private string GetPhoneticValue(string word, PhoneticTable phoneticTable)
{
string phoneticValue = phoneticTable.GetPhoneticValueOf(word);
if (phoneticValue == null)
phoneticValue = GetPhoneticValueOfWordUsingSimilarWordEnding(word, phoneticTable);
return phoneticValue;
}
public void Expand(PhoneticTable phoneticTable, WordListFile frequentWordListFile)
{
string phoneticValue;
int countBefore;
do
{
countBefore = phoneticTable.Count;
foreach (string word in frequentWordListFile)
{
if (!phoneticTable.Contains(word))
{
phoneticValue = null;
if (phoneticValue == null)
phoneticValue = phoneticConcatenator.TryConcatenate(word, phoneticTable);
if (phoneticValue == null)
phoneticValue = phoneticSplitter.TrySplit(word, phoneticTable);
if (phoneticValue != null)
phoneticTable.Add(word, phoneticValue);
}
}
} while (countBefore != phoneticTable.Count);
}
private string TryConcatenate(string word, string wordStart, string wordEnd, PhoneticTable phoneticTable)
{
if (word == wordStart + wordEnd)
{
return phoneticTable.GetPhoneticValueOf(wordStart) + " [dash] " + phoneticTable.GetPhoneticValueOf(wordEnd);
}
else
{
return null;
}
}
private List<string> GetListStartsWith(string word, PhoneticTable phoneticTable)
{
List<string> listStartsWith = new List<string>();
foreach (string currentWord in phoneticTable.EnglishWordList)
{
if (currentWord.StartsWith(word))
{
listStartsWith.Add(currentWord);
}
}
return listStartsWith;
}
private string GetPhoneticValueOfWordUsingSimilarWordEnding(string word, PhoneticTable phoneticTable)
{
string englishEnding;
if (word.Length >= howManyEnglishLetterForEnding)
englishEnding = word.Substring(word.Length - howManyEnglishLetterForEnding);
else
englishEnding = word;
string similarWord = GetWordEndsWith(englishEnding, phoneticTable);
if (similarWord == null)
return null;
return phoneticTable.GetPhoneticValueOf(similarWord);
}
public void Build(PhoneticTable phoneticTable, WordListFile frequentWordListFile, string rhymeChartFile)
{
string phoneticValue;
using (StreamWriter streamWriter = new StreamWriter(rhymeChartFile))
{
foreach (string word in frequentWordListFile)
{
phoneticValue = GetPhoneticValue(word,phoneticTable);
if (phoneticValue != null)
{
streamWriter.WriteLine(word + " : " + GetPhoneticEnding(phoneticValue));
}
}
}
}
/// <summary>
/// Get phonetic value by concatenation or null if fails
/// </summary>
/// <param name="word">word for which we need phonetic value</param>
/// <param name="phoneticTable">phonetic table</param>
/// <returns>phonetic value by concatenation or null if fails</returns>
public string TryConcatenate(string word, PhoneticTable phoneticTable)
{
List<string> listWordStartsWith = GetListStartsWith(word, phoneticTable);
List<string> listWordEndsWith = GetListEndsWith(word, phoneticTable);
string concatenation = null;
foreach (string wordStart in listWordStartsWith)
{
foreach (string wordEnd in listWordEndsWith)
{
concatenation = TryConcatenate(word, wordStart, wordEnd, phoneticTable);
if (concatenation != null)
return concatenation;
}
}
return null;
}
/// <summary>
/// Repair phonetic table
/// </summary>
/// <param name="phoneticTable">phonetic table</param>
public void Repair(PhoneticTable phoneticTable)
{
string phoneticEnding;
string trimmedPhoneticValue;
foreach (HomophoneGroup homophoneGroup in new List<HomophoneGroup>(phoneticTable))
{
foreach (string wordVariant in new HashSet<string>(homophoneGroup))
{
if (IsMatchEndingType(wordVariant, homophoneGroup))
{
if (IsMatchWordVariantEnding(wordVariant))
{
trimmedPhoneticValue = RemoveUndesiredEnding(homophoneGroup.PhoneticValue);
