[Treebase-guts] SF.net SVN: treebase:[73] trunk/treebase-core/src/main/java/org/cipres/ treebase/domain/nexus/nexml

[<rv...@us...>]

Revision: 73
          http://treebase.svn.sourceforge.net/treebase/?rev=73&view=rev
Author:   rvos
Date:     2009-06-23 10:52:08 +0000 (Tue, 23 Jun 2009)

Log Message:
-----------
Now emit valid nexml.

Modified Paths:
--------------
    trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlDocumentConverter.java
    trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlMatrixConverter.java
    trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlOTUConverter.java
    trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlObjectConverter.java
    trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlTreeBlockConverter.java

Modified: trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlDocumentConverter.java
===================================================================
--- trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlDocumentConverter.java	2009-06-23 10:50:48 UTC (rev 72)
+++ trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlDocumentConverter.java	2009-06-23 10:52:08 UTC (rev 73)
@@ -1,6 +1,7 @@
 package org.cipres.treebase.domain.nexus.nexml;
 
 import org.cipres.treebase.domain.admin.Person;
+import org.cipres.treebase.domain.matrix.CharacterMatrix;
 import org.cipres.treebase.domain.nexus.NexusDataSet;
 import org.cipres.treebase.domain.study.ArticleCitation;
 import org.cipres.treebase.domain.study.Citation;
@@ -69,7 +70,9 @@
 		
 		NexmlMatrixConverter nmc = new NexmlMatrixConverter(getStudy(),getTaxonLabelHome(),getDocument());
 		for (org.cipres.treebase.domain.matrix.Matrix matrix : pNexusDataSet.getMatrices() ) {
-			nmc.fromTreeBaseToXml(matrix);
+			if ( matrix instanceof CharacterMatrix ) {
+				nmc.fromTreeBaseToXml((CharacterMatrix)matrix);
+			}
 		}
 		
 		NexmlTreeBlockConverter ntc = new NexmlTreeBlockConverter(getStudy(),getTaxonLabelHome(),getDocument());
@@ -87,22 +90,40 @@
 	 * @param document
 	 */
 	private void copyCitationMetadata(Citation citation,Document document) {
-		attachAnnotation("prism:publicationDate",citation.getPublishYear().toString(),mPrismURI,document);
-		attachAnnotation("prism:doi",citation.getDoi(),mPrismURI,document);
-		String[] pages = citation.getPages().split("-");
-		if ( pages.length > 2 ) {
-			attachAnnotation("prism:startingPage",pages[0],mPrismURI,document);
-			attachAnnotation("prism:endingPage",pages[1],mPrismURI,document);
-			attachAnnotation("prism:pageRange",citation.getPages(),mPrismURI,document);			
+		if ( null != citation.getTitle() ) {
+			attachAnnotation("dc:title",citation.getTitle(),mPrismURI,document);
 		}
-		String[] keywords = citation.getKeywords().split(", ");
-		for ( int i = 0; i < keywords.length; i++ ) {
-			attachAnnotation("prism:keyword",keywords[i],mPrismURI,document);
-		}		
+		if ( null != citation.getPublishYear() ) {
+			attachAnnotation("prism:publicationDate",citation.getPublishYear().toString(),mPrismURI,document);
+		}
+		if ( null != citation.getDoi() ) {
+			attachAnnotation("prism:doi",citation.getDoi(),mPrismURI,document);
+		}
+		if ( null != citation.getPages() ) {
+			String[] pages = citation.getPages().split("\\-");
+			if ( pages.length == 2 ) {
+				attachAnnotation("prism:startingPage",pages[0],mPrismURI,document);
+				attachAnnotation("prism:endingPage",pages[1],mPrismURI,document);
+				attachAnnotation("prism:pageRange",citation.getPages(),mPrismURI,document);			
+			}
+		}
+		if ( null != citation.getKeywords() ) {
+			String[] keywords = citation.getKeywords().split(", ");
+			for ( int i = 0; i < keywords.length; i++ ) {
+				attachAnnotation("prism:keyword",keywords[i],mPrismURI,document);
+			}		
+		}
 		if ( citation instanceof ArticleCitation ) {
-			attachAnnotation("prism:publicationName",((ArticleCitation)citation).getJournal(),mPrismURI,document);
-			attachAnnotation("prism:volume",((ArticleCitation)citation).getVolume(),mPrismURI,document);
-			attachAnnotation("prism:number",((ArticleCitation)citation).getIssue(),mPrismURI,document);
+			ArticleCitation ac = (ArticleCitation)citation;
+			if ( null != ac.getJournal() ) {
+				attachAnnotation("prism:publicationName",ac.getJournal(),mPrismURI,document);
+			}
+			if ( null != ac.getVolume() ) {
+				attachAnnotation("prism:volume",ac.getVolume(),mPrismURI,document);
+			}
+			if ( null != ac.getIssue() ) {
+				attachAnnotation("prism:number",ac.getIssue(),mPrismURI,document);
+			}
 		}		
 	}
 	
