/// <summary>
/// takes the two sources and performs the GST
/// </summary>
/// <param name="sourcePath1"></param>
/// <param name="sourcePath2"></param>
public ComparisonModel(string name, Int64 evalRunID, string sourcePath1, string sourcePath2)
{
Name = name;
EvaluationRunID = evalRunID;
var directory = Path.GetDirectoryName(Path.GetFullPath(sourcePath1));
bool tmplFileExists = File.Exists(Path.Combine(directory, "template.c"));
var watch = Stopwatch.StartNew();
var tmplFile = Directory.GetFiles(directory, "template.c").FirstOrDefault();
var path1 = tmplFileExists ? TemplatingHelper.StripTemplateFromSourceFile(sourcePath1, tmplFile) : sourcePath1;
var path2 = tmplFileExists ? TemplatingHelper.StripTemplateFromSourceFile(sourcePath2, tmplFile) : sourcePath2;
var factory = new MutexTokenFactory();
var tokens1 = factory.GetTokenWrapperListFromFile(path1);
var tokens2 = factory.GetTokenWrapperListFromFile(path2);
cLogger.DebugFormat("TokenStream Length: {0} -- {1}", tokens1.Count(), tokens2.Count());
var algo = new GSTAlgorithm<GSTToken<TokenWrapper>>(
tokens1.ToGSTTokenList<TokenWrapper>(),
tokens2.ToGSTTokenList<TokenWrapper>());
algo.RunToCompletion();
Result = algo.Similarity;
Source1 = new SourceModel(Path.GetFileNameWithoutExtension(sourcePath1), factory.GetJoinedTokenString(tokens1));
Source2 = new SourceModel(Path.GetFileNameWithoutExtension(sourcePath2), factory.GetJoinedTokenString(tokens2));
SQLFacade.Instance.CreateComparison(name, Result, watch.ElapsedMilliseconds, evalRunID, Source1.ID, Source2.ID);
}
/// <summary>
/// takes the two sources and performs the GST
/// </summary>
/// <param name="sourcePath1"></param>
/// <param name="sourcePath2"></param>
public ComparisonModel(string name, Int64 evalRunID, string sourcePath1, string sourcePath2)
{
Name = name;
EvaluationRunID = evalRunID;
var directory = Path.GetDirectoryName(Path.GetFullPath(sourcePath1));
bool tmplFileExists = File.Exists(Path.Combine(directory, "template.c"));
var watch = Stopwatch.StartNew();
var tmplFile = Directory.GetFiles(directory, "template.c").FirstOrDefault();
var path1 = tmplFileExists ? TemplatingHelper.StripTemplateFromSourceFile(sourcePath1, tmplFile) : sourcePath1;
var path2 = tmplFileExists ? TemplatingHelper.StripTemplateFromSourceFile(sourcePath2, tmplFile) : sourcePath2;
var factory = new MutexTokenFactory();
var tokens1 = factory.GetTokenWrapperListFromFile(path1);
var tokens2 = factory.GetTokenWrapperListFromFile(path2);
cLogger.DebugFormat("TokenStream Length: {0} -- {1}", tokens1.Count(), tokens2.Count());
var algo = new GSTAlgorithm <GSTToken <TokenWrapper> >(
tokens1.ToGSTTokenList <TokenWrapper>(),
tokens2.ToGSTTokenList <TokenWrapper>());
algo.RunToCompletion();
Result = algo.Similarity;
Source1 = new SourceModel(Path.GetFileNameWithoutExtension(sourcePath1), factory.GetJoinedTokenString(tokens1));
Source2 = new SourceModel(Path.GetFileNameWithoutExtension(sourcePath2), factory.GetJoinedTokenString(tokens2));
SQLFacade.Instance.CreateComparison(name, Result, watch.ElapsedMilliseconds, evalRunID, Source1.ID, Source2.ID);
}
public void RunToCompletion()
{
Assert.False(Algorithm.Finished);
Algorithm.RunToCompletion();
Assert.True(Algorithm.Finished);
}
/// <summary>
/// evaluation of accuracy only.
/// DO NOT TRY THIS AT HOME!
