/// <summary>
/// Extract a sub lexicon from a word list and return those words which not in the main lexicon.
/// </summary>
/// <param name="words">Words list to extract.</param>
/// <param name="missedLexWords">Words that not in the main lexicon.</param>
/// <returns>New sub lexicon.</returns>
public Lexicon ExtractSubLexicon(List<string> words, List<string> missedLexWords)
{
Lexicon newLex = new Lexicon();
newLex.Language = Language;
newLex.Encoding = Encoding;
newLex.PhoneSet = PhoneSet;
newLex.PosSet = PosSet;
Dictionary<string, object> missedWords = null;
if (missedLexWords != null)
{
missedLexWords.Clear();
missedWords = new Dictionary<string, object>(StringComparer.InvariantCulture);
}
foreach (string word in words)
{
if (string.IsNullOrEmpty(word) || newLex.Items.ContainsKey(word))
{
continue;
}
// First do case sensitive lookup; if not found, do case insensitive lookup.
LexicalItem wordItem = newLex.Lookup(word);
if (wordItem == null)
{
wordItem = Lookup(word, true);
}
if (wordItem != null)
{
newLex.Items.Add(word, wordItem);
}
else
{
if (missedWords != null && !missedWords.ContainsKey(word))
{
missedWords.Add(word, null);
}
}
}
if (missedLexWords != null)
{
missedLexWords.AddRange(missedWords.Keys);
}
return newLex;
}
/// <summary>
/// Merge lexicon. when facing conflict word, use the attributes in the first lexicon.
/// </summary>
/// <param name="mergedLexicon">Main lexicon and merged lexicon.</param>
/// <param name="subLexicon">Sub lexicon.</param>
private void MergeLexiconWithKeepFirstOne(Lexicon mergedLexicon, Lexicon subLexicon)
{
Collection<string> skippedPronMessage = new Collection<string>();
int addedWord = 0;
foreach (KeyValuePair<string, LexicalItem> subLexiconItem in subLexicon.Items)
{
string word = subLexiconItem.Key;
// If the sub lexicon item doesn't exist in merged lexicon, then add it.
LexicalItem originalItem = mergedLexicon.Lookup(word, true);
if (originalItem == null)
{
mergedLexicon.Items.Add(subLexiconItem.Key, subLexiconItem.Value);
addedWord++;
continue;
}
foreach (LexiconPronunciation newPron in subLexiconItem.Value.Pronunciations)
{
if (!originalItem.ContainsPronunciation(newPron.Symbolic))
{
skippedPronMessage.Add(Helper.NeutralFormat(
"Pronunciation for word [{0}] has been skipped: [{1}]",
subLexiconItem.Key, newPron.Symbolic));
}
}
}
// Log the message
{
Log("---------------------------------");
Log("Totally:");
Log("[{0}] pronunciations have been skipped.",
skippedPronMessage.Count);
Log(Helper.NeutralFormat("[{0}] new words have been added by the latter lexicon", addedWord));
}
}
/// <summary>
/// Create the lexicon from Xml Script file.
