GetParagraphSentences(IList <SentenceItems> paragraphSentenceItems, int paragraphStartPosition)
{
List <TextParsing.Sentence> result = new List <TextParsing.Sentence>();
TextParsing.Sentence parent = null;
foreach (var sentenceItems in paragraphSentenceItems)
{
Entity lastEntity = sentenceItems.Entities.LastOrDefault();
TextParsing.Sentence current = sentenceItems.Sentence;
if (parent != null)
{
parent.AppendNextPart(current);
current = parent;
parent = null;
}
int sentenceEndPosition = paragraphStartPosition + current.StartPosition + current.Text.Length;
if ((lastEntity != null) && (sentenceEndPosition < lastEntity.PositionInfo.End))
{
parent = current;
}
else
{
result.Add(current);
}
}
return(new Tuple <TextParsing.Sentence[], TextParsing.Sentence>(result.ToArray(), parent));
}
public SentenceItems(TextParsing.Sentence sentence, Entity[] entities, int contentIndex)
: this()
{
Sentence = sentence;
Entities = entities;
ContentIndex = contentIndex;
}
/// <summary>
/// Задание элементов предложения
/// </summary>
/// <param name="sentence">предложение</param>
/// <param name="entities">сущности</param>
/// <param name="currentContentIndex">индекс, с которого идут контенты данного предложения</param>
/// <returns>элементы предложения</returns>
private SentenceItems SetSentenceItems(TextParsing.Sentence sentence, IList <Entity> entities, ref int currentContentIndex)
{
var result = new SentenceItems(sentence, entities.ToArray(), currentContentIndex);
GetContentIndexesInsideSentence(sentence, ref currentContentIndex);
entities.Clear();
return(result);
}
/// <summary>
/// Получение списка индексов контентов начиная с текущего, которые лежат в заданном предложении
/// </summary>
/// <param name="textSentence">текстовое предложение</param>
/// <param name="currentContentIndex">индекс текущего контента</param>
/// <param name="sentenceShift">сдвиг предложения</param>
/// <returns>список индексов контентов в заданном предложении</returns>
private List <int> GetContentIndexesInsideSentence(TextParsing.Sentence textSentence, ref int currentContentIndex, int sentenceShift = 0)
{
List <int> result = new List <int>();
int sentenceStart = textSentence.GetFullStartPosition() + sentenceShift;
int sentenceEnd = textSentence.GetFullEndPosition() + sentenceShift;
while (IsSentenceContainContent(currentContentIndex, textSentence.Text, sentenceStart, sentenceEnd))
{
result.Add(currentContentIndex);
++currentContentIndex;
}
return(result);
}
/// <summary>
/// Получение текста предложения с дочерними частями
/// </summary>
/// <param name="sentence">предложение</param>
/// <param name="startPosition">начальная позиция предложения</param>
/// <param name="isOnlyPotencial">флаг, что нужно выбирать только потенциальные части</param>
/// <param name="contentIndex">индекс текущего контента</param>
/// <returns>текст предложения с дочерними частями</returns>
private string GetSentenceWithChildrenParts(TextParsing.Sentence sentence, int startPosition, bool isOnlyPotencial, ref int contentIndex)
{
StringBuilder result = new StringBuilder(sentence.Text);
var indexes = GetContentIndexesInsideSentence(sentence, ref contentIndex);
int partsLength = 0;
foreach (int index in indexes)
{
ChildContent child = Children[index];
if (!isOnlyPotencial || child.IsPotencialParentPart)
{
result.Insert(GetContentCorrectedStart(index) - startPosition + partsLength, child.GetTextWithChildrenParts(isOnlyPotencial));
partsLength += child.GetFullLength();
}
}
return(result.ToString());
}
/// <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 нового параграфа
/// </summary>
/// <param name="sentenceItemsCollection">коллекция элементов предложений</param>
/// <param name="startPosition">начальная позиция</param>
/// <param name="contentInfo">информация о корректируемом контенте</param>
/// <returns>новый параграф</returns>
private Paragraph GetNewParagraph(IEnumerable <SentenceItems> sentenceItemsCollection,
int startPosition, CorrectedContentInfo contentInfo)
