private static long TimeCutPage(string fileName)
{
SautinSoft.PdfFocus f = new PdfFocus();
long timeFor1 = 0;
long timeFor2 = 0;
f.OpenPdf(fileName);
f.ImageOptions.Dpi = 320;
pages = f.PageCount;
sw.Start();
f.ToImage(prefixFilePath, 1);
sw.Stop();
timeFor1 = sw.ElapsedMilliseconds;
sw.Start();
f.ToImage(prefixFilePath, 2);
sw.Stop();
timeFor2 = sw.ElapsedMilliseconds;
long allTime = timeFor1 + ((timeFor2 - timeFor1) * (f.PageCount - 1));
f.ClosePdf();
timeForJpeg = allTime;
sw.Reset();
return(timeForJpeg);
}
private string OpenPDF(string path) //метод конвертирования PDF
{
//Создания экземпляра класса для конвертации pdf в jpg
string txtFile = DirectoryTemporary + "textOCR.txt";
PdfFocus f = new PdfFocus();
f.OpenPdf(path);
Console.WriteLine($"Колличество страниц: {f.PageCount}");
if (f.PageCount > 0)
{
f.ImageOptions.ImageFormat = System.Drawing.Imaging.ImageFormat.Jpeg;
f.ImageOptions.Dpi = 300;
//Деление каждой страницы pdf на изображения
f.ToImage(DirectoryTemporary, "page");
}
using (FileStream fileStream = File.Open(txtFile, FileMode.Create))
{
using (StreamWriter strWrite = new StreamWriter(fileStream))
{
for (int i = 1; i < f.PageCount + 1; i++)
{
string pathPage = DirectoryTemporary + $"page{i}.jpg";
try
{
using (TesseractEngine engine = new TesseractEngine(@"tessdata", "rus", EngineMode.Default))
{
using (var img = Pix.LoadFromFile(pathPage))
{
using (var page = engine.Process(img))
{
Console.WriteLine("Качество скана: {0}", page.GetMeanConfidence());
var text = page.GetText();
strWrite.WriteLine($"{text}");
}
}
}
}
catch (Exception ex)
{
Console.WriteLine("Error: {0}", ex.Message);
}
}
}
}
return(txtFile);
}
/// <summary>
/// 将PDF文档转换为图片的方法
/// </summary>
/// <param name="pdfInputPath">PDF文件路径</param>
/// <param name="imageOutputPath">图片输出路径</param>
/// <param name="imageName">生成图片的名字</param>
/// <param name="startPageNum">从PDF文档的第几页开始转换</param>
/// <param name="endPageNum">从PDF文档的第几页开始停止转换</param>
/// <param name="imageFormat">设置所需图片格式</param>
/// <param name="definition">设置图片的清晰度，数字越大越清晰</param>
public static void ConvertPDF2Image(string pdfInputPath, string imageOutputPath,
string imageName, int startPageNum, int endPageNum, ImageFormat imageFormat, Definition definition)
{
PdfFocus pdfFocus = new PdfFocus();
pdfFocus.OpenPdf(pdfInputPath);
if (!Directory.Exists(imageOutputPath))
{
Directory.CreateDirectory(imageOutputPath);
}
// validate pageNum
if (startPageNum <= 0)
{
startPageNum = 1;
}
if (endPageNum > pdfFocus.PageCount)
{
endPageNum = pdfFocus.PageCount;
}
if (startPageNum > endPageNum)
{
int tempPageNum = startPageNum;
startPageNum = endPageNum;
endPageNum = startPageNum;
}
// start to convert each page
for (int i = startPageNum; i <= endPageNum; i++)
{
byte[] img = pdfFocus.ToImage(i);
using (FileStream fs1 = File.Create(imageOutputPath + imageName + i.ToString() + "." + imageFormat.ToString()))
{
fs1.Write(img, 0, img.Length);
}
}
pdfFocus.ClosePdf();
}
public static void ConvertToImg(string path, ref ProgressBar progressBar, ref double time, ref Label label)
{
//Stopwatch sw = new Stopwatch();
// sw.Start();
SautinSoft.PdfFocus f = new PdfFocus();
f.OpenPdf(path);
int pageCount = f.PageCount;
string prefixFilePath = "PDF\\";
f.ImageOptions.Dpi = 320;
Stopwatch sw = new Stopwatch();
Cutter.setMaximumAndStep(progressBar, f.PageCount * 2);
for (int i = 1; i <= f.PageCount; i++)
{
sw.Start();
f.ToImage(prefixFilePath + i.ToString() + ".jpeg", i);
sw.Stop();
time -= sw.ElapsedMilliseconds;
TimeCalc.MinuteSeconds(time, label);
sw.Reset();
}
f.ClosePdf();
for (int i = 1; i <= pageCount; i++)
{
sw.Start();
Cutter.Cut(prefixFilePath + i.ToString() + ".jpeg", i, progressBar);
sw.Stop();
time -= sw.ElapsedMilliseconds;
TimeCalc.MinuteSeconds(time, label);
sw.Reset();
}
}