phoneticEnding = BuildPhoneticEnding(homophoneGroup.PhoneticValue);
homophoneGroup.Remove(wordVariant);
phoneticTable.Add(wordVariant, trimmedPhoneticValue + " " + phoneticEnding);
}
}
}
}
}
public void ReplaceEnding(PhoneticTable phoneticTable, string fromEnglish, string fromPhonetic, string toEnglish, string toPhonetic)
{
string phoneticValue;
foreach (HomophoneGroup homophoneGroup in new List<HomophoneGroup>(phoneticTable))
{
foreach (string wordVariant in new HashSet<string>(homophoneGroup))
{
if (wordVariant.EndsWith(toEnglish))
{
if (homophoneGroup.GetShortestVariant(wordVariant).EndsWith(fromEnglish) && wordVariant != homophoneGroup.GetShortestVariant(wordVariant))
{
if (homophoneGroup.PhoneticValue.EndsWith(fromPhonetic))
{
phoneticValue = ReplaceEnding(homophoneGroup.PhoneticValue, fromPhonetic, toPhonetic);
homophoneGroup.Remove(wordVariant);
phoneticTable.Add(wordVariant, phoneticValue);
}
}
}
}
}
}
private string GetWordEndsWith(string englishEnding, PhoneticTable phoneticTable)
{
foreach (HomophoneGroup homophoneGroup in phoneticTable)
{
foreach (string currentWord in homophoneGroup)
{
if (currentWord.EndsWith(englishEnding))
{
return phoneticTable.GetPhoneticValueOf(currentWord);
}
}
}
return null;
}
public void Repair(PhoneticTable phoneticTable)
{
int countBeforeRepair;
do
{
countBeforeRepair = phoneticTable.Count;
endingRepairerS.Repair(phoneticTable);
endingRepairerEd.Repair(phoneticTable);
endingRepairerLy.Repair(phoneticTable);
endingRepairerIsmToIst.Repair(phoneticTable);
endingRepairerIstToIsm.Repair(phoneticTable);
endingRepairerYtoIes.Repair(phoneticTable);
endingRepairerAteToAtion.Repair(phoneticTable);
endingRepairerAteToAting.Repair(phoneticTable);
endingRepairerAteToAtive.Repair(phoneticTable);
endingRepairerAteToAtor.Repair(phoneticTable);
endingRepairerYToIc.Repair(phoneticTable);
endingReplacer.ReplaceEnding(phoneticTable, "y", "[emacr]", "ically", "[ibreve] [dash] [kreg] [schwa] [lreg] [emacr]");
endingReplacer.ReplaceEnding(phoneticTable, "ic", "[ibreve] [kreg]", "ically", "[ibreve] [dash] [kreg] [schwa] [lreg] [emacr]");
endingReplacer.ReplaceEnding(phoneticTable, "ate", "[amacr] [treg] [lprime]", "ations", "[amacr] [prime] [sreg] [hreg] [schwa] [nreg] [zreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ate", "[amacr] [treg] [lprime]", "ators", "[amacr] [lprime] [treg] [schwa] [rreg] [zreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ize", "[imacr] [zreg] [lprime]", "izing", "[imacr] [lprime] [zreg] [ibreve] [nreg] [greg]");
endingReplacer.ReplaceEnding(phoneticTable, "ize", "[imacr] [zreg] [lprime]", "ising", "[imacr] [lprime] [zreg] [ibreve] [nreg] [greg]");
endingReplacer.ReplaceEnding(phoneticTable, "ize", "[imacr] [zreg] [lprime]", "ised", "[imacr] [zreg] [lprime] [dreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ism", "[ibreve] [zreg] [lprime] [schwa] [mreg]", "istic", "[ibreve] [sreg] [treg] [ibreve] [kreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ism", "[ibreve] [zreg] [lprime] [schwa] [mreg]", "istically", "[ibreve] [sreg] [treg] [ibreve] [kreg] [schwa] [dash] [lreg] [emacr]");
endingReplacer.ReplaceEnding(phoneticTable, "p", "[preg] [prime]", "pped", "[preg] [treg]");
endingReplacer.ReplaceEnding(phoneticTable, "p", "[preg] [prime]", "pper", "[preg] [prime] [schwa] [rreg]");
endingReplacer.ReplaceEnding(phoneticTable, "p", "[preg] [prime]", "ppers", "[preg] [prime] [schwa] [rreg] [zreg]");
endingReplacer.ReplaceEnding(phoneticTable, "p", "[preg] [prime]", "pping", "[preg] [ibreve] [nreg] [greg]");
