public async Task ProcessDocumentsAsync()
{
_criteriaFilePaths = new List <string>();
ClassifyEnabled = false;
var tempDirectoryInfo = new DirectoryInfo(Common.TempStorage);
var tempFiles = tempDirectoryInfo.GetFiles();
foreach (var file in tempFiles)
{
File.Delete(file.FullName);
}
var pdfFiles = new List <string> {
PdfPath
};
var pdfImages = await Common.ConvertPdfsToImagesAsync(pdfFiles);
var pngImages = await Common.CopyImagesToTempFolderAsync(pdfFiles);
PdfImages = new Dictionary <string, string>(pdfImages);
var types = new List <DocumentTypes>();
List <DocumentCriteria> documentCriteria = null;
using (var context = new ClassifierContext())
{
var dTypes = context.DocumentTypes.ToList();
foreach (var o in dTypes)
{
if (DocumentSelectionList.First(c => c.DocumentTypeId == o.Id).Selected)
{
types.Add(o);
}
}
documentCriteria = context.DocumentCriteria.ToList();
}
await Common.CreateCriteriaFilesAsync(documentCriteria, types);
var criteriaFolder = new DirectoryInfo(Common.CriteriaStorage);
_criteriaFilePaths = new List <string>();
var criteriaFiles = criteriaFolder.GetFiles();
foreach (var o in criteriaFiles)
{
_criteriaFilePaths.Add(o.FullName);
}
tempDirectoryInfo = new DirectoryInfo(Common.TempStorage);
var files = tempDirectoryInfo.GetFiles();
await ProcessSelectedDocumentsAsync(files.ToList());
ClassifyEnabled = true;
}
public Task ProcessSelectedDocumentsAsync(IProgress <TaskProgress> prog, CancellationToken token, List <DocumentTypes> types, EtaCalculator pc, List <FileInfo> files)
{
var currentFile = 0.0;
var fileCount = Convert.ToDouble(files.Count);
return(Task.Run(() =>
{
foreach (var file in files)
{
var criteriaMatches = types.Select(o => new CriteriaMatchModel {
DocumentType = o
}).ToList();
using (var observedImage = CvInvoke.Imread(file.FullName))
{
Parallel.ForEach(_criteriaImages, (criteriaImage) =>
{
var criteriaFile = criteriaImage.Info;
var criteriaFileSplit = criteriaFile.Name.Split('-');
var type = types.First(c => c.DocumentType == criteriaFileSplit[0]);
var score = Classify(criteriaImage, observedImage);
var existingModel = criteriaMatches.First(c => c.DocumentType == type);
existingModel.Score += score;
existingModel.PdfFile = file.FullName;
});
}
if (token.IsCancellationRequested)
{
return;
}
var matchedCriteria = criteriaMatches.First(c => c.Score == criteriaMatches.Max(p => p.Score));
Console.WriteLine($"Score: {matchedCriteria.Score}");
if (matchedCriteria.Score >= matchedCriteria.DocumentType.AverageScore)
{
DocumentSelectionList.First(c => c.DocumentType == matchedCriteria.DocumentType.DocumentType).Matches += 1;
System.Windows.Application.Current.Dispatcher.Invoke(() =>
{
SelectionViewSource.Refresh();
});
var matchedFileName = file.Name.Substring(0, file.Name.Length - 4).Split('.')[0];
var matchedFile = PdfFiles.First(c => c.Contains(matchedFileName));
var matchedFileExtension = matchedFile.Substring(matchedFile.Length - 3);
if (matchedFileExtension.Equals("pdf", StringComparison.CurrentCultureIgnoreCase))
{
ExtractPageFromPdf(file, matchedCriteria.DocumentType.DocumentType, NamingModels);
}
else
{
CreatePdfFromImage(file, matchedCriteria.DocumentType.DocumentType, NamingModels);
}
}
currentFile++;
var rawProgress = (currentFile / fileCount);
var progress = rawProgress * 100;
var progressFloat = (float)rawProgress;
pc.Update(progressFloat);
if (pc.ETAIsAvailable)
{
var timeRemaining = pc.ETR.ToString(@"dd\.hh\:mm\:ss");
prog.Report(new TaskProgress
{
ProgressText = file.Name,
ProgressPercentage = progress,
ProgressText2 = timeRemaining
});
}
else
{
prog.Report(new TaskProgress
{
ProgressText = file.Name,
ProgressPercentage = progress,
ProgressText2 = "Calculating..."
});
}
}
}));
}
public async Task ProcessDocumentsAsync()
{
CancelEnabled = true;
ClassifyEnabled = false;
var tempDirectory = new DirectoryInfo(Common.TempStorage);
var tempFiles = tempDirectory.GetFiles();
foreach (var file in tempFiles)
{
File.Delete(file.FullName);
}
CancelTokenSource = new CancellationTokenSource();
var token = CancelTokenSource.Token;
var prog = new Progress <TaskProgress>();
prog.ProgressChanged += (sender, exportProgress) =>
{
ProgressPercentage = Math.Round(exportProgress.ProgressPercentage, 2);
ProgressText = exportProgress.ProgressText;
ProgressText2 = exportProgress.ProgressText2;
};
ProgressText = "Creating PDF Files.";
var eta = new EtaCalculator(1, 30);
var pdfImages = await Common.ConvertPdfsToImagesAsync(PdfFiles, prog, eta);
var pngImages = await Common.CopyImagesToTempFolderAsync(PdfFiles);
PdfImages = new Dictionary <string, string>(pdfImages);
foreach (var png in pngImages)
{
PdfImages.Add(png.Key, png.Value);
}
var types = new List <DocumentTypes>();
List <DocumentCriteria> documentCriteria = null;
using (var context = new ClassifierContext())
{
var dTypes = context.DocumentTypes.ToList();
foreach (var o in dTypes)
{
if (DocumentSelectionList.First(c => c.DocumentTypeId == o.Id).Selected)
{
types.Add(o);
}
}
documentCriteria = context.DocumentCriteria.ToList();
}
ProgressText = "Finding Matches";
ProgressPercentage = 0.0;
await Common.CreateCriteriaFilesAsync(documentCriteria, types);
var criteriaAndNaming = Common.SetNamingAndCriteria(NamingSpreadsheetPath);
NamingModels = new List <FileNamingModel>(criteriaAndNaming.Item1);
_criteriaImages = new List <CriteriaImageModel>(criteriaAndNaming.Item2);
var files = tempDirectory.GetFiles();
eta = new EtaCalculator(3, 30);
await ProcessSelectedDocumentsAsync(prog, token, types, eta, files.ToList());
DialogTitle = "Complete";
DialogText = "The documents you selected have been classified.";
DialogVisible = true;
}