/// </summary>
/// <param name="scriptFile">Xml script file.</param>
/// <param name="defaultPos">Part of Speech String.</param>
/// <param name="mainLexicon">MainLexicon.</param>
/// <returns>Lexicon.</returns>
public static Lexicon CreateFromXmlScriptFile(XmlScriptFile scriptFile, string defaultPos, Lexicon mainLexicon)
{
if (scriptFile == null)
{
throw new ArgumentNullException("scriptFile");
}
if (string.IsNullOrEmpty(defaultPos))
{
throw new ArgumentNullException("defaultPos");
}
Lexicon lexicon = new Lexicon(scriptFile.Language);
foreach (ScriptItem item in scriptFile.Items)
{
foreach (ScriptWord scriptWord in item.AllPronouncedWords)
{
string word = scriptWord.Grapheme;
// Create LexiconPronunciaton Node
LexiconPronunciation pron = new LexiconPronunciation(lexicon.Language);
pron.Symbolic = scriptWord.Pronunciation;
if (mainLexicon != null)
{
LexicalItem mainLexiconItem = mainLexicon.Lookup(word, true);
if (mainLexiconItem != null)
{
LexiconPronunciation lexPron = mainLexiconItem.FindPronunciation(pron.Symbolic, true);
if (lexPron != null)
{
pron.Symbolic = lexPron.Symbolic;
}
}
}
LexiconItemProperty property = new LexiconItemProperty();
if (string.IsNullOrEmpty(scriptWord.PosString))
{
property.PartOfSpeech = new PosItem(defaultPos);
}
else
{
property.PartOfSpeech = new PosItem(scriptWord.PosString);
}
pron.Properties.Add(property);
if (!lexicon.Items.ContainsKey(word))
{
LexicalItem lexicalItem = new LexicalItem(lexicon.Language);
lexicalItem.Grapheme = word;
lexicalItem.Pronunciations.Add(pron);
lexicon.Items.Add(word, lexicalItem);
}
else
{
bool needAdd = true;
foreach (LexiconPronunciation pronunciation in lexicon.Items[word].Pronunciations)
{
if (pronunciation.Symbolic.Equals(pron.Symbolic, StringComparison.InvariantCultureIgnoreCase))
{
needAdd = false;
if (!pronunciation.Properties.Contains(property))
{
pronunciation.Properties.Add(property);
}
}
}
if (needAdd)
{
lexicon.Items[word].Pronunciations.Add(pron);
}
}
}
}
return lexicon;
}
/// <summary>
/// Merge lexicon, when facing conflict word, use the attributes in the last lexicon.
/// </summary>
/// <param name="mergedLexicon">Main lexicon and merged lexicon.</param>
/// <param name="subLexicon">Sub lexicon.</param>
private void MergeLexiconWithKeepLastOne(Lexicon mergedLexicon, Lexicon subLexicon)
{
Collection<string> replacedPronMessage = new Collection<string>();
Collection<string> replacedPropertyMessage = new Collection<string>();
Collection<string> existedWords = new Collection<string>();
Collection<string> existedWordsInLower = new Collection<string>();
// Dump the conflict pronunciations or properties from the merged lexicon
foreach (KeyValuePair<string, LexicalItem> mergedLexiconItem in mergedLexicon.Items)
{
string word = mergedLexiconItem.Key;
LexicalItem newItem = subLexicon.Lookup(word, true);
// If the sub lexicon contain the same grapheme, then delete the one in original lexicon.
if (newItem != null)
{
existedWords.Add(word);
existedWordsInLower.Add(word.ToLowerInvariant());
foreach (LexiconPronunciation originalPron in mergedLexiconItem.Value.Pronunciations)
{
LexiconPronunciation existedPron = newItem.FindPronunciation(originalPron.Symbolic);
if (existedPron == null)
{
replacedPronMessage.Add(Helper.NeutralFormat(
"Pronunciation for word [{0}] has been removed: [{1}]",
mergedLexiconItem.Key, originalPron.Symbolic));
}
else
{
foreach (LexiconItemProperty subItemProperty in originalPron.Properties)
{
bool hasProperty = false;
// Find old properties in new(sub) item.
foreach (LexiconItemProperty itemProperty in existedPron.Properties)
{
if (itemProperty.Equals(subItemProperty))
{
hasProperty = true;
break;
}
}
// Add the property if doesn't contains it.
if (!hasProperty)
{
replacedPropertyMessage.Add(Helper.NeutralFormat(
"Property has been replaced for word [{0}]'s pronunciation [{1}] : [{2}]",
word, originalPron.Symbolic, subItemProperty.ToString()));
}
}
}
}
}
}
// Remove the duplicate word entries
foreach (string word in existedWords)
{
mergedLexicon.Items.Remove(word);
}
// Add new word entries into merged lexicon.
int newWord = 0;
foreach (KeyValuePair<string, LexicalItem> subLexiconItem in subLexicon.Items)
{
mergedLexicon.Items.Add(subLexiconItem.Key, subLexiconItem.Value);
if (!existedWordsInLower.Contains(subLexiconItem.Key.ToLowerInvariant()))
{
newWord++;
}
}
// Log the Message
{
Log("---------------------------------");
Log("Totally:");
Log(Helper.NeutralFormat("[{0}] words have been replaced by the latter lexicon", replacedPronMessage.Count));
Log("[{0}] properties have been replaced.", replacedPropertyMessage.Count);
Log(Helper.NeutralFormat("[{0}] new words have been added by the latter lexicon", newWord));
}
}