@@ -112,22 +133,7 @@
 	 * @return
 	 */
 	public Document fromTreeBaseToXml(Study pStudy) {
-		attachTreeBaseID(getDocument(), pStudy);
-		attachAnnotation("dc:title", pStudy.getName(), mDCURI, getDocument());
-		attachAnnotation("dc:abstract",pStudy.getCitation().getAbstract(), mDCURI,getDocument());		
-		attachAnnotation(
-				"dc:creator",
-				pStudy.getSubmission().getSubmitter().getPerson().getFullNameCitationStyle(),
-				mDCURI,
-				getDocument()
-		);
-		for ( Person person : pStudy.getAuthors() ) {
-			String personName = person.getFullNameCitationStyle();
-			attachAnnotation("dc:contributor",personName,mDCURI,getDocument());
-		}
-		attachAnnotation("prism:creationDate",pStudy.getSubmission().getCreateDate().toString(),mPrismURI,getDocument());
-		attachAnnotation("prism:embargoDate",pStudy.getReleaseDate().toString(),mPrismURI,getDocument());
-		copyCitationMetadata(pStudy.getCitation(),getDocument());		
+		copyMetadata(pStudy);		
 		
 		NexmlOTUConverter noc = new NexmlOTUConverter(getStudy(),getTaxonLabelHome(),getDocument());
 		for ( TaxonLabelSet taxonLabelSet : pStudy.getTaxonLabelSets() ) {
@@ -136,14 +142,48 @@
 		
 		NexmlMatrixConverter nmc = new NexmlMatrixConverter(getStudy(),getTaxonLabelHome(),getDocument());
 		for (org.cipres.treebase.domain.matrix.Matrix matrix : pStudy.getMatrices() ) {
-			nmc.fromTreeBaseToXml(matrix);
+			if ( matrix instanceof CharacterMatrix ) {
+				nmc.fromTreeBaseToXml((CharacterMatrix)matrix);
+			}
 		}
 		
 		NexmlTreeBlockConverter ntc = new NexmlTreeBlockConverter(getStudy(),getTaxonLabelHome(),getDocument());
 		for ( org.cipres.treebase.domain.tree.TreeBlock treeBlock : pStudy.getTreeBlocks() ) {
 			ntc.fromTreeBaseToXML(treeBlock);
 		}		
+		
 		return getDocument();
 	}
 
+	private void copyMetadata(Study pStudy) {
+		attachTreeBaseID(getDocument(), pStudy,Study.class);
+		if ( null != pStudy.getName() ) {
+			attachAnnotation("dc:title", pStudy.getName(), mDCURI, getDocument());
+		}		
+		for ( Person person : pStudy.getAuthors() ) {
+			String personName = person.getFullNameCitationStyle();
+			attachAnnotation("dc:contributor",personName,mDCURI,getDocument());
+		}
+		if ( null != pStudy.getReleaseDate() ) {
+			attachAnnotation("prism:embargoDate",pStudy.getReleaseDate().toString(),mPrismURI,getDocument());
+		}
+		if ( null != pStudy.getSubmission() ) {
+			if ( null != pStudy.getSubmission().getSubmitter() ) {
+				attachAnnotation(
+						"dc:creator",
+						pStudy.getSubmission().getSubmitter().getPerson().getFullNameCitationStyle(),
+						mDCURI,
+						getDocument()
+				);		
+			}
+			if ( null != pStudy.getSubmission().getCreateDate() ) {
+				attachAnnotation("prism:creationDate",pStudy.getSubmission().getCreateDate().toString(),mPrismURI,getDocument());
+			}
+		}		
+		if ( null != pStudy.getCitation() ) {			
+			//attachAnnotation("dc:abstract",forXML(pStudy.getCitation().getAbstract()), mDCURI,getDocument());
+			copyCitationMetadata(pStudy.getCitation(),getDocument());
+		}
+	}
+
 }

Modified: trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlMatrixConverter.java
===================================================================
--- trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlMatrixConverter.java	2009-06-23 10:50:48 UTC (rev 72)
+++ trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlMatrixConverter.java	2009-06-23 10:52:08 UTC (rev 73)
@@ -2,29 +2,45 @@
 
 import java.util.ArrayList;
 import java.util.List;
+
 import org.cipres.treebase.dao.jdbc.ContinuousMatrixElementJDBC;
 import org.cipres.treebase.dao.jdbc.ContinuousMatrixJDBC;
 import org.cipres.treebase.dao.jdbc.DiscreteMatrixElementJDBC;
 import org.cipres.treebase.dao.jdbc.DiscreteMatrixJDBC;
 import org.cipres.treebase.dao.jdbc.MatrixColumnJDBC;
+import org.cipres.treebase.domain.matrix.CharacterMatrix;
 import org.cipres.treebase.domain.matrix.ContinuousChar;
+import org.cipres.treebase.domain.matrix.ContinuousMatrix;
+import org.cipres.treebase.domain.matrix.ContinuousMatrixElement;
 import org.cipres.treebase.domain.matrix.DiscreteChar;
+import org.cipres.treebase.domain.matrix.DiscreteCharState;
 import org.cipres.treebase.domain.matrix.DiscreteMatrix;
+import org.cipres.treebase.domain.matrix.DiscreteMatrixElement;
 import org.cipres.treebase.domain.matrix.Matrix;
+import org.cipres.treebase.domain.matrix.MatrixColumn;
+import org.cipres.treebase.domain.matrix.MatrixDataType;
+import org.cipres.treebase.domain.matrix.MatrixElement;
+import org.cipres.treebase.domain.matrix.MatrixRow;
+import org.cipres.treebase.domain.matrix.PhyloChar;
+import org.cipres.treebase.domain.matrix.StandardMatrix;
 import org.cipres.treebase.domain.study.Study;
 import org.cipres.treebase.domain.taxon.TaxonLabelHome;
 import org.cipres.treebase.domain.taxon.TaxonLabelSet;
 import org.nexml.model.Annotatable;
 import org.nexml.model.CategoricalMatrix;
 import org.nexml.model.CharacterState;
-import org.nexml.model.ContinuousMatrix;
+import org.nexml.model.CharacterStateSet;
 import org.nexml.model.Document;
 import org.nexml.model.MatrixCell;
+import org.nexml.model.MolecularMatrix;
 import org.nexml.model.OTUs;
 import org.nexml.model.OTU;
 
