X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2FSequenceIdMatcher.java;h=4b1900cd618f9696051ef58e6eadb0b503936a9b;hb=797df64fa2a0a30773d0f48f5494d4155e5a8be3;hp=71b27549cd285e8dde5d43c30552202f3318dca4;hpb=e75297d4a787f7544d462656d5ccca2692b8a029;p=jalview.git

diff --git a/src/jalview/analysis/SequenceIdMatcher.java b/src/jalview/analysis/SequenceIdMatcher.java
index 71b2754..4b1900c 100755
--- a/src/jalview/analysis/SequenceIdMatcher.java
+++ b/src/jalview/analysis/SequenceIdMatcher.java
@@ -1,6 +1,6 @@
 /*
- * Jalview - A Sequence Alignment Editor and Viewer (Version 2.6)
- * Copyright (C) 2010 J Procter, AM Waterhouse, G Barton, M Clamp, S Searle
+ * Jalview - A Sequence Alignment Editor and Viewer (Version 2.7)
+ * Copyright (C) 2011 J Procter, AM Waterhouse, J Engelhardt, LM Lui, G Barton, M Clamp, S Searle
  * 
  * This file is part of Jalview.
  * 
@@ -22,27 +22,10 @@ import java.util.*;
 import jalview.datamodel.*;
 
 /**
- * <p>
- * Title:
- * </p>
- * SequenceIdMatcher
- * <p>
- * Description:
- * </p>
- * Routine which does approximate Sequence Id resolution by name using string
+ * Routines for approximate Sequence Id resolution by name using 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>
- * 
- * @author not attributable
- * @version 1.0
  */
 public class SequenceIdMatcher
 {
@@ -53,7 +36,8 @@ public class SequenceIdMatcher
     names = new Hashtable();
     for (int i = 0; i < seqs.length; i++)
     {
-      names.put(new SeqIdName(seqs[i].getName()), seqs[i]);
+      // TODO: deal with ID collisions - SequenceI should be appended to list associated with this key.
+      names.put(new SeqIdName(seqs[i].getDisplayId(true)), seqs[i]);
       // add in any interesting identifiers
       if (seqs[i].getDBRef() != null)
       {
@@ -83,6 +67,22 @@ public class SequenceIdMatcher
    */
   private SequenceI pickbestMatch(SeqIdName candName, Vector matches)
   {
+    SequenceI[] st= pickbestMatches(candName, matches);
+    return st==null || st.length==0 ? null : st[0];
+  }
+  /**
+   * 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 Object[] { SequenceI closest SequenceI to SeqIdName, SequenceI[] ties }
+   */
+  private SequenceI[] pickbestMatches(SeqIdName candName, Vector matches)
+  {
+    ArrayList best=new ArrayList();
     SequenceI match = null;
     if (candName == null || matches == null || matches.size() == 0)
     {
@@ -90,6 +90,7 @@ public class SequenceIdMatcher
     }
     match = (SequenceI) matches.elementAt(0);
     matches.removeElementAt(0);
+    best.add(match);
     names.put(new SeqIdName(match.getName()), match);
     int matchlen = match.getName().length();
     int namlen = candName.id.length();
@@ -97,17 +98,26 @@ public class SequenceIdMatcher
     {
       // look through for a better one.
       SequenceI cand = (SequenceI) matches.elementAt(0);
+      matches.remove(0);
       names.put(new SeqIdName(cand.getName()), cand);
-      int candlen = cand.getName().length();
+      int q,w,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)
+      if ((q=Math.abs(matchlen - namlen)) > (w=Math.abs(candlen - namlen))
               && candlen > matchlen)
       {
+        best.clear();
         match = cand;
         matchlen = candlen;
+        best.add(match);
+      }
+      if (q==w && candlen==matchlen)
+      {
+        // record any ties
+        best.add(cand);
       }
     }
-    return match;
+    if (best.size()==0) { return null; };
+    return (SequenceI[]) best.toArray(new SequenceI[0]);
   }
 
   /**
@@ -130,6 +140,17 @@ public class SequenceIdMatcher
   }
 
   /**
+   *  Find all matches for a given sequence name.
+   *  @param seqnam string to query Matcher with.
+   */
+  public SequenceI[] findAllIdMatches(String seqnam)
+  {
+
+    SeqIdName nam = new SeqIdName(seqnam);
+    return findAllIdMatches(nam);
+  }
+
+  /**
    * findIdMatch
    * 
    * Return pointers to sequences (or sequence object containers) which have
@@ -183,6 +204,25 @@ public class SequenceIdMatcher
     }
     return pickbestMatch(nam, matches);
   }
+  /**
+   * core findIdMatch search method for finding all equivalent matches
+   * 
+   * @param nam
+   *          SeqIdName
+   * @return SequenceI[]
+   */
+  private SequenceI[] findAllIdMatches(
+          jalview.analysis.SequenceIdMatcher.SeqIdName nam)
+  {
+    Vector matches = new Vector();
+    while (names.containsKey(nam))
+    {
+      matches.addElement(names.remove(nam));
+    }
+    SequenceI[] r=pickbestMatches(nam, matches);
+    return r;
+  }
+
 
   private class SeqIdName
   {