inprogress

[jalview.git] / forester / java / src / org / forester / analysis / TaxonomyDataManager.java

diff --git a/forester/java/src/org/forester/analysis/TaxonomyDataManager.java b/forester/java/src/org/forester/analysis/TaxonomyDataManager.java
index c2c625b..cddaa4f 100644 (file)
--- a/forester/java/src/org/forester/analysis/TaxonomyDataManager.java
+++ b/forester/java/src/org/forester/analysis/TaxonomyDataManager.java
@@ -22,7 +22,7 @@
 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
 //
 // Contact: phylosoft @ gmail . com
-// WWW: www.phylosoft.org/forester
+// WWW: https://sites.google.com/site/cmzmasek/home/software/forester
 
 package org.forester.analysis;
 
@@ -38,6 +38,7 @@ import javax.swing.JOptionPane;
 
 import org.forester.archaeopteryx.MainFrameApplication;
 import org.forester.archaeopteryx.TreePanel;
+import org.forester.archaeopteryx.tools.AncestralTaxonomyInferrer;
 import org.forester.archaeopteryx.tools.RunnableProcess;
 import org.forester.io.parsers.phyloxml.PhyloXmlDataFormatException;
 import org.forester.phylogeny.Phylogeny;
@@ -46,8 +47,8 @@ import org.forester.phylogeny.data.Identifier;
 import org.forester.phylogeny.data.Taxonomy;
 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
 import org.forester.util.ForesterUtil;
-import org.forester.ws.uniprot.UniProtTaxonomy;
-import org.forester.ws.uniprot.UniProtWsTools;
+import org.forester.ws.seqdb.SequenceDbWsTools;
+import org.forester.ws.seqdb.UniProtTaxonomy;
 
 public final class TaxonomyDataManager extends RunnableProcess {
 
@@ -126,8 +127,8 @@ public final class TaxonomyDataManager extends RunnableProcess {
     }
 
     private final static UniProtTaxonomy obtainTaxonomy( final HashMap<String, UniProtTaxonomy> cache,
-                                                        final Object query,
-                                                        final QUERY_TYPE qt ) throws IOException,
+                                                         final Object query,
+                                                         final QUERY_TYPE qt ) throws IOException,
             AncestralTaxonomyInferenceException {
         if ( cache.containsKey( query ) ) {
             return cache.get( query ).copy();
@@ -175,19 +176,19 @@ public final class TaxonomyDataManager extends RunnableProcess {
     }
 
     private final static List<UniProtTaxonomy> getTaxonomiesFromCommonName( final String query ) throws IOException {
-        return UniProtWsTools.getTaxonomiesFromCommonNameStrict( query, MAX_TAXONOMIES_TO_RETURN );
+        return SequenceDbWsTools.getTaxonomiesFromCommonNameStrict( query, MAX_TAXONOMIES_TO_RETURN );
     }
 
     private final static List<UniProtTaxonomy> getTaxonomiesFromId( final String query ) throws IOException {
-        return UniProtWsTools.getTaxonomiesFromId( query, MAX_TAXONOMIES_TO_RETURN );
+        return SequenceDbWsTools.getTaxonomiesFromId( query, MAX_TAXONOMIES_TO_RETURN );
     }
 
     private final static List<UniProtTaxonomy> getTaxonomiesFromScientificName( final String query ) throws IOException {
-        return UniProtWsTools.getTaxonomiesFromScientificNameStrict( query, MAX_TAXONOMIES_TO_RETURN );
+        return SequenceDbWsTools.getTaxonomiesFromScientificNameStrict( query, MAX_TAXONOMIES_TO_RETURN );
     }
 
     private final static List<UniProtTaxonomy> getTaxonomiesFromTaxonomyCode( final String query ) throws IOException {
-        return UniProtWsTools.getTaxonomiesFromTaxonomyCode( query, MAX_TAXONOMIES_TO_RETURN );
+        return SequenceDbWsTools.getTaxonomiesFromTaxonomyCode( query, MAX_TAXONOMIES_TO_RETURN );
     }
 
