/// <summary>
/// Used to initialize the routine asynchronously
/// </summary>
/// <param name="languageId">Language ID</param>
/// <returns>IEnumerator</returns>
private IEnumerator InitializeRoutine(string languageId)
{
// Copies all tess data to persistent directory
yield return CoroutineRunner.RunCoroutine(InitializeTessdataRoutine());
string[] languageSplit = languageId.Split('+');
// Check whether we have stored the correct tesseract language data
for(int i = 0; i < languageSplit.Length; i++)
{
if(!m_AvailableDataSet.Contains(languageSplit[i]))
{
Debug.LogWarning(string.Format("Tesseract language id is not available. Please check your settings for the correct data import. [{0}]", languageSplit[i]));
}
}
if(m_Handle != IntPtr.Zero)
{
// Perform initialization from the native library
if(TesseractOCRBridge.Initialize(m_Handle, GetTessdataPersistentPath(), languageId) == 0)
{
// Copies the configuration file to persistent directory and sets the config file to use
yield return CoroutineRunner.RunCoroutine(InitializeConfigFileRoutine(true));
Debug.Log("Initialize TesseractOCR Success!");
}
else
{
Debug.Log("Initialize TesseractOCR Failed!");
}
}
}
/// <summary>
/// Sets the page segmentation mode.
/// </summary>
/// <param name="mode">Segmentation Mode</param>
private void SetPageSegmentationMode(SegmentationMode mode)
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
TesseractOCRBridge.SetPageSegmentationMode(m_Handle, mode);
}
}
/// <summary>
/// Library cleanup
/// </summary>
public void Dispose()
{
if(m_Handle != IntPtr.Zero)
{
TesseractOCRBridge.EndTesseractHandle(m_Handle);
TesseractOCRBridge.DeleteTesseractHandle(m_Handle);
m_Handle = IntPtr.Zero;
}
}
/// <summary>
/// Clears tesseract engine's persistent cache
/// </summary>
private void ClearPersistentCache()
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
TesseractOCRBridge.ClearPersistentCache(m_Handle);
}
}
/// <summary>
/// Used to initialize Tesseract OCR library
/// </summary>
/// <param name="languageId">Language ID</param>
public void Initialize(string languageId)
{
if(m_Handle == IntPtr.Zero)
{
m_Handle = TesseractOCRBridge.CreateTesseractHandle();
}
CoroutineRunner.RunCoroutine(InitializeRoutine(languageId));
}
/// <summary>
/// Recognizes the image if the library can parse it
/// Monitor handle can be nulled
/// </summary>
/// <returns>0 = success, -1 = failed</returns>
private int Recognize()
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
return TesseractOCRBridge.Recognize(m_Handle, IntPtr.Zero);
}
return -1;
}
/// <summary>
/// Gets the page segmentation mode
/// </summary>
/// <returns>Segmentation Mode</returns>
private SegmentationMode GetPageSegmentationMode()
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
return TesseractOCRBridge.GetPageSegmentationMode(m_Handle);
}
return SegmentationMode.SINGLE_BLOCK;
}
/// <summary>
/// Gets the OCR Engine Mode
/// </summary>
/// <returns>OCR Engine Mode</returns>
private OCREngineMode GetOCREngineMode()
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
return TesseractOCRBridge.GetTesseractEngineMode(m_Handle);
}
return OCREngineMode.DEFAULT;
}
/// <summary>
/// Sets the image to be used
/// </summary>
/// <param name="imageData">Image data buffer</param>
/// <param name="width">Image Width</param>
/// <param name="height">Image Height</param>
public void SetImage(byte[] imageData, int width, int height)
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
Color32[] colorBuffer = TesseractOCRUtility.ImageFlipVertical(imageData, width, height);
imageData = TesseractOCRUtility.Color32ToBytes(colorBuffer);
TesseractOCRBridge.SetImageData(m_Handle, imageData, width, height, BYTES_PER_PIXEL, BYTES_PER_PIXEL * width);
}
}
/// <summary>
/// Gets the version of Tesseract OCR library
/// </summary>
/// <returns>TesseractOCR Version</returns>
public string GetVersion()
{
IntPtr versionPtr = TesseractOCRBridge.GetTesseractVersion();
Debug.Assert(versionPtr != IntPtr.Zero, "Version must not be nulled");
if(versionPtr != IntPtr.Zero)
{
return Marshal.PtrToStringAnsi(versionPtr);
}
return "";
}
/// <summary>
/// Sets the configuration file
/// </summary>
/// <param name="filename">Config Filename</param>
/// <param name="isDebug">Is Debug</param>
public void SetConfigurationFile(string filename, bool isDebug = false)
{
if(m_Handle != IntPtr.Zero)
{
if(!isDebug)
{
TesseractOCRBridge.SetConfigurationFile(m_Handle, filename);
}
else
{
TesseractOCRBridge.SetDebugConfigurationFile(m_Handle, filename);
}
}
}
/// <summary>
/// Gets a type double configuration variable
/// </summary>
/// <param name="name">Parameter Name</param>
/// <returns>Double Result</returns>
private double GetConfigDoubleVariable(string name)
{
double result = 0;
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
