private void bSave_Click(object sender, EventArgs e)
{
int VectorType = 0;
if (rbAddress.Checked)
{
VectorType = 1;
}
else if (rbInfoBlock.Checked)
{
VectorType = 2;
}
else if (rbRGLine.Checked)
{
VectorType = 3;
}
if (VectorType > 0)
{
DataBase.DB DocObjects = new DataBase.DB();
DocObjects.SaveVectorSet(VectorType, FieldDataDetails);
this.Dispose();
}
else
{
MessageBox.Show("Please indicate a type for this object.");
}
//start here with fixing up a data model to house my raw vector data
}
private void ArchiveTrainingData()
{
DataBase.DB DocObjects = new DataBase.DB();
List <string[]> FieldDataDetails = new List <string[]>();
FieldDataDetails = DocObjects.GetDocObjects("Extract");
int GridRow = 0;
foreach (string[] FieldDataDetail in FieldDataDetails)
{
DocObjects.ArchiveObjectRow(
FieldDataDetail[0], // Object_Type
FieldDataDetail[1], //Native_Value
FieldDataDetail[2], //OCR_Value
FieldDataDetail[3], //XLTD_Value
Convert.ToDouble(FieldDataDetail[4]), //X1
Convert.ToDouble(FieldDataDetail[5]), //Y1
Convert.ToDouble(FieldDataDetail[6]), //X2
Convert.ToDouble(FieldDataDetail[7]), //Y2
Convert.ToDouble(FieldDataDetail[8]), //X3
Convert.ToDouble(FieldDataDetail[9]), //Y3
Convert.ToDouble(FieldDataDetail[10]), //X4
Convert.ToDouble(FieldDataDetail[11]), //Y4
PDFFileName);
GridRow++;
}
}
private void MainForm_Load(object sender, EventArgs e)
{
AppDomain.CurrentDomain.ProcessExit += new EventHandler(OnProcessExit);
DataBase.DB DocObjects = new DataBase.DB();
DocObjects.TruncateObjects(); // delete any data that might be in database
string[] args = Environment.GetCommandLineArgs();
if (args.Length > 1)
{
AutoprocessDocs(args[1]);
}
//C3PO_Text
//Key 1 415260fc79c1481986027718fa4c45ac
//Key 2 f6c0e985e2854e3c9db4ddf385ea5d89
//C3PO_LUIS
//Key 1 92d1791c0dff4a4fa6953b89818b8ccd
//Key 2 3dd2d4a6235d4fb4bc1fd7fe372831c3
/*ServiceClients.LuisClient LC = new ServiceClients.LuisClient("c68ba480925a4dc4be6ac44d35ace6cf");
* string results = LC.ReadText("Purchase Order").Result;
* ServiceClients.LUISResponse LUISResults = new ServiceClients.LUISResponse();
* LUISResults = Newtonsoft.Json.JsonConvert.DeserializeObject<ServiceClients.LUISResponse>(results);
*/
}
private void bSegCorrection_Click(object sender, EventArgs e)
{
DataBase.DB DocObjects = new DataBase.DB();
DocObjects.FormLineObjects();
PopulateGridView("LinedText");
int MeanSpacing = DocObjects.FindMeanSpacing();
}
private void bBlackBlob_Click(object sender, EventArgs e)
{
ImageProcessor EvalImage = new ImageProcessor();
DataBase.DB DocObjects = new DataBase.DB();
int ImageWidth = 0;
int ImageHeight = 0;
List <Rectangle> BlackBoxes = EvalImage.BlackBlob(PDFFileName + ".png", out ImageWidth, out ImageHeight);
double ImageScale = PDFPageWidth / ImageWidth;
DocObjects.ExtractTextFromRect(BlackBoxes, ImageScale);
PopulateGridView("Extract");
}
private void PopulateGridView(string ProcessField)
{
dgvDocObjects.Rows.Clear();
DataBase.DB DocObjects = new DataBase.DB();
List <string[]> FieldDataDetails = new List <string[]>();
FieldDataDetails = DocObjects.GetDocObjects(ProcessField);
//size the data grid to the number of rows in the data
int RowNum = FieldDataDetails.Count();
if (dgvDocObjects.Rows.Count < RowNum)
{
int NewRows = RowNum - dgvDocObjects.Rows.Count;
dgvDocObjects.Rows.Add(NewRows);
}
int GridRow = 0;
foreach (string[] FieldDataDetail in FieldDataDetails)
{
dgvDocObjects.Rows[GridRow].Cells[0].Value = FieldDataDetail[0]; // Object_Type
dgvDocObjects.Rows[GridRow].Cells[1].Value = FieldDataDetail[1]; //Native_Value
dgvDocObjects.Rows[GridRow].Cells[2].Value = FieldDataDetail[2]; //OCR_Value