endingReplacer.ReplaceEnding(phoneticTable, "p", "[preg]", "pped", "[preg] [treg]");
endingReplacer.ReplaceEnding(phoneticTable, "p", "[preg]", "pper", "[preg] [prime] [schwa] [rreg]");
endingReplacer.ReplaceEnding(phoneticTable, "p", "[preg]", "ppers", "[preg] [prime] [schwa] [rreg] [zreg]");
endingReplacer.ReplaceEnding(phoneticTable, "p", "[preg]", "pping", "[preg] [ibreve] [nreg] [greg]");
endingReplacer.ReplaceEnding(phoneticTable, "fy", "[freg] [imacr] [lprime]", "fied", "[freg] [imacr] [dreg] [lprime]");
endingReplacer.ReplaceEnding(phoneticTable, "fy", "[freg] [imacr] [lprime]", "fication", "[freg] [ibreve] [dash] [kreg] [amacr] [prime] [sreg] [hreg] [schwa] [nreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ist", "[ibreve] [dash] [sreg] [treg]", "istical", "[ibreve] [sreg] [prime] [treg] [ibreve] [dash] [kreg] [schwa] [lreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ist", "[ibreve] [sreg] [treg]", "istical", "[ibreve] [sreg] [prime] [treg] [ibreve] [dash] [kreg] [schwa] [lreg]");
endingReplacer.ReplaceEnding(phoneticTable, "er", "[schwa] [rreg]", "erous", "[schwa] [rreg] [dash] [schwa] [sreg]");
endingReplacer.ReplaceEnding(phoneticTable, "er", "[schwa] [rreg]", "ering", "[schwa] [rreg] [ibreve] [nreg] [greg]");
endingReplacer.ReplaceEnding(phoneticTable, "er", "[schwa] [rreg]", "erings", "[schwa] [rreg] [ibreve] [nreg] [greg] [zreg]");
endingReplacer.ReplaceEnding(phoneticTable, "y", "[emacr]", "iness", "[emacr] [nreg] [ebreve] [sreg]");
endingReplacer.ReplaceEnding(phoneticTable, "y", "[emacr]", "ier", "[imacr] [lprime] [schwa] [rreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ant", "[schwa] [nreg] [treg]", "ance", "[schwa] [nreg] [sreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ent", "[schwa] [nreg] [treg]", "ence", "[schwa] [nreg] [sreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ant", "[schwa] [nreg] [treg]", "ancy", "[schwa] [nreg] [dash] [sreg] [emacr]");
endingReplacer.ReplaceEnding(phoneticTable, "ent", "[schwa] [nreg] [treg]", "ency", "[schwa] [nreg] [dash] [sreg] [emacr]");
endingReplacer.ReplaceEnding(phoneticTable, "ance", "[schwa] [nreg] [sreg]", "ant", "[schwa] [nreg] [treg]");
endingReplacer.ReplaceEnding(phoneticTable, "ence", "[schwa] [nreg] [sreg]", "ent", "[schwa] [nreg] [treg]");
endingReplacer.ReplaceEnding(phoneticTable, "ancy", "[schwa] [nreg] [dash] [sreg] [emacr]", "ant", "[schwa] [nreg] [treg]");
endingReplacer.ReplaceEnding(phoneticTable, "ency", "[schwa] [nreg] [dash] [sreg] [emacr]", "ent", "[schwa] [nreg] [treg]");
endingReplacer.ReplaceEnding(phoneticTable, "tion", "[sreg] [hreg] [schwa] [nreg]", "tive", "[treg] [ibreve] [vreg]");
endingReplacer.ReplaceEnding(phoneticTable, "tive", "[treg] [ibreve] [vreg]", "tion", "[sreg] [hreg] [schwa] [nreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ia", "[emacr] [dash] [schwa]", "ic", "[ibreve] [kreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ic", "[ibreve] [kreg]", "ia", "[emacr] [dash] [schwa]");
endingReplacer.ReplaceEnding(phoneticTable, "ists", "[ibreve] [sreg] [treg] [sreg]", "isms", "[ibreve] [zreg] [lprime] [schwa] [mreg] [zreg]");
endingReplacer.ReplaceEnding(phoneticTable, "isms", "[ibreve] [zreg] [lprime] [schwa] [mreg] [zreg]", "ists", "[ibreve] [sreg] [treg] [sreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ist", "[ibreve] [sreg] [treg]", "isms", "[ibreve] [zreg] [lprime] [schwa] [mreg] [zreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ism", "[ibreve] [zreg] [lprime] [schwa] [mreg]", "ists", "[ibreve] [sreg] [treg] [sreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ble", "[breg] [schwa] [lreg]", "bility", "[breg] [ibreve] [lreg] [prime] [ibreve] [dash] [treg] [emacr]");
