X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;ds=inline;f=src%2Fjalview%2Fanalysis%2FSequenceIdMatcher.java;h=9d33996333997931356a5f3ca6b7530e7cb0ce77;hb=b242374a54c1288647de25dcf02163da168044a0;hp=1efe4982810ecb2a79b57a85834178374108926a;hpb=efc31b4a8d5cee63555586804a2b79c06bdb5a14;p=jalview.git

diff --git a/src/jalview/analysis/SequenceIdMatcher.java b/src/jalview/analysis/SequenceIdMatcher.java
index 1efe498..9d33996 100755
--- a/src/jalview/analysis/SequenceIdMatcher.java
+++ b/src/jalview/analysis/SequenceIdMatcher.java
@@ -26,7 +26,8 @@ import jalview.datamodel.*;
  * <p>Title: </p>
  * SequenceIdMatcher
  * <p>Description: </p>
- * Routine which does approximate Sequence Id resolution by name using string containment rather than equivalence
+ * Routine which does approximate Sequence Id resolution by name using
+ * string containment (on word boundaries) rather than equivalence
  * <p>Copyright: Copyright (c) 2004</p>
  *
  * <p>Company: Dundee University</p>
@@ -41,7 +42,6 @@ public class SequenceIdMatcher
   public SequenceIdMatcher(SequenceI[] seqs)
   {
     names = new Hashtable();
-
     for (int i = 0; i < seqs.length; i++)
     {
       names.put(new SeqIdName(seqs[i].getName()), seqs[i]);
@@ -83,13 +83,13 @@ public class SequenceIdMatcher
    */
   SequenceI[] findIdMatch(SequenceI[] seqs)
   {
-    SequenceI[] namedseqs = new SequenceI[seqs.length];
-
+    SequenceI[] namedseqs = null;
     int i = 0;
     SeqIdName nam;
 
     if (seqs.length > 0)
     {
+      namedseqs = new SequenceI[seqs.length];
       do
       {
         nam = new SeqIdName(seqs[i].getName());
@@ -103,7 +103,7 @@ public class SequenceIdMatcher
           namedseqs[i] = null;
         }
       }
-      while (i++ < seqs.length);
+      while (++i < seqs.length);
     }
 
     return namedseqs;
@@ -115,12 +115,15 @@ public class SequenceIdMatcher
 
     SeqIdName(String s)
     {
-      id = new String(s);
+      if (s!=null)
+        id = new String(s);
+      else
+        id = "";
     }
 
     public int hashCode()
     {
-      return (id.substring(0, 4).hashCode());
+      return ((id.length()>=4) ? id.substring(0, 4).hashCode() : id.hashCode());
     }
 
     public boolean equals(Object s)
@@ -140,24 +143,45 @@ public class SequenceIdMatcher
       return false;
     }
 
+    /**
+     * Characters that define the end of a unique sequence ID at
+     * the beginning of an arbitrary ID string
+     * JBPNote: This is a heuristic that will fail for arbritrarily extended sequence id's
+     * (like portions of an aligned set of repeats from one sequence)
+     */
+    private String WORD_SEP="~. |#\\/<>!\"£$%^*)}[@',?";
+
+   /**
+    * matches if one ID properly contains another at a whitespace boundary.
+    * TODO: (JBPNote) These are not efficient. should use char[] for speed
+    * todo: (JBPNote) Set separator characters appropriately
+    * @param s SeqIdName
+    * @return boolean
+    */
     public boolean equals(SeqIdName s)
     {
-      if (id.startsWith(s.id) || s.id.startsWith(id))
-      {
-        return true;
-      }
-
-      return false;
+      if (id.length()>s.id.length()) {
+        return id.startsWith(s.id) ?
+            (WORD_SEP.indexOf(id.charAt(s.id.length()))>-1)
+            : false;
+      } else
+        return s.id.startsWith(id) ?
+            (s.id.equals(id) ? true :
+             (WORD_SEP.indexOf(s.id.charAt(id.length()))>-1))
+            : false;
     }
 
     public boolean equals(String s)
     {
-      if (id.startsWith(s) || s.startsWith(id))
-      {
-        return true;
-      }
-
-      return false;
+      if (id.length()>s.length()) {
+        return id.startsWith(s) ?
+            (WORD_SEP.indexOf(id.charAt(s.length()))>-1)
+            : false;
+      } else
+        return s.startsWith(id) ?
+            (s.equals(id) ? true :
+             (WORD_SEP.indexOf(s.charAt(id.length()))>-1))
+            : false;
     }
   }
 }