dgvDocObjects.Rows[GridRow].Cells[3].Value = FieldDataDetail[3]; //XLTD_Value
dgvDocObjects.Rows[GridRow].Cells[4].Value = FieldDataDetail[4]; //X1
dgvDocObjects.Rows[GridRow].Cells[5].Value = FieldDataDetail[5]; //Y1
dgvDocObjects.Rows[GridRow].Cells[6].Value = FieldDataDetail[6]; //X2
dgvDocObjects.Rows[GridRow].Cells[7].Value = FieldDataDetail[7]; //Y2
dgvDocObjects.Rows[GridRow].Cells[8].Value = FieldDataDetail[8]; //X3
dgvDocObjects.Rows[GridRow].Cells[9].Value = FieldDataDetail[9]; //Y3
dgvDocObjects.Rows[GridRow].Cells[10].Value = FieldDataDetail[10]; //X4
dgvDocObjects.Rows[GridRow].Cells[11].Value = FieldDataDetail[11]; //Y4
dgvDocObjects.Rows[GridRow].Cells[12].Value = FieldDataDetail[12]; //Horizontal_Axis
dgvDocObjects.Rows[GridRow].Cells[13].Value = FieldDataDetail[13]; //Vertical_Axis
dgvDocObjects.Rows[GridRow].Cells[14].Value = FieldDataDetail[14]; //Span
dgvDocObjects.Rows[GridRow].Cells[15].Value = FieldDataDetail[15]; //Object_ID
GridRow++;
}
}
private void pbPDFImage_MouseUp(object sender, MouseEventArgs e)
{
if (drawing)
{
drawing = false;
var rc = getRectangle();
rectangles.Clear();
if (rc.Width > 0 && rc.Height > 0)
{
rectangles.Add(rc);
}
pbPDFImage.Invalidate();
//convert the image box coordinates to the document coordinates
Rectangle display = pbPDFImage.DisplayRectangle;
double X1Doc = rc.X * Convert.ToDouble(PDFPageWidth) / Convert.ToDouble(display.Width);
double Y1Doc = rc.Y * Convert.ToDouble(PDFPageHeight) / Convert.ToDouble(display.Height);
double WidthDoc = rc.Width * Convert.ToDouble(PDFPageWidth) / Convert.ToDouble(display.Width);
double HeightDoc = rc.Height * Convert.ToDouble(PDFPageHeight) / Convert.ToDouble(display.Height);
//get all Lined Text Objects that fall within this box
DataBase.DB DocObjects = new DataBase.DB();
List <string[]> FieldDataDetails = new List <string[]>();
FieldDataDetails = DocObjects.GetDocLTObjectsInBox(X1Doc, Y1Doc, WidthDoc, HeightDoc);
//highlight the fields on the screen
foreach (string[] FieldData in FieldDataDetails)
{
DrawRect(Convert.ToDouble(FieldData[4]), Convert.ToDouble(FieldData[9]), Convert.ToDouble(FieldData[5]), Convert.ToDouble(FieldData[6])); //X1, Y1, Y3, X2
}
Classifier UserAction = new Classifier(FieldDataDetails);
UserAction.ShowDialog(); //show a modal dialog box here for manual classification
}
}
private void btnObjectAnalysis_Click(object sender, EventArgs e)
{
//use the PDF library to break out all of the PDF Objects
Readers.PDFReader Rdr = new Readers.PDFReader();
Rdr.getObjectsFromPDF(PDFFileName, out PDFPageWidth, out PDFPageHeight);
//break up text objects with multiple words into a single word per object
DataBase.DB DocObjects = new DataBase.DB();
DocObjects.LocateKeywords();
PdfFixedDocument doc = new PdfFixedDocument(PDFFileName);
PdfPageRenderer renderer = new PdfPageRenderer(doc.Pages[0]);
FileStream PNGStream = File.OpenWrite(PDFFileName + ".png");
renderer.ConvertPageToImage(dpi, PNGStream, PdfPageImageFormat.Png);
PNGStream.Flush();
PNGStream.Close();
LoadImageBox();
PopulateGridView("Text");
}
private void AutoprocessDocs(string Filename)
{
//load the imagebox
PDFFileName = Filename;
//Autoprocess
//use the PDF library to break out all of the PDF Objects
Readers.PDFReader Rdr = new Readers.PDFReader();
Rdr.getObjectsFromPDF(Filename, out PDFPageWidth, out PDFPageHeight);
//break up text objects with multiple words into a single word per object
DataBase.DB DocObjects = new DataBase.DB();
DocObjects.LocateKeywords();
ImageProcessor EvalImage = new ImageProcessor();
int ImageWidth = 0;
int ImageHeight = 0;