 public class NexmlMatrixConverter extends NexmlObjectConverter {
 
+	private static final int MAX_GRANULAR_NCHAR = 1000;
+	private static final int MAX_GRANULAR_NTAX = 30;
+
 	/**
 	 * 
 	 * @param study
@@ -56,6 +72,7 @@
 			columnJDBCs.add(aColumnJDBC);
 			int rowIndex = 0;
 			for ( OTU xmlOTU : xmlMatrix.getOTUs().getAllOTUs() ) {
+				@SuppressWarnings("unused")
 				MatrixCell<CharacterState> xmlCell = xmlMatrix.getCell(xmlOTU, xmlCharacter);
 				DiscreteMatrixElementJDBC element = new DiscreteMatrixElementJDBC();
 				//element.setValue(xmlCell.getValue()); // XXX nested stateset lookup song & dance here
@@ -76,9 +93,8 @@
 	 * @param xmlMatrix
 	 * @return
 	 */
-	public Matrix fromXmlToTreeBase(ContinuousMatrix xmlMatrix) {		
-		org.cipres.treebase.domain.matrix.ContinuousMatrix tbMatrix = 
-			new org.cipres.treebase.domain.matrix.ContinuousMatrix();		
+	public Matrix fromXmlToTreeBase(org.nexml.model.ContinuousMatrix xmlMatrix) {		
+		ContinuousMatrix tbMatrix = new ContinuousMatrix();		
 		ContinuousMatrixJDBC matrixJDBC = new ContinuousMatrixJDBC(tbMatrix, xmlMatrix, this);		
 		List<MatrixColumnJDBC> columnJDBCs = new ArrayList<MatrixColumnJDBC>();
 		long[] colIds = matrixJDBC.getColIDs();
@@ -119,7 +135,7 @@
 		OTUs xmlOTUs = xmlMatrix.getOTUs();
 		Matrix tbMatrix = null;
 		TaxonLabelSet tbTaxa = null;
-		Long tbTaxonLabelSetID = readTreeBaseID((Annotatable) xmlOTUs,TaxonLabelSet.class);
+		Long tbTaxonLabelSetID = readTreeBaseID((Annotatable) xmlOTUs);
 		if ( null != tbTaxonLabelSetID ) {
 			tbTaxa = getTaxonLabelHome()
 				.findPersistedObjectByID(TaxonLabelSet.class, tbTaxonLabelSetID);
@@ -127,11 +143,11 @@
 		if ( xmlMatrix instanceof CategoricalMatrix ) {
 			tbMatrix = fromXmlToTreeBase((CategoricalMatrix) xmlMatrix);
 		}
-		else if ( xmlMatrix instanceof ContinuousMatrix ) {
-			tbMatrix = fromXmlToTreeBase((ContinuousMatrix) xmlMatrix);
+		else if ( xmlMatrix instanceof org.nexml.model.ContinuousMatrix ) {
+			tbMatrix = fromXmlToTreeBase((org.nexml.model.ContinuousMatrix) xmlMatrix);
 		}
 		if ( null != tbMatrix ) {
-			attachTreeBaseID((Annotatable) xmlMatrix, tbMatrix);
+			attachTreeBaseID((Annotatable) xmlMatrix, tbMatrix,Matrix.class);
 			if ( null != tbTaxa ) {
 				tbMatrix.setTaxa(tbTaxa);
 			}
@@ -142,21 +158,192 @@
 		return tbMatrix;		
 	}
 	
