protected override void OnPrintPage(PrintPageEventArgs e)
{
if (_currentPage < _document.PageCount)
{
_document.Render(
_currentPage++,
e.Graphics,
e.Graphics.DpiX,
e.Graphics.DpiY,
new Rectangle(
0,
0,
(int)((e.PageBounds.Width / 100.0) * e.Graphics.DpiX),
(int)((e.PageBounds.Height / 100.0) * e.Graphics.DpiY)
),
true
);
}
int pageCount =
PrinterSettings.ToPage == 0
? _document.PageCount
: Math.Min(PrinterSettings.ToPage, _document.PageCount);
e.HasMorePages = _currentPage < pageCount;
}
private static void SavePng(PdfDocument document, int pageNumber, string destFileName)
{
SizeF sizeInPoints = document.PageSizes[pageNumber];
int widthInPixels = (int)Math.Round(sizeInPoints.Width * (float)dpi / 72F);
int heightInPixels = (int)Math.Round(sizeInPoints.Height * (float)dpi / 72F);
using (Image image = document.Render(pageNumber, widthInPixels, heightInPixels, dpi, dpi, true)) {
image.Save(destFileName, ImageFormat.Png);
}
}
private void DrawPageImage(Graphics graphics, int page, Rectangle pageBounds)
{
_document.Render(page, graphics, graphics.DpiX, graphics.DpiY, pageBounds, false);
}
protected override void OnPrintPage(PrintPageEventArgs e)
{
if (_currentPage < _document.PageCount)
{
var pageOrientation = GetOrientation(_document.PageSizes[_currentPage]);
var printOrientation = GetOrientation(e.PageBounds.Size);
e.PageSettings.Landscape = pageOrientation == Orientation.Landscape;
double left;
double top;
double width;
double height;
if (_printMode == PdfPrintMode.ShrinkToMargin)
{
left = 0;
top = 0;
width = e.PageBounds.Width - e.PageSettings.HardMarginX * 2;
height = e.PageBounds.Height - e.PageSettings.HardMarginY * 2;
}
else
{
left = -e.PageSettings.HardMarginX;
top = -e.PageSettings.HardMarginY;
width = e.PageBounds.Width;
height = e.PageBounds.Height;
}
if (pageOrientation != printOrientation)
{
double tmp = width;
width = height;
height = tmp;
tmp = left;
left = top;
top = tmp;
}
_document.Render(
_currentPage++,
e.Graphics,
e.Graphics.DpiX,
e.Graphics.DpiY,
new Rectangle(
AdjustDpi(e.Graphics.DpiX, left),
AdjustDpi(e.Graphics.DpiY, top),
AdjustDpi(e.Graphics.DpiX, width),
AdjustDpi(e.Graphics.DpiY, height)
),
PdfRenderFlags.ForPrinting | PdfRenderFlags.Annotations
);
}
int pageCount =
PrinterSettings.ToPage == 0
? _document.PageCount
: Math.Min(PrinterSettings.ToPage, _document.PageCount);
e.HasMorePages = _currentPage < pageCount;
}
public static string ExtractTextFromRawPdf(byte[] pdfFile, out ExtractedAuditLetterText extractedAuditLetterText, TimeSpan ts, DateTime dt1, int num)
{
//const string testReadPath = @"C:\Users\t-holu\Documents\AuditLetter\JixiProjectData\testread.txt";
//const string testimageReadPath = @"C:\Users\t-holu\Documents\AuditLetter\JixiProjectData\testimageread.txt";
List <MergedTraversedLine> txtFormattedLines = new List <MergedTraversedLine>();
List <MergedTraversedLine> imageFormattedLines = new List <MergedTraversedLine>();
List <int> txtFormattedNum = new List <int>();
List <int> imageFormattedNum = new List <int>();
//try {
// pdfRead.pdfObject.PdfDocument document = new pdfRead.pdfObject.PdfDocument(pdfFile);
//} catch(Exception e) {
// throw e;
//}
//Merge image reader
Spire.Pdf.PdfDocument imgDoc = new Spire.Pdf.PdfDocument(pdfFile);
pdfRead.pdfObject.PdfDocument doc = new pdfRead.pdfObject.PdfDocument(pdfFile);
List <PdfTextLine> textLines;
doc.PageTextLine(doc.pages.Count - 1);
for (int i = 0; i < doc.pages.Count; i++)
{
textLines = doc.PageTextLine(i);
if (textLines.Count > 2)
{
txtFormattedNum.Add(i);
}
else
{
imageFormattedNum.Add(i);
}
}
ExtractedAuditLetterText extractedTexts = new ExtractedAuditLetterText();
ExtractedAuditLetterText extractedImageTexts = new ExtractedAuditLetterText();
/// <summary>
/// Get text from txt-formatted pdf
/// </summary>
if (txtFormattedNum.Count != 0)
{
//Get lines from pdf and they are grouped by their fontsizes.
