public GitDataCollector()
{
// Gets the Calendar instance associated with a CultureInfo.
_cultureInfo = new CultureInfo("en-US");
_calendar = _cultureInfo.Calendar;
ActivityByHourOfDay = new DictionaryWithDefault <int, decimal>();
ActivityByDayOfWeek = new DictionaryWithDefault <int, int>();
ActivityByMonthOfYear = new DictionaryWithDefault <int, int>();
ActivityByHourOfWeek = new DictionaryWithDefault <int, DictionaryWithDefault <int, int> >();
ActivityByHourOfDayBusiest = 0;
ActivityByHourOfWeekBusiest = 0;
ActivityByYearWeek = new DictionaryWithDefault <string, int>();
ActivityByYearWeekPeak = 0;
Authors = new DictionaryWithDefault <string, Author>();
Domains = new DictionaryWithDefault <string, Domain>();
AuthorOfMonth = new DictionaryWithDefault <string, DictionaryWithDefault <string, int> >();
AuthorOfYear = new DictionaryWithDefault <int, DictionaryWithDefault <string, int> >();
CommitsByMonth = new DictionaryWithDefault <string, int>();
CommitsByYear = new DictionaryWithDefault <int, int>();
ActiveDays = new List <DateTime>();
TotalLines = 0;
TotalLinesAdded = 0;
TotalLinesRemoved = 0;
CommitsByTimezone = new DictionaryWithDefault <string, int>();
Tags = new DictionaryWithDefault <string, Tag>();
}
private void GetExtensions()
{
Extensions = new DictionaryWithDefault <string, Extension>();
var lines = GitStats.GetPipeOutput(new[]
{
"git ls-tree -r -z HEAD"
}).Split("\0").ToList().Where(s => !string.IsNullOrWhiteSpace(s)).ToArray();
;
TotalFiles = lines.Length;
Console.WriteLine($"Processing {TotalFiles} files in repo");
foreach (var line in lines)
{
if (line.Length == 0)
{
continue;
}
var parts = Regex.Split(line, "\\s+", RegexOptions.None);
var sha1 = parts[2];
var filename = parts[3];
string ext;
if (filename.IndexOf(".", StringComparison.Ordinal) == -1 ||
filename.IndexOf(".", StringComparison.Ordinal) == 0)
{
ext = "";
}
else
{
ext = filename.Substring(filename.LastIndexOf(".", StringComparison.Ordinal) + 1);
}
if (ext.Length > _configuration.MaxExtensionLength)
{
ext = "";
}
if (!Extensions.ContainsKey(ext))
{
Extensions[ext] = new Extension();
}
Extensions[ext].Files += 1;
try
{
Extensions[ext].Lines += GetLinesInBlob(sha1);
}
catch
{
Console.WriteLine("Warning: Could not count lines for file \"%s\"", line);
}
}
}
private void GetFilesByStampAndTotalCommits(List <string> lines)
{
// TODO Optimize this, it's the worst bottleneck
// outputs "<stamp> <files>" for each revision
lines.Clear();
FilesByStamp = new DictionaryWithDefault <DateTime, int>();
var revLines = GitStats.GetPipeOutput(new[]
{
"git rev-list --remotes --pretty=format:\"%at %T\" HEAD",
"grep -v ^commit"
}).Trim().Split("\n");
foreach (var revLine in revLines)
{
var tup2 = revLine.Split(" ");
var time = tup2[0];
var rev = tup2[1];
var lineCount = GetFilesInCommit(rev);
lines.Add($"{Convert.ToInt32(time)} {lineCount}");
}
TotalCommits = lines.Count;
foreach (var line in lines)
{
var parts = line.Split(" ");
if (parts.Length != 2)
{
continue;
}
try
{
FilesByStamp[DateTimeOffset.FromUnixTimeSeconds(Convert.ToInt32(parts[0])).DateTime] = Convert.ToInt32(parts[1]);
}
catch (FormatException)
{
Console.WriteLine("Warning: failed to parse line \"%s\"", line);
}
}
}
private void GetChangesByDateAndTotalLines()
{
// line statistics
// outputs:
// N files changed, N insertions (+), N deletions(-)
// <stamp> <author>
ChangesByDate = new DictionaryWithDefault <DateTime, Change>();
var lines = GitStats.GetPipeOutput(new[] { "git log --shortstat --pretty=format:\"%at %an\"" }).Split("\n")
.ToList();
lines.Reverse();
var files = 0;
var inserted = 0;
var deleted = 0;
var totalLines = 0;
foreach (var line in lines)
{
if (line.Length == 0)
{
files = 0;
inserted = 0;
deleted = 0;
continue;
}
if (line.IndexOf(" changed,", StringComparison.CurrentCultureIgnoreCase) == -1)
{
var pos = line.IndexOf(" ", StringComparison.CurrentCultureIgnoreCase);
if (pos != -1)
{
try
{
var datetime = DateTimeOffset.FromUnixTimeSeconds(Convert.ToInt32(line.Substring(0, pos))).DateTime;
var author = line.Substring(pos + 1);
ChangesByDate[datetime] = new Change
{
Files = files,
Inserted = inserted,
Deleted = deleted,
TotalLines = totalLines
};
if (!Authors.ContainsKey(author))
{
Authors[author] = new Author();
}