-	public org.nexml.model.Matrix<?> fromTreeBaseToXml(Matrix tbMatrix) {
-		TaxonLabelSet taxonLabelSet = tbMatrix.getTaxa();
-		Long taxonLabelSetId = taxonLabelSet.getId();
-		OTUs xmlOTUs = null;
-		for ( OTUs otus : getDocument().getOTUsList() ) {
-			Long annotatedID = readTreeBaseID(otus, TaxonLabelSet.class);
-			if ( taxonLabelSetId == annotatedID ) {
-				xmlOTUs = otus;
-				break;
+	/**
+	 * Creates and populates characters (i.e. columns) with their annotations,
+	 * and state sets, with their annotations
+	 * 
+	 * @param tbMatrix
+	 * @return an xml matrix with empty rows
+	 */
+	public CategoricalMatrix fromTreeBaseToXml(StandardMatrix tbMatrix) {
+		OTUs xmlOTUs = getOTUsById(tbMatrix.getTaxa().getId());
+		CategoricalMatrix xmlMatrix = getDocument().createCategoricalMatrix(xmlOTUs);
+		List<List<DiscreteCharState>> tbStateLabels = tbMatrix.getStateLabels();
+		List<MatrixColumn> tbColumns = tbMatrix.getColumnsReadOnly();
+		for ( int i = 0; i < tbColumns.size(); i++ ) {
+			CharacterStateSet xmlStateSet = xmlMatrix.createCharacterStateSet();
+			for ( DiscreteCharState tbState : tbStateLabels.get(i) ) {
+				CharacterState xmlState = xmlStateSet.createCharacterState(tbState.getSymbol().toString());
+				if ( null != tbState.getDescription() ) {
+					xmlState.setLabel(tbState.getDescription());
+				}	
+				if ( null != tbState.getNotes() ) {
+					((Annotatable)xmlState).addAnnotationValue("dc:description", mDCURI, tbState.getNotes());
+				}
+				attachTreeBaseID((Annotatable)xmlState,tbState,DiscreteCharState.class);
+			}			
+			org.nexml.model.Character xmlCharacter = xmlMatrix.createCharacter(xmlStateSet);
+			PhyloChar tbCharacter = tbColumns.get(i).getCharacter();
+			if ( null != tbCharacter.getDescription() ) {
+				xmlCharacter.setLabel(tbCharacter.getDescription());
+			}			
+			attachTreeBaseID((Annotatable)xmlCharacter,tbCharacter,PhyloChar.class);
+		}		
+		return xmlMatrix;
+	}
+	
+	/**
+	 * Creates and populates characters (i.e. columns) with their annotations,
+	 * and state sets, with their annotations
+	 * 
+	 * @param tbMatrix
+	 * @return an xml matrix with empty rows
+	 */	
+	public MolecularMatrix fromTreeBaseToXml(DiscreteMatrix tbMatrix) {
+		OTUs xmlOTUs = getOTUsById(tbMatrix.getTaxa().getId());
+		String tbDataType = tbMatrix.getDataType().getDescription();
+		MolecularMatrix xmlMatrix = null;
+		CharacterStateSet xmlStateSet = null;
+		if ( tbDataType.equals(MatrixDataType.MATRIX_DATATYPE_DNA) ) {
+			xmlMatrix = getDocument().createMolecularMatrix(xmlOTUs, MolecularMatrix.DNA);
+			xmlStateSet = ((MolecularMatrix)xmlMatrix).getDNACharacterStateSet();
+		}
+		else if ( tbDataType.equals(MatrixDataType.MATRIX_DATATYPE_RNA) ) {
+			xmlMatrix = getDocument().createMolecularMatrix(xmlOTUs, MolecularMatrix.RNA);
+			xmlStateSet = ((MolecularMatrix)xmlMatrix).getRNACharacterStateSet();
+		}
+		else if ( tbDataType.equals(MatrixDataType.MATRIX_DATATYPE_PROTEIN) ) {
+			xmlMatrix = getDocument().createMolecularMatrix(xmlOTUs, MolecularMatrix.Protein);
+			xmlStateSet = ((MolecularMatrix)xmlMatrix).getProteinCharacterStateSet();
+		}
+		for ( MatrixColumn tbColumn : tbMatrix.getColumnsReadOnly() ) {
+			org.nexml.model.Character xmlCharacter = xmlMatrix.createCharacter(xmlStateSet);
+			PhyloChar tbCharacter = tbColumn.getCharacter();
+			if ( null != tbCharacter.getDescription() && ! tbCharacter.getDescription().equals(tbDataType) ) {
+				((Annotatable)xmlCharacter).addAnnotationValue("dc:description", mDCURI, tbCharacter.getDescription());
+			}			
+			if ( null != tbCharacter.getId() && tbCharacter.getId() != 2 ) { // XXX is PhyloChar.id 2 some sort of magic number?
+				attachTreeBaseID((Annotatable)xmlCharacter,tbCharacter,PhyloChar.class);
 			}
 		}
-		org.nexml.model.Matrix<?> xmlMatrix = getDocument().createCategoricalMatrix(xmlOTUs);
+		return xmlMatrix;
+	}
+	
+	/**
+	 * Creates and populates characters (i.e. columns) with their annotations,
+	 * and state sets, with their annotations
+	 * 
+	 * @param tbMatrix
+	 * @return an xml matrix with empty rows
+	 */		
+	public org.nexml.model.ContinuousMatrix fromTreeBaseToXml(ContinuousMatrix tbMatrix) {
+		OTUs xmlOTUs = getOTUsById(tbMatrix.getTaxa().getId());
+		org.nexml.model.ContinuousMatrix xmlMatrix = getDocument().createContinuousMatrix(xmlOTUs);
+		for ( MatrixColumn tbColumn : tbMatrix.getColumnsReadOnly() ) {
+			org.nexml.model.Character xmlCharacter = xmlMatrix.createCharacter();
+			PhyloChar tbCharacter = tbColumn.getCharacter();
+			if ( null != tbCharacter.getDescription() ) {
+				((Annotatable)xmlCharacter).addAnnotationValue("dc:description", mDCURI, tbCharacter.getDescription());
+			}			
+			attachTreeBaseID((Annotatable)xmlCharacter,tbCharacter,PhyloChar.class);			
+		}		
+		return xmlMatrix;
+	}
+	
+
+// XXX doesn't handle the following data types: 	
+//	public static final String MATRIX_DATATYPE_NUCLEOTIDE = "Nucleotide";
+//	public static final String MATRIX_DATATYPE_DISTANCE = "Distance";
+//	public static final String MATRIX_DATATYPE_MIXED = "Mixed";		
+	@SuppressWarnings("unchecked")
+	public org.nexml.model.Matrix<?> fromTreeBaseToXml(CharacterMatrix tbMatrix) {
+		org.nexml.model.Matrix<?> xmlMatrix = null;
+		if ( tbMatrix instanceof DiscreteMatrix ) {
+			if ( tbMatrix.getDataType().getDescription().equals(MatrixDataType.MATRIX_DATATYPE_STANDARD) ) {
+				xmlMatrix = fromTreeBaseToXml((StandardMatrix) tbMatrix);
+			}
+			else {
+				xmlMatrix = fromTreeBaseToXml((DiscreteMatrix) tbMatrix);				
+			}
+			populateXmlMatrix((org.nexml.model.Matrix<CharacterState>)xmlMatrix,(DiscreteMatrix)tbMatrix);
+		}
+		else if ( tbMatrix instanceof ContinuousMatrix ) {
+			xmlMatrix = fromTreeBaseToXml((ContinuousMatrix) tbMatrix);			
+			populateXmlMatrix((org.nexml.model.ContinuousMatrix)xmlMatrix,(ContinuousMatrix)tbMatrix);
+		}
 		xmlMatrix.setLabel(tbMatrix.getTitle());
-		attachTreeBaseID((Annotatable)xmlMatrix, tbMatrix);
+		attachTreeBaseID((Annotatable)xmlMatrix, tbMatrix,Matrix.class);
 		return xmlMatrix;
 	}
 