Dictionary <double, List <PdfTextLine> > linesKeyValue = doc.lineFontSize;
int index = 0;
foreach (var key in linesKeyValue)
{
foreach (var value in key.Value)
{
var bdcLines = new MergedTraversedLine();
bdcLines.Index = index;
bdcLines.Text = value.text;
index++;
txtFormattedLines.Add(bdcLines);
}
}
ExtractedAuditLetterText extractedTxtedTexts = PdfExtractor.Utilities.ConvertToExtractedAuditLetterTexts(txtFormattedLines);
var littleImages = doc.ExtractImages();
foreach (var img in littleImages)
{
var lines = Utilities.ExtractLinesFromImage(img);
foreach (var line in lines)
{
string other = string.Join(" ", line.Words.Select(r => r.Text));
if (other == "auren")
{
other = "Auren";
}
extractedTexts.Others.Add(other);
//Console.WriteLine(other);
}
}
//foreach(int i in txtFormattedNum) {
// var littleImages = imgDoc.Pages[i].ExtractImages();
// foreach(var img in littleImages) {
// var lines = Utilities.ExtractLinesFromImage(img);
// foreach(var line in lines) {
// extractedTexts.Others.Add(string.Join(" ", line.Words.Select(r => r.Text)));
// Console.WriteLine(string.Join(" ", line.Words.Select(r => r.Text)));
// }
// }
//}
//string serialzedAuditLetterTxtedText = extractedTxtedTexts.SerializeExtractedAuditLetterText();
//extractedAuditLetterText = extractedTxtedTexts;
//return serialzedAuditLetterTxtedText;
}
/// <summary>
/// Get text from txt-image-formatted pdf
/// </summary>
if (imageFormattedNum.Count != 0)
{
int titlePageStartIndexNum = 0;
int titlePageEndIndexNum = -1;
List <Image> imagePdf = new List <Image>();
PdfiumViewer.PdfDocument document = PdfiumViewer.PdfDocument.Load(new MemoryStream(pdfFile));
foreach (int i in imageFormattedNum)
{
Image image = document.Render(i, Consts.DpiX, Consts.DpiY, PdfRenderFlags.CorrectFromDpi);
imagePdf.Add(image);
}
List <List <Line> > linesOfImagePdf = new List <List <Line> >();
foreach (var img in imagePdf)
{
linesOfImagePdf.Add(Utilities.ExtractLinesFromImage(img));
}
extractedTexts.Others.AddRange(Utilities.GetRedundantLines(imagePdf, linesOfImagePdf));
StringBuilder sb = new StringBuilder();
for (int i = 0; i < imagePdf.Count; i++)
{
//Add small region text to ExtractedAuditLetterText.others
extractedTexts.Others.AddRange(Utilities.RemoveSmallRegion(imagePdf[i]));
//Get raw context in main body
List <TraversedLine> rawImageTexts = PdfExtractor.Utilities.GetContents(imagePdf[i]);
foreach (var line in rawImageTexts)
{
if (line.Text.IndexOf("sha-", StringComparison.OrdinalIgnoreCase) >= 0 || (line.Text.IndexOf("thumb", StringComparison.OrdinalIgnoreCase) >= 0))
{
line.Text = line.Text.Replace("O", "0").Replace("o", "0").Replace("i", "1").Replace("I", "1").Replace("l", "1");
}
}
//Merge raw context by paragraph
List <MergedTraversedLine> mergedImageTexts = PdfExtractor.Utilities.MergeTraversedLines(rawImageTexts);
titlePageEndIndexNum += mergedImageTexts.Count;
List <MergedTraversedLine> titleLines = mergedImageTexts.Where(x => x.IsTitle == true).ToList <MergedTraversedLine>();
//If this is a title page, add title page start index and end index to ExtractedAuditLetterText.TitlePageRanges
//If this is a title page, add the title index to ExtractedAuditLetterText.TitleRanges
