";
idx = search.Search(src, sub);
Assert.IsTrue(idx >= 0);
Assert.IsTrue(idx == src.Length - sub.Length);
sub = "tttt";
idx = search.Search(src, sub);
Assert.IsTrue(idx < 0);
}
private static void ThreeBillionRkSearch()
{
RkSearch rksearch = new RkSearch();
SeqUtils util = new SeqUtils();
SimpleSearch ssearch = new SimpleSearch();
// We generate really large strings, which total up to 3 billion in the end.
// The substring which we are searching for is 27 characters in length
string sub = util.GenerateSequence(27);
List<int> rkmatch = new List<int>();
List<int> ssmatch = new List<int>();
TimeSpan rkspan = TimeSpan.Zero, sspan = TimeSpan.Zero;
DateTime start, end;
for (int i = 0; i < 10; i++)
{
Console.WriteLine("Iteration {0}", i);
for (int j = 0; j < 3000; j++)
{
string src = util.GenerateSequence(1000000);
start = DateTime.Now;
rkmatch.AddRange(rksearch.SearchWithDiffAll(src, sub, 5));
end = DateTime.Now;
rkspan += (end - start);
start = DateTime.Now;
ssmatch.AddRange(ssearch.SearchWithDiffAll(src, sub, 5));
end = DateTime.Now;
sspan += (end - start);
src = null;
}
}
Console.WriteLine("Rabin Karp Time taken: {0} s", rkspan.TotalSeconds);
for (int i = 0; i < rkmatch.Count; i++)
{
Console.WriteLine(rkmatch[i]);
}
Console.WriteLine("Simple Search Time Taken: {0} s", sspan.TotalSeconds);
for (int i = 0; i < rkmatch.Count; i++)
{
Console.WriteLine(ssmatch[i]);
}
bool equal = true;
for (int i = 0; i < rkmatch.Count; i++)
{
if (rkmatch[i] != ssmatch[i])
{
equal = false;
break;
}
}
Console.WriteLine("Rabin Karp and Simple Search matches are equal: {0}", equal);
}
public void RkSearchWithDiff()
{
SeqUtils util = new SeqUtils();
string src = util.GenerateSequence(1000);
string sub = src.Substring(900, 23);
RkSearch search = new RkSearch();
sub = util.InsertDiff(sub, 5);
int idx = search.SearchWithDiff(src, sub, 5);
Assert.IsTrue(idx > 0);
}