+	/**
+	 * 
+	 * @param xmlMatrix
+	 * @param tbMatrix
+	 * @param xmlOTUs
+	 * @param stateSet
+	 */
+	private void populateXmlMatrix(
+			org.nexml.model.Matrix<CharacterState> xmlMatrix,
+			DiscreteMatrix tbMatrix) {	
+		OTUs xmlOTUs = xmlMatrix.getOTUs();
+		List<org.nexml.model.Character> characterList = xmlMatrix.getCharacters();
+		for ( MatrixRow row : tbMatrix.getRowsReadOnly() ) {
+			OTU xmlOTU = getOTUById(xmlOTUs, row.getTaxonLabel().getId());
+			int charIndex = 0;
+			if ( characterList.size() <= MAX_GRANULAR_NCHAR && xmlOTUs.getAllOTUs().size() <= MAX_GRANULAR_NTAX ) {
+				for ( MatrixElement tbCell : row.getElements() ) {
+					org.nexml.model.Character xmlCharacter = characterList.get(charIndex);
+					MatrixCell<CharacterState> xmlCell = xmlMatrix.getCell(xmlOTU, xmlCharacter);				
+					DiscreteCharState tbState = ((DiscreteMatrixElement)tbCell).getCharState();
+					String tbSymbolString = ( null == tbState ) ? "?" : tbState.getSymbol().toString();
+					CharacterState xmlState = xmlCharacter.getCharacterStateSet().lookupCharacterStateBySymbol(tbSymbolString);								
+					xmlCell.setValue(xmlState);								
+					attachTreeBaseID((Annotatable)xmlCell,tbCell,DiscreteMatrixElement.class);
+					charIndex++;
+				}
+			}
+			else {
+				String seq = row.buildElementAsString();
+				if ( tbMatrix.getDataType().getDescription().equals(MatrixDataType.MATRIX_DATATYPE_STANDARD) ) {
+					StringBuilder sb = new StringBuilder();
+					for ( int i = 0; i < seq.length(); i++ ) {
+						sb.append(seq.charAt(i));
+						if ( i < seq.length() - 1 ) {
+							sb.append(' ');
+						}
+					}
+				}				
+				xmlMatrix.setSeq(seq,xmlOTU);
+			}
+		}	
+	}
+	
+	/**
+	 * 
+	 * @param xmlMatrix
+	 * @param tbMatrix
+	 */
+	private void populateXmlMatrix(org.nexml.model.ContinuousMatrix xmlMatrix,
+			ContinuousMatrix tbMatrix) {
+		List<org.nexml.model.Character> characterList = xmlMatrix.getCharacters();
+		OTUs xmlOTUs = xmlMatrix.getOTUs();
+		for ( MatrixRow row : tbMatrix.getRowsReadOnly() ) {
+			List<MatrixElement> elements = row.getElements();
+			OTU xmlOTU = getOTUById(xmlOTUs, row.getTaxonLabel().getId());
+			if ( characterList.size() <= MAX_GRANULAR_NCHAR && xmlOTUs.getAllOTUs().size() <= MAX_GRANULAR_NTAX ) {
+				for ( int elementIndex = 0; elementIndex < tbMatrix.getnChar(); elementIndex++ ) {
+					ContinuousMatrixElement tbCell = (ContinuousMatrixElement)elements.get(elementIndex);
+					MatrixCell<Double> xmlCell = xmlMatrix.getCell(xmlOTU, characterList.get(elementIndex));
+					xmlCell.setValue(tbCell.getValue());
+					attachTreeBaseID((Annotatable)xmlCell,tbCell,DiscreteMatrixElement.class);
+				}
+			}
+			else {
+				String seq = row.buildElementAsString();
+				xmlMatrix.setSeq(seq,xmlOTU);
+			}
+		}		
+	}
+	
 }

