/// <summary>
/// Converts Pdf to MS Doc type.
/// </summary>
/// <param name="FileFullPath"></param>
/// <param name="OutPutFileFullPath"></param>
public static void PDFToWord(string FileFullPath, string OutPutFileFullPath)
{
iTextSharp.text.pdf.parser.ITextExtractionStrategy strategy = new iTextSharp.text.pdf.parser.LocationTextExtractionStrategy();
PdfReader pdfReader = new PdfReader(FileFullPath);
for (int page = 1; page <= pdfReader.NumberOfPages; page++)
{
string Pagestring;
Pagestring = iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(pdfReader, page, strategy);
FileStream richTextBox1fs = new FileStream(OutPutFileFullPath, FileMode.Create, FileAccess.Write, FileShare.None);
StreamWriter sw = new StreamWriter(richTextBox1fs);
sw.WriteLine(Pagestring);
sw.Flush();
sw.Close();
}
//AcroPDDoc pdfd = new AcroPDDoc();
//pdfd.Open(FileFullPath);
//Object jsObj = pdfd.GetJSObject();
//Type jsType = pdfd.GetType();
////have to use acrobat javascript api because, acrobat
//object[] saveAsParam = { "newFile.doc", "com.adobe.acrobat.doc", "", false, false };
//jsType.InvokeMember("saveAs", BindingFlags.InvokeMethod | BindingFlags.Public | BindingFlags.Instance, null, jsObj, saveAsParam, CultureInfo.InvariantCulture);
}
public static string GetTextFromAllPages(iTextSharp.text.pdf.PdfReader reader)
{
// ITextExtractionStrategy strategy = new LocationTextExtractionStrategy();
var strategy = new iTextSharp.text.pdf.parser.LocationTextExtractionStrategy();
var output = new System.IO.StringWriter();
for (int i = 1; i <= reader.NumberOfPages; i++)
{
var text = iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(reader, i, strategy);
text = Convert(text);
output.WriteLine(text);
}
return(output.ToString());
}
public void verify()
{
addresscollection aic = new addresscollection();
addressitem ai = null;
string sourceFileName = _sourcefilename;
FileStream x = new FileStream(sourceFileName, FileMode.Open);
iTextSharp.text.pdf.PdfReader reader = new iTextSharp.text.pdf.PdfReader(x);
//AWESOME!!
x.Close();
x.Dispose();
string s = "";
string s1 = "";
System.Xml.XmlNode ep = null;
System.Xml.XmlNode batch = null;
System.Xml.XmlNode startingpage = null;
System.Xml.XmlNode envelope = null;
int pages = reader.NumberOfPages;
int i;
for (i = 0; i <= reader.NumberOfPages - 1; i++) {
this.Label2.Invoke(new updatetext(updatelabel1text), new object[] { "Processing Page " + Convert.ToString(i + 1) + " of " + pages });
DataRow dr = _dtt.Rows[0];
System.util.RectangleJ rect1 = new System.util.RectangleJ(Convert.ToInt32( dr["x"]), Convert.ToInt32( dr["y"]), Convert.ToInt32( dr["width"]), Convert.ToInt32( dr["height"]));
iTextSharp.text.pdf.parser.RegionTextRenderFilter rf = new iTextSharp.text.pdf.parser.RegionTextRenderFilter(rect1);
iTextSharp.text.pdf.parser.LocationTextExtractionStrategy mystrat = new iTextSharp.text.pdf.parser.LocationTextExtractionStrategy();
iTextSharp.text.pdf.parser.RegionTextRenderFilter[] rtrf = new iTextSharp.text.pdf.parser.RegionTextRenderFilter[2];
rtrf[0] = rf;
//Dim rect2 As New System.util.RectangleJ(0, 700, 800, 140)
//Dim rf2 As New iTextSharp.text.pdf.parser.RegionTextRenderFilter(rect2)
iTextSharp.text.pdf.parser.FilteredTextRenderListener textExtractionStrategy = new iTextSharp.text.pdf.parser.FilteredTextRenderListener(mystrat, rtrf);
DataRow dr1 = _dtt.Rows[1];
System.util.RectangleJ rect2 = new System.util.RectangleJ(Convert.ToInt32( dr1["x"]), Convert.ToInt32( dr1["y"]), Convert.ToInt32( dr1["width"]), Convert.ToInt32( dr1["height"]));
iTextSharp.text.pdf.parser.RegionTextRenderFilter rf1 = new iTextSharp.text.pdf.parser.RegionTextRenderFilter(rect2);
iTextSharp.text.pdf.parser.LocationTextExtractionStrategy mystrat1 = new iTextSharp.text.pdf.parser.LocationTextExtractionStrategy();
iTextSharp.text.pdf.parser.RegionTextRenderFilter[] rtrf1 = new iTextSharp.text.pdf.parser.RegionTextRenderFilter[2];
rtrf1[0] = rf1;
//Dim rect2 As New System.util.RectangleJ(0, 700, 800, 140)
//Dim rf2 As New iTextSharp.text.pdf.parser.RegionTextRenderFilter(rect2)
iTextSharp.text.pdf.parser.FilteredTextRenderListener textExtractionStrategy1 = new iTextSharp.text.pdf.parser.FilteredTextRenderListener(mystrat1, rtrf1);
s = iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(reader, i + 1, textExtractionStrategy);
if (!string.IsNullOrEmpty(s)) {
s1 = iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(reader, i + 1, textExtractionStrategy1);
