/// <summary>
/// Initializes the <see cref="PinglishConverter"/> class.
/// </summary>
public PinglishConverter()
{
#region Preprocess Replacements
//TODO: It's better to load them from a file --> In progress -> DONE
PreprocessElementInfo entryInfo = null;
///////////////////////////////////////////////////////////////
entryInfo = new PreprocessElementInfo("U", true);
entryInfo.Equivalents.Add("too");
this.m_preprocessReplacements.Add(entryInfo.PinglishString, entryInfo);
///////////////////////////////////////////////////////////////
entryInfo = new PreprocessElementInfo("I", true);
entryInfo.Equivalents.Add("man");
this.m_preprocessReplacements.Add(entryInfo.PinglishString, entryInfo);
/////////////////////////////////////////////////////////////////
entryInfo = new PreprocessElementInfo(QuotationMark.RightSingleQuotationMark.ToString(), false);
entryInfo.Equivalents.Add(QuotationMark.SingleQuotationMark.ToString());
this.m_preprocessReplacements.Add(entryInfo.PinglishString, entryInfo);
/////////////////////////////////////////////////////////////////
entryInfo = new PreprocessElementInfo(QuotationMark.Prime.ToString(), false);
entryInfo.Equivalents.Add(QuotationMark.SingleQuotationMark.ToString());
this.m_preprocessReplacements.Add(entryInfo.PinglishString, entryInfo);
/////////////////////////////////////////////////////////////////
entryInfo = new PreprocessElementInfo(QuotationMark.SingleHighReveresed9QuotationMark.ToString(), false);
entryInfo.Equivalents.Add(QuotationMark.SingleQuotationMark.ToString());
this.m_preprocessReplacements.Add(entryInfo.PinglishString, entryInfo);
/////////////////////////////////////////////////////////////////
entryInfo = new PreprocessElementInfo("k", true);
entryInfo.Equivalents.Add("kei");
this.m_preprocessReplacements.Add(entryInfo.PinglishString, entryInfo);
/////////////////////////////////////////////////////////////////
#region m_bigLettersAtBeginOrEnd
this.m_bigLettersAtBeginOrEnd.Add('B', new[] { "b", "bi" });
this.m_bigLettersAtBeginOrEnd.Add('C', new[] { "c", "si" });
this.m_bigLettersAtBeginOrEnd.Add('D', new[] { "d", "di" });
this.m_bigLettersAtBeginOrEnd.Add('E', new[] { "e", "ee" });
this.m_bigLettersAtBeginOrEnd.Add('F', new[] { "f", "ef" });
this.m_bigLettersAtBeginOrEnd.Add('G', new[] { "g", "ji" });
this.m_bigLettersAtBeginOrEnd.Add('K', new [] { "k", "kei" });
this.m_bigLettersAtBeginOrEnd.Add('M', new [] { "m", "em" });
this.m_bigLettersAtBeginOrEnd.Add('N', new [] { "n", "en" });
this.m_bigLettersAtBeginOrEnd.Add('P', new [] { "p", "pi" });
this.m_bigLettersAtBeginOrEnd.Add('Q', new [] { "q", "kioo" });
this.m_bigLettersAtBeginOrEnd.Add('S', new [] { "s", "es" });
this.m_bigLettersAtBeginOrEnd.Add('T', new [] { "t", "ti" });
this.m_bigLettersAtBeginOrEnd.Add('U', new [] { "u", "yoo" });
this.m_bigLettersAtBeginOrEnd.Add('V', new [] { "v", "vi", });
this.m_bigLettersAtBeginOrEnd.Add('Y', new [] { "y", "vay" });
this.m_bigLettersAtBeginOrEnd.Add('Z', new [] { "z", "zed" });
#endregion
#region m_oneLetterWords
m_oneLetterWords.Add('4', new string[] { "baraye" });
m_oneLetterWords.Add('K', new string[] { "kei" });
m_oneLetterWords.Add('U', new string[] { "to" });
/*m_oneLetterWords.Add('', new string[] { "" });
* m_oneLetterWords.Add('', new string[] { "" });
* m_oneLetterWords.Add('', new string[] { "" });
* m_oneLetterWords.Add('', new string[] { "" });
* m_oneLetterWords.Add('', new string[] { "" });
* m_oneLetterWords.Add('', new string[] { "" });
* m_oneLetterWords.Add('', new string[] { "" });*/
foreach (char key in m_oneLetterWords.Keys)
{
Debug.Assert(StringUtil.OneLetterPinglishWords.Contains(char.ToLower(key)));
}
#endregion
#endregion
}
/// <summary>
/// Loads Pinglish preprocess elements from a file.
/// </summary>
public static List <PreprocessElementInfo> LoadPreprocessElements(string filePath)
{
string[] elementWithInfos;
var preprocessElementInfos = new List <PreprocessElementInfo>();
PreprocessElementInfo elementInfo;
if (File.Exists(filePath))
{
StreamReader reader = null;
try
{
reader = new StreamReader(filePath);
string line;
while (!reader.EndOfStream)
{
line = reader.ReadLine();
if (string.IsNullOrEmpty(line) || line.StartsWith("#")) // It's for comments
{
continue;
}
elementWithInfos = line.Split(PreprocessElementInfoSeparators, StringSplitOptions.RemoveEmptyEntries);
if (elementWithInfos.Length <= 1)
{
continue;
}
string pinglishStr = elementWithInfos[0];
bool isWholeWord = false;
bool isExactWord = false;
var position = TokenPosition.None;
var equivalents = new List <string>();
for (int i = 1; i < elementWithInfos.Length; i++)
{
if (!elementWithInfos[i].StartsWith("#"))
{
equivalents.Add(elementWithInfos[i]);
}
else
{
string info = elementWithInfos[i].TrimStart('#');
switch (info.ToLower())
{
case "start":
position = position.Set(TokenPosition.StartOfWord);
break;
case "middle":
position = position.Set(TokenPosition.MiddleOfWord);
break;
case "end":
position = position.Set(TokenPosition.EndOfWord);
break;
case "wholeword":
isWholeWord = true;
break;
case "exact":
isExactWord = true;
break;
}
}
}
if (position == TokenPosition.None || isWholeWord)
{
position = TokenPosition.Any;
}
elementInfo = new PreprocessElementInfo(pinglishStr, isWholeWord, isExactWord, position);
elementInfo.Equivalents.AddRange(equivalents);
preprocessElementInfos.Add(elementInfo);
}
}
catch (Exception ex)
{
Debug.WriteLine(ex);
}
finally
{
if (reader != null)
{
reader.Close();
}
}
}
return(preprocessElementInfos);
}