private static void ProcessWithPage(object st)
{
OCRPage p = ProcessWithPageAsync((InternalOCRProcessorStatus)st).Result;
return;
}
public void AddPage(OCRPage page)
{
_Pages.Add(page);
}
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);
}
}