if (parseaddress(s, s1, ref batch, i + 1, ref ep, ref envelope, ref startingpage, ref ai)) {
//Des not catch a single first page, anytime this is true it's a new first page
// ep.InnerText = i + 1
// Console.WriteLine(ai.Address1)
aic.Add(ai);
}
}
if (i == reader.NumberOfPages - 1 & (ep != null)) {
ep.InnerText = Convert.ToString(i + 1);
}
if (i == reader.NumberOfPages - 1 & (ai != null)) {
ai.endpage = Convert.ToInt32( i + 1);
}
// CurrentPage = CurrentPage + 1
}
reader.Close();
//Me.RichTextBox1.Invoke(New updatert(AddressOf updaterichtext), New Object() {XMLDOC.OuterXml, badaddress})
this.RichTextBox1.Invoke(new updatert(updaterichtext), new object[] {
XMLDOC,
badaddress
});
//Me.RichTextBox1.Invoke(New updatert(AddressOf updaterichtext), New Object() {badaddress})
Invoke(new updatedatagrid(updatedatagridonMail), new object[] { aic });
Invoke(new updatecomplete(updatecompleted), new object[]{});
//rtrf(0) = rf2
//textExtractionStrategy = New iTextSharp.text.pdf.parser.FilteredTextRenderListener(mystrat, rtrf)
//MsgBox(iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(reader, 1, textExtractionStrategy))
// iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(reader, 1, textExtractionStrategy)
}
public string ExtractTextFromRegionOfPdf(string sourceFileName)
{
FileStream x = new FileStream(sourceFileName, FileMode.Open);
iTextSharp.text.pdf.PdfReader reader = new iTextSharp.text.pdf.PdfReader(x);
//AWESOME!!
System.util.RectangleJ rect1 = new System.util.RectangleJ(Rect.X, System.Math.Abs(this.PictureBox1.Height - Rect.Y) - Rect.Height, Rect.Width, Rect.Height);
iTextSharp.text.pdf.parser.RegionTextRenderFilter rf = new iTextSharp.text.pdf.parser.RegionTextRenderFilter(rect1);
iTextSharp.text.pdf.parser.LocationTextExtractionStrategy mystrat = new iTextSharp.text.pdf.parser.LocationTextExtractionStrategy();
iTextSharp.text.pdf.parser.RegionTextRenderFilter[] rtrf = new iTextSharp.text.pdf.parser.RegionTextRenderFilter[2];
rtrf[0] = rf;
//Dim rect2 As New System.util.RectangleJ(0, 700, 800, 140)
//Dim rf2 As New iTextSharp.text.pdf.parser.RegionTextRenderFilter(rect2)
iTextSharp.text.pdf.parser.FilteredTextRenderListener textExtractionStrategy = new iTextSharp.text.pdf.parser.FilteredTextRenderListener(mystrat, rtrf);
//rtrf(0) = rf2
//textExtractionStrategy = New iTextSharp.text.pdf.parser.FilteredTextRenderListener(mystrat, rtrf)
//MsgBox(iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(reader, 1, textExtractionStrategy))
// iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(reader, 1, textExtractionStrategy)
x.Close();
x.Dispose();
reader.Close();
if (_mode == 1) {
if (this.loadedbool) {
string s = iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(reader, CurrentPage + 1, textExtractionStrategy);
DialogResult y = MessageBox.Show("This field is showing : " + s + "\r\n" + "Is this the correct variable on this page?","Confirm" , MessageBoxButtons.YesNo );
if (y == DialogResult.Yes ) {
_dtt.Rows[_CurrentCount]["x"] = Rect.X;
_dtt.Rows[_CurrentCount]["y"] = System.Math.Abs(this.PictureBox1.Height - Rect.Y) - Rect.Height;
_dtt.Rows[_CurrentCount]["width"] = Rect.Width;
_dtt.Rows[_CurrentCount]["height"] = Rect.Height;
if (_CurrentCount == 0) {
DialogResult xx = MessageBox.Show("There is an optional Parimeter where you can select something that only appears on the first page, do you want to add this. It can be part of a string like Page 1 of XX?", "Confirm", MessageBoxButtons.YesNo);
if (xx == DialogResult.No)
{
_CurrentCount = 2;
}
}
if (_dtt.Rows[_CurrentCount]["FieldName"] == "FirstPageConstant") {
_validatetext = Interaction.InputBox("Enter Charectors to match, if you enter \"1 of \" it will be true for anything after the of");
_dtt.Rows[_CurrentCount + 1]["misc"] = _validatetext;
_CurrentCount += 1;
}
_CurrentCount += 1;
if (_CurrentCount == 3) {
_CurrentCount = 0;
this.Label2.Text = "OK, you have completed the template, if you wish to start over simply do it again and start by selecting the area with:" + _dtt.Rows[0]["fieldname"];
startover = 1;
drawrectangles();
MessageBox.Show("Make sure you save this if you want to use it in the future.");
} else {
this.Label2.Text = "Now Select : " + _dtt.Rows[_CurrentCount]["fieldname"];
}
}
}
} else {
this.Label2.Text = "OK, you have completed the template, if you wish to start over simply do it again and start by selecting the area with:" + _dtt.Rows[0]["fieldname"];
}
return "";
}