if(TesseractOCRBridge.GetVariable(m_Handle, name, out result))
{
return result;
}
}
return result;
}
/// <summary>
/// Gets a type boolean configuration variable
/// </summary>
/// <param name="name">Parameter Name</param>
/// <returns>Integer Result</returns>
private bool GetConfigBoolVariable(string name)
{
bool result = false;
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
if(TesseractOCRBridge.GetVariable(m_Handle, name, out result))
{
return result;
}
}
return result;
}
/// <summary>
/// Sets a configuration variable
/// </summary>
/// <param name="name">Parameter Name</param>
/// <param name="value">Parameter Value</param>
/// <param name="isDebug">Is Debug</param>
private void SetConfigVariable(string name, string value, bool isDebug = false)
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
if(!isDebug)
{
TesseractOCRBridge.SetVariable(m_Handle, name, value);
}
else
{
TesseractOCRBridge.SetDebugVariable(m_Handle, name, value);
}
}
}
/// <summary>
/// Gets a type string configuration variable
/// </summary>
/// <param name="name">Parameter Name</param>
/// <returns>String Result</returns>
private string GetConfigStringVariable(string name)
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
IntPtr stringPtr = TesseractOCRBridge.GetVariable(m_Handle, name);
Debug.Assert(stringPtr != IntPtr.Zero, "Handle must not be nulled!");
if(stringPtr != IntPtr.Zero)
{
return Marshal.PtrToStringAnsi(stringPtr);
}
}
return "";
}
/// <summary>
/// Gets the text from the image in UTF-8 format
/// </summary>
/// <returns>UTF-8 Text</returns>
public string GetText()
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
IntPtr textPtr = TesseractOCRBridge.GetTextData(m_Handle);
Debug.Assert(textPtr != IntPtr.Zero, "Text must not be nulled!");
if(textPtr != IntPtr.Zero)
{
return Marshal.PtrToStringAnsi(textPtr);
}
}
return "";
}
/// <summary>
/// Gets the language id used by the library
/// </summary>
/// <returns>Language ID</returns>
public string GetLanguageId()
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
IntPtr languagePtr = TesseractOCRBridge.GetTesseractLanguage(m_Handle);
Debug.Assert(languagePtr != IntPtr.Zero, "Datapath must not be nulled!");
if(languagePtr != IntPtr.Zero)
{
return Marshal.PtrToStringAnsi(languagePtr);
}
}
return "";
}
/// <summary>
/// Gets the datapath where the tessdata is stored
/// </summary>
/// <returns>Tess data directory path</returns>
public string GetDataPath()
{
Debug.Assert(m_Handle != IntPtr.Zero, "Handle must not be nulled!");
if(m_Handle != IntPtr.Zero)
{
IntPtr dataPathPtr = TesseractOCRBridge.GetTesseractDataPath(m_Handle);
Debug.Assert(dataPathPtr != IntPtr.Zero, "Datapath must not be nulled!");
if(dataPathPtr != IntPtr.Zero)
{
return Marshal.PtrToStringAnsi(dataPathPtr);
}
}
return "";
}
/// <summary>
/// Initializes tesseract's configuration list
/// </summary>
private void InitializeTesseractConfiguration()
{
IntPtr handle = TesseractOCRBridge.CreateTesseractHandle();
if(TesseractOCRBridge.Initialize(handle, Application.persistentDataPath + "/tessdata", "eng") == 0)
{
TesseractOCRBridge.PrintVariablesToFile(handle, "config");
}
TesseractOCRBridge.EndTesseractHandle(handle);
TesseractOCRBridge.DeleteMonitorHandle(handle);
string defaultConfigPath = Path.Combine(Directory.GetCurrentDirectory(), "config");
if(File.Exists(defaultConfigPath))
{
string[] lines = File.ReadAllLines(defaultConfigPath);
m_TesseractDefaultConfigInfo = new TesseractConfigInfo[lines.Length];
for(int i = 0; i < lines.Length; i++)
{
string[] split = lines[i].Split('\t');
m_TesseractDefaultConfigInfo[i] = new TesseractConfigInfo()
{
name = split[0],
value = split[1],
description = split[2]
};
}
IComparer<TesseractConfigInfo> comparer = new AscendingCompare();
Array.Sort<TesseractConfigInfo>(m_TesseractDefaultConfigInfo, comparer);
int defaultConfigLen = m_TesseractDefaultConfigInfo.Length;
m_TesseractConfigInfo = new TesseractConfigInfo[defaultConfigLen];
Array.Copy(m_TesseractDefaultConfigInfo, m_TesseractConfigInfo, defaultConfigLen);
File.Delete(defaultConfigPath);
}
string userDefinedConfigPath = "Assets/StreamingAssets/tessdata/configs";
if(Directory.Exists(userDefinedConfigPath))
{
string[] configs = Directory.GetFiles(userDefinedConfigPath, "*.", SearchOption.TopDirectoryOnly);
if(configs.Length == 1)
{
string[] lines = File.ReadAllLines(configs[0]);
for(int i = 0; i < lines.Length; i++)
{
string[] split = lines[i].Split('\t');
int index = Array.FindIndex<TesseractConfigInfo>(m_TesseractConfigInfo, a => a.name == split[0]);
if(index >= 0)
{
m_TesseractConfigInfo[index].value = split[1];
}
}
IComparer<TesseractConfigInfo> comparer = new AscendingCompare();
Array.Sort<TesseractConfigInfo>(m_TesseractDefaultConfigInfo, comparer);
}
}
}