/// <summary>
/// Идентификация всех сущностей в подпредложении
/// </summary>
/// <param name="subSentence">подпредложение</param>
private void IdentifyEntities(SubSentence subSentence)
{
if (subSentence.SubTextInfo != null)
{
IdentifyEntities(subSentence.SubTextInfo);
}
else
{
int childIndex = 0;
foreach (var unit in subSentence.Units)
{
while ((childIndex < subSentence.Children.Length) &&
(subSentence.Children[childIndex].StartPosition < unit.PositionInfo.Start))
{
IdentifyEntities(subSentence.Children[childIndex]);
++childIndex;
}
if (unit.IsEntity)
{
IdentifyEntity((Entity)unit);
}
}
for (; childIndex < subSentence.Children.Length; ++childIndex)
{
IdentifyEntities(subSentence.Children[childIndex]);
}
}
}
/// <summary>
/// Создание подпредложения из xml-элемента
/// </summary>
/// <param name="subSentence">xml-элемент</param>
/// <param name="language">язык предложения</param>
/// <returns>подпредложение</returns>
private static SubSentence SubSentenceFromXElement(XElement subSentence, string language)
{
#region [.defense.]
if (!subSentence.IsName(UnitTextType.SUB_SENT))
{
throw new ArgumentException("element isn't SUB_SENT");
}
#endregion
SubSentenceType type = (SubSentenceType)Enum.Parse(typeof(SubSentenceType), subSentence.GetAttributeValue(RDF.Attribute.Type));
SubSentence result = new SubSentence(type, _positionInSentence);
switch (type)
{
case SubSentenceType.AdjunctPhrase:
case SubSentenceType.DanglingParticiple:
case SubSentenceType.Default:
case SubSentenceType.Introductory:
case SubSentenceType.Participial:
case SubSentenceType.Subordinate:
SimpleSubSentenceFromXElement(result, subSentence, language);
break;
case SubSentenceType.ApproximatingForceps:
case SubSentenceType.DirectSpeech:
SubTextFromXElement(result, subSentence, language);
break;
#region [.defense.]
default:
throw new InvalidOperationException("Unknown subSentence type");
#endregion
}
return(result);
}
/// <summary>
/// Создание подтекста из xml-элемента
/// </summary>
/// <param name="subSentence">родительское подпредложение</param>
/// <param name="subSentenceXML">xml-элемент подтекста</param>
/// <param name="language">язык предложения</param>
private static void SubTextFromXElement(SubSentence subSentence, XElement subSentenceXML, string language)
{
int position = _positionInSentence;
var entities = _sentenceEntities;
SubTextInfo subText = new SubTextInfo(
GetSymbolFromXElement(subSentenceXML.Elements().First()),
GetSymbolFromXElement(subSentenceXML.Elements().Last()));
subText.Sentences = TextFromXElement(subSentenceXML).Sentences;
subSentence.SubTextInfo = subText;
_sentenceEntities = entities;
_positionInSentence = position + GetFullValue(subSentenceXML).Length;
}
/// <summary>
/// Пополнение заданной коллекции заданным подпредложением и всеми его подпредложениями-потомками
/// </summary>
/// <param name="parentSubSentence">заданное подпредложение</param>
/// <param name="result">пополняемая коллекция</param>
private void GetFlattenSubSentences(SubSentence parentSubSentence, ref List <SubSentence> result)
{
result.Add(parentSubSentence);
if (parentSubSentence.SubTextInfo == null)
{
var children = ((SubSentence)parentSubSentence).Children;
if (children == null)
{
return;
}
foreach (var subSentence in children)
{
GetFlattenSubSentences(subSentence, ref result);
}
}
}
/// <summary>
/// Создание объекта подпредложения
/// </summary>
/// <param name="parentSentence">родительское предложение</param>
/// <param name="textSentence">текстовое предложение</param>
/// <param name="entities">список сущностей предложения</param>
/// <param name="contentIndexes">индексы контентов в предложении</param>