endingReplacer.ReplaceEnding(phoneticTable, "ize", "[imacr] [zreg] [lprime]", "ization", "[ibreve] [dash] [zreg] [amacr] [prime] [sreg] [hreg] [schwa] [nreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ia", "[emacr] [dash] [schwa]", "ian", "[emacr] [dash] [schwa] [nreg]");
endingReplacer.ReplaceEnding(phoneticTable, "ian", "[emacr] [dash] [schwa] [nreg]", "ia", "[emacr] [dash] [schwa]");
RepairEnding(phoneticTable, "ing", "[ibreve] [nreg] [greg]");
RepairEnding(phoneticTable, "er", "[schwa] [rreg]");
RepairEnding(phoneticTable, "ers", "[schwa] [rreg] [zreg]");
RepairEnding(phoneticTable, "r", "[schwa] [rreg]");
RepairEnding(phoneticTable, "or", "[schwa] [rreg]");
RepairEnding(phoneticTable, "ic", "[ibreve] [kreg]");
RepairEnding(phoneticTable, "ity", "[ibreve] [dash] [treg] [emacr]");
RepairEnding(phoneticTable, "ism", "[ibreve] [zreg] [lprime] [schwa] [mreg]");
RepairEnding(phoneticTable, "ist", "[ibreve] [dash] [sreg] [treg]");
RepairEnding(phoneticTable, "ness", "[nreg] [ebreve] [sreg]");
RepairEnding(phoneticTable, "less", "[lreg] [ebreve] [sreg]");
RepairEnding(phoneticTable, "est", "[ebreve] [sreg] [treg]");
RepairEnding(phoneticTable, "st", "[ebreve] [sreg] [treg]");
RepairEnding(phoneticTable, "istic", "[ibreve] [sreg] [treg] [ibreve] [kreg]");
RepairEnding(phoneticTable, "ically", "[ibreve] [dash] [kreg] [schwa] [lreg] [emacr]");
RepairEnding(phoneticTable, "ation", "[amacr] [prime] [sreg] [hreg] [schwa] [nreg]");
RepairEnding(phoneticTable, "ations", "[amacr] [prime] [sreg] [hreg] [schwa] [nreg] [zreg]");
RepairEnding(phoneticTable, "al", "[lreg]");
RepairEnding(phoneticTable, "ally", "[schwa] [dash] [lreg] [emacr]");
RepairEnding(phoneticTable, "ment", "[mreg] [ebreve] [nreg] [prime] [treg]");
} while (phoneticTable.Count != countBeforeRepair);
#warning Implement Repair() for other cases
//endingRepairerLy (l or no l at the end)
//ist to ism and vice versae even when no ist or ism counterpart available
//involuntary ->involuntarily
//invulnerable -> invulnerability
//evict -> eviction
/*
//RepairEnding(phoneticTable, "ator");
RepairEnding(phoneticTable, "ly");
RepairEnding(phoneticTable, "ion");
RepairEnding(phoneticTable, "es");
RepairEnding(phoneticTable, "ence", "ent");*/
//militant -> militance
//militant -> militancy
}
private void RepairEnding(PhoneticTable phoneticTable, string englishEnding, string phoneticEnding)
{
phoneticEnding = phoneticEnding.Trim();
foreach (HomophoneGroup homophoneGroup in new List<HomophoneGroup>(phoneticTable))
{
foreach (string wordVariant in new HashSet<string>(homophoneGroup))
{
if (wordVariant == homophoneGroup.GetShortestVariant(wordVariant) + englishEnding && wordVariant != homophoneGroup.GetShortestVariant(wordVariant))
{
homophoneGroup.Remove(wordVariant);
phoneticTable.Add(wordVariant, homophoneGroup.PhoneticValue + " " + phoneticEnding);
}
}
}
}
public void Trim(PhoneticTable phoneticTable)
{
foreach (HomophoneGroup homophoneGroup in phoneticTable)
Trim(homophoneGroup);
}
/// <summary>
/// Get phonetic value by splitting or null if fails
/// </summary>
/// <param name="word">word for which we need phonetic value</param>
/// <param name="phoneticTable">phonetic table</param>
/// <returns>phonetic value by splitting or null if fails</returns>
public string TrySplit(string word, PhoneticTable phoneticTable)
{
#warning Implement TrySplit() (return null only if fails)
return null;
}
