public static OcrResult Load(OcrImage image, string language, OcrEngineType targetType)
{
try
{
string ip = null;
foreach (KeyValuePair <string, Dictionary <OcrEngineType, List <string> > > remoteType in RemoteTypes)
{
if (remoteType.Value.ContainsKey(targetType))
{
ip = remoteType.Key;
break;
}
}
if (string.IsNullOrWhiteSpace(ip))
{
return(OcrResult.Create(OcrResultType.NotInstalled));
}
using (WebClient webClient = new WebClient())
{
string url = "http://" + ip + ":" + port + "/" + RunOcrCommand;
webClient.Proxy = null;
webClient.Encoding = Encoding.UTF8;
string response = webClient.UploadString(url, string.Format("type={0}&language={1}&data={2}",
targetType.ToString(), language, HttpUtility.UrlEncode(Convert.ToBase64String(image.ToBinary()))));
return(OcrResult.FromBinary(Convert.FromBase64String(HttpUtility.UrlDecode(response))));
}
}
catch (Exception e)
{
return(OcrResult.Create(OcrResultType.Exception, e.ToString()));
}
}
public OcrResult LoadFile(string file, string language)
{
try
{
if (!File.Exists(file))
{
return(OcrResult.Create(OcrResultType.InvalidFilePath));
}
OcrImage image = new OcrImage();
try
{
using (Image bitmap = Bitmap.FromFile(file))
{
image.Width = bitmap.Width;
image.Height = bitmap.Height;
}
image.Data = File.ReadAllBytes(file);
// Path is mostly ignored, remove it for now to stop it being sent over the network
//image.Path = file;
}
catch (Exception e)
{
return(OcrResult.Create(OcrResultType.InvalidFile, e.ToString()));
}
return(Load(image, language));
}
catch (Exception e)
{
return(OcrResult.Create(OcrResultType.Exception, e.ToString()));
}
}
public override OcrResult Load(OcrImage image, string language, string apiLanguage)
{
string inputPath = null;
string outputPath = null;
try
{
inputPath = GetTempPath();
outputPath = GetTempPath(false);
string spacing = AllowsSpacing(language) ? string.Empty : "--nospacing";
string newLines = AllowsNewLines(language) ? string.Empty : "--nolines";
File.WriteAllBytes(inputPath, image.Data);
RunCommand("\"{0}\" \"{1}\" \"{2}\" {3} {4}", apiLanguage, inputPath, outputPath, spacing, newLines);
if (File.Exists(outputPath))
{
string[] lines = File.ReadAllLines(outputPath);
OcrResult result = new OcrResult();
result.ResultType = OcrResultType.Success;
foreach (string line in lines)
{
result.AddLine(line);
}
return(result);
}
else
{
return(OcrResult.Create(OcrResultType.Failed));
}
}
catch (Exception e)
{
return(OcrResult.Create(OcrResultType.Exception, e.ToString()));
}
finally
{
try
{
if (!string.IsNullOrEmpty(inputPath))
{
File.Delete(inputPath);
}
}
catch { }
try
{
if (!string.IsNullOrEmpty(outputPath))
{
File.Delete(outputPath);
}
}
catch { }
}
}
public static void LoadFileAsync <T>(string file, string language, Action <OcrResult> callback) where T : OcrEngine, new()
{
Task.Factory.StartNew(() =>
{
OcrResult ocrResult = LoadFile <T>(file, language);
if (callback != null)
{
callback(ocrResult);
}
});
}
public void LoadFileAsync(string file, string language, Action <OcrResult> callback)
{
Task.Factory.StartNew(() =>
{
OcrResult ocrResult = LoadFile(file, language);
if (callback != null)
{
callback(ocrResult);
}
});
}
public OcrResult Load(OcrImage image, string language)
{
try
{
OcrResult result = null;
if (GetType() != typeof(RemoteOcr) && RemoteOcr.IsRemote(GetType()))
{
OcrEngineType targetType = GetType(GetType());
RemoteOcr remoteOcr = new RemoteOcr(targetType);
result = remoteOcr.Load(image, language, language);
}
else if (!IsInstalled)
{
return(OcrResult.Create(OcrResultType.NotInstalled));
}
else
{
