/// <summary>
/// Traverse script file.
/// </summary>
/// <param name="visitor">Script item processor interface.</param>
/// <param name="importDataPathList">File list for importing data model.</param>
/// <param name="inScriptFile">Input script file.</param>
/// <param name="outScriptFile">Output script file.</param>
/// <param name="logger">Text logger.</param>
/// <param name="parameters">Parameters.</param>
public static void TraverseScriptFile(IScriptItemProcessor visitor, Dictionary<string, string> importDataPathList,
string inScriptFile, string outScriptFile, TextLogger logger,
object parameters)
{
if (string.IsNullOrEmpty(inScriptFile))
{
throw new ArgumentNullException("inScriptFile");
}
if (!File.Exists(inScriptFile))
{
throw new FileNotFoundException(inScriptFile);
}
XmlScriptFile script = new XmlScriptFile();
string message = Helper.NeutralFormat("Processing file : {0}", inScriptFile);
Console.WriteLine(message);
string fileName = Path.GetFileName(inScriptFile);
try
{
script.Load(inScriptFile);
List<string> listDiscard = new List<string>();
foreach (ScriptItem scriptItem in script.Items)
{
try
{
// When the mode is "Import F0/Power/Segment", the importDataPathList will not be null.
if (importDataPathList == null || importDataPathList.ContainsKey(scriptItem.Id))
{
visitor.ProcessItem(scriptItem, parameters);
}
else
{
listDiscard.Add(scriptItem.Id);
}
}
catch (InvalidDataException exception)
{
if (logger != null)
{
logger.LogLine("ERROR : [File {0}][Item {1}]{2}", fileName,
scriptItem.Id, exception.Message);
}
}
}
foreach (string id in listDiscard)
{
script.Remove(id);
}
if (!string.IsNullOrEmpty(outScriptFile))
{
Helper.TestWritable(outScriptFile);
script.Save(outScriptFile, Encoding.Unicode);
}
}
catch (InvalidDataException exception)
{
if (logger != null)
{
logger.LogLine("ERROR : [File {0}]{1}", fileName, exception.Message);
}
}
}
/// <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>
/// Load script.
/// </summary>
/// <param name="scriptPath">ScriptPath.</param>
/// <param name="errorSet">ErrorSet.</param>
private void LoadScript(string scriptPath, ErrorSet errorSet)
{
XmlScriptFile xmlScript = new XmlScriptFile();
XmlScriptFile.ContentControler controler = new XmlScriptFile.ContentControler();
controler.LoadComments = _loadComments;
xmlScript.Load(scriptPath, controler);
if (_language == xmlScript.Language)
{
_xmlScriptFiles.Add(xmlScript);
foreach (Error error in xmlScript.ErrorSet.Errors)
{
errorSet.Add(new Error(ScriptError.ScriptCollectionError,
Path.GetFileName(scriptPath),
error.ToString()));
}
}
else
{
errorSet.Add(ScriptError.InvalidLanguage,
Localor.LanguageToString(xmlScript.Language),
Localor.LanguageToString(_language),
Path.GetFileName(scriptPath));
}
}
/// <summary>
/// Extract Domain Lexicon from script.
/// </summary>
/// <param name="scriptFolder">Script Folder.</param>
/// <param name="domainListFile">Domain List File.</param>
/// <param name="inMainLex">Input Main Lexicon.</param>
/// <param name="defaultPartOfSpeech">Default Part of Speech.</param>
/// <param name="mergeMode">Merging Mode for Lexicon.</param>
/// <param name="phoneSet">Phone set.</param>
/// <param name="attribSchema">Lexical attribute schema.</param>
/// <returns>Lexicon.</returns>
private Lexicon ExtractDomainLexicon(string scriptFolder, string domainListFile,
Lexicon inMainLex, string defaultPartOfSpeech, MergeMode mergeMode,
TtsPhoneSet phoneSet, LexicalAttributeSchema attribSchema)
{
if (attribSchema != null)
{
if (PosItem.Validate(defaultPartOfSpeech, null, attribSchema).Count > 0)
{
Log("Default Part of speech {0} is unrecognized according to attribute schema, extraction breaks",
defaultPartOfSpeech);
return null;
}
}
Lexicon outLex = null;
foreach (string domainName in Helper.FileLines(domainListFile))
{
string domainFilePath = Path.Combine(scriptFolder, domainName);
XmlScriptFile scriptFile = new XmlScriptFile();
scriptFile.Load(domainFilePath);
if (outLex != null && outLex.Language != scriptFile.Language)
{
throw new InvalidDataException(Helper.NeutralFormat(
"Found inconsistent language \"{0}\" against previous one \"{1}\" in the file of \"{2}\"",
scriptFile.Language.ToString(),
outLex.Language.ToString(), domainFilePath));
}
Lexicon lexicon = Lexicon.CreateFromXmlScriptFile(scriptFile, defaultPartOfSpeech, inMainLex);
if (phoneSet != null && attribSchema != null)
{
lexicon.Validate(phoneSet, attribSchema);
if (lexicon.ErrorSet.Count > 0)
{
Console.Error.WriteLine("The script file {0} contains {1} errors, skip!",
domainFilePath, lexicon.ErrorSet.Count);
Log("The script file {0} contains {1} errors:",
domainFilePath, lexicon.ErrorSet.Count);
foreach (Error error in lexicon.ErrorSet.Errors)
{
Log(error.ToString());
}
// Skip this domain lexicon
continue;
}
}
if (outLex == null)
{
outLex = lexicon;
}
else
{
MergeLexicon(outLex, lexicon, mergeMode);
}
}
if (outLex.Items.Count == 0)
{
Log("The final lexicon is empty.");
}
return outLex;
}
/// <summary>
/// Sort XML script file items.
/// </summary>
/// <param name="scriptFile">Script file path.</param>
/// <param name="targetFilePath">Target file path.</param>
private static void ExportSortedCleanXmlScriptFiles(XmlScriptFile scriptFile, string targetFilePath)
{
XmlScriptFile sortedCleanScriptFile = new XmlScriptFile();
sortedCleanScriptFile.Language = scriptFile.Language;
sortedCleanScriptFile.Encoding = scriptFile.Encoding;
SortedDictionary<string, ScriptItem> sortedItems = new SortedDictionary<string, ScriptItem>();
foreach (KeyValuePair<string, ScriptItem> pair in scriptFile.ItemDic)
{
sortedItems.Add(pair.Key, pair.Value);
}
foreach (KeyValuePair<string, ScriptItem> pair in sortedItems)
{
sortedCleanScriptFile.Items.Add(pair.Value);
sortedCleanScriptFile.ItemDic.Add(pair.Key, pair.Value);
}
XmlScriptFile.ContentControler controler = new XmlScriptFile.ContentControler();
controler.SaveComments = false;
sortedCleanScriptFile.Save(targetFilePath, scriptFile.Encoding, controler);
}