/// </summary>
private static void doEvaluate()
{
var doneList = new List <string>();
var args = Environment.GetCommandLineArgs();
var dir = new DirectoryInfo(@"myDir");
if (args.Length > 1)
{
dir = new DirectoryInfo(args[1]);
}
Console.WriteLine("directory: {0}", dir.FullName);
foreach (var subdir in dir.GetDirectories().OrderBy(sd => sd.Name))
{
DirectoryInfo subdir1 = subdir;
var otherFiles = dir.GetDirectories().Where(od => subdir1.Name != od.Name).OrderBy(od => od.Name).Select(od => new FileInfo(Path.Combine(od.FullName, "main.c")));
var fileA = new FileInfo(Path.Combine(subdir.FullName, "main.c"));
var tokensA = GetTokens(fileA);
foreach (var fileB in otherFiles)
{
var format1 = string.Format("{0}-{1}", fileA.Directory.Name, fileB.Directory.Name);
var format2 = string.Format("{1}-{0}", fileA.Directory.Name, fileB.Directory.Name);
if (doneList.Any(str => str.Equals(format1) || str.Equals(format2)))
{
continue;
}
doneList.Add(format1);
var tokensB = GetTokens(fileB);
var algo = new GSTAlgorithm <GSTToken <TokenWrapper> >(tokensA, tokensB)
{
MinimumMatchLength = 5
};
algo.RunToCompletion();
Console.WriteLine("{0}-{1}:{2}", fileA.Directory.Name, fileB.Directory.Name, algo.Similarity);
}
}
Console.WriteLine("FINISHED EVAL");
Environment.Exit(0);
}
private static void Calculate(int len)
{
var string1 = src1.Substring(0, len);
var string2 = src2.Substring(0, len);
//len += 200; // double length in size
var watch = Stopwatch.StartNew();
//Console.WriteLine("String length: A = {0}, B = {1}", string1.Length, string2.Length);
for (int i = testRuns; i >= 0; i--)
{
var algo = new HashingGSTAlgorithm <GSTToken <char> >(
GSTHelper.FromString(string1),
GSTHelper.FromString(string2));
algo.RunToCompletion();
//Console.WriteLine("finished hashing run {0} in {1}", i, watch.Elapsed);
}
var runtimeHashing = watch.Elapsed;
//Console.WriteLine("runtime hashing: {0}", ((int)runtimeHashing.TotalMilliseconds / testRuns));
watch = Stopwatch.StartNew();
for (int i = testRuns; i >= 0; i--)
{
var algo = new GSTAlgorithm <GSTToken <char> >(
GSTHelper.FromString(string1),
GSTHelper.FromString(string2));
algo.RunToCompletion();
}
var ratio = (double)watch.ElapsedTicks / runtimeHashing.Ticks;
Console.WriteLine("runtimes: {0}, {1}, ratio: {2} for {3}",
watch.ElapsedMilliseconds / testRuns,
runtimeHashing.TotalMilliseconds / testRuns,
ratio,
len);
ResultList.Sort(String.CompareOrdinal);
ResultList.Add(string.Format("{3:000};{0};{1};{2}",
(watch.Elapsed.TotalMilliseconds / testRuns),
(runtimeHashing.TotalMilliseconds / testRuns),
ratio,
len));
}
/// <summary>
/// evaluation of GST speed only.
/// DO NOT TRY THIS AT HOME!
/// </summary>
private static void EvaluateGSTSpeed()
{
var files = Directory.GetFiles(@"C:\Quelltexte", @"main.c", SearchOption.AllDirectories).Select(GetTokens).ToList();
//Console.WriteLine(string.Format("total product count: {0}", cartesianProduct.Count()));
long cRuntime = 0;
int similarity = 0;
IEnumerable <GSTTokenList <GSTToken <TokenWrapper> >[]> product = null;
for (int i = 100; i >= 0; i--)
{
var cartesianProduct = from first in files
from second in files
select new[] { first, second };
GSTAlgorithm <GSTToken <TokenWrapper> > algorithm = null;
foreach (var set in cartesianProduct)
{
var alg = new GSTAlgorithm <GSTToken <TokenWrapper> >(set[0], set[1])
{
MinimumMatchLength = 5
};
algorithm = alg;
var watch = Stopwatch.StartNew();
alg.RunToCompletion();
cRuntime += watch.ElapsedTicks;
watch.Stop();
}
similarity += algorithm.Similarity;
product = cartesianProduct;
}
Console.WriteLine("Number: {0}, Similarity: {1}", product.Count(), similarity);
Console.WriteLine("finished in {0} seconds", TimeSpan.FromTicks(cRuntime).TotalSeconds / 100);
Console.ReadLine();
Environment.Exit(0);
}
/// <summary>
/// evaluation of accuracy only.
