public static LuceneDocument ToLucene(this MarcellDocument sourceDocument)
{
LuceneDocument result = new LuceneDocument();
//Add basic document data
result.AddStringField("Id", sourceDocument.Id ?? sourceDocument.InternalId.ToString("N"), Field.Store.YES);
result.AddStringField("InternalId", sourceDocument.InternalId.ToString("N"), Field.Store.YES);
result.AddDateField("ApprovalDate", sourceDocument.ApprovalDate);
result.AddDateField("DocumentDate", sourceDocument.DocumentDate);
result.AddDateField("EffectiveDate", sourceDocument.EffectiveDate);
result.AddStringList("DocumentToken", sourceDocument.DocumentSimilarityData.ConsolidatedTokens);
result.AddStringList("DocumentTopic", sourceDocument.DocumentSimilarityData.ConsolidatedTopics);
result.AddInt32Field("TokenCount", sourceDocument.TokenCount, Field.Store.YES);
result.AddTextField("DocumentType", sourceDocument.DocumentType ?? "", Field.Store.YES);
result.AddTextField("OriginalType", sourceDocument.OriginalType ?? "", Field.Store.YES);
result.AddTextField("Issuer", sourceDocument.Issuer ?? "", Field.Store.YES);
result.AddStringField("Language", sourceDocument.Language, Field.Store.YES);
result.AddStringField("Url", sourceDocument.Url ?? "", Field.Store.YES);
result.AddScoredDoubleField("RecognitionQuality", sourceDocument.RecognitionQuality);
result.AddBoolField("IsStructured", sourceDocument.IsStructured);
if (sourceDocument.FileName != null)
{
result.AddStringField("FileName", sourceDocument.FileName, Field.Store.YES);
}
return(result);
}
public static LuceneDocument ToLucene(this Section sourceSection, MarcellDocument parentDocument)
{
LuceneDocument result = new LuceneDocument();
//Add basic document data
result.AddStringField("Id", sourceSection.Id ?? sourceSection.InternalId.ToString("N"), Field.Store.YES);
result.AddStringField("InternalId", sourceSection.InternalId.ToString("N"), Field.Store.YES);
result.AddStringList("DocumentToken", sourceSection.DocumentSimilarityData.ConsolidatedTokens);
result.AddStringList("DocumentTopic", sourceSection.DocumentSimilarityData.ConsolidatedTopics);
result.AddStringList("SectionToken", sourceSection.SectionSimilarityData.ConsolidatedTokens);
result.AddStringList("SectionTopic", sourceSection.SectionSimilarityData.ConsolidatedTopics);
result.AddInt32Field("TokenCount", sourceSection.TokenCount, Field.Store.YES);
result.AddStringField("Language", sourceSection.Language, Field.Store.YES);
result.AddScoredDoubleField("RecognitionQuality", sourceSection.RecognitionQuality);
result.AddStringField("Type", sourceSection.Type.ToString(), Field.Store.YES);
string sectionText = sourceSection.Text ?? "";
result.AddTextField("Text", sectionText, Field.Store.YES);
//Add reference to parent document
if (parentDocument != null)
{
result.AddStringField("ParentDocumentId", parentDocument.InternalId.ToString("N"), Field.Store.YES);
}
return(result);
}
public static LuceneDocument ToLucene(this Sentence sourceSentence, MarcellDocument parentDocument, Section parentSection, Paragraph parentParagraph)
{
LuceneDocument result = new LuceneDocument();
//Add basic document data
result.AddStringField("Id", sourceSentence.Id ?? sourceSentence.InternalId.ToString("N"), Field.Store.YES);
result.AddStringField("InternalId", sourceSentence.InternalId.ToString("N"), Field.Store.YES);
result.AddStringList("DocumentToken", sourceSentence.DocumentSimilarityData.ConsolidatedTokens);
result.AddStringList("DocumentTopic", sourceSentence.DocumentSimilarityData.ConsolidatedTopics);
result.AddStringList("SectionToken", sourceSentence.SectionSimilarityData.ConsolidatedTokens);
result.AddStringList("SectionTopic", sourceSentence.SectionSimilarityData.ConsolidatedTopics);
