|
public static LoadMap ( string dangAmbigsFile ) : string>.Dictionary |
||
| dangAmbigsFile | string | |
| return | string>.Dictionary | |
public static string PostProcess(string text, string langCode, string dangAmbigsPath, bool dangAmbigsOn)
{
// postprocessor
StringBuilder strB = new StringBuilder(PostProcess(text, langCode));
if (!dangAmbigsOn)
{
return(strB.ToString());
}
// replace text based on entries read from a DangAmbigs.txt file
Dictionary <string, string> replaceRules = TextUtilities.LoadMap(Path.Combine(dangAmbigsPath, langCode + ".DangAmbigs.txt"));
if (replaceRules.Count == 0 && langCode.Length > 3)
{
replaceRules = TextUtilities.LoadMap(Path.Combine(dangAmbigsPath, langCode.Substring(0, 3) + ".DangAmbigs.txt")); // fall back on base
}
Dictionary <string, string> .KeyCollection.Enumerator enumer = replaceRules.Keys.GetEnumerator();
while (enumer.MoveNext())
{
string key = enumer.Current;
string value = replaceRules[key];
strB = strB.Replace(key, value);
}
return(strB.ToString());
}
public static string PostProcess(string text, string langCode, string dangAmbigsPath, bool dangAmbigsOn, bool replaceHyphens)
{
if (text.Trim().Length == 0)
{
return(text);
}
if (replaceHyphens)
{
text = Net.SourceForge.Vietpad.Utilities.TextUtilities.ReplaceHyphensWithSoftHyphens(text);
}
// correct using external x.DangAmbigs.txt file first, if enabled
if (dangAmbigsOn)
{
StringBuilder strB = new StringBuilder(text);
// replace text based on entries read from an x.DangAmbigs.txt file
List <Dictionary <string, string> > replaceRules = TextUtilities.LoadMap(Path.Combine(dangAmbigsPath, langCode + ".DangAmbigs.txt"));
if (replaceRules.Count == 0 && langCode.Length > 3)
{
replaceRules = TextUtilities.LoadMap(Path.Combine(dangAmbigsPath, langCode.Substring(0, 3) + ".DangAmbigs.txt")); // fall back on base
}
if (replaceRules.Count == 0)
{
throw new NotSupportedException(langCode);
}
Dictionary <string, string> replaceRulesPlain = replaceRules[PLAIN];
Dictionary <string, string> .KeyCollection.Enumerator enumer = replaceRulesPlain.Keys.GetEnumerator();
while (enumer.MoveNext())
{
string key = enumer.Current;
string value = replaceRulesPlain[key];
strB = strB.Replace(key, value);
}
text = strB.ToString();
Dictionary <string, string> replaceRulesRegex = replaceRules[REGEX];
enumer = replaceRulesRegex.Keys.GetEnumerator();
while (enumer.MoveNext())
{
string key = enumer.Current;
string value = replaceRulesRegex[key];
text = Regex.Replace(text, key, value);
}
}
// postprocessor
text = PostProcess(text, langCode);
// correct letter cases
return(TextUtilities.CorrectLetterCases(text));
}
public static string PostProcess(string text, string langCode, string dangAmbigsPath, bool dangAmbigsOn)
{
if (text.Trim().Length == 0)
{
return(text);
}
// correct using external x.DangAmbigs.txt file first, if enabled
if (dangAmbigsOn)
{
StringBuilder strB = new StringBuilder(text);
// replace text based on entries read from an x.DangAmbigs.txt file
Dictionary <string, string> replaceRules = TextUtilities.LoadMap(Path.Combine(dangAmbigsPath, langCode + ".DangAmbigs.txt"));
if (replaceRules.Count == 0 && langCode.Length > 3)
{
replaceRules = TextUtilities.LoadMap(Path.Combine(dangAmbigsPath, langCode.Substring(0, 3) + ".DangAmbigs.txt")); // fall back on base
}
if (replaceRules.Count == 0)
{
throw new NotSupportedException(langCode);
}
Dictionary <string, string> .KeyCollection.Enumerator enumer = replaceRules.Keys.GetEnumerator();
while (enumer.MoveNext())
{
string key = enumer.Current;
string value = replaceRules[key];
strB = strB.Replace(key, value);
}
text = strB.ToString();
}
// postprocessor
text = PostProcess(text, langCode);
// correct common errors caused by OCR
text = TextUtilities.CorrectOCRErrors(text);
// correct letter cases
return(TextUtilities.CorrectLetterCases(text));
}