{
List <TextParsing.Sentence> sentences = new List <TextParsing.Sentence>();
int sentenceStartPosition = 0;
foreach (var sentenceItems in sentenceItemsCollection)
{
var sentence = new TextParsing.Sentence(GetNewSentenceText(sentenceItems, contentInfo), sentenceStartPosition);
sentences.Add(sentence);
sentenceStartPosition = sentence.EndPosition;
}
Paragraph result = new Paragraph(sentences.GetText(), startPosition);
result.SetSentences(sentences);
return(result);
}
/// <summary>
/// Создание объекта предложения
/// </summary>
/// <param name="textSentence">текстовое предложение</param>
/// <param name="entities">список сущностей предложения</param>
/// <param name="currentContentIndex">индекс текущего контента</param>
/// <returns>объект предложения</returns>
private Sentence CreateSentenceObject(TextParsing.Sentence textSentence, IEnumerable <Entity> entities, ref int currentContentIndex)
{
var result = new Sentence(textSentence.GetFullStartPosition() + GetFullContentsLength(currentContentIndex));
if (entities.Any())
{
var entityWithLang = entities.FirstOrDefault(_ => !string.IsNullOrEmpty(_.Language));
if (entityWithLang != null)
{
result.Language = entityWithLang.Language;
}
else
{
result.Language = "EN";
}
}
var subSentence = CreateSubSentenceObject(result, textSentence, entities, GetContentIndexesInsideSentence(textSentence, ref currentContentIndex));
result.SetSubSentencesHeirarchy(new SubSentence[] { subSentence });
return(result);
}
/// <summary>
/// Проставка для предложений статуса обработки
/// </summary>
//---WTF-?!?!?!---//
/*
* public void SetSentenceProcessedStatus()
* {
*
* foreach ( var paragraph in _paragraphs )
* {
* foreach ( var sentence in paragraph.Sentences )
* {
* if ( GarbageSentenceSelector.IsGarbage( sentence.Text ) )
* {
* sentence.IsProcessed = false;
* }
* }
* }
*
* }*/
/// <summary>
/// Объединение предложений по сущностям
/// </summary>
/// <param name="entities">сущности</param>
public void UnionSentencesByEntities(IEnumerable <Entity> entities)
{
var sentenceItems = GetSentencesItems(entities);
int sentenceIndex = 0;
List <Paragraph> paragraphs = new List <Paragraph>(_paragraphs.Length);
TextParsing.Sentence nextParagraphSentence = null;
for (int i = 0; i < _paragraphs.Length; ++i)
{
var paragraphSentenceItems = sentenceItems.GetRange(sentenceIndex, _paragraphs[i].Sentences.Length);
sentenceIndex += _paragraphs[i].Sentences.Length;
var sentences = GetParagraphSentences(paragraphSentenceItems, _paragraphs[i].StartPosition);
var paragraph = GetParagraph(_paragraphs[i], nextParagraphSentence, sentences.Item1, sentences.Item2);
nextParagraphSentence = sentences.Item2;
if (paragraph != null)
{
paragraphs.Add(paragraph);
}
}
_paragraphs = paragraphs.ToArray();
}
/// <summary>
/// Получение параграфа
/// </summary>
/// <param name="paragraph">текущий параграф</param>
/// <param name="firstSentence">первое предложение текущего параграфа (наследуется от предыдущего)</param>
/// <param name="sentences">предложения текущего параграфа</param>
/// <param name="nextParagraphSentence">предложение следующего параграфа</param>
/// <returns>параграф</returns>
private Paragraph GetParagraph(
Paragraph paragraph,
TextParsing.Sentence firstSentence,
TextParsing.Sentence[] sentences,
TextParsing.Sentence nextParagraphSentence)
{
Paragraph result = paragraph;
if ((firstSentence != null) || (nextParagraphSentence != null))
{
int startPosition = paragraph.StartPosition;
string paragraphText = string.Empty;
if (firstSentence != null)
{
if (sentences.Any())
{
sentences[0].InsertPreviousPart(firstSentence);
startPosition = firstSentence.Parent.StartPosition + firstSentence.StartPosition;
}
else
{
nextParagraphSentence.InsertPreviousPart(firstSentence);
}
}
if (!sentences.Any())
{
result = null;
}
else
{
result = new Paragraph(sentences.GetText(), startPosition);
}
}
if (result != null)
{
result.SetSentences(sentences);
}
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);
}
