JAL-2797 added constructor including embedded/standalone boolean

[jalview.git] / forester / java / src / org / forester / sequence / BasicSequence.java

diff --git a/forester/java/src/org/forester/sequence/BasicSequence.java b/forester/java/src/org/forester/sequence/BasicSequence.java
index c14277c..de3084e 100644 (file)
--- a/forester/java/src/org/forester/sequence/BasicSequence.java
+++ b/forester/java/src/org/forester/sequence/BasicSequence.java
@@ -22,29 +22,58 @@
 // 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.sequence;
 
-public class BasicSequence implements Sequence {
+import org.forester.util.ForesterUtil;
+
+public class BasicSequence implements MolecularSequence {
 
     private final char[] _mol_sequence;
-    private final String _identifier;
+    private String       _identifier;
     private final TYPE   _type;
 
-    private BasicSequence( final String identifier, final String mol_sequence, final TYPE type ) {
+    /**
+     * Only use if you know what you are doing!
+     *
+     */
+    public BasicSequence( final String identifier, final String mol_sequence, final TYPE type ) {
+        check( identifier, mol_sequence );
         _mol_sequence = mol_sequence.toCharArray();
         _identifier = identifier;
         _type = type;
     }
 
-    // Only use if you know what you are doing!
+    private static final void check( final String identifier, final String mol_sequence ) {
+        if ( ForesterUtil.isEmpty( identifier ) ) {
+            throw new IllegalArgumentException( "identifier of sequence cannot be empty" );
+        }
+        if ( ForesterUtil.isEmpty( mol_sequence ) ) {
+            throw new IllegalArgumentException( "molecular sequence cannot be empty" );
+        }
+    }
+
+    /**
+     * Only use if you know what you are doing!
+     *
+     */
     public BasicSequence( final String identifier, final char[] mol_sequence, final TYPE type ) {
+        if ( ForesterUtil.isEmpty( identifier ) ) {
+            throw new IllegalArgumentException( "identifier of sequence cannot be empty" );
+        }
+        if ( ( mol_sequence == null ) || ( mol_sequence.length < 1 ) ) {
+            throw new IllegalArgumentException( "molecular sequence cannot be empty" );
+        }
         _mol_sequence = mol_sequence;
         _identifier = identifier;
         _type = type;
     }
 
+    public void setIdentifier( final String id ) {
+        _identifier = id;
+    }
+
     @Override
     public String getIdentifier() {
         return _identifier;
@@ -73,8 +102,8 @@ public class BasicSequence implements Sequence {
     @Override
     public int getNumberOfGapResidues() {
         int gaps = 0;
-        for( int i = 0; i < _mol_sequence.length; ++i ) {
-            if ( _mol_sequence[ i ] == GAP ) {
+        for( final char element : _mol_sequence ) {
+            if ( element == GAP ) {
                 ++gaps;
             }
         }
@@ -82,15 +111,35 @@ public class BasicSequence implements Sequence {
     }
 
     @Override
+    public boolean equals( final Object obj ) {
+        if ( obj == null ) {
+            return false;
+        }
+        if ( obj.getClass() != getClass() ) {
+            return false;
+        }
+        final MolecularSequence other = ( MolecularSequence ) obj;
+        if ( getMolecularSequenceAsString().equals( other.getMolecularSequenceAsString() ) ) {
+            return true;
+        }
+        return false;
+    }
+
+    @Override
+    public int hashCode() {
+        return getMolecularSequenceAsString().hashCode();
+    }
+
+    @Override
     public String toString() {
         final StringBuffer sb = new StringBuffer();
         sb.append( _identifier.toString() );
-        sb.append( " " );
-        sb.append( new String( _mol_sequence ) );
+        sb.append( ": " );
+        sb.append( getMolecularSequenceAsString() );
         return sb.toString();
     }
 