Authors[author].LinesAdded += inserted;
Authors[author].LinesRemoved += deleted;
}
catch (Exception)
{
Console.WriteLine($"Warning: unexpected line \"{line}\"");
}
}
else
{
Console.WriteLine($"Warning: unexpected line \"{line}\"");
}
}
else
{
files = GetIntFromStartOfRegex(line, "\\d+ file");
inserted = GetIntFromStartOfRegex(line, "\\d+ insertion");
deleted = GetIntFromStartOfRegex(line, "\\d+ delet");
totalLines += inserted;
totalLines -= deleted;
TotalLinesAdded += inserted;
TotalLinesRemoved += deleted;
}
}
TotalLines = totalLines;
}
// Collect revision statistics
// Outputs "<stamp> <date> <time> <timezone> <author> '<' <mail> '>'"
private List <string> GetActivityDataAndAuthors()
{
var lines = GitStats.GetPipeOutput(new[]
{
"git rev-list --remotes --pretty=format:\"%at %ai %an <%aE>\" HEAD",
"grep -v ^commit"
}, PipingLevel.Full, true).Split("\n").ToList();
foreach (var line in lines)
{
var parts = Regex.Split(line, "([01-9-:+]+ )").Where(x => !string.IsNullOrEmpty(x)).Select(s => s.Trim())
.ToArray();
DateTime stamp;
try
{
stamp = DateTimeOffset.FromUnixTimeSeconds(Convert.ToInt32(parts[0])).DateTime;
}
catch (FormatException)
{
stamp = DateTime.MinValue;
}
var timezone = parts[3];
var tup1 = parts[4].Split("<");
var author = tup1[0];
var mail = tup1[1];
author = author.TrimEnd();
mail = mail.TrimEnd('>');
var domain = "?";
if (mail.IndexOf("@", StringComparison.CurrentCultureIgnoreCase) != -1)
{
domain = mail.Split("@")[1];
}
var date = stamp;
// First and last commit stamp
if (LastCommitStamp == DateTime.MinValue)
{
LastCommitStamp = stamp;
}
FirstCommitStamp = stamp;
// activity
// hour
var hour = date.Hour;
ActivityByHourOfDay[hour] = ActivityByHourOfDay[hour] + 1;
// most active hour?
if (ActivityByHourOfDay[hour] > ActivityByHourOfDayBusiest)
{
ActivityByHourOfDayBusiest = ActivityByHourOfDay[hour];
}
// day of week
var day = (int)date.DayOfWeek - 1;
ActivityByDayOfWeek[day] = ActivityByDayOfWeek[day] + 1;
// domain stats
if (!Domains.ContainsKey(domain))
{
Domains[domain] = new Domain();
}
// commits
Domains[domain].Commits = Domains[domain].Commits + 1;
// hour of week
if (!ActivityByHourOfWeek.ContainsKey(day))
{
ActivityByHourOfWeek[day] = new DictionaryWithDefault <int, int>();
}
ActivityByHourOfWeek[day][hour] = ActivityByHourOfWeek[day][hour] + 1;
// most active hour?
if (ActivityByHourOfWeek[day][hour] > ActivityByHourOfWeekBusiest)
{
ActivityByHourOfWeekBusiest = ActivityByHourOfWeek[day][hour];
}
// month of year
var month = date.Month;
ActivityByMonthOfYear[month] = ActivityByMonthOfYear[month] + 1;
// yearly/weekly activity
var yyw =
$"{date.Year}-{_calendar.GetWeekOfYear(date, _cultureInfo.DateTimeFormat.CalendarWeekRule, _cultureInfo.DateTimeFormat.FirstDayOfWeek)}";
ActivityByYearWeek[yyw] = ActivityByYearWeek[yyw] + 1;
if (ActivityByYearWeekPeak < ActivityByYearWeek[yyw])
{
ActivityByYearWeekPeak = ActivityByYearWeek[yyw];
}
// author stats
if (!Authors.ContainsKey(author))
{
Authors[author] = new Author();
}
// commits
if (Authors[author].LastCommitStamp == DateTime.MinValue)
{
Authors[author].LastCommitStamp = stamp;
}
Authors[author].FirstCommitStamp = stamp;
Authors[author].Commits = Authors[author].Commits + 1;
// author of the month/year
var yymm = $"{date.Year}-{date.Month:D2}";
if (AuthorOfMonth.ContainsKey(yymm))
{
AuthorOfMonth[yymm][author] = AuthorOfMonth[yymm][author] + 1;
}
else
{
AuthorOfMonth[yymm] = new DictionaryWithDefault <string, int> {
[author] = 1
}
};
CommitsByMonth[yymm] = CommitsByMonth[yymm] + 1;
var yy = date.Year;
if (AuthorOfYear.ContainsKey(yy))
{
AuthorOfYear[yy][author] = AuthorOfYear[yy][author] + 1;
}
else
{
AuthorOfYear[yy] = new DictionaryWithDefault <string, int> {
[author] = 1
}
};
CommitsByYear[yy] = CommitsByYear[yy] + 1;
// authors: active days
var yymmdd = date;
if (Authors[author].LastActiveDay == DateTime.MinValue)
{
Authors[author].LastActiveDay = yymmdd;
Authors[author].ActiveDays = 1;
}
else if (yymmdd != Authors[author].LastActiveDay)
{
Authors[author].LastActiveDay = yymmdd;
Authors[author].ActiveDays += 1;
}
// project: active days
if (yymmdd != LastActiveDay)
{
LastActiveDay = yymmdd;
ActiveDays.Add(yymmdd);
}
// timezone
CommitsByTimezone[timezone] = CommitsByTimezone[timezone] + 1;
}
return(lines);
}