public void Find(OSPCResult result, List <CompareResult> compareResult)
{
var friends = new Dictionary <Submission, FriendOf>();
var min_friend_finder_similarity = _cfg.MIN_FRIEND_FINDER_SIMILARITY >= 0
? _cfg.MIN_FRIEND_FINDER_SIMILARITY
: result.POI_Similarity - 0.2;
min_friend_finder_similarity = Math.Max(min_friend_finder_similarity, _cfg.MIN_SIMILARITY);
// if result.POI_Similarity is 0, min_friend_finder_similarity may become negative.
// So every submission is a group and every match is in that group.
if (min_friend_finder_similarity > 0)
{
foreach (var cp in compareResult)
{
ProcessMatch(friends, cp, cp.A, cp.SimilarityB, min_friend_finder_similarity);
ProcessMatch(friends, cp, cp.B, cp.SimilarityA, min_friend_finder_similarity);
}
}
result.Friends = friends.Values
.Where(i => i.InMatches.Count > 1)
.OrderByDescending(i => i.SumSimilarity)
.ToList();
}
public static OSPCResult Create(List <CompareResult> compareResult)
{
var result = new OSPCResult();
result.Results = compareResult
.Where(r => r.MatchCount > 0)
.OrderByDescending(r => Math.Max(r.SimilarityA, r.SimilarityB))
.ThenByDescending(r => Math.Min(r.SimilarityA, r.SimilarityB))
.Select((item, idx) =>
{
item.Seal(idx);
return(item);
})
.ToList();
double[] lst = result.Results.SelectMany(i => new[] { i.SimilarityA, i.SimilarityB }).OrderBy(i => i).ToArray();
if (lst.Length > 0)
{
var poiidx = lst.CalcDerv2().MaxIndex();
result.AVG_Similarity = lst.Average();
result.POI_Similarity = poiidx >= 0 ? lst[poiidx] : 0;
}
lst = result.Results.Select(i => (double)i.TokenCount).OrderBy(i => i).ToArray();
if (lst.Length > 0)
{
var poiidx = lst.CalcDerv2().MaxIndex();
result.AVG_TokenCount = lst.Average();
result.POI_TokenCount = poiidx >= 0 ? lst[poiidx] : 0;
}
lst = result.Results.Select(i => (double)i.TokenCount / (double)i.MatchCount).OrderBy(i => i).ToArray();
if (lst.Length > 0)
{
var poiidx = lst.CalcDerv2().MaxIndex();
result.AVG_TokenPerMatch = lst.Average();
result.POI_TokenPerMatch = poiidx >= 0 ? lst[poiidx] : 0;
}
return(result);
}
public static OSPCResult Create(List<CompareResult> compareResult)
{
var result = new OSPCResult();
result.Results = compareResult
.Where(r => r.MatchCount > 0)
.OrderByDescending(r => Math.Max(r.SimilarityA, r.SimilarityB))
.ThenByDescending(r => Math.Min(r.SimilarityA, r.SimilarityB))
.Select((item, idx) =>
{
item.Seal(idx);
return item;
})
.ToList();
double[] lst = result.Results.SelectMany(i => new[] { i.SimilarityA, i.SimilarityB }).OrderBy(i => i).ToArray();
if (lst.Length > 0)
{
result.AVG_Similarity = lst.Average();
result.POI_Similarity = lst[lst.CalcDerv2().MaxIndex()];
}
lst = result.Results.Select(i => (double)i.TokenCount).OrderBy(i => i).ToArray();
if (lst.Length > 0)
{
result.AVG_TokenCount = lst.Average();
result.POI_TokenCount = lst[lst.CalcDerv2().MaxIndex()];
}
lst = result.Results.Select(i => (double)i.TokenCount / (double)i.MatchCount).OrderBy(i => i).ToArray();
if (lst.Length > 0)
{
result.AVG_TokenPerMatch = lst.Average();
result.POI_TokenPerMatch = lst[lst.CalcDerv2().MaxIndex()];
}
return result;
}
public void Find(OSPCResult result, List<CompareResult> compareResult)
{
var friends = new Dictionary<Submission, FriendOf>();
var min_friend_finder_similarity = _cfg.MIN_FRIEND_FINDER_SIMILARITY >= 0
? _cfg.MIN_FRIEND_FINDER_SIMILARITY
: result.POI_Similarity - 0.2;
min_friend_finder_similarity = Math.Max(min_friend_finder_similarity, _cfg.MIN_SIMILARITY);
// if result.POI_Similarity is 0, min_friend_finder_similarity may become negative.