Modified: trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlOTUConverter.java
===================================================================
--- trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlOTUConverter.java	2009-06-23 10:50:48 UTC (rev 72)
+++ trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlOTUConverter.java	2009-06-23 10:52:08 UTC (rev 73)
@@ -27,9 +27,11 @@
 	 */
 	public TaxonLabelSet fromXmlToTreeBase (OTUs xmlOTUs) {
 		TaxonLabelSet labelSet = new TaxonLabelSet();
-		attachTreeBaseID(xmlOTUs,labelSet);
+		attachTreeBaseID(xmlOTUs,labelSet,TaxonLabelSet.class);
 		labelSet.setStudy(getStudy());
-		labelSet.setTitle(xmlOTUs.getLabel());
+		if ( null != xmlOTUs.getLabel() ) {
+			labelSet.setTitle(xmlOTUs.getLabel());
+		}
 		labelSet.setTaxa(true);
 		for ( OTU xmlOTU : xmlOTUs.getAllOTUs() ) {
 			TaxonLabel taxonLabel = fromXmlToTreeBase(xmlOTU);
@@ -46,7 +48,7 @@
 	public OTUs fromTreeBaseToXml(TaxonLabelSet taxonLabelSet) {
 		OTUs xmlOTUs = getDocument().createOTUs();
 		xmlOTUs.setLabel(taxonLabelSet.getTitle());
-		attachTreeBaseID(xmlOTUs,taxonLabelSet);
+		attachTreeBaseID(xmlOTUs,taxonLabelSet,TaxonLabelSet.class);
 		for ( TaxonLabel taxonLabel : taxonLabelSet.getTaxonLabelsReadOnly() ) {
 			fromTreeBaseToXml(taxonLabel,xmlOTUs);
 		}
@@ -61,7 +63,7 @@
 	public TaxonLabel fromXmlToTreeBase(OTU xmlOTU) {
 		TaxonLabel taxonLabel = getTaxonLabelHome().getByDescriptionAndStudy(xmlOTU.getLabel(), getStudy());
 		taxonLabel.setStudy(getStudy());
-		attachTreeBaseID(xmlOTU,taxonLabel);
+		attachTreeBaseID(xmlOTU,taxonLabel,TaxonLabel.class);
 		return taxonLabel;		
 	}
 	
@@ -73,8 +75,10 @@
 	 */
 	public OTU fromTreeBaseToXml(TaxonLabel taxonLabel,OTUs xmlOTUs) {
 		OTU xmlOTU = xmlOTUs.createOTU();
-		xmlOTU.setLabel(taxonLabel.getTaxonLabel());
-		attachTreeBaseID(xmlOTU,taxonLabel);
+		if ( null != taxonLabel.getTaxonLabel() ) {
+			xmlOTU.setLabel(taxonLabel.getTaxonLabel());
+		}
+		attachTreeBaseID(xmlOTU,taxonLabel,TaxonLabel.class);
 		if ( null != taxonLabel.getNcbiTaxID() ) {
 			attachAnnotation("dc:identifier", "NCBI:" + taxonLabel.getNcbiTaxID(), mDCURI, xmlOTU);
 		}

Modified: trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlObjectConverter.java
===================================================================
--- trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlObjectConverter.java	2009-06-23 10:50:48 UTC (rev 72)
+++ trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlObjectConverter.java	2009-06-23 10:52:08 UTC (rev 73)
@@ -2,16 +2,22 @@
 
 import java.net.URI;
 import java.net.URISyntaxException;
+import java.text.CharacterIterator;
+import java.text.StringCharacterIterator;
 import java.util.Iterator;
 import java.util.Set;
 
 import org.cipres.treebase.TreebaseIDString;
+import org.cipres.treebase.NamespacedGUID;
+import org.cipres.treebase.TreebaseIDString.MalformedTreebaseIDString;
 import org.cipres.treebase.domain.TBPersistable;
 import org.cipres.treebase.domain.nexus.AbstractNexusConverter;
 import org.cipres.treebase.domain.study.Study;
 import org.cipres.treebase.domain.taxon.TaxonLabelHome;
 import org.nexml.model.Annotatable;
 import org.nexml.model.Document;
+import org.nexml.model.OTU;
+import org.nexml.model.OTUs;
 
 public class NexmlObjectConverter extends AbstractNexusConverter {
 	protected static URI mDCURI;
@@ -44,14 +50,9 @@
 	 * @param tbPersistable
 	 * @return
 	 */
-	public String makeNamespacedID (TBPersistable tbPersistable) {
-		StringBuilder sb = new StringBuilder();
-		sb
-			.append(TreeBASE2Prefix)
-			.append(':')
-			.append(TreebaseIDString.getPrefixForClass(tbPersistable.getClass()))
-			.append(tbPersistable.getId());
-		return sb.toString();
+	private String makeNamespacedID (TBPersistable tbPersistable,Class<?> persistableClass) {
+		TreebaseIDString tbIDString = new TreebaseIDString(persistableClass,tbPersistable.getId());
+		return tbIDString.getNamespacedGUID().toString();
 	}
 	
