public DeepExtractionStrategy(ref PageMetrics pageMetrics)
{
this.pageMetrics = pageMetrics;
charMetrices = new List <CharMetrics>();
startOfNewline = true;
descent = pageMetrics.Height;
}
public void AddPage(PageMetrics page)
{
pages.Add(page);
nbPages++;
if (page.Width > pageMaxWidth)
{
pageMaxWidth = page.Width;
}
if (page.Height > pageMaxHeight)
{
pageMaxHeight = page.Height;
}
}
public static void ExtractPdf(string input, string output)
{
PdfDocument pdfDoc = new PdfDocument(
new PdfReader(input), new PdfWriter(output));
EncodingProvider codePages = CodePagesEncodingProvider.Instance;
Encoding.RegisterProvider(codePages);
pdfDoc.GetNumberOfPages();
book = new BookMetrics();
book.Author = pdfDoc.GetDocumentInfo().GetAuthor();
book.Title = pdfDoc.GetDocumentInfo().GetTitle();
book.Publisher = pdfDoc.GetDocumentInfo().GetProducer();
int nbPages = pdfDoc.GetNumberOfPages();
for (int i = 0; i < nbPages; i++)
{
Rectangle size = pdfDoc.GetPage(i + 1).GetPageSize();
PageMetrics page = new PageMetrics(i + 1);
page.Width = size.GetWidth();
page.Height = size.GetHeight();
page.Rotation = pdfDoc.GetPage(i + 1).GetRotation();
DeepExtractionStrategy strategy = new DeepExtractionStrategy(ref page);
Console.WriteLine("Processing page {0}", i + 1);
PdfTextExtractor.GetTextFromPage(pdfDoc.GetPage(i + 1), strategy);
page.SortLines();
page.WordMetrices = WordMetrics.FromChars(strategy.CharMetrices);
page.CharMetrices = strategy.CharMetrices;
book.AddPage(page);
}
pdfDoc.Close();
}