if (titleLines.Count > 0)
{
extractedImageTexts.TitlePageRanges.Add(new KeyValuePair <int, int>(titlePageStartIndexNum, titlePageEndIndexNum));
foreach (MergedTraversedLine title in titleLines)
{
extractedImageTexts.TitleRanges.Add(new KeyValuePair <int, int>(title.Index + titlePageStartIndexNum, title.Index + titlePageStartIndexNum));
}
}
titlePageStartIndexNum += mergedImageTexts.Count;
//Add titles, contents
ExtractedAuditLetterText tempText = PdfExtractor.Utilities.ConvertToExtractedAuditLetterTexts(mergedImageTexts);
extractedImageTexts.Titles.AddRange(tempText.Titles);
extractedImageTexts.Contents.AddRange(tempText.Contents);
}
}
if (imageFormattedNum.Count == 0)
{
ExtractedAuditLetterText tempText = PdfExtractor.Utilities.ConvertToExtractedAuditLetterTexts(txtFormattedLines);
extractedTexts.Titles.AddRange(tempText.Titles);
extractedTexts.Contents.AddRange(tempText.Contents);
StringBuilder sb = new StringBuilder();
sb.Append(num + "Ocr: " + ts.TotalMilliseconds.ToString() + " ");
string serialzedAuditLetterText2 = extractedTexts.SerializeExtractedAuditLetterText();
extractedAuditLetterText = extractedTexts;
DateTime dt2 = System.DateTime.Now;
ts = dt2.Subtract(dt1);
sb.Append(" Reader: " + ts.TotalMilliseconds.ToString() + "\n");
string fileTime = @"C:\Users\t-holu\Documents\AuditLetter\JixiProjectData\comTextTime.txt";
File.AppendAllText(fileTime, sb.ToString());
return(serialzedAuditLetterText2);
}
else if (txtFormattedNum.Count == 0)
{
extractedTexts.Titles.AddRange(extractedImageTexts.Titles);
extractedTexts.Contents.AddRange(extractedImageTexts.Contents);
extractedTexts.TitleRanges.AddRange(extractedImageTexts.TitleRanges);
}
else
{
if (txtFormattedNum[0] < imageFormattedNum[0])
{
ExtractedAuditLetterText tempText = PdfExtractor.Utilities.ConvertToExtractedAuditLetterTexts(txtFormattedLines);
extractedTexts.Titles.AddRange(tempText.Titles);
extractedTexts.Contents.AddRange(tempText.Contents);
foreach (var line in extractedImageTexts.Titles)
{
extractedTexts.Titles.Add(new KeyValuePair <int, string>(line.Key, line.Value));
}
foreach (var line in extractedImageTexts.Contents)
{
extractedTexts.Contents.Add(new KeyValuePair <int, string>(line.Key, line.Value));
}
foreach (var line in extractedImageTexts.TitleRanges)
{
extractedTexts.TitleRanges.Add(new KeyValuePair <int, int>(line.Key, line.Value));
}
}
else
{
foreach (var line in extractedImageTexts.Titles)
{
extractedTexts.Titles.Add(new KeyValuePair <int, string>(line.Key, line.Value));
}
foreach (var line in extractedImageTexts.Contents)
{
extractedTexts.Contents.Add(new KeyValuePair <int, string>(line.Key, line.Value));
}
foreach (var line in extractedImageTexts.TitleRanges)
{
extractedTexts.TitleRanges.Add(new KeyValuePair <int, int>(line.Key, line.Value));
}
ExtractedAuditLetterText tempText = PdfExtractor.Utilities.ConvertToExtractedAuditLetterTexts(txtFormattedLines);
extractedTexts.Titles.AddRange(tempText.Titles);
extractedTexts.Contents.AddRange(tempText.Contents);
}
}
string serialzedAuditLetterText = extractedTexts.SerializeExtractedAuditLetterText();
extractedAuditLetterText = extractedTexts;
return(serialzedAuditLetterText);
}