 	/**
@@ -59,8 +60,10 @@
 	 * @param annotatable
 	 * @param tbPersistable
 	 */
-	protected void attachTreeBaseID(Annotatable annotatable,TBPersistable tbPersistable) {
-		attachAnnotation(mDCIdentifier,makeNamespacedID(tbPersistable),mDCURI,annotatable);
+	protected void attachTreeBaseID(Annotatable annotatable,TBPersistable tbPersistable,Class<?> persistableClass) {
+		if ( null != tbPersistable.getId() ) {
+			attachAnnotation(mDCIdentifier,makeNamespacedID(tbPersistable,persistableClass),mDCURI,annotatable);
+		}
 	}
 	
 	/**
@@ -71,7 +74,7 @@
 	 * @param annotatable
 	 */
 	protected void attachAnnotation(String key,String value,URI namespace,Annotatable annotatable) {
-		annotatable.addAnnotationValue(key, namespace, value);
+		annotatable.addAnnotationValue(key, namespace, value); // FIXME! Attaches meta element as last child
 	}
 	
 	/**
@@ -80,22 +83,49 @@
 	 * @param persistableClass
 	 * @return
 	 */
-	protected Long readTreeBaseID(Annotatable annotatable, Class<? extends TBPersistable> persistableClass) {
-		Set<Object> values = annotatable.getAnnotationValues(mDCIdentifier);
-		Iterator<Object> objectIterator = values.iterator();
+	protected Long readTreeBaseID(Annotatable annotatable) {
+		
+		// this will return the value object associated with a
+		// dc:identifier predicate in a nexml meta annotation,
+		// e.g. <meta property="dc:identifier" content="TB2:Tr231"/>
+		// this will return something that stringifies to TB2:Tr231
+		Set<Object> dublinCoreIdentifierObjects = annotatable.getAnnotationValues(mDCIdentifier);
+		Iterator<Object> objectIterator = dublinCoreIdentifierObjects.iterator();
 		while ( objectIterator.hasNext() ) {
-			String namespacedId = objectIterator.next().toString();
-			StringBuffer prefix = new StringBuffer();
-			prefix
-				.append(TreeBASE2Prefix)
-				.append(':')
-				.append(TreebaseIDString.getPrefixForClass(persistableClass));
-			if ( namespacedId.startsWith(prefix.toString()) ) {
-				return Long.parseLong(namespacedId.substring(prefix.length()));
+			TreebaseIDString treebaseIDString = null;
+			NamespacedGUID namespacedGUID = null;
+			try {
+				namespacedGUID = new NamespacedGUID(objectIterator.next().toString());
+				treebaseIDString = namespacedGUID.getTreebaseIDString();
+				return treebaseIDString.getId();
+			} catch ( MalformedTreebaseIDString e ) {
+				// XXX do nothing, it's OK, it means we're 
+				// parsing an id from a different naming
+				// authority, e.g. uBio or NCBI
 			}
 		}
 		return null;
 	}
+	
+	protected OTUs getOTUsById(Long taxonLabelSetId) {
+		for ( OTUs otus : getDocument().getOTUsList() ) {
+			Long annotatedID = readTreeBaseID(otus);
+			if ( taxonLabelSetId.equals(annotatedID) ) {
+				return otus;
+			}
+		}	
+		return null;
+	}
+	
+	protected OTU getOTUById(OTUs otus,Long taxonLabelId) {
+		for ( OTU otu : otus.getAllOTUs() ) {
+			Long annotatedID = readTreeBaseID(otu);
+			if ( taxonLabelId.equals(annotatedID) ) {
+				return otu;
+			}
+		}
+		return null;
+	}
 
 	public Document getDocument() {
 		return mDocument;
@@ -104,4 +134,29 @@
 	public void setDocument(Document document) {
 		mDocument = document;
 	}
+	
+	public static String forXML(String aText){
+		final StringBuilder result = new StringBuilder();
+		final StringCharacterIterator iterator = new StringCharacterIterator(aText);
+		char character = iterator.current();
+		while (character != CharacterIterator.DONE) {
+			if (character == '<') {
+				result.append("&lt;");
+			} else if (character == '>') {
+				result.append("&gt;");
+			} else if (character == '\"') {
+				result.append("&quot;");
+			} else if (character == '\'') {
+				result.append("&#039;");
+			} else if (character == '&') {
+				result.append("&amp;");
+			} else {
+				//the char is not a special one
+				//add it to the result as is
+				result.append(character);
+			}
+			character = iterator.next();
+		}
+		return result.toString();
+	}	
 }

Modified: trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlTreeBlockConverter.java
===================================================================
--- trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlTreeBlockConverter.java	2009-06-23 10:50:48 UTC (rev 72)
+++ trunk/treebase-core/src/main/java/org/cipres/treebase/domain/nexus/nexml/NexmlTreeBlockConverter.java	2009-06-23 10:52:08 UTC (rev 73)
@@ -17,6 +17,7 @@
 import org.nexml.model.IntEdge;
 import org.nexml.model.Network;
 import org.nexml.model.Node;
+import org.nexml.model.OTU;
 import org.nexml.model.OTUs;
 import org.nexml.model.Tree;
 