-    public static Sequence copySequence( final Sequence seq ) {
+    public static MolecularSequence copySequence( final MolecularSequence seq ) {
         final char[] s = new char[ seq.getMolecularSequence().length ];
         for( int i = 0; i < seq.getMolecularSequence().length; i++ ) {
             s[ i ] = seq.getMolecularSequence()[ i ];
@@ -98,18 +147,61 @@ public class BasicSequence implements Sequence {
         return new BasicSequence( new String( seq.getIdentifier() ), s, seq.getType() );
     }
 
-    public static Sequence createAaSequence( final String identifier, final String mol_sequence ) {
+    public static MolecularSequence createSequence( final String identifier, final String mol_sequence ) {
+        check( identifier, mol_sequence );
+        final TYPE type = ForesterUtil.guessMolecularSequenceType( mol_sequence );
+        final String re;
+        final char repl;
+        if ( type == TYPE.AA ) {
+            re = AA_REGEXP;
+            repl = UNSPECIFIED_AA;
+        }
+        else if ( type == TYPE.DNA ) {
+            re = DNA_REGEXP;
+            repl = UNSPECIFIED_NUC;
+        }
+        else if ( type == TYPE.RNA ) {
+            re = RNA_REGEXP;
+            repl = UNSPECIFIED_NUC;
+        }
+        else {
+            throw new IllegalArgumentException( "could not determine sequence type for: " + mol_sequence);
+        }
         return new BasicSequence( identifier, mol_sequence.toUpperCase().replaceAll( "\\.", GAP_STR )
-                .replaceAll( AA_REGEXP, Character.toString( UNSPECIFIED_AA ) ), TYPE.AA );
+                                  .replaceAll( re, Character.toString( repl ) ), type );
+    }
+    
+    public static MolecularSequence createGeneralSequence( final String identifier, final String mol_sequence ) {
+        check( identifier, mol_sequence );
+        return new BasicSequence( identifier, mol_sequence.toUpperCase().replaceAll( "\\.", GAP_STR 
+                                  ), TYPE.GENERAL );
+    }
+    
+    public static MolecularSequence createAaSequence( final String identifier, final String mol_sequence ) {
+        check( identifier, mol_sequence );
+        return new BasicSequence( identifier, mol_sequence.toUpperCase().replaceAll( "\\.", GAP_STR )
+                                  .replaceAll( AA_REGEXP, Character.toString( UNSPECIFIED_AA ) ), TYPE.AA );
     }
 
-    public static Sequence createDnaSequence( final String identifier, final String mol_sequence ) {
+    public static MolecularSequence createDnaSequence( final String identifier, final String mol_sequence ) {
+        check( identifier, mol_sequence );
         return new BasicSequence( identifier, mol_sequence.toUpperCase().replaceAll( "\\.", GAP_STR )
-                .replaceAll( DNA_REGEXP, Character.toString( UNSPECIFIED_NUC ) ), TYPE.DNA );
+                                  .replaceAll( DNA_REGEXP, Character.toString( UNSPECIFIED_NUC ) ), TYPE.DNA );
     }
 
-    public static Sequence createRnaSequence( final String identifier, final String mol_sequence ) {
+    public static MolecularSequence createRnaSequence( final String identifier, final String mol_sequence ) {
+        check( identifier, mol_sequence );
         return new BasicSequence( identifier, mol_sequence.toUpperCase().replaceAll( "\\.", GAP_STR )
-                .replaceAll( RNA_REGEXP, Character.toString( UNSPECIFIED_NUC ) ), TYPE.RNA );
+                                  .replaceAll( RNA_REGEXP, Character.toString( UNSPECIFIED_NUC ) ), TYPE.RNA );
+    }
+
+    @Override
+    public String getMolecularSequenceAsString() {
+        return new String( getMolecularSequence() );
+    }
+
+    @Override
+    public boolean isGapAt( final int position ) {
+        return getResidueAt( position ) == GAP;
     }
 }