/// <returns>объект подпредложения</returns>
private SubSentence CreateSubSentenceObject(
Sentence parentSentence,
TextParsing.Sentence textSentence,
IEnumerable <Entity> entities,
IEnumerable <int> contentIndexes)
{
var result = new SubSentence(SubSentenceType.Default, 0)
{
ParentObject = parentSentence
};
var units = GetSubSentenceUnits(result, textSentence, entities);
var subSentences = GetChildContentSubSentences(contentIndexes, units, parentSentence);
result.SetUnits(units);
result.AppendSubSentences(subSentences);
return(result);
}
/// <summary>
/// Получение коллекции элементов подпредложения
/// </summary>
/// <param name="parentSubSentence">объект подпредложения</param>
/// <param name="textSentence">текстовое подпредложение</param>
/// <param name="entities">список сущностей</param>
/// <returns>коллекция элементов предложения</returns>
private List <UnitTextBase> GetSubSentenceUnits(SubSentence parentSubSentence, TextParsing.Sentence textSentence, IEnumerable <Entity> entities)
{
List <UnitTextBase> result = new List <UnitTextBase>();
int currentTextPos = 0;
foreach (Entity entity in entities)
{
entity.MovePosition(textSentence.GetFullStartPosition());
if (entity.PositionInfo.Start != 0)
{
result.Add(SetParentSubSentence(CreateUnmarkedText(textSentence.Text, currentTextPos, entity.PositionInfo.Start), parentSubSentence));
}
result.Add(SetParentSubSentence(entity, parentSubSentence));
currentTextPos = entity.PositionInfo.End;
}
if (currentTextPos < textSentence.Text.Length)
{
result.Add(SetParentSubSentence(CreateUnmarkedText(textSentence.Text, currentTextPos, textSentence.Text.Length), parentSubSentence));
}
return(result);
}
/// <summary>
/// Создание простого подпредложения из xml-элемента
/// </summary>
/// <param name="subSentence">xml-элемент</param>
/// <param name="language">язык предложения</param>
/// <returns>простое подпредложение</returns>
private static void SimpleSubSentenceFromXElement(SubSentence subSentence, XElement subSentenceXML, string language)
{
List <UnitTextBase> units = new List <UnitTextBase>();
List <SubSentence> subSentences = new List <SubSentence>();
foreach (XElement element in subSentenceXML.Elements())
{
if (element.IsName(UnitTextType.U))
{
units.Add(UnmarkedFromXElement(element));
}
else if (element.IsName(UnitTextType.SUB_SENT))
{
subSentences.Add(SubSentenceFromXElement(element, language));
}
else if (IsEntity(element))
{
units.Add(EntityFromXElement(element, language, subSentence.Type));
}
#region [.defense.]
else
{
throw new InvalidOperationException("unknown xml element");
}
#endregion
}
if (subSentenceXML.IsContainAttribute(RDF.Attribute.Subject))
{
subSentence.Subject = _sentenceEntities[subSentenceXML.GetAttributeValue(RDF.Attribute.Subject)];
}
if (subSentenceXML.IsContainAttribute(RDF.Attribute.IsPassiveSubject))
{
subSentence.IsPassiveSubject = bool.Parse(subSentenceXML.GetAttributeValue(RDF.Attribute.IsPassiveSubject));
}
subSentence.AppendSubSentences(subSentences);
subSentence.SetUnits(units);
}
/// <summary>
/// Задание родительского подпредложения для заданного юнита
/// </summary>
/// <param name="unit">юнит</param>
/// <param name="parentSubSentence">родительское предложение</param>
/// <returns>юнит</returns>
private UnitTextBase SetParentSubSentence(UnitTextBase unit, SubSentence parentSubSentence)
{
unit.ParentObject = parentSubSentence;
return(unit);
}
/// <summary>
/// Получение идентификатора заданного подпредложения
/// </summary>
/// <param name="subSentence">подпредложение</param>
/// <returns>идентификатор подпредложения</returns>
public static int GetID(SubSentence subSentence)
{
return(++_currentSubSentenceID);
}