// So every submission is a group and every match is in that group.
if (min_friend_finder_similarity > 0)
{
foreach (var cp in compareResult)
{
ProcessMatch(friends, cp, cp.A, cp.SimilarityB, min_friend_finder_similarity);
ProcessMatch(friends, cp, cp.B, cp.SimilarityA, min_friend_finder_similarity);
}
}
result.Friends = friends.Values
.Where(i => i.InMatches.Count > 1)
.OrderByDescending(i => i.SumSimilarity)
.ToList();
}
static void Main(string[] args)
{
try
{
var cfg = new Configuration();
bool showHelp = false;
// common dependencies
var progress = new ConsoleProgressReporter();
// changable through config or arguments
Reporter.IReporter html = null;
Reporter.IReporter console = new Reporter.ConsoleReporter();
Tokenizer.ITokenizer tokenizer = new Tokenizer.CLikeTokenizer();
var p = new OptionSet()
{
{ "h|?|help", "Prints this help", v => showHelp = true },
{ "c=", "Reads the given configuration. Note, this switch should be the first argument as it overrides any other argument parsed yet.", v => cfg = LoadConfig(v) },
{ "write-config=", "Write the current configuration to the given file. Note, this switch should be the last argument.", v => SaveConfig(cfg, v) },
{ "f=", "File filter. If -d is specified, then -f defaults to \"*.*.\"", v => cfg.Filter.Add(v) },
{ "d=", "Specifies a directory where the filer applies. If -f is specified, then -d defaults to \".\"", v => cfg.Dirs.Add(v) },
{ "include=", "Specifies a regular expression that every file must match. More than one expression is allowed. A file must match any of these expressions.", v => cfg.Include.Add(v) },
{ "exclude=", "Specifies a regular expression to exclude files. More than one expression is allowed. If a file must match any of these expressions it will be excluded.", v => cfg.Exclude.Add(v) },
{ "recurse", "Traversals all directories recurse. Use include-dir and exclude-dir to have more control over the selected directories.", v => cfg.Recurse = true },
{ "include-dir=", "Specifies a regular expression that every file must match. More than one expression is allowed. A file must match any of these expressions.", v => cfg.IncludeDir.Add(v) },
{ "exclude-dir=", "Specifies a regular expression to exclude files. More than one expression is allowed. If a file must match any of these expressions it will be excluded.", v => cfg.ExcludeDir.Add(v) },
{ "againstdir", "Compares the files specified as arguments against the directory set with -d", v => cfg.OneAgainstDir = true },
{ "detailed", "Print a detailed report to the console", v => console = new Reporter.DetailedConsoleReporter() },
{ "summary", "Print only a summay to the console. Usefull if --html is used.", v => console = new Reporter.SummaryConsoleReporter() },
{ "html:", "Saves a html report to the specified directory. Defaults to \"report\"", v => html = new Reporter.Html.HtmlReporter(v, progress) },
{ "min-similarity=", "Minimum similarity of match reports (0 - 1). Default is 0.5 (=50%).", v => cfg.MIN_SIMILARITY = double.Parse(v, System.Globalization.CultureInfo.InvariantCulture) },
{ "min-match-length=", "Minimum count of matching tokens, including non-matching tokens.", v => cfg.MIN_MATCH_LENGTH = int.Parse(v) },
{ "max-match-distance=", "Maximum distance between tokens to count as a match. 1 = exact match.", v => cfg.MAX_MATCH_DISTANCE = int.Parse(v) },
{ "min-common-token=", "Percent of token that must match to count as a match. 1 = every token must match.", v => cfg.MIN_COMMON_TOKEN = double.Parse(v, System.Globalization.CultureInfo.InvariantCulture) },
{ "v|verbose", "Verbose output.", v => cfg.Verbose = true },
};
var extraFiles = p.Parse(args);
// split that, as during parse the reference of cfg may change.
// The compiler might save the reference to the old, overridden
// config instance.
cfg.ExtraFiles = extraFiles;
if (showHelp)
{
ShowHelp(p);
return;
}
var comparer = new Comparer(cfg, progress);
var friendfinder = new FriendFinder(cfg);
var watch = Stopwatch.StartNew();
if (cfg.OneAgainstDir)
{
var dirFiles = CollectFiles(cfg, tokenizer, true, false);
var givenFiles = CollectFiles(cfg, tokenizer, false, true);
var compareResult = comparer.Compare(givenFiles, dirFiles);
var result = OSPCResult.Create(compareResult);
friendfinder.Find(result, compareResult);
CreateReports(cfg, html, console, result);
}
else
{
var files = CollectFiles(cfg, tokenizer);
if (files.Length != 0)
{
var compareResult = comparer.Compare(files);
var result = OSPCResult.Create(compareResult);
friendfinder.Find(result, compareResult);
CreateReports(cfg, html, console, result);
}
}
}
catch (Exception ex)
{
Console.WriteLine();
Console.WriteLine("********** Error occurred, exiting **********");
Console.WriteLine(ex.ToString());
Environment.Exit(1);
}
}
private static void CreateReports(Configuration cfg, Reporter.IReporter html, Reporter.IReporter console, OSPCResult result)
{
if (html != null)
{
html.Create(cfg, result);
}
console.Create(cfg, result);
}
private static void CreateReports(Configuration cfg, Reporter.IReporter html, Reporter.IReporter console, OSPCResult result)