/// DO NOT TRY THIS AT HOME!
/// </summary>
private static void doEvaluate()
{
var doneList = new List<string>();
var args = Environment.GetCommandLineArgs();
var dir = new DirectoryInfo(@"myDir");
if(args.Length > 1)
dir = new DirectoryInfo(args[1]);
Console.WriteLine("directory: {0}", dir.FullName);
foreach (var subdir in dir.GetDirectories().OrderBy(sd => sd.Name))
{
DirectoryInfo subdir1 = subdir;
var otherFiles = dir.GetDirectories().Where(od => subdir1.Name != od.Name).OrderBy(od => od.Name).Select(od => new FileInfo(Path.Combine(od.FullName, "main.c")));
var fileA = new FileInfo(Path.Combine(subdir.FullName, "main.c"));
var tokensA = GetTokens(fileA);
foreach(var fileB in otherFiles)
{
var format1 = string.Format("{0}-{1}", fileA.Directory.Name, fileB.Directory.Name);
var format2 = string.Format("{1}-{0}", fileA.Directory.Name, fileB.Directory.Name);
if (doneList.Any(str => str.Equals(format1) || str.Equals(format2)))
continue;
doneList.Add(format1);
var tokensB = GetTokens(fileB);
var algo = new GSTAlgorithm<GSTToken<TokenWrapper>>(tokensA, tokensB) {MinimumMatchLength = 5};
algo.RunToCompletion();
Console.WriteLine("{0}-{1}:{2}", fileA.Directory.Name, fileB.Directory.Name, algo.Similarity);
}
}
Console.WriteLine("FINISHED EVAL");
Environment.Exit(0);
}
/// <summary>
/// compares the two algorithms against each other over a rather large file (~8000 chars)
/// </summary>
private static void EvaluateSpeed()
{
XmlConfigurator.Configure(new FileInfo("log4net.xml"));
src1 = File.ReadAllText(@"test\default_set\TPLV04-S01-02\main-01.c");
src2 = File.ReadAllText(@"test\default_set\TPLV04-S01-02\main-02.c");
var testRuns = 2;
int len = 100;
// JIT compile both algorithms before we start
AbstractGSTAlgorithm <GSTToken <char> > initalgo = new HashingGSTAlgorithm <GSTToken <char> >(
GSTHelper.FromString(src1.Substring(0, 100)),
GSTHelper.FromString(src2.Substring(0, 100)));
initalgo.RunToCompletion();
initalgo = new GSTAlgorithm <GSTToken <char> >(GSTHelper.FromString(src1.Substring(0, 100)),
GSTHelper.FromString(src2.Substring(0, 100)));
initalgo.RunToCompletion();
var liLen1 = new List <int>();
var liLen2 = new List <int>();
var liLen3 = new List <int>();
int c = 0;
while (len < src1.Length && len < src2.Length)
{
switch (c++ % 3)
{
case 0:
liLen1.Add(len);
break;
case 1:
liLen2.Add(len);
break;
default:
liLen3.Add(len);
break;
}
len += 100;
}
Length1 = liLen1.ToArray();
Length2 = liLen2.ToArray();
Length3 = liLen3.ToArray();
var t1 = new Thread(Start1);
t1.Start();
var t2 = new Thread(Start2);
t2.Start();
var t3 = new Thread(Start3);
t3.Start();
Console.WriteLine("Main thread joining");
t1.Join();
t2.Join();
t3.Join();
File.WriteAllLines(@"test\default_set\performance.txt", ResultList);
Console.WriteLine("finished all runs");
Console.ReadLine();
}
/// <summary>
/// evaluation of GST speed only.
/// DO NOT TRY THIS AT HOME!
