private static void BuildStopWordIndex(Iterable <RevCommit> log, Repository repository, Git git)
{
Console.WriteLine("Index build start -" + System.DateTime.Now);
DateTime startTime = DateTime.Now;
for (Iterator <RevCommit> iterator = log.Iterator(); iterator.HasNext();)
{
RevCommit rev = iterator.Next();
var dt = new DateTime(1970, 1, 1, 0, 0, 0, 0).AddSeconds(rev.CommitTime).ToLocalTime();
DiffFormatter df = new DiffFormatter(DisabledOutputStream.INSTANCE);
df.SetRepository(repository);
df.SetDiffComparator(RawTextComparator.DEFAULT);
df.SetDetectRenames(true);
List <String> files = new List <string>();
if (rev.ParentCount > 0)
{
List <DiffEntry> diffs = df.Scan(rev.GetParent(0).Tree, rev.Tree).ToList();
foreach (DiffEntry diff in diffs)
{
string filename = diff.GetNewPath();
if (Utils.AllowedFileExtentions(filename))
{
string data = GitHelper.GetCommitDiff(repository, git, rev, filename);
StopWords.StopWordIndexBuilder.BuildWordIndex(filename, data);
}
}
}
if (DateTime.Now.Subtract(startTime).TotalMinutes > 60)
{
break;
}
}
Console.WriteLine("Index build end -" + System.DateTime.Now);
StopWords.StopWordIndexBuilder.BuildStopWordsIndex();
StopWords.StopWordIndexBuilder.WriteToFile();
Console.WriteLine("Stop word keyword index build completed..");
}
static void Main(string[] args)
{
FileRepositoryBuilder builder = new FileRepositoryBuilder();
Repository repository = builder.SetGitDir(new FilePath(@"D:\Personal\E Books\Higher Education\Research\SampleProjects\NopCommerce\.git"))
//Repository repository = builder.SetGitDir(new FilePath(@"C:\Users\neemathu\Documents\GitHub\angular.js\.git"))
// Repository repository = builder.SetGitDir(new FilePath(@"D:\Personal\E Books\Higher Education\RefactoringActivity\ganttproject\.git"))
.ReadEnvironment() // scan environment GIT_* variables
.FindGitDir() // scan up the file system tree
.Build();
HashSet <String> uniqueFile = new HashSet <string>();
Dictionary <String, int> logicalCoupling = new Dictionary <string, int>();
RevWalk rw = new RevWalk(repository);
Git git = new Git(repository);
Iterable <RevCommit> log = git.Log().Call();
for (Iterator <RevCommit> iterator = log.Iterator(); iterator.HasNext();)
{
RevCommit rev = iterator.Next();
//RevWalk revWalk = new RevWalk(git.GetRepository());
//RevTree revTree = revWalk.ParseTree(rev.Tree.Id);
//TreeWalk treeWalk = new TreeWalk(git.GetRepository());
//treeWalk.AddTree(revTree);
//while (treeWalk.Next())
//{
// //compare treeWalk.NameString yourself
// byte[] bytes = treeWalk.ObjectReader.Open(treeWalk.GetObjectId(0)).GetBytes();
// string result1 = System.Text.Encoding.UTF8.GetString(bytes);
//}
// Sharpen.OutputStream os = new Sharpen.OutputStream();
//rev.CopyRawTo(os);
//System.Console.WriteLine("Author: "+rev.GetAuthorIdent().GetName());
//System.Console.WriteLine("ID:" + rev.Id);
var dt = new DateTime(1970, 1, 1, 0, 0, 0, 0).AddSeconds(rev.CommitTime).ToLocalTime();
//var ts = new TimeSpan(DateTime.UtcNow.Ticks - rev.CommitTime);
//System.Console.WriteLine("Date:" + dt.ToString());
//System.Console.WriteLine("Description:" + rev.GetFullMessage());
DiffFormatter df = new DiffFormatter(DisabledOutputStream.INSTANCE);
df.SetRepository(repository);
df.SetDiffComparator(RawTextComparator.DEFAULT);
df.SetDetectRenames(true);
List <String> files = new List <string>();
if (rev.ParentCount > 0)
{
List <DiffEntry> diffs = df.Scan(rev.GetParent(0).Tree, rev.Tree).ToList();
foreach (DiffEntry diff in diffs)
{
// Fetch data from the commited new file
//ObjectLoader loader = repository.Open(diff.GetNewId().ToObjectId());
//OutputStream @out = new ByteArrayOutputStream();
////loader.CopyTo(@out);
////Fetch diffrence of commit
//DiffCommand diff1 = git.Diff().SetPathFilter(PathFilter.Create(diff.GetNewPath())).SetOldTree(GetTreeIterator(rev.GetParent(0).Tree.Name, repository)).SetNewTree(GetTreeIterator(rev.Tree.Name, repository)).SetOutputStream(@out);
//IList<DiffEntry> entries = diff1.Call();
//string data = @out.ToString();
files.Add(diff.GetNewPath());
uniqueFile.Add(diff.GetNewPath());
//System.Console.WriteLine(String.Format("FilePath: {0} {1}", diff.GetNewMode().GetBits(), diff.GetNewPath()));
}
}
if (isContainFile(rev, files))
{
//System.Console.WriteLine(rev.Id);
//System.Console.WriteLine(dt);
//System.Console.WriteLine(rev.GetAuthorIdent().GetName());
//System.Console.WriteLine(rev.GetFullMessage());
tfIdfBeforeData.Add(rev.Id.Name, new Dictionary <string, Dictionary <string, double> >());
foreach (String file in files)
{