     static final boolean isHasAppropriateId( final Taxonomy tax ) {
@@ -236,13 +237,23 @@ public final class TaxonomyDataManager extends RunnableProcess {
                     uniprot_tax = obtainUniProtTaxonomy( tax, null, qt );
                 }
                 else {
-                    uniprot_tax = obtainUniProtTaxonomy( node.getName(), null, qt );
+                    uniprot_tax = obtainUniProtTaxonomy( node.getName(), qt );
                 }
                 if ( uniprot_tax != null ) {
+                    if ( tax == null ) {
+                        tax = new Taxonomy();
+                        node.getNodeData().addTaxonomy( tax );
+                        node.setName( "" );
+                    }
                     updateTaxonomy( qt, node, tax, uniprot_tax );
                 }
                 else {
-                    not_found.add( tax.toString() );
+                    if ( tax != null ) {
+                        not_found.add( tax.toString() );
+                    }
+                    else {
+                        not_found.add( node.getName() );
+                    }
                     if ( delete && node.isExternal() ) {
                         not_found_external_nodes.add( node );
                     }
@@ -254,7 +265,7 @@ public final class TaxonomyDataManager extends RunnableProcess {
                 phy.deleteSubtree( node, true );
             }
             phy.externalNodesHaveChanged();
-            phy.hashIDs();
+            phy.clearHashIdToNodeMap();
             phy.recalculateNumberOfExternalDescendants( true );
         }
         return not_found;
@@ -294,20 +305,20 @@ public final class TaxonomyDataManager extends RunnableProcess {
         }
     }
 
-    public final static UniProtTaxonomy obtainUniProtTaxonomy( final String simple_name, Object query, QUERY_TYPE qt )
+    public final static UniProtTaxonomy obtainUniProtTaxonomy( final String simple_name, QUERY_TYPE qt )
             throws IOException, AncestralTaxonomyInferenceException {
         if ( ForesterUtil.isEmpty( simple_name ) ) {
             throw new IllegalArgumentException( "illegal attempt to use empty simple name" );
         }
         qt = QUERY_TYPE.SN;
-        UniProtTaxonomy ut = obtainTaxonomy( TaxonomyDataManager.getSnTaxCacheMap(), query, qt );
+        UniProtTaxonomy ut = obtainTaxonomy( TaxonomyDataManager.getSnTaxCacheMap(), simple_name, qt );
         if ( ut == null ) {
             qt = QUERY_TYPE.CODE;
-            ut = obtainTaxonomy( TaxonomyDataManager.getCodeTaxCacheMap(), query, qt );
+            ut = obtainTaxonomy( TaxonomyDataManager.getCodeTaxCacheMap(), simple_name, qt );
         }
         if ( ut == null ) {
             qt = QUERY_TYPE.CN;
-            ut = obtainTaxonomy( TaxonomyDataManager.getCnTaxCacheMap(), query, qt );
+            ut = obtainTaxonomy( TaxonomyDataManager.getCnTaxCacheMap(), simple_name, qt );
         }
         return ut;
     }
@@ -326,13 +337,15 @@ public final class TaxonomyDataManager extends RunnableProcess {
                 for( final UniProtTaxonomy up_taxonomy : up_taxonomies ) {
                     boolean match = true;
                     I: for( int i = 0; i < lineage.size(); ++i ) {
-                        if ( !lineage.get( i ).equalsIgnoreCase( up_taxonomy.getLineage().get( i ) ) ) {
+                        if ( ( i == up_taxonomy.getLineage().size() )
+                                || !lineage.get( i ).equalsIgnoreCase( up_taxonomy.getLineage().get( i ) ) ) {
                             match = false;
                             break I;
                         }
                     }
                     if ( match ) {
                         if ( up_tax != null ) {
+                            //TODO this is dead code?!
                             throw new AncestralTaxonomyInferenceException( "lineage \""
                                     + ForesterUtil.stringListToString( lineage, " > " ) + "\" is not unique" );
                         }
@@ -364,7 +377,8 @@ public final class TaxonomyDataManager extends RunnableProcess {
     synchronized final private static void updateTaxonomy( final QUERY_TYPE qt,
                                                            final PhylogenyNode node,
                                                            final Taxonomy tax,
-                                                           final UniProtTaxonomy up_tax ) {
+                                                           final UniProtTaxonomy up_tax )
+            throws PhyloXmlDataFormatException {
         if ( ( qt != QUERY_TYPE.SN ) && !ForesterUtil.isEmpty( up_tax.getScientificName() )
                 && ForesterUtil.isEmpty( tax.getScientificName() ) ) {
             tax.setScientificName( up_tax.getScientificName() );
@@ -511,7 +525,7 @@ public final class TaxonomyDataManager extends RunnableProcess {
     }
 
     private final String getBaseUrl() {
-        return UniProtWsTools.BASE_URL;
+        return AncestralTaxonomyInferrer.getBaseUrl();
     }
 
     @Override