added AGRESSIVE tax extraction ^^

[jalview.git] / forester / java / src / org / forester / phylogeny / PhylogenyMethods.java

diff --git a/forester/java/src/org/forester/phylogeny/PhylogenyMethods.java b/forester/java/src/org/forester/phylogeny/PhylogenyMethods.java
index dab7523..27f2a33 100644 (file)
--- a/forester/java/src/org/forester/phylogeny/PhylogenyMethods.java
+++ b/forester/java/src/org/forester/phylogeny/PhylogenyMethods.java
@@ -21,7 +21,7 @@
 // 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.phylogeny;
 
@@ -327,10 +327,14 @@ public class PhylogenyMethods {
 
     public static final HashMap<String, PhylogenyNode> createNameToExtNodeMap( final Phylogeny phy ) {
         final HashMap<String, PhylogenyNode> nodes = new HashMap<String, PhylogenyNode>();
-        for( final PhylogenyNodeIterator iter = phy.iteratorExternalForward(); iter.hasNext(); ) {
-            final PhylogenyNode n = iter.next();
+        final List<PhylogenyNode> ext = phy.getExternalNodes();
+        for( final PhylogenyNode n : ext ) {
             nodes.put( n.getName(), n );
         }
+        // for( final PhylogenyNodeIterator iter = phy.iteratorExternalForward(); iter.hasNext(); ) {
+        //     final PhylogenyNode n = iter.next();
+        //     nodes.put( n.getName(), n );
+        // }
         return nodes;
     }
 
@@ -408,7 +412,7 @@ public class PhylogenyMethods {
         final ArrayList<PhylogenyNode> to_delete = new ArrayList<PhylogenyNode>();
         for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
             final PhylogenyNode n = iter.next();
-            if ( ( !n.isExternal() ) && ( !n.isRoot() ) && ( n.getNumberOfDescendants() == 1 ) ) {
+            if ( ( !n.isExternal() ) && ( n.getNumberOfDescendants() == 1 ) ) {
                 to_delete.add( n );
             }
         }
@@ -439,7 +443,7 @@ public class PhylogenyMethods {
 
     public static List<PhylogenyNode> getAllDescendants( final PhylogenyNode node ) {
         final List<PhylogenyNode> descs = new ArrayList<PhylogenyNode>();
-        final Set<Integer> encountered = new HashSet<Integer>();
+        final Set<Long> encountered = new HashSet<Long>();
         if ( !node.isExternal() ) {
             final List<PhylogenyNode> exts = node.getAllExternalDescendants();
             for( PhylogenyNode current : exts ) {
@@ -764,25 +768,6 @@ public class PhylogenyMethods {
     }
 
     /**
-     * Returns the set of distinct taxonomies of
-     * all external nodes of node.
-     * If at least one the external nodes has no taxonomy,
-     * null is returned.
-     * 
-     */
-    public static Set<Taxonomy> obtainDistinctTaxonomies( final PhylogenyNode node ) {
-        final List<PhylogenyNode> descs = node.getAllExternalDescendants();
-        final Set<Taxonomy> tax_set = new HashSet<Taxonomy>();
-        for( final PhylogenyNode n : descs ) {
-            if ( !n.getNodeData().isHasTaxonomy() || n.getNodeData().getTaxonomy().isEmpty() ) {
-                return null;
-            }
-            tax_set.add( n.getNodeData().getTaxonomy() );
-        }
-        return tax_set;
-    }
-
-    /**
      * Returns a map of distinct taxonomies of
      * all external nodes of node.
      * If at least one of the external nodes has no taxonomy,
@@ -882,7 +867,7 @@ public class PhylogenyMethods {
             return;
         }
         phy.setIdToNodeMap( null );
-        int i = PhylogenyNode.getNodeCount();
+        long i = PhylogenyNode.getNodeCount();
         for( final PhylogenyNodeIterator it = phy.iteratorPreorder(); it.hasNext(); ) {
             it.next().setId( i++ );
         }
@@ -913,20 +898,19 @@ public class PhylogenyMethods {
     }
 