/// </summary>
private static void EvaluateGSTSpeed()
{
var files = Directory.GetFiles(@"C:\Quelltexte", @"main.c", SearchOption.AllDirectories).Select(GetTokens).ToList();
//Console.WriteLine(string.Format("total product count: {0}", cartesianProduct.Count()));
long cRuntime = 0;
int similarity = 0;
IEnumerable<GSTTokenList<GSTToken<TokenWrapper>>[]> product = null;
for (int i = 100; i >= 0; i--)
{
var cartesianProduct = from first in files
from second in files
select new[] { first, second };
GSTAlgorithm<GSTToken<TokenWrapper>> algorithm = null;
foreach (var set in cartesianProduct)
{
var alg = new GSTAlgorithm<GSTToken<TokenWrapper>>(set[0], set[1]) {MinimumMatchLength = 5};
algorithm = alg;
var watch = Stopwatch.StartNew();
alg.RunToCompletion();
cRuntime += watch.ElapsedTicks;
watch.Stop();
}
similarity += algorithm.Similarity;
product = cartesianProduct;
}
Console.WriteLine("Number: {0}, Similarity: {1}", product.Count(), similarity);
Console.WriteLine("finished in {0} seconds", TimeSpan.FromTicks(cRuntime).TotalSeconds / 100);
Console.ReadLine();
Environment.Exit(0);
}
private static void Calculate(int len)
{
var string1 = src1.Substring(0, len);
var string2 = src2.Substring(0, len);
//len += 200; // double length in size
var watch = Stopwatch.StartNew();
//Console.WriteLine("String length: A = {0}, B = {1}", string1.Length, string2.Length);
for (int i = testRuns; i >= 0; i--)
{
var algo = new HashingGSTAlgorithm<GSTToken<char>>(
GSTHelper.FromString(string1),
GSTHelper.FromString(string2));
algo.RunToCompletion();
//Console.WriteLine("finished hashing run {0} in {1}", i, watch.Elapsed);
}
var runtimeHashing = watch.Elapsed;
//Console.WriteLine("runtime hashing: {0}", ((int)runtimeHashing.TotalMilliseconds / testRuns));
watch = Stopwatch.StartNew();
for (int i = testRuns; i >= 0; i--)
{
var algo = new GSTAlgorithm<GSTToken<char>>(
GSTHelper.FromString(string1),
GSTHelper.FromString(string2));
algo.RunToCompletion();
}
var ratio = (double)watch.ElapsedTicks / runtimeHashing.Ticks;
Console.WriteLine("runtimes: {0}, {1}, ratio: {2} for {3}",
watch.ElapsedMilliseconds / testRuns,
runtimeHashing.TotalMilliseconds / testRuns,
ratio,
len);
ResultList.Sort(String.CompareOrdinal);
ResultList.Add(string.Format("{3:000};{0};{1};{2}",
(watch.Elapsed.TotalMilliseconds / testRuns),
(runtimeHashing.TotalMilliseconds / testRuns),
ratio,
len));
}
/// <summary>
/// compares the two algorithms against each other over a rather large file (~8000 chars)
/// </summary>
private static void EvaluateSpeed()
{
XmlConfigurator.Configure(new FileInfo("log4net.xml"));
src1 = File.ReadAllText(@"test\default_set\TPLV04-S01-02\main-01.c");
src2 = File.ReadAllText(@"test\default_set\TPLV04-S01-02\main-02.c");
var testRuns = 2;
int len = 100;
// JIT compile both algorithms before we start
AbstractGSTAlgorithm<GSTToken<char>> initalgo = new HashingGSTAlgorithm<GSTToken<char>>(
GSTHelper.FromString(src1.Substring(0, 100)),
GSTHelper.FromString(src2.Substring(0, 100)));
initalgo.RunToCompletion();
initalgo = new GSTAlgorithm<GSTToken<char>>(GSTHelper.FromString(src1.Substring(0, 100)),
GSTHelper.FromString(src2.Substring(0, 100)));
initalgo.RunToCompletion();
var liLen1 = new List<int>();
var liLen2 = new List<int>();
var liLen3 = new List<int>();
int c = 0;
while(len < src1.Length && len < src2.Length)
{
switch (c++ % 3)
{
case 0:
liLen1.Add(len);
break;
case 1:
liLen2.Add(len);
break;
default:
liLen3.Add(len);
break;
}
len += 100;
}
Length1 = liLen1.ToArray();
Length2 = liLen2.ToArray();
Length3 = liLen3.ToArray();
var t1 = new Thread(Start1);
t1.Start();
var t2 = new Thread(Start2);
t2.Start();
var t3 = new Thread(Start3);
t3.Start();
Console.WriteLine("Main thread joining");
t1.Join();
t2.Join();
t3.Join();
File.WriteAllLines(@"test\default_set\performance.txt", ResultList);
Console.WriteLine("finished all runs");
Console.ReadLine();
}