public void LoadMainResource(string rpLanguage)
{
if (!InstalledLanguages.Any(r => r.Directory == rpLanguage))
{
Preference.Instance.Language.Value = GetDefaultLanguage().Directory;
}
LoadMainResourceCore(rpLanguage);
}
public bool CanProcess(string langCode)
{
if (string.IsNullOrEmpty(langCode) || !IsInstalled || !IsSupported)
{
return(false);
}
// Support multiple specified languages (e.g. "eng+fra")
return(langCode.Split('+').All(code => InstalledLanguages.Any(x => x.Code == code)));
}
public override async Task InitializeAsync()
{
await base.InitializeAsync();
InstalledLanguages.SyncWith(
InteractiveInkServices.LanguageService.Languages.Where(x =>
LanguageService.IsInstalled(x) || LanguageService.IsInstalling(x)));
InstalledLanguages.SortBy(x => x.Id);
InteractiveInkServices.LanguageService.Downloading += OnDownloading;
InteractiveInkServices.LanguageService.Initialized += OnInitialized;
InteractiveInkServices.LanguageService.Initializing += OnInitializing;
}
public OcrResult ProcessImage(string imagePath, OcrParams ocrParams, CancellationToken cancelToken)
{
string tempHocrFilePath = Path.Combine(Paths.Temp, Path.GetRandomFileName());
string tempHocrFilePathWithExt = tempHocrFilePath + TesseractHocrExtension;
try
{
var runInfo = TesseractRunInfo(ocrParams);
var startInfo = new ProcessStartInfo
{
FileName = Path.Combine(TesseractBasePath, TesseractExePath),
Arguments = $"\"{imagePath}\" \"{tempHocrFilePath}\" -l {ocrParams.LanguageCode} {runInfo.Arguments} hocr",
UseShellExecute = false,
CreateNoWindow = true,
RedirectStandardOutput = true,
RedirectStandardError = true
};
if (runInfo.PrefixPath != null)
{
startInfo.EnvironmentVariables["TESSDATA_PREFIX"] = Path.Combine(TesseractBasePath, runInfo.PrefixPath);
}
if (runInfo.DataPath != null)
{
var tessdata = new DirectoryInfo(Path.Combine(TesseractBasePath, runInfo.DataPath));
EnsureHocrConfigExists(tessdata);
}
var tesseractProcess = Process.Start(startInfo);
if (tesseractProcess == null)
{
// Couldn't start tesseract for some reason
Log.Error("Couldn't start OCR process.");
return(null);
}
var timeout = (int)(appConfigManager.Config.OcrTimeoutInSeconds * 1000);
if (timeout == 0)
{
timeout = DEFAULT_TIMEOUT;
}
var stopwatch = Stopwatch.StartNew();
while (!tesseractProcess.WaitForExit(CHECK_INTERVAL))
{
if (stopwatch.ElapsedMilliseconds >= timeout || cancelToken.IsCancellationRequested)
{
if (stopwatch.ElapsedMilliseconds >= timeout)
{
Log.Error("OCR process timed out.");
}
try
{
tesseractProcess.Kill();
// Wait a bit to give the process time to release its file handles
Thread.Sleep(200);
}
catch (Exception e)
{
Log.ErrorException("Error killing OCR process", e);
}
return(null);
}
}
#if DEBUG && DEBUGTESS
Debug.WriteLine("Tesseract stopwatch: " + stopwatch.ElapsedMilliseconds);
var output = tesseractProcess.StandardOutput.ReadToEnd();
if (output.Length > 0)
{
Log.Error("Tesseract stdout: {0}", output);
}
output = tesseractProcess.StandardError.ReadToEnd();
if (output.Length > 0)
{
Log.Error("Tesseract stderr: {0}", output);
}
#endif
XDocument hocrDocument = XDocument.Load(tempHocrFilePathWithExt);
return(new OcrResult
{
PageBounds = hocrDocument.Descendants()
.Where(x => x.Attributes("class").Any(y => y.Value == "ocr_page"))
.Select(x => GetBounds(x.Attribute("title")))
.First(),
Elements = hocrDocument.Descendants()
.Where(x => x.Attributes("class").Any(y => y.Value == "ocrx_word"))
.Select(x => new OcrResultElement {
Text = x.Value, Bounds = GetBounds(x.Attribute("title"))
}),
RightToLeft = InstalledLanguages.Where(x => x.Code == ocrParams.LanguageCode).Select(x => x.RTL).FirstOrDefault()
});
}
catch (Exception e)
{
Log.ErrorException("Error running OCR", e);
return(null);
}
finally
{
try
{
File.Delete(tempHocrFilePathWithExt);
}
catch (Exception e)
{
Log.ErrorException("Error cleaning up OCR temp files", e);
}
}
}
