public static List <Tuple <Shred, int> > LoadShredsRandomized(List <string> filenames, bool runOcr = true)
{
var randomizedFilenames = Shuffle(filenames);
var shreds = Shred.Factory(randomizedFilenames.Select(pair => pair.Item1).ToList(), runOcr);
return(randomizedFilenames.Zip(shreds, (pair, shred) => Tuple.Create(shred, pair.Item2)).ToList());
}
public void NaiveKruskalArtificial()
{
Shred.BUFFER = 0;
Shred.SAMPLE_SIZE = 4;
var path = Path.Combine(Drive.GetDriveRoot(), Dir.ArtificialTestDirectory, Dir.NaiveKruskalHttpDocument);
var shreds = Shred.Factory("image", path, false);
var results = Reconstructor.NaiveKruskalAlgorithm(shreds);
var checker = Helpers.BuildChecker(Path.Combine(path, Helpers.CheckFile));
shreds.ForEach(shred => Console.Write(" " + shred.Id + ", "));
Console.WriteLine();
results.ForEach(shred => Console.Write(" " + shred.Id + ", "));
Console.WriteLine();
var indicies = results.Select((t, pos) => new Tuple <int, Shred>(pos, t)).ToList();
var result = indicies.Select(pair =>
{
var filename = Path.GetFileName(pair.Item2.Filepath);
Assert.IsNotNull(filename);
Assert.IsNotNull(checker[filename]);
var expected = checker[filename];
var actual = pair.Item1;
Console.WriteLine("Actual " + actual + " vs. " + expected + " | File: " + filename);
return(actual.ToString(CultureInfo.InvariantCulture) == expected);
}).ToList();
// var diff = Differ.DiffShredByOrder(results.Select(shred => shred.Id).ToList(),
//Enumerable.Range(0, results.Count).Select(ii => (long)ii).ToList());
// Console.WriteLine("Difference : " + diff);
ExportResult((Cluster)shreds.First().Root(), "../../visualizer/NaiveKruskalArtifical.png");
}
public void StitcherArtificialTest()
{
var shreds = Shred.Factory("image", Path.Combine(Drive.GetDriveRoot(), Dir.ArtificialTestDirectory, Dir.ArtificialHttpDocument), false);
var bitmap = Stitcher.Merge(shreds);
bitmap.Save("StitcherArtificialTest.png", ImageFormat.Png);
}
public void StitcherPrimitiveTest()
{
var shreds = Shred.Factory("Shred", Path.Combine(Drive.GetDriveRoot(), Dir.StitcherTestDirectory, "PrimitiveTest"), false);
var bitmap = Stitcher.Merge(shreds);
bitmap.Save("StitcherPrimitiveTest.png", ImageFormat.Png);
}
public static List <Shred> InitializeShreds(bool runOcr = false)
{
Console.WriteLine("Building Shreds");
var path = Path.Combine(Drive.GetDriveRoot(), Dir.CarusoTestDirectory, Dir.FewTestMaterials);
var shreds = Shred.Factory("image", path, runOcr);
return(shreds);
}
public static List <Shred> BootstrapPrimitiveThree()
{
Console.WriteLine("Building Shreds");
var path = Path.Combine(Drive.GetDriveRoot(), Dir.PrimitiveTestDirectory, Dir.PrimitiveTestThreeNormal);
var shreds = Shred.Factory("Shred", path, false);
return(shreds);
}
public static void Reconstruct(string prefix, string dir, bool display)
{
Console.WriteLine("Loading Shreds");
var shreds = Shred.Factory(prefix, dir, false);
Console.WriteLine("Comparing And Clusering");
var results = Reconstructor.NaiveKruskalAlgorithm(shreds);
Console.WriteLine("Exporting Results");
NaiveKruskalTests.ExportResult((Cluster)results.First().Root(), Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Desktop), "output.png"),
Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Desktop), "json.js"));
}
public void OcrFullOrientationTest()
{
const string docRev = @"Documents\EnergyInfrastructure\Full2";
const string docReg = @"Documents\IncomeTax_Ingalls\Full1";
var driveRev = new Drive(docRev, Drive.Reason.Read);
var driveReg = new Drive(docReg, Drive.Reason.Read);
var revs = Shred.Factory(driveRev.Files("image").ToList(), true);
var regs = Shred.Factory(driveReg.Files("image").ToList(), true);
var resV = revs.Select(shred =>
{
Console.Write(" | Expected:" + Orientation.Reversed);
Console.Write(" | Actual: " + (shred.TrueOrienation.ToString()));
Console.Write(" | Orientation Confidence: " + shred.OrientationConfidence);
if (shred.TrueOrienation == null)
{
Console.WriteLine(" | Empty? =" + shred.IsEmpty);
Console.WriteLine("Correct? Indeterminable");
return(true);
}
var passed = (Orientation.Reversed == shred.TrueOrienation || shred.IsEmpty);
Console.WriteLine(" | Empty? =" + shred.IsEmpty);
Console.WriteLine("Correct? = " + passed);
return(passed);
}).ToList();
var resG = regs.Select(shred =>
{
Console.Write(" | Expected:" + Orientation.Regular);
Console.Write(" | Actual: " + (shred.TrueOrienation.ToString()));
Console.Write(" | Orientation Confidence: " + shred.OrientationConfidence);
if (shred.TrueOrienation == null)
{
Console.WriteLine(" | Empty? =" + shred.IsEmpty);
Console.WriteLine("Correct? Indeterminable");