@@ -38,7 +39,7 @@
 	 */
 	public TreeBlock fromXmlToTreeBase(org.nexml.model.TreeBlock xmlTreeBlock) {
 		OTUs xmlOTUs = xmlTreeBlock.getOTUs();
-		Long tbTaxonLabelSetId = readTreeBaseID(xmlOTUs,TaxonLabelSet.class);
+		Long tbTaxonLabelSetId = readTreeBaseID(xmlOTUs);
 		TreeBlock tbTreeBlock = new TreeBlock();
 		tbTreeBlock.setTitle(xmlTreeBlock.getLabel());
 		if ( null != tbTaxonLabelSetId ) {
@@ -80,8 +81,10 @@
 	 */
 	public Tree<?> fromTreeBaseToXml(PhyloTree phyloTree,org.nexml.model.TreeBlock xmlTreeBlock) {
 		Tree<FloatEdge> xmlTree = xmlTreeBlock.createFloatTree();
-		xmlTree.setLabel(phyloTree.getLabel());
-		attachTreeBaseID(xmlTree, phyloTree);
+		if ( null != phyloTree.getLabel() ) {
+			xmlTree.setLabel(phyloTree.getLabel());
+		}
+		attachTreeBaseID(xmlTree, phyloTree,PhyloTree.class);
 		copyTreeBaseTree(phyloTree, xmlTree);
 		return xmlTree;
 	}
@@ -93,18 +96,12 @@
 	 */
 	public org.nexml.model.TreeBlock fromTreeBaseToXML(TreeBlock treeBlock) {
 		TaxonLabelSet taxonLabelSet = treeBlock.getTaxonLabelSet();
-		OTUs xmlOTUs = null;
-		Long taxonLabelSetId = taxonLabelSet.getId();
-		for ( OTUs otus : getDocument().getOTUsList() ) {
-			Long otusId = readTreeBaseID(otus, TaxonLabelSet.class);
-			if ( taxonLabelSetId == otusId ) {
-				xmlOTUs = otus;
-				break;
-			}
-		}
+		OTUs xmlOTUs = getOTUsById(taxonLabelSet.getId());
 		org.nexml.model.TreeBlock xmlTreeBlock = getDocument().createTreeBlock(xmlOTUs);
-		xmlTreeBlock.setLabel(treeBlock.getTitle());
-		attachTreeBaseID((Annotatable)xmlTreeBlock,treeBlock);
+		if ( null != treeBlock.getTitle() ) {
+			xmlTreeBlock.setLabel(treeBlock.getTitle());
+		}
+		attachTreeBaseID((Annotatable)xmlTreeBlock,treeBlock,TreeBlock.class);
 		for ( PhyloTree phyloTree : treeBlock.getTreeList() ) {
 			fromTreeBaseToXml(phyloTree,xmlTreeBlock);
 		}
@@ -128,25 +125,27 @@
 	 * @param xmlTree
 	 */
 	private void traverseTreeBaseTree(PhyloTree tbTree,PhyloTreeNode tbNode,Node xmlNode,Tree<FloatEdge> xmlTree) {
-		xmlNode.setLabel(tbNode.getName());
-		attachTreeBaseID(xmlNode, tbNode);	
+		if ( null != tbNode.getName() ) {
+			xmlNode.setLabel(tbNode.getName());
+		}
+		attachTreeBaseID(xmlNode, tbNode,PhyloTreeNode.class);	
 		TaxonLabel taxonLabel = tbNode.getTaxonLabel();
 		if ( null != taxonLabel ) {
 			Long taxonId = taxonLabel.getId();
 			for ( OTUs xmlOTUs : getDocument().getOTUsList() ) {
-				for ( org.nexml.model.OTU xmlOTU : xmlOTUs.getAllOTUs() ) {
-					Long currentTaxonId = readTreeBaseID(xmlOTU, TaxonLabel.class);
-					if ( taxonId == currentTaxonId ) {
-						xmlNode.setOTU(xmlOTU);
-						break;
-					}
+				OTU xmlOTU = getOTUById(xmlOTUs, taxonId);
+				if ( null != xmlOTU ) {
+					xmlNode.setOTU(xmlOTU);
+					break;
 				}
 			}
 		}
 		for ( PhyloTreeNode tbChildNode : tbNode.getChildNodes() ) {
 			Node xmlChildNode = xmlTree.createNode();
 			FloatEdge xmlEdge = xmlTree.createEdge(xmlNode, xmlChildNode);
-			xmlEdge.setLength(tbChildNode.getBranchLength());
+			if ( null != tbChildNode.getBranchLength() ) {
+				xmlEdge.setLength(tbChildNode.getBranchLength());
+			}
 			traverseTreeBaseTree(tbTree, tbChildNode, xmlChildNode, xmlTree);
 		}
 	}
@@ -176,7 +175,7 @@
 	private void traverseXmlTree(Tree<?> xmlTree, StringBuilder sb, Node xmlNode, PhyloTreeNode tbNode, PhyloTree tbTree) {
 		tbTree.addTreeNode(tbNode);
 		if ( null != xmlNode.getOTU() ) {
-			Long tbTaxonLabelId = readTreeBaseID(xmlNode.getOTU(),TaxonLabel.class);
+			Long tbTaxonLabelId = readTreeBaseID(xmlNode.getOTU());
 			if ( null != tbTaxonLabelId ) {
 				TaxonLabel tbTaxonLabel = getTaxonLabelHome()
 					.findPersistedObjectByID(TaxonLabel.class, tbTaxonLabelId);


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.