string apiLanguage = language == null ? null : GetSupportedLanguageName(language);
if (string.IsNullOrEmpty(apiLanguage))
{
return(OcrResult.Create(OcrResultType.LanguageNotSupported));
}
result = Load(image, language, apiLanguage);
}
if (result != null && result.Rect == OcrRect.Empty)
{
result.Rect = new OcrRect(0, 0, image.Width, image.Height);
}
return(result);
}
catch (Exception e)
{
return(OcrResult.Create(OcrResultType.Exception, e.ToString()));
}
}
private static void RunGoogleOcrTests()
{
while (true)
{
OcrEngine abbyy = OcrEngine.Create(OcrEngineType.Google);//OcrEngineType.ABBYY);
MessageBox.Show("Installed: " + abbyy.IsInstalled);
Stopwatch sw1 = new Stopwatch();
sw1.Start();
if (abbyy.IsInstalled)
{
OcrResult result = abbyy.LoadFile("image1.png", "ko");
if (result != null)
{
sw1.Stop();
if (!string.IsNullOrEmpty(result.Text))
{
Clipboard.SetText(result.Text);
}
MessageBox.Show(result.ResultType + Environment.NewLine + result.Text + " " + sw1.Elapsed);
}
}
}
}
private static void RunNicomsoftOcrTests()
{
while (true)
{
OcrEngine engine = OcrEngine.Create(OcrEngineType.Nicomsoft);
MessageBox.Show("Installed: " + engine.IsInstalled);
Stopwatch sw1 = new Stopwatch();
sw1.Start();
if (engine.IsInstalled)
{
OcrResult result = engine.LoadFile("image1.png", "ko");
if (result != null)
{
sw1.Stop();
if (!string.IsNullOrEmpty(result.Text))
{
Clipboard.SetText(result.Text);
}
MessageBox.Show(result.ResultType + Environment.NewLine + result.Text + " " + result.Error + " " + sw1.Elapsed);
}
}
}
}
public static OcrResult FromBinary(byte[] buffer)
{
try
{
using (BinaryReader reader = new BinaryReader(new MemoryStream(buffer)))
{
OcrResult result = new OcrResult();
result.Rect = ReadRect(reader);
result.ResultType = (OcrResultType)reader.ReadByte();
result.Error = reader.ReadString();
if (string.IsNullOrEmpty(result.Error))
{
result.Error = null;
}
int areaCount = reader.ReadInt32();
for (int i = 0; i < areaCount; i++)
{
result.AddArea(ReadArea(reader));
}
int lineCount = reader.ReadInt32();
for (int i = 0; i < lineCount; i++)
{
result.AddLine(ReadLine(reader));
}
return(result);
}
}
catch (Exception e)
{
return(OcrResult.Create(OcrResultType.Exception, e.ToString()));
}
}
public override OcrResult Load(OcrImage image, string language, string apiLanguage)
{
try
{
if (!SaveLoadableImage(image))
{
return(OcrResult.Create(OcrResultType.InvalidFilePath));
}
Program.ClipSync.Text = null;
if (automation.Run())
{
string text = WaitForClipSync();
if (!string.IsNullOrEmpty(text))
{
return(new OcrResult(text));
}
return(OcrResult.Create(OcrResultType.AutomationFailed));
}
else
{
//System.Windows.Forms.MessageBox.Show("Failed at index: " + automation.Index);
}
return(OcrResult.Create(OcrResultType.AutomationFailed));
}
finally
{
for (int i = 0; i < 10; i++)
{
automation.SendKey(Keys.Escape);
System.Threading.Thread.Sleep(10);
}
}
}
private void Process(HttpListenerContext context)
{
try
{
string url = context.Request.Url.OriginalString;
string response = string.Empty;
if (url.Contains("favicon.ico"))
{
context.Response.StatusCode = (int)HttpStatusCode.InternalServerError;
context.Response.OutputStream.Close();
return;
}
if (url.Contains(GetEnginesCommand))
{
if (Program.Engines != null)
{
Dictionary <OcrEngineType, OcrEngine> engines = new Dictionary <OcrEngineType, OcrEngine>(Program.Engines);
foreach (OcrEngine engine in engines.Values)
{
if (engine.IsInstalled)
{
string languages = string.Empty;
foreach (string language in engine.SupportedLanguages.Keys)
{
languages += language + ",";
}
languages = languages.TrimEnd(',');
response += engine.Name + "," + languages + "|";
