X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2FSequenceIdMatcher.java;h=ed90422403ec00eb40940f1cc545ae7f4c0f651f;hb=174230b4233d9ce80f94527768d2cd2f76da11ab;hp=b88a03f0636d71f8456a905a1c3379e9041b049b;hpb=df3d63b97cc55ee0e7d078d050775f7f58ab3c42;p=jalview.git

diff --git a/src/jalview/analysis/SequenceIdMatcher.java b/src/jalview/analysis/SequenceIdMatcher.java
index b88a03f..ed90422 100755
--- a/src/jalview/analysis/SequenceIdMatcher.java
+++ b/src/jalview/analysis/SequenceIdMatcher.java
@@ -1,6 +1,6 @@
 /*
  * Jalview - A Sequence Alignment Editor and Viewer
- * Copyright (C) 2005 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle
+ * Copyright (C) 2006 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle
  *
  * This program is free software; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License
@@ -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)
@@ -141,20 +144,30 @@ public class SequenceIdMatcher
     }
 
     /**
-     * matches if one ID properly contains another at a whitespace boundary.
-     * TODO: (JBPNote) These are not efficient. should use char[] for speed
-     * @param s SeqIdName
-     * @return boolean
+     * 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.length()>s.id.length()) {
         return id.startsWith(s.id) ?
-            (id.equals(s.id) ? true : 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 : s.id.startsWith(id+" "))
+            (s.id.equals(id) ? true :
+             (WORD_SEP.indexOf(s.id.charAt(id.length()))>-1))
             : false;
     }
 
@@ -162,11 +175,12 @@ public class SequenceIdMatcher
     {
       if (id.length()>s.length()) {
         return id.startsWith(s) ?
-            (id.equals(s) ? true : id.startsWith(s+" "))
+            (WORD_SEP.indexOf(id.charAt(s.length()))>-1)
             : false;
       } else
         return s.startsWith(id) ?
-            (s.equals(id) ? true : s.startsWith(id+" "))
+            (s.equals(id) ? true :
+             (WORD_SEP.indexOf(s.charAt(id.length()))>-1))
             : false;
     }
   }