List <Rectangle> BlackBoxes = EvalImage.BlackBox(Filename + ".png", out ImageWidth, out ImageHeight, true);
double ImageScale = PDFPageWidth / ImageWidth;
DocObjects.ExtractTextFromRect(BlackBoxes, ImageScale);
List <Rectangle> BlackBlobs = EvalImage.BlackBlob(Filename + ".png", out ImageWidth, out ImageHeight, true);
ImageScale = PDFPageWidth / ImageWidth;
DocObjects.ExtractTextFromRect(BlackBoxes, ImageScale);
DocObjects.TranslateText();
ArchiveTrainingData();
PopulateGridView("Extract");
}
private void bIdentifyData_Click(object sender, EventArgs e)
{
DataBase.DB DocObjects = new DataBase.DB();
DocObjects.ProcessIntents();
}
private void btranslate_click(object sender, EventArgs e)
{
DataBase.DB DocObjects = new DataBase.DB();
DocObjects.TranslateText();
PopulateGridView("Extract");
}
public void getObjectsFromPDF(string PDFFileName, out double PDFPageWidth, out double PDFPageHeight)
{
FileStream pageObjectsInput = new FileStream(PDFFileName, FileMode.Open, FileAccess.Read, FileShare.Read);
PdfFixedDocument document = new PdfFixedDocument(pageObjectsInput);
PdfPage DisplayPage = document.Pages[0];
PDFPageWidth = document.Pages[0].Width;
PDFPageHeight = document.Pages[0].Height;
PdfContentExtractor ce = new PdfContentExtractor(DisplayPage);
PdfVisualObjectCollection voc = ce.ExtractVisualObjects(false);
DataBase.DB DocObjects = new DataBase.DB();
DocObjects.TruncateObjects(); // delete any data that might be in database
for (int i = 0; i < voc.Count; i++)
{
switch (voc[i].Type)
{
case PdfVisualObjectType.Image:
PdfImageVisualObject ivo = voc[i] as PdfImageVisualObject;
//add image ocr here
DocObjects.InsertObjectRow("Image", ivo.Image.ImageID, "", "",
ivo.Image.ImageCorners[0].X, ivo.Image.ImageCorners[0].Y,
ivo.Image.ImageCorners[1].X, ivo.Image.ImageCorners[1].Y,
ivo.Image.ImageCorners[2].X, ivo.Image.ImageCorners[2].Y,
ivo.Image.ImageCorners[3].X, ivo.Image.ImageCorners[3].Y);
break;
case PdfVisualObjectType.Text:
PdfTextVisualObject tvo = voc[i] as PdfTextVisualObject;
string Translated_Text = "";
int Object_ID = DocObjects.InsertObjectRow("Text", tvo.TextFragment.Text, "", Translated_Text,
tvo.TextFragment.FragmentCorners[0].X, tvo.TextFragment.FragmentCorners[0].Y,
tvo.TextFragment.FragmentCorners[1].X, tvo.TextFragment.FragmentCorners[1].Y,
tvo.TextFragment.FragmentCorners[2].X, tvo.TextFragment.FragmentCorners[2].Y,
tvo.TextFragment.FragmentCorners[3].X, tvo.TextFragment.FragmentCorners[3].Y);
// int CharCtr = 0;
// foreach (Xfinium.Pdf.Content.PdfTextGlyph Character in tvo.TextFragment.Glyphs)
//{
// DocObjects.InsertInvoiceObjectDetail(Object_ID, CharCtr, Character.Text, Character.GlyphCorners[0].X, Character.Width, Character.GlyphCorners[0].Y, tvo.TextFragment.FragmentCorners[3].Y);
// CharCtr++;
// }
break;
case PdfVisualObjectType.Path:
PdfPathVisualObject pvo = voc[i] as PdfPathVisualObject;
// Examine all the path points and determine the minimum rectangle that bounds the path.
double minX = 999999, minY = 999999, maxX = -999999, maxY = -999999;
for (int j = 0; j < pvo.PathItems.Count; j++)
{
PdfPathItem pi = pvo.PathItems[j];
if (pi.Points != null)
{
for (int k = 0; k < pi.Points.Length; k++)
{
if (minX >= pi.Points[k].X)
{
minX = pi.Points[k].X;
}
if (minY >= pi.Points[k].Y)
{
minY = pi.Points[k].Y;
}
if (maxX <= pi.Points[k].X)
{
maxX = pi.Points[k].X;
}
if (maxY <= pi.Points[k].Y)
{
maxY = pi.Points[k].Y;
}
}
}
DocObjects.InsertObjectRow("Path", pvo.ToString(), "", "", minX, minY, maxX, minY, minX, maxY, maxX, maxY);
}
break;
}
}
//DocObjects.FormTextObjects();
//DocObjects.FormLineObjects();
//DocObjects.FormCenteredObjects();
}