}
}
response = response.TrimEnd('|', ',');
}
}
else if (url.Contains(RunOcrCommand))
{
Dictionary <string, string> postData = GetPostData(context);
OcrEngineType type = (OcrEngineType)Enum.Parse(typeof(OcrEngineType), postData["type"]);
string language = postData["language"];
byte[] data = Convert.FromBase64String(postData["data"]);
OcrImage image = OcrImage.FromBinary(data);
OcrResult result = null;
try
{
result = OcrEngine.Create(type).Load(image, language);
}
catch (Exception e)
{
result = OcrResult.Create(OcrResultType.Exception, e.ToString());
}
try
{
response = HttpUtility.UrlEncode(Convert.ToBase64String(result.ToBinary()));
}
catch
{
}
}
byte[] resposeBuffer = Encoding.UTF8.GetBytes(response.ToString());
context.Response.ContentType = "text/html";
context.Response.ContentEncoding = Encoding.UTF8;
context.Response.ContentLength64 = resposeBuffer.Length;
context.Response.AddHeader("Date", DateTime.Now.ToString("r"));
context.Response.AddHeader("Last-Modified", DateTime.Now.ToString("r"));
context.Response.OutputStream.Write(resposeBuffer, 0, resposeBuffer.Length);
context.Response.OutputStream.Flush();
context.Response.StatusCode = (int)HttpStatusCode.OK;
}
catch
{
}
context.Response.StatusCode = (int)HttpStatusCode.InternalServerError;
context.Response.OutputStream.Close();
}
public override OcrResult Load(OcrImage image, string language, string apiLanguage)
{
string inputPath = null;
string outputPath = null;
string outputPath2 = null;
string outputPath3 = null;
try
{
inputPath = GetTempPath();
outputPath = GetTempPath(false);
outputPath2 = outputPath + (IncludeTextRegions ? ".hocr" : ".txt");
outputPath3 = outputPath + (IncludeTextRegions ? ".txt" : ".hocr");
File.WriteAllBytes(inputPath, image.Data);
RunCommand("\"{0}\" \"{1}\" -l {2} {3}", inputPath, outputPath, apiLanguage, IncludeTextRegions ? "-c tessedit_create_hocr=1" : string.Empty);
OcrResult result = null;
if (IncludeTextRegions)
{
result = OcrXml.FromFile(outputPath2);
}
else
{
string[] lines = File.ReadAllLines(outputPath2);
result = new OcrResult();
result.ResultType = OcrResultType.Success;
foreach (string line in lines)
{
result.AddLine(line);
}
}
return(result);
}
catch (Exception e)
{
return(OcrResult.Create(OcrResultType.Exception, e.ToString()));
}
finally
{
try
{
if (!string.IsNullOrEmpty(inputPath))
{
File.Delete(inputPath);
}
}
catch { }
try
{
if (!string.IsNullOrEmpty(outputPath))
{
File.Delete(outputPath);
}
}
catch { }
try
{
if (!string.IsNullOrEmpty(outputPath2))
{
File.Delete(outputPath2);
}
}
catch { }
try
{
if (!string.IsNullOrEmpty(outputPath3))
{
File.Delete(outputPath3);
}
}
catch { }
}
}
static OcrHelper()
{
thread = new Thread(delegate()
{
byte[] image = null;
OcrResult result = null;
while (Program.Running)
{
image = null;
result = null;
lock (locker)
{
image = currentImage;
}
if (image != null)
{
string path = GetTempPath(".png");
try
{
if (Start != null)
{
Start();
}
}
catch
{
}
try
{
File.WriteAllBytes(path, currentImage);
result = Program.ActiveOcrEngine.LoadFile(path, Program.ActiveLanguageFrom);
}
finally
{
try
{
File.Delete(path);
}
catch
{
}
}
}
if (image != null && result != null && Complete != null)
{
Complete(result);
}
lock (locker)
{
if (image != null)
{
currentImage = nextImage;
nextImage = null;
}
}
Thread.Sleep(idleWait);
}
});
thread.SetApartmentState(ApartmentState.STA);
thread.Start();
}
public static OcrResult Parse(string xml)
{
Dictionary <string, string> meta = new Dictionary <string, string>();
List <Page> pages = new List <Page>();
XmlReaderSettings settings = new XmlReaderSettings();