     public static void removeNode( final PhylogenyNode remove_me, final Phylogeny phylogeny ) {
-        if ( remove_me.isRoot() && remove_me.getNumberOfDescendants() != 1 ) {
-            throw new IllegalArgumentException( "attempt to remove a root node with more than one descendants" );
-        }
-        
-        if ( remove_me.isRoot() && remove_me.getNumberOfDescendants() == 1 ) {
-            final PhylogenyNode desc = remove_me.getDescendants().get( 0 );
-           
-            desc.setDistanceToParent( addPhylogenyDistances( remove_me.getDistanceToParent(),
-                                                             desc.getDistanceToParent() ) );
-            desc.setParent( null );
-            phylogeny.setRoot( desc );
-            phylogeny.clearHashIdToNodeMap();
+        if ( remove_me.isRoot() ) {
+            if ( remove_me.getNumberOfDescendants() == 1 ) {
+                final PhylogenyNode desc = remove_me.getDescendants().get( 0 );
+                desc.setDistanceToParent( addPhylogenyDistances( remove_me.getDistanceToParent(),
+                                                                 desc.getDistanceToParent() ) );
+                desc.setParent( null );
+                phylogeny.setRoot( desc );
+                phylogeny.clearHashIdToNodeMap();
+            }
+            else {
+                throw new IllegalArgumentException( "attempt to remove a root node with more than one descendants" );
+            }
         }
-        
         else if ( remove_me.isExternal() ) {
             phylogeny.deleteSubtree( remove_me, false );
             phylogeny.clearHashIdToNodeMap();
@@ -1395,13 +1379,16 @@ public class PhylogenyMethods {
             if ( !n.getNodeData().isHasTaxonomy() ) {
                 throw new IllegalArgumentException( "no taxonomic data in node: " + n );
             }
-            //  ref_ext_taxo.add( getSpecies( n ) );
             if ( !ForesterUtil.isEmpty( n.getNodeData().getTaxonomy().getScientificName() ) ) {
                 ref_ext_taxo.add( n.getNodeData().getTaxonomy().getScientificName() );
             }
             if ( !ForesterUtil.isEmpty( n.getNodeData().getTaxonomy().getTaxonomyCode() ) ) {
                 ref_ext_taxo.add( n.getNodeData().getTaxonomy().getTaxonomyCode() );
             }
+            if ( ( n.getNodeData().getTaxonomy().getIdentifier() != null )
+                    && !ForesterUtil.isEmpty( n.getNodeData().getTaxonomy().getIdentifier().getValue() ) ) {
+                ref_ext_taxo.add( n.getNodeData().getTaxonomy().getIdentifier().getValuePlusProvider() );
+            }
         }
         final ArrayList<PhylogenyNode> nodes_to_delete = new ArrayList<PhylogenyNode>();
         for( final PhylogenyNodeIterator it = to_be_stripped.iteratorExternalForward(); it.hasNext(); ) {
@@ -1410,7 +1397,9 @@ public class PhylogenyMethods {
                 nodes_to_delete.add( n );
             }
             else if ( !( ref_ext_taxo.contains( n.getNodeData().getTaxonomy().getScientificName() ) )
-                    && !( ref_ext_taxo.contains( n.getNodeData().getTaxonomy().getTaxonomyCode() ) ) ) {
+                    && !( ref_ext_taxo.contains( n.getNodeData().getTaxonomy().getTaxonomyCode() ) )
+                    && !( ( n.getNodeData().getTaxonomy().getIdentifier() != null ) && ref_ext_taxo.contains( n
+                            .getNodeData().getTaxonomy().getIdentifier().getValuePlusProvider() ) ) ) {
                 nodes_to_delete.add( n );
             }
         }