X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2FSequenceIdMatcher.java;h=0744937743bcba699e34bc0421f06133144b08ca;hb=7bc226b58110fa26d9dbd3f0c78095d06909ffc3;hp=9d33996333997931356a5f3ca6b7530e7cb0ce77;hpb=15af442616b10419a3ec3eb080f8378879688fd1;p=jalview.git

diff --git a/src/jalview/analysis/SequenceIdMatcher.java b/src/jalview/analysis/SequenceIdMatcher.java
index 9d33996..0744937 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) 2007 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
@@ -27,7 +27,9 @@ import jalview.datamodel.*;
  * SequenceIdMatcher
  * <p>Description: </p>
  * Routine which does approximate Sequence Id resolution by name using
- * string containment (on word boundaries) rather than equivalence
+ * string containment (on word boundaries) rather than equivalence. It also
+ * attempts to resolve ties where no exact match is available by picking the
+ * the id closest to the query.
  * <p>Copyright: Copyright (c) 2004</p>
  *
  * <p>Company: Dundee University</p>
@@ -48,28 +50,57 @@ public class SequenceIdMatcher
     }
   }
 
-  SequenceI findIdMatch(SequenceI seq)
+  /**
+   * returns the closest SequenceI in matches to SeqIdName and returns all the matches
+   * to the names hash.
+   * @param candName SeqIdName
+   * @param matches Vector of SequenceI objects
+   * @return SequenceI closest SequenceI to SeqIdName
+   */
+  private SequenceI pickbestMatch(SeqIdName candName, Vector matches)
   {
-    SeqIdName nam = new SeqIdName(seq.getName());
-
-    if (names.containsKey(nam))
+    SequenceI match = null;
+    if (candName == null || matches == null || matches.size() == 0)
     {
-      return (SequenceI) names.get(nam);
+      return null;
     }
+    match = (SequenceI) matches.elementAt(0);
+    matches.removeElementAt(0);
+    names.put(new SeqIdName(match.getName()), match);
+    int matchlen = match.getName().length();
+    int namlen = candName.id.length();
+    while (matches.size() > 0)
+    {
+      // look through for a better one.
+      SequenceI cand = (SequenceI) matches.elementAt(0);
+      names.put(new SeqIdName(cand.getName()), cand);
+      int candlen = cand.getName().length();
+      // keep the one with an id 'closer' to the given seqnam string
+      if (Math.abs(matchlen - namlen) > Math.abs(candlen - namlen) &&
+          candlen > matchlen)
+      {
+        match = cand;
+        matchlen = candlen;
+      }
+    }
+    return match;
+  }
 
-    return null;
+  /**
+   * get SequenceI with closest SequenceI.getName() to seq.getName()
+   * @param seq SequenceI
+   * @return SequenceI
+   */
+  SequenceI findIdMatch(SequenceI seq)
+  {
+    SeqIdName nam = new SeqIdName(seq.getName());
+    return findIdMatch(nam);
   }
 
   SequenceI findIdMatch(String seqnam)
   {
     SeqIdName nam = new SeqIdName(seqnam);
-
-    if (names.containsKey(nam))
-    {
-      return (SequenceI) names.get(nam);
-    }
-
-    return null;
+    return findIdMatch(nam);
   }
 
   /**
@@ -96,7 +127,7 @@ public class SequenceIdMatcher
 
         if (names.containsKey(nam))
         {
-          namedseqs[i] = (SequenceI) names.get(nam);
+          namedseqs[i] = findIdMatch(nam);
         }
         else
         {
@@ -109,21 +140,41 @@ public class SequenceIdMatcher
     return namedseqs;
   }
 
+  /**
+   * core findIdMatch search method
+   * @param nam SeqIdName
+   * @return SequenceI
+   */
+  private SequenceI findIdMatch(jalview.analysis.SequenceIdMatcher.SeqIdName
+                                nam)
+  {
+    Vector matches = new Vector();
+    while (names.containsKey(nam))
+    {
+      matches.addElement(names.remove(nam));
+    }
+    return pickbestMatch(nam, matches);
+  }
+
   private class SeqIdName
   {
     String id;
 
     SeqIdName(String s)
     {
-      if (s!=null)
+      if (s != null)
+      {
         id = new String(s);
+      }
       else
+      {
         id = "";
+      }
     }
 
     public int hashCode()
     {
-      return ((id.length()>=4) ? id.substring(0, 4).hashCode() : id.hashCode());
+      return ( (id.length() >= 4) ? id.substring(0, 4).hashCode() : id.hashCode());
     }
 
     public boolean equals(Object s)
@@ -149,39 +200,47 @@ public class SequenceIdMatcher
      * 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="~. |#\\/<>!\"£$%^*)}[@',?";
-
-   /**
-    * matches if one ID properly contains another at a whitespace boundary.
-    * TODO: (JBPNote) These are not efficient. should use char[] for speed
-    * todo: (JBPNote) Set separator characters appropriately
-    * @param s SeqIdName
-    * @return boolean
-    */
+    private String WORD_SEP = "~. |#\\/<>!\"£$%^*)}[@',?_";
+
+    /**
+     * 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()) {
+      if (id.length() > s.id.length())
+      {
         return id.startsWith(s.id) ?
-            (WORD_SEP.indexOf(id.charAt(s.id.length()))>-1)
+            (WORD_SEP.indexOf(id.charAt(s.id.length())) > -1)
             : false;
-      } else
+      }
+      else
+      {
         return s.id.startsWith(id) ?
             (s.id.equals(id) ? true :
-             (WORD_SEP.indexOf(s.id.charAt(id.length()))>-1))
+             (WORD_SEP.indexOf(s.id.charAt(id.length())) > -1))
             : false;
+      }
     }
 
     public boolean equals(String s)
     {
-      if (id.length()>s.length()) {
+      if (id.length() > s.length())
+      {
         return id.startsWith(s) ?
-            (WORD_SEP.indexOf(id.charAt(s.length()))>-1)
+            (WORD_SEP.indexOf(id.charAt(s.length())) > -1)
             : false;
-      } else
+      }
+      else
+      {
         return s.startsWith(id) ?
             (s.equals(id) ? true :
-             (WORD_SEP.indexOf(s.charAt(id.length()))>-1))
+             (WORD_SEP.indexOf(s.charAt(id.length())) > -1))
             : false;
+      }
     }
   }
 }