{
if (html != null)
{
html.Create(cfg, result);
}
console.Create(cfg, result);
}
static void Main(string[] args)
{
try
{
var cfg = new Configuration();
bool showHelp = false;
// common dependencies
var progress = new ConsoleProgressReporter();
// changable through config or arguments
Reporter.IReporter html = null;
Reporter.IReporter console = new Reporter.ConsoleReporter();
Tokenizer.ITokenizer tokenizer = new Tokenizer.CLikeTokenizer();
Console.WriteLine("Open Software Plagiarism Checker");
Console.WriteLine("================================");
Console.WriteLine("Version {0}", typeof(OSPC.Program).Assembly.GetName().Version);
Console.WriteLine();
Console.WriteLine("Copyright (C) 2015 Arthur Zaczek at the UAS Technikum Wien, GPL V3");
Console.WriteLine("This program comes with ABSOLUTELY NO WARRANTY; see the GPL V3 for details.");
Console.WriteLine("This is free software, and you are welcome to redistribute it");
Console.WriteLine("under certain conditions; see the GPL V3 for details.");
Console.WriteLine();
var p = new OptionSet()
{
{ "h|?|help", "Prints this help", v => showHelp = true },
{ "c=", "Reads the given configuration. Note, this switch should be the first argument as it overrides any other argument parsed yet.", v => cfg = LoadConfig(v) },
{ "write-config=", "Write the current configuration to the given file. Note, this switch should be the last argument.", v => SaveConfig(cfg, v) },
{ "f=", "File filter. If -d is specified, then -f defaults to \"*.*.\"", v => cfg.Filter.Add(v) },
{ "d=", "Specifies a directory where the filer applies. If -f is specified, then -d defaults to \".\"", v => cfg.Dirs.Add(v) },
{ "include=", "Specifies a regular expression that every file must match. More than one expression is allowed. A file must match any of these expressions.", v => cfg.Include.Add(v) },
{ "exclude=", "Specifies a regular expression to exclude files. More than one expression is allowed. If a file must match any of these expressions it will be excluded.", v => cfg.Exclude.Add(v) },
{ "recurse", "Traversals all directories recurse. Use include-dir and exclude-dir to have more control over the selected directories.", v => cfg.Recurse = true },
{ "include-dir=", "Specifies a regular expression that every file must match. More than one expression is allowed. A file must match any of these expressions.", v => cfg.IncludeDir.Add(v) },
{ "exclude-dir=", "Specifies a regular expression to exclude files. More than one expression is allowed. If a file must match any of these expressions it will be excluded.", v => cfg.ExcludeDir.Add(v) },
{ "detailed", "Print a detailed report to the console", v => console = new Reporter.DetailedConsoleReporter() },
{ "summary", "Print only a summay to the console. Usefull if --html is used.", v => console = new Reporter.SummaryConsoleReporter() },
{ "html:", "Saves a html report to the specified directory. Defaults to \"report\"", v => html = new Reporter.Html.HtmlReporter(v, progress) },
{ "min-similarity=", "Minimum similarity of match reports (0 - 1). Default is 0.5 (=50%).", v => cfg.MIN_SIMILARITY = double.Parse(v, System.Globalization.CultureInfo.InvariantCulture) },
{ "min-match-length=", "Minimum count of matching tokens, including non-matching tokens.", v => cfg.MIN_MATCH_LENGTH = int.Parse(v) },
{ "max-match-distance=", "Maximum distance between tokens to count as a match. 1 = exact match.", v => cfg.MAX_MATCH_DISTANCE = int.Parse(v) },
{ "min-common-token=", "Percent of token that must match to count as a match. 1 = every token must match.", v => cfg.MIN_COMMON_TOKEN = double.Parse(v, System.Globalization.CultureInfo.InvariantCulture) },
{ "v|verbose", "Verbose output.", v => cfg.Verbose = true },
};
var extraFiles = p.Parse(args);
// split that, as during parse the reference of cfg may change.
// The compiler might save the reference to the old, overridden
// config instance.
cfg.ExtraFiles = extraFiles;
if (showHelp)
{
ShowHelp(p);
return;
}
var comparer = new Comparer(cfg, progress);
var friendfinder = new FriendFinder(cfg);
var watch = Stopwatch.StartNew();
Console.WriteLine("Collecting files");
var files = CollectFiles(cfg, tokenizer);
Console.WriteLine(" finished; time: {0:n2} sec.", watch.Elapsed.TotalSeconds);
if (files.Length == 0)
{
Console.WriteLine("No files found to compare!");
}
else
{
Console.Write("Comparing {0} files ", files.Length);
var compareResult = comparer.Compare(files);
Console.WriteLine(" finished; time: {0:n2} sec.", watch.Elapsed.TotalSeconds);
Console.WriteLine("Creating statistics");
var result = OSPCResult.Create(compareResult);
friendfinder.Find(result, compareResult);
Console.WriteLine(" finished; time: {0:n2} sec.", watch.Elapsed.TotalSeconds);
Console.WriteLine("Creating reports");
CreateReports(cfg, html, console, result);
Console.WriteLine(" finished in total {0:n2} sec.", watch.Elapsed.TotalSeconds);
}
}
catch (Exception ex)
{
Console.WriteLine();
Console.WriteLine("********** Error occurred, exiting **********");
Console.WriteLine(ex.ToString());
Environment.Exit(1);
}
}