String fileName = file.Substring(file.LastIndexOf("/") + 1, file.Length - file.LastIndexOf("/") - 1);
if (IsFileExtentionAllowed(fileName))
{
string data = GetCommitDiff(repository, git, rev, file);
Dictionary <string, double> tokensTF = GetTokensWithTF(data);
tfIdfBeforeData[rev.Id.Name].Add(file, tokensTF);
//System.Console.WriteLine("File path: " + file);
//System.Console.WriteLine(data);
//System.Console.WriteLine("------------------");
if (!logicalCoupling.ContainsKey(fileName))
{
logicalCoupling.Add(fileName, 1);
}
else
{
logicalCoupling[fileName] += 1;
}
}
}
//System.Console.WriteLine("###################################");
}
//foreach (var item in uniqueFile)
//{
// System.Console.WriteLine(item);
//}
//System.Console.WriteLine("--------------------");
//http://stackoverflow.com/questions/11869412/jgit-using-revwalk-to-get-revcommit-returns-nothing
////ObjectId head = repository.Resolve("master");
//RevWalk walk = new RevWalk(repository);
//foreach (var commit in walk)
//{
// String email = commit.GetAuthorIdent().GetEmailAddress();
//}
}
CalculateTfIdfScore("defaultResources.nopres.xml");
CalculateLogicalDependency(logicalCoupling);
System.Console.WriteLine("----------Done----------");
System.Console.ReadLine();
}
static void Main(string[] args)
{
//string input = "group, and test but not testing. But yes to test";
//string val="Group";
//string pattern = @"\b" + val + @"\b";
//string replace = " ";
//string result = System.Text.RegularExpressions.Regex.Replace(input, pattern, replace, System.Text.RegularExpressions.RegexOptions.IgnoreCase);
//Console.WriteLine(result);
FileRepositoryBuilder builder = new FileRepositoryBuilder();
Repository repository = builder.SetGitDir(new FilePath(@"D:\Personal\E Books\Higher Education\Research\SampleProjects\NopCommerce\.git"))
//Repository repository = builder.SetGitDir(new FilePath(@"C:\Users\neemathu\Documents\GitHub\angular.js\.git"))
// Repository repository = builder.SetGitDir(new FilePath(@"D:\Personal\E Books\Higher Education\RefactoringActivity\ganttproject\.git"))
.ReadEnvironment() // scan environment GIT_* variables
.FindGitDir() // scan up the file system tree
.Build();
RevWalk rw = new RevWalk(repository);
Git git = new Git(repository);
Iterable <RevCommit> log = git.Log().Call();
if (args.Length > 0)
{
switch (args[0])
{
case "buildstopwordindex":
BuildStopWordIndex(log, repository, git);
return;
break;
default:
break;
}
}
// Iterat over revisions
for (Iterator <RevCommit> iterator = log.Iterator(); iterator.HasNext();)
{
RevCommit rev = iterator.Next();
var dt = new DateTime(1970, 1, 1, 0, 0, 0, 0).AddSeconds(rev.CommitTime).ToLocalTime();
DiffFormatter df = new DiffFormatter(DisabledOutputStream.INSTANCE);
df.SetRepository(repository);
df.SetDiffComparator(RawTextComparator.DEFAULT);
df.SetDetectRenames(true);
List <String> files = new List <string>();
if (rev.ParentCount > 0)
{
List <DiffEntry> diffs = df.Scan(rev.GetParent(0).Tree, rev.Tree).ToList();
foreach (DiffEntry diff in diffs)
{
// Fetch data from the commited new file
//ObjectLoader loader = repository.Open(diff.GetNewId().ToObjectId());
//OutputStream @out = new ByteArrayOutputStream();
////loader.CopyTo(@out);
////Fetch diffrence of commit
//DiffCommand diff1 = git.Diff().SetPathFilter(PathFilter.Create(diff.GetNewPath())).SetOldTree(GetTreeIterator(rev.GetParent(0).Tree.Name, repository)).SetNewTree(GetTreeIterator(rev.Tree.Name, repository)).SetOutputStream(@out);
//IList<DiffEntry> entries = diff1.Call();
//string data = @out.ToString();
string filePath = diff.GetNewPath();
//if (fileCount.ContainsKey(filePath))
//{
// fileCount[filePath] = fileCount[filePath] + 1;
//}
//else
//{
// fileCount.Add(filePath, 1);
//}
files.Add(filePath);
//System.Console.WriteLine(String.Format("FilePath: {0} {1}", diff.GetNewMode().GetBits(), diff.GetNewPath()));
}
//continue;
}
if (GitHelper.HasFile(files))
{
foreach (String file in files)
{
String FileName = file.Substring(file.LastIndexOf("/") + 1, file.Length - file.LastIndexOf("/") - 1);
if (Utils.AllowedFileExtentions(FileName))
{
string DiffContent = GitHelper.GetCommitDiff(repository, git, rev, file);
//data = Common.Utils.RemoveStopWords(data);
LogicalDependency.LogicalDependency.AddArtifact(rev.Id.Name, file, DiffContent);
//StopWords.StopWordIndexBuilder.BuildWordIndex(file, data);
}
}
}
}
//var sortedElements = fileCount.OrderByDescending(kvp => kvp.Value);
//foreach (var item in sortedElements)
//{
// Console.WriteLine(item.Key + ": " + item.Value);
//}
LogicalDependency.LogicalDependency.CalculateSimilarityIndex();
}