From fbec1b33d0fc169d72be059a5d2cf12b248270e1 Mon Sep 17 00:00:00 2001
From: jprocter <Jim Procter>
Date: Wed, 17 Aug 2005 15:20:10 +0000
Subject: [PATCH] Fixed basic name to sequence ID matching (bug
 http://dragonfly.compbio.dundee.ac.uk/mantis/view.php?id=9227)
 and improved code robustness when dealing with badly formed
 IDs (null strings).

---
 src/jalview/analysis/SequenceIdMatcher.java |   32 ++++++++++++++++++---------
 1 file changed, 22 insertions(+), 10 deletions(-)

diff --git a/src/jalview/analysis/SequenceIdMatcher.java b/src/jalview/analysis/SequenceIdMatcher.java
index b88a03f..f4f305b 100755
--- a/src/jalview/analysis/SequenceIdMatcher.java
+++ b/src/jalview/analysis/SequenceIdMatcher.java
@@ -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 +144,28 @@ 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
      */
+    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 +173,12 @@ 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;
     }
   }
-- 
1.7.10.2