settings.DtdProcessing = DtdProcessing.Parse;
settings.XmlResolver = new XmlPreloadedResolver(XmlKnownDtds.Xhtml10);
using (StringReader stringReader = new StringReader(xml))
using (XmlReader reader = XmlReader.Create(stringReader, settings))
{
while (reader.Read())
{
switch (reader.NodeType)
{
case XmlNodeType.Element:
switch (reader.Name)
{
case "meta":
{
string metaName = reader.GetAttribute("name");
string metaContent = reader.GetAttribute("content");
string metaValue = reader.GetAttribute("value");
if (!string.IsNullOrEmpty(metaName))
{
meta[metaName] = metaContent == null ? metaValue : metaContent;
}
}
break;
case "div":
case "span":
case "p":
{
switch (reader.GetAttribute("class"))
{
case "ocr_page":
pages.Add(Parse <Page>(reader, null));
break;
}
}
break;
}
break;
case XmlNodeType.EndElement:
break;
}
}
}
OcrResult result = new OcrResult();
foreach (Page page in pages)
{
foreach (Area area in page.Areas)
{
OcrArea ocrArea = new OcrArea(area.Rect.ToRect());
List <Line> lines = new List <Line>();
lines.AddRange(area.Lines);
foreach (Paragraph paragraph in area.Paragraphs)
{
lines.AddRange(paragraph.Lines);
}
foreach (Line line in lines)
{
OcrLine ocrLine = new OcrLine();
ocrLine.Rect = line.Rect.ToRect();
foreach (Word word in line.Words)
{
OcrWord ocrWord = new OcrWord();
ocrWord.Rect = word.Rect.ToRect();
ocrWord.Confidence = word.Confidence;
ocrWord.Text = word.Text;
ocrLine.Words.Add(ocrWord);
}
ocrArea.Lines.Add(ocrLine);
}
if (!string.IsNullOrEmpty(ocrArea.Text.Trim()))
{
result.AddArea(ocrArea);
}
}
}
return(result);
}
public override OcrResult Load(OcrImage image, string language, string apiLanguage)
{
try
{
Automation automation = new Automation();
automation.UseImageCaching = true;
automation.Path = System.IO.Path.Combine("Images", "ABBYY");
// Task icon
automation.AddReferenceImage("AbbyyReference.png");
// Task icon
automation.AddStartImage("AbbyyStart.png");
// Click on language text box
automation.AddRelativeClick(450, 50);
//automation.AddWait(50);
// Select language text box text
automation.AddSelectAll();
automation.AddWait(50);
automation.AddText(apiLanguage, true);
automation.AddWait(50);
automation.AddKey(Keys.Enter);
automation.AddWait(50);
automation.AddKey(Keys.Enter);
// Click open image button
automation.AddRelativeClick(140, 30);
// Wait for the open image dialog
automation.AddWaitImage("OpenImage.png");
// Type the file name
automation.AddOpenImage();
// Wait for the complete dialog
automation.AddWaitImage("Complete.png");
// Press escape to close complete dialog
automation.AddKey(Keys.Escape);
// Seems to freeze if we don't do this
for (int i = 0; i < 15; i++)
{
// Click the ocr text box
automation.AddMoveMouseImageOffset("OutputOffset.png", 50 + ((i + 1) * 1), 0 + ((i + 1) * 1));
automation.AddWait(100);
}
automation.AddClickableImageOffset("OutputOffset.png", 50, 0);
//automation.AddWait(50);
automation.AddSelectAll();
//automation.AddWait(50);
automation.AddCopy();
if (!SaveLoadableImage(image))
{
return(OcrResult.Create(OcrResultType.InvalidFilePath));
}
// Clear the clipboard
Clipboard.Clear();
if (automation.Run())
{
string text = WaitForClipboard();
if (!string.IsNullOrEmpty(text))
{
return(new OcrResult(text));
}
return(OcrResult.Create(OcrResultType.AutomationFailed));
}
else
{
//System.Windows.Forms.MessageBox.Show("Failed at index: " + automation.Index);
}
}
catch (Exception e)
{
//MessageBox.Show(e.ToString());
return(OcrResult.Create(OcrResultType.Exception, e.ToString()));
}
return(OcrResult.Create(OcrResultType.AutomationFailed));
}