/// <summary>
/// cSpell correction process, output to an ArrayList of TokenObj by using
/// funcMode and rankMode from configuratin files, with debug print option.
/// </summary>
/// <param name="inText"> input text to be corrected </param>
/// <param name="debugFlag"> boolean flag for debug print </param>
/// <returns> an ArrayList of TokenObj </returns>
public virtual List <TokenObj> ProcessToTokenObj(string inText, bool debugFlag)
{
DebugPrint.Println("====== SpellApi.Process( ), funcMode: " + funcMode_ + ", rankMode: " + rankMode_ + " ======", debugFlag);
// non-dictionary and dictionary base correction
List <TokenObj> inTokenList = TextObj.TextToTokenList(inText);
List <TokenObj> outTokenList = CorrectionApi.ProcessByTokenObj(inTokenList, this, debugFlag);
return(outTokenList);
}
public virtual void AddDictionaries(string inFilePaths, bool debugFlag)
{
DebugPrint.Println("- Dictionary Files: [" + inFilePaths + "].", debugFlag);
string[] inFileStrArray = inFilePaths.Split(":", true);
foreach (string inFileStr in inFileStrArray)
{
DebugPrint.Println("--- Add Dictionary: [" + inFileStr + "].", debugFlag);
AddDictionary(inFileStr);
}
}
public virtual void AddDictionaries2(string inFiles, bool debugFlag)
{
DebugPrint.Println("- Dictionary Files: [" + inFiles + "].", debugFlag);
string[] inFileStrArray = inFiles.Split("|", true);
foreach (var item in inFileStrArray)
{
DebugPrint.Println("--- Add Dictionary: [" + item + "].", debugFlag);
AddDictionary(item);
}
}
public virtual void AddDictionaries(string inFiles, string rootPath, bool debugFlag)
{
DebugPrint.Println("- Dictionary Files: [" + inFiles + "].", debugFlag);
string[] inFileStrArray = inFiles.Split(":", true);
foreach (string inFileStr in inFileStrArray)
{
string inDicFile = rootPath + inFileStr;
DebugPrint.Println("--- Add Dictionary: [" + inDicFile + "].", debugFlag);
AddDictionary(inDicFile);
}
}
private static bool CheckRealWord1To1Rules(ContextScore topContextScore, string inStr, int tarPos, int tarSize, List <TokenObj> nonSpaceTokenList, Word2Vec word2VecIm, Word2Vec word2VecOm, bool word2VecSkipWord, int contextRadius, double rw1To1Factor, bool debugFlag)
{
// return false if no topCand found
if ((topContextScore == null) || (topContextScore.GetTerm().Equals(inStr)))
{
return(false);
}
// 1. get the word2Vec score for the org inStr b4 one-to-one
// 1.1 wordVec for context
DoubleVec contextVec = Word2VecContext.GetContextVec(tarPos, tarSize, nonSpaceTokenList, word2VecIm, contextRadius, word2VecSkipWord, debugFlag);
// 1.2 wordVec for the original words before one-to-one
ContextScore orgCs = new ContextScore(inStr, contextVec, word2VecOm);
DebugPrint.Println("--- Real-Word One-To-One Context Score Detail: ---", debugFlag);
DebugPrint.Println("- Score - orgTerm: " + orgCs.ToString(), debugFlag);
DebugPrint.Println("- Score - top 1-to-1: " + topContextScore.ToString(), debugFlag);
DebugPrint.Println("- rw1To1Factor: " + rw1To1Factor, debugFlag);
// Score rules for one-to-one
double orgScore = orgCs.GetScore();
double topScore = topContextScore.GetScore();
bool flag = false;
// 2.1 no one-to-one correction if orgScore is 0.0d, no word2Vec information
if (orgScore < 0.0d)
{
// 2.2a one-to-one if the org score is negative and top score is positive
if (topScore > 0.0d)
{
// another rule for word2Vec on real-word
if (((topScore - orgScore) > 0.085) && (orgScore > -0.085)) // help from 0.6812 to 0.6877
{
flag = true;
}
}
// 2.2b one-to-one if the org score is negative and top score is better
else if ((topScore < 0.0d) && (topScore > orgScore * rw1To1Factor))
{
flag = true;
}
}
else if (orgScore > 0.0d)
{
// 2.3a merge if the org score is positive and better 0.01*topScore
if (topScore * rw1To1Factor > orgScore)
{
flag = true;
}
}
return(flag);
}