return(true);
}
var passed = Orientation.Regular == shred.TrueOrienation || shred.IsEmpty;
Console.WriteLine(" | Empty? =" + shred.IsEmpty);
Console.WriteLine("Correct? = " + passed);
return(passed);
}).ToList();
Assert.IsTrue(resV.All(item => item));
Assert.IsTrue(resG.All(item => item));
}
public void NaiveKruskalAuthentic()
{
var path = Path.Combine(Drive.GetDriveRoot(), Dir.PdfRequiremnetTestDirectory,
Dir.PdfRequiremnetTestFullOne);
var shreds = Shred.Factory("image", path, false);
var results = Reconstructor.NaiveKruskalAlgorithm(shreds);
shreds.ForEach(shred => Console.Write(" " + shred.Id + ", "));
Console.WriteLine();
results.ForEach(shred => Console.Write(" " + shred.Id + ", "));
Console.WriteLine();
var diff = Differ.DiffShredByOrder(results.Select(shred => shred.Id).ToList(), Enumerable.Range(0, results.Count).Select(ii => (long)ii).ToList());
Console.WriteLine("Difference : " + diff);
ExportResult((Cluster)results.First().Root(), "../../visualizer/NaiveKruskalAuthentic.png");
}
public void TestQueue()
{
var max = 1.00;
var path = Path.Combine(Drive.GetDriveRoot(), Dir.PdfRequiremnetTestDirectory,
Dir.PdfRequiremnetTestFullOne);
var shreds = Shred.Factory("image", path, false);
var queue = Reconstructor.BuildQueue(shreds);
while (queue.Count > 0)
{
var next = queue.Dequeue().ChamferSimilarity;
Assert.IsTrue(next <= max);
Console.WriteLine(next);
max = next;
}
}
public void ShredFactoryTest()
{
Console.WriteLine("Loading File from TestDrive And From Factory Independently");
var relativeDirectory = Path.Combine(CarusoTestDirectory, CarusoTestMaterialsFolder);
var fullDirectory = Path.Combine(Drive.GetDriveRoot(), relativeDirectory);
var mydrive = new Drive(relativeDirectory, Drive.Reason.Read);
var shreds = Shred.Factory("image", fullDirectory);
Assert.IsTrue(shreds.Count == mydrive.FileCount("image"));
foreach (Shred shred in shreds)
{
Assert.IsTrue(File.Exists(shred.Filepath));
}
Console.WriteLine("Factory Test Successful");
}
public string Reconstruct(string prefix, string dir, bool display)
{
StringBuilder sb = new StringBuilder();
WriteToLog("Loading Shreds");
var shreds = Shred.Factory(prefix, dir, false);
WriteToLog("Comparing And Clusering");
var results = Reconstructor.NaiveKruskalAlgorithm(shreds);
WriteToLog("Exporting Results");
string currentDir = Directory.GetCurrentDirectory();
NaiveKruskalTests.ExportResult((Cluster)results.First().Root(), Path.Combine(currentDir, "output.png"),
Path.Combine(currentDir, "json.js"));
WriteToLog("Export Complete");
return(sb.ToString());
}
public void OcrEmptyDetectionTest()
{
//Load Shreds
var abspath = Path.Combine(Drive.GetDriveRoot(), Dir.OcrDirectory, Dir.OcrEmptyTestDirectory);
var shreds = Shred.Factory("empty", abspath, true);
var checker = Helpers.BuildChecker(Path.Combine(abspath, Helpers.CheckFile));
var results = shreds.Select(shred =>
{
var filename = Path.GetFileName(shred.Filepath);
Assert.IsTrue(filename != null);
Console.WriteLine("--------------------------------------");
Console.WriteLine("Shred Name" + Path.GetFileNameWithoutExtension(shred.Filepath));
Console.WriteLine("IsEmpty? = " + shred.IsEmpty);
Console.WriteLine("Is Really Empty? = " + checker[filename]);
Console.WriteLine("Correct: " + (checker[filename] == (shred.IsEmpty ? "y" : "n")));
Console.WriteLine();
return(checker[filename] == (shred.IsEmpty ? "y" : "n"));
});
results.ToList().ForEach(Assert.IsTrue);
}
public void NaiveKruskalPrimitive()
{
Shred.BUFFER = 0;
Shred.SAMPLE_SIZE = 1;
var folder = Path.Combine(Drive.GetDriveRoot(), Dir.NaiveKruskalTestDirectory);
var paths = new List <string>();
paths.Add(Path.Combine(folder, NaiveThree));
paths.Add(Path.Combine(folder, NaiveSix));
paths.Add(Path.Combine(folder, NaiveTen));
foreach (var path in paths)
{
var shreds = Shred.Factory("Shred", path, false);
var checker = Helpers.BuildChecker(Path.Combine(path, Helpers.CheckFile));
var results = Reconstructor.NaiveKruskalAlgorithm(shreds);
var indicies = results.Select((t, pos) => new Tuple <int, Shred>(pos, t)).ToList();
Console.WriteLine(" ---- " + path + "----");
var result = indicies.Select(pair =>
{
var filename = Path.GetFileName(pair.Item2.Filepath);
Assert.IsNotNull(filename);
Assert.IsNotNull(checker[filename]);
var expected = checker[filename];
var actual = pair.Item1;
Console.WriteLine("Actual " + actual + " vs. " + expected + " | File: " + filename);
return(actual.ToString(CultureInfo.InvariantCulture) == expected);
}).ToList();
Exporter.ExportJson(results.First().Root());
Console.WriteLine();
result.ForEach(Assert.IsTrue);
}
}
