/// <summary>
/// Adds a new named entity instance into the list, while avoid the duplicated ones.
/// </summary>
/// <param name="entity">The named entity instance to add.</param>
/// <returns>Whether succeeded added.</returns>
public bool AddNamedEntity(ScriptNamedEntity entity)
{
bool existed = false;
foreach (var item in NamedEntities)
{
if (item.Equals(entity))
{
existed = true;
break;
}
}
if (!existed)
{
NamedEntities.Add(entity);
_namedEntities = NamedEntities.SortBy(e => e.StartIndex).ToList();
}
return !existed;
}
/// <summary>
/// Converts one word instance to script named entity.
/// </summary>
/// <param name="word">The word instance to convert.</param>
/// <returns>The converted script named entity instance.</returns>
private ScriptNamedEntity ToScriptNamedEntity(ScriptWord word)
{
Helper.ThrowIfNull(word);
if (string.IsNullOrEmpty(word.NamedEntityTypeString))
{
throw new InvalidDataException(Helper.NeutralFormat(
"The type of the named entity [{0}] should not be empty.", word.NamedEntityTypeString));
}
ScriptNamedEntity entity = new ScriptNamedEntity();
entity.Text = word.Grapheme;
entity.PosString = word.PosString;
entity.Type = word.NamedEntityTypeString;
entity.Start = word.SubWords[0];
entity.End = word.SubWords[word.SubWords.Count - 1];
return entity;
}
/// <summary>
/// Converts one named entity to script word instance.
/// </summary>
/// <param name="entity">The named entity instance to convert.</param>
/// <returns>The converted word instance.</returns>
private ScriptWord ToScriptWord(ScriptNamedEntity entity)
{
Helper.ThrowIfNull(entity);
if (string.IsNullOrEmpty(entity.Type))
{
throw new InvalidDataException(Helper.NeutralFormat(
"The type of the named entity [{0}] should not be empty.", entity.Text));
}
ScriptWord word = new ScriptWord(entity.Start.Language);
word.Break = entity.End.Break;
word.Grapheme = entity.Text;
word.PosString = entity.PosString;
if (word.PosString == ScriptNamedEntity.DefaultEmptyPosString)
{
word.PosString = ScriptNamedEntity.DefaultEntityPosString;
}
word.NamedEntityTypeString = entity.Type;
word.Sentence = entity.Start.Sentence;
word.SubWords = new Collection<ScriptWord>();
StringBuilder pronunciation = new StringBuilder();
for (int i = Words.IndexOf(entity.Start); i <= Words.IndexOf(entity.End); i++)
{
word.SubWords.Add(Words[i]);
if (!string.IsNullOrEmpty(Words[i].Pronunciation))
{
if (pronunciation.Length != 0)
{
pronunciation.AppendFormat(" {0} ", Pronunciation.WordPronBoundaryString);
}
pronunciation.Append(Words[i].Pronunciation);
}
}
word.Pronunciation = pronunciation.ToString();
return word;
}
/// <summary>
/// Load one script named entity from the xml text reader.
/// </summary>
/// <param name="reader">The XML reader instance to read data from.</param>
/// <param name="sentence">Script sentence.</param>
/// <param name="scriptContentController">ContentControler.</param>
/// <returns>ScriptNamedEntity instance that read.</returns>
public static ScriptNamedEntity LoadNamedEntity(XmlTextReader reader,
ScriptSentence sentence, ContentControler scriptContentController)
{
Debug.Assert(reader != null);
Debug.Assert(scriptContentController != null);
ScriptNamedEntity entity = new ScriptNamedEntity();
entity.Type = reader.GetAttribute("type");
entity.Text = reader.GetAttribute("v");
string pos = reader.GetAttribute("pos");
if (!string.IsNullOrEmpty(pos))
{
entity.PosString = pos;
}
Debug.Assert(sentence.Words.Count > 0);
int startIndex = int.Parse(reader.GetAttribute("s"), CultureInfo.InvariantCulture);
int endIndex = int.Parse(reader.GetAttribute("e"), CultureInfo.InvariantCulture);
Collection<ScriptWord> graphemeWords = sentence.TextWords;
if (startIndex < 0 && startIndex >= graphemeWords.Count)
{
throw new InvalidDataException(Helper.NeutralFormat(
"Invalid start index for sentence [{0}] : [{1}]",
sentence.ScriptItem.GetSentenceId(sentence), startIndex));
}
entity.Start = graphemeWords[startIndex];
if (endIndex < 0 || endIndex >= graphemeWords.Count)
{
throw new InvalidDataException(Helper.NeutralFormat(
"Invalid end index for sentence [{0}] : [{1}]",
sentence.ScriptItem.GetSentenceId(sentence), endIndex));
}
entity.End = graphemeWords[endIndex];
return entity;
}