result.AddStringList("ParagraphToken", sourceSentence.ParagraphSimilarityData.ConsolidatedTokens);
result.AddStringList("ParagraphTopic", sourceSentence.ParagraphSimilarityData.ConsolidatedTopics);
result.AddStringList("SentenceToken", sourceSentence.SentenceSimilarityData.ConsolidatedTokens);
result.AddStringList("SentenceTopic", sourceSentence.SentenceSimilarityData.ConsolidatedTopics);
result.AddInt32Field("TokenCount", sourceSentence.TokenCount, Field.Store.YES);
result.AddStringField("Language", sourceSentence.Language, Field.Store.YES);
result.AddScoredDoubleField("RecognitionQuality", sourceSentence.RecognitionQuality);
result.AddInt32Field("Order", sourceSentence.Order, Field.Store.YES);
string sentencehText = sourceSentence.Text ?? "";
result.AddTextField("Text", sentencehText, Field.Store.YES);
//Add reference to parent document
if (parentDocument != null)
{
result.AddStringField("ParentDocumentId", parentDocument.InternalId.ToString("N"), Field.Store.YES);
}
//Add reference to parent section
if (parentSection != null)
{
result.AddStringField("ParentSectionId", parentSection.InternalId.ToString("N"), Field.Store.YES);
}
//Add reference to parent section
if (parentParagraph != null)
{
result.AddStringField("ParentParagraphId", parentParagraph.InternalId.ToString("N"), Field.Store.YES);
}
result.AddStringList("ContainedTokenEV", sourceSentence.Tokens.SelectMany(t => t.EuroVocEntities), Field.Store.NO);
result.AddStringList("ContainedTokenIATE", sourceSentence.Tokens.SelectMany(t => t.IateEntities), Field.Store.NO);
return(result);
}
public static LuceneDocument ToLucene(this Paragraph sourceParagraph, MarcellDocument parentDocument, Section parentSection)
{
LuceneDocument result = new LuceneDocument();
//Add basic document data
result.AddStringField("Id", sourceParagraph.Id ?? sourceParagraph.InternalId.ToString("N"), Field.Store.YES);
result.AddStringField("InternalId", sourceParagraph.InternalId.ToString("N"), Field.Store.YES);
result.AddStringList("DocumentToken", sourceParagraph.DocumentSimilarityData.ConsolidatedTokens);
result.AddStringList("DocumentTopic", sourceParagraph.DocumentSimilarityData.ConsolidatedTopics);
result.AddStringList("SectionToken", sourceParagraph.SectionSimilarityData.ConsolidatedTokens);
result.AddStringList("SectionTopic", sourceParagraph.SectionSimilarityData.ConsolidatedTopics);
result.AddStringList("ParagraphToken", sourceParagraph.ParagraphSimilarityData.ConsolidatedTokens);
result.AddStringList("ParagraphTopic", sourceParagraph.ParagraphSimilarityData.ConsolidatedTopics);
result.AddInt32Field("TokenCount", sourceParagraph.TokenCount, Field.Store.YES);
result.AddStringField("Language", sourceParagraph.Language, Field.Store.YES);
result.AddScoredDoubleField("RecognitionQuality", sourceParagraph.RecognitionQuality);
result.AddInt32Field("Order", sourceParagraph.Order, Field.Store.YES);
result.AddStringField("ParagraphType", sourceParagraph.ParagraphType.ToString(), Field.Store.YES);
result.AddStringField("ParagraphNumber", sourceParagraph.ParagraphNumber ?? "", Field.Store.YES);
result.AddStringField("PointNumber", sourceParagraph.PointNumber ?? "", Field.Store.YES);
string paragraphText = sourceParagraph.Text ?? "";
result.AddTextField("Text", paragraphText, Field.Store.YES);
//Add reference to parent document
if (parentDocument != null)
{
result.AddStringField("ParentDocumentId", parentDocument.InternalId.ToString("N"), Field.Store.YES);
}
//Add reference to parent section
if (parentSection != null)
{
result.AddStringField("ParentSectionId", parentSection.InternalId.ToString("N"), Field.Store.YES);
}
return(result);
}