public static async Task <OCRResult> GetTextFromStreamsAsync(Stream[] streams, OctarineEngine.IEngine engine, OCRStatus status)
{
var result = new OCRResult();
try {
foreach (Stream stream in streams)
{
(OCRPage page, OctarineError error, string errorMessage) = await engine.GetTextFromStreamAsync(stream);
if (page == null)
{
status.Error = error;
status.ErrorMessage = errorMessage;
return(null);
}
if (status.OCRCancellationToken.IsCancellationRequested)
{
status.Error = OctarineError.CancellationRequested;
return(null);
}
result.AddPage(page);
}
return(result);
} catch (Exception ex) {
status.Error = OctarineError.WrongFileFormat;
status.ErrorMessage = ex.Message;
return(null);
}
}
public void PrepareOCR(Stream[] streams, string name = "Plik")
{
if (updateWorker != null && !updateWorker.IsCompleted)
{
updateWorkerCTS.Cancel();
try {
updateWorker.Wait();
} catch {}
updateWorker = null;
}
LoadingWindow wnd_waiter = new LoadingWindow("Rozpoznawanie");
updateWorkerCTS = new CancellationTokenSource();
updateWorkerCT = updateWorkerCTS.Token;
updateWorker = Task.Factory.StartNew(() => {
wnd_waiter.SetStatus("Przygotowywanie");
OCRStatus status = new OCRStatus();
status.OCRCancellationToken = updateWorkerCT;
status.PageCurrentChanged += (sender, e) => {
wnd_waiter.SetStatus($"Rozpoznawanie strony {status.PageCurrent} z {status.PageCount}");
wnd_waiter.SetPercentage((int)(100 * status.PageCurrent / status.PageCount));
};
OCRResult result = OCR.GetTextFromStreamsAsync(streams, this.Engine, status).Result;
if (result != null)
{
wnd.SetResult(name, result);
}
else
if (status.Error != OctarineError.CancellationRequested)
{
wnd.ShowError(status.Error, status.ErrorMessage);
}
wnd_waiter.Close();
updateWorker = null;
}, updateWorkerCT);
wnd_waiter.ShowDialog(wnd);
if (updateWorker != null && !updateWorker.IsCompleted)
{
updateWorkerCTS.Cancel();
try {
updateWorker.Wait();
} catch {}
updateWorker = null;
}
}
public void SetResult(string file, OCRResult result)
{
this.Text = Path.GetFileName(file) + " - Octarine";
this.file = file;
edt_result.Text = result.Text;
}
public void PrepareOCR(string file)
{
if (updateWorker != null && !updateWorker.IsCompleted)
{
updateWorkerCTS.Cancel();
try {
updateWorker.Wait();
} catch {}
updateWorker = null;
}
int pageFirst = 0, pageLast = 0;
if (Path.GetExtension(file).ToLower() == ".pdf")
{
int cnt = (int)OCR.GetPDFPagesCount(file);
if (cnt > 1)
{
var wnd_range = new PageRangeWindow(cnt);
wnd_range.ShowDialog(wnd);
if (wnd_range.Cancelled)
{
return;
}
pageFirst = wnd_range.PageFirst;
pageLast = wnd_range.PageLast;
}
}
LoadingWindow wnd_waiter = new LoadingWindow("Rozpoznawanie");
updateWorkerCTS = new CancellationTokenSource();
updateWorkerCT = updateWorkerCTS.Token;
updateWorker = Task.Factory.StartNew(() => {
wnd_waiter.SetStatus("Przygotowywanie");
OCRStatus status = new OCRStatus();
status.OCRCancellationToken = updateWorkerCT;
status.PageCurrentChanged += (sender, e) => {
wnd_waiter.SetStatus($"Rozpoznawanie strony {status.PageCurrent} z {status.PageCount}");
int prc = (int)(100 * status.PageCurrent / status.PageCount);
if (prc > 100)
{
prc = 100;
}
wnd_waiter.SetPercentage(prc);
};
OCRResult result = OCR.GetTextFromFileAsync(file, this.Engine, status, pageFirst, pageLast).Result;
if (result != null)
{
wnd.SetResult(file, result);
}
else
if (status.Error != OctarineError.CancellationRequested)
{
wnd.ShowError(status.Error, status.ErrorMessage);
}
wnd_waiter.Close();
updateWorker = null;
}, updateWorkerCT);
wnd_waiter.ShowDialog(wnd);
if (updateWorker != null && !updateWorker.IsCompleted)
{
updateWorkerCTS.Cancel();
try {
updateWorker.Wait();
} catch {}
updateWorker = null;
}
}
public static async Task <OCRResult> GetTextFromFileAsync(string filePath, OctarineEngine.IEngine engine, OCRStatus status, int pageFirst = 0, int pageLast = 0)
{
var result = new OCRResult(filePath);
try {
var file = await StorageFile.GetFileFromPathAsync(filePath);
if (Path.GetExtension(filePath).ToLower() == ".pdf")
{
var pdfDoc = await PdfDocument.LoadFromFileAsync(file);
if (pageFirst <= 0)
{
pageFirst = 1;
}
if (pageLast <= 0 || pageLast > pdfDoc.PageCount)
{
pageLast = (int)pdfDoc.PageCount;
}
uint pagesToRecognize = (uint)(pageLast - pageFirst + 1);
status.PageCount = pagesToRecognize;
Stream[] streams = new Stream[pagesToRecognize];
int pagesRunning = 0;
int numThreads = Environment.ProcessorCount;
for (int p = pageFirst - 1; p < pageLast; p++)
{
while (pagesRunning > numThreads)
{
await Task.Delay(10);
}
using (PdfPage pdfPage = pdfDoc.GetPage((uint)p)) {
int i = (int)p - pageFirst + 1;
_ = Task.Run(async() => {
++pagesRunning;
var stream = new InMemoryRandomAccessStream();
await pdfPage.RenderToStreamAsync(stream);
streams[i] = stream.AsStream();
--pagesRunning;
});
}
await Task.Delay(10);
}
while (pagesRunning > 0)
{
await Task.Delay(200);
}
OCRPage[] pages = new OCRPage[status.PageCount];
for (int i = pageFirst - 1; i < pageLast; i++)
{
if (status.Error != OctarineError.Success)
{
break;
}
while (status.ActiveWorkers > numThreads)
{
await Task.Delay(200);
}
var st = new InternalOCRProcessorStatus(engine, status, streams[i - pageFirst + 1], i - pageFirst + 1, pages);
_ = Task.Run(() => ProcessWithPageAsync(st));
await Task.Delay(250);
}
if (status.Error != OctarineError.Success)
{
return(null);
}
while (status.ActiveWorkers > 0)
{
await Task.Delay(200);
}
if (status.Error != OctarineError.Success)
{
return(null);
}
foreach (OCRPage page in pages)
{
result.AddPage(page);
}
return(result);
}
else
{
var stream = await file.OpenAsync(FileAccessMode.Read);
(OCRPage page, OctarineError error, string errorMessage) = await engine.GetTextFromStreamAsync(stream.AsStream());
if (page == null)
{
status.Error = error;
status.ErrorMessage = errorMessage;
return(null);
}
if (status.OCRCancellationToken.IsCancellationRequested)
{
status.Error = OctarineError.CancellationRequested;
return(null);
}
result.AddPage(page);
return(result);
}
} catch (Exception ex) {
status.Error = OctarineError.WrongFileFormat;
status.ErrorMessage = ex.Message;
return(null);
}
}
