/*
- * Jalview - A Sequence Alignment Editor and Viewer (Version 2.8.2)
- * Copyright (C) 2014 The Jalview Authors
+ * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
+ * Copyright (C) $$Year-Rel$$ The Jalview Authors
*
* This file is part of Jalview.
*
*/
package jalview.analysis;
-import java.util.*;
+import java.util.Locale;
-import jalview.datamodel.*;
+import jalview.datamodel.DBRefEntry;
+import jalview.datamodel.SequenceI;
+
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Vector;
/**
* Routines for approximate Sequence Id resolution by name using string
*/
public class SequenceIdMatcher
{
- private Hashtable names;
+ private HashMap<SeqIdName, SequenceI> names;
+
+ public SequenceIdMatcher(List<SequenceI> seqs)
+ {
+ names = new HashMap<SeqIdName, SequenceI>();
+ addAll(seqs);
+ }
- public SequenceIdMatcher(SequenceI[] seqs)
+ /**
+ * Adds sequences to this matcher
+ *
+ * @param seqs
+ */
+ public void addAll(List<SequenceI> seqs)
{
- names = new Hashtable();
- for (int i = 0; i < seqs.length; i++)
+ for (SequenceI seq : seqs)
{
- // 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]);
- SequenceI dbseq = seqs[i];
- while (dbseq.getDatasetSequence()!=null)
- {
- dbseq = dbseq.getDatasetSequence();
- }
- // add in any interesting identifiers
- if (dbseq.getDBRef() != null)
+ add(seq);
+ }
+ }
+
+ /**
+ * Adds one sequence to this matcher
+ *
+ * @param seq
+ */
+ public void add(SequenceI seq)
+ {
+ // TODO: deal with ID collisions - SequenceI should be appended to list
+ // associated with this key.
+ names.put(new SeqIdName(seq.getDisplayId(true)), seq);
+ SequenceI dbseq = seq;
+ while (dbseq.getDatasetSequence() != null)
+ {
+ dbseq = dbseq.getDatasetSequence();
+ }
+ // add in any interesting identifiers
+ List<DBRefEntry> dbr = dbseq.getDBRefs();
+ if (dbr != null)
+ {
+ SeqIdName sid = null;
+ for (int r = 0, nr = dbr.size(); r < nr; r++)
{
- DBRefEntry dbr[] = dbseq.getDBRef();
- SeqIdName sid = null;
- for (int r = 0; r < dbr.length; r++)
+ sid = new SeqIdName(dbr.get(r).getAccessionId());
+ if (!names.containsKey(sid))
{
- sid = new SeqIdName(dbr[r].getAccessionId());
- if (!names.contains(sid))
- {
- names.put(sid, seqs[i]);
- }
+ names.put(sid, seq);
}
}
}
}
/**
+ * convenience method to make a matcher from concrete array
+ *
+ * @param sequences
+ */
+ public SequenceIdMatcher(SequenceI[] sequences)
+ {
+ this(Arrays.asList(sequences));
+ }
+
+ /**
* 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
+ * List of SequenceI objects
* @return SequenceI closest SequenceI to SeqIdName
*/
- private SequenceI pickbestMatch(SeqIdName candName, Vector matches)
+ private SequenceI pickbestMatch(SeqIdName candName,
+ List<SequenceI> matches)
{
- SequenceI[] st = pickbestMatches(candName, matches);
- return st == null || st.length == 0 ? null : st[0];
+ List<SequenceI> st = pickbestMatches(candName, matches);
+ return st == null || st.size() == 0 ? null : st.get(0);
}
/**
* @return Object[] { SequenceI closest SequenceI to SeqIdName, SequenceI[]
* ties }
*/
- private SequenceI[] pickbestMatches(SeqIdName candName, Vector matches)
+ private List<SequenceI> pickbestMatches(SeqIdName candName,
+ List<SequenceI> matches)
{
- ArrayList best = new ArrayList();
- SequenceI match = null;
+ ArrayList<SequenceI> best = new ArrayList<SequenceI>();
if (candName == null || matches == null || matches.size() == 0)
{
return null;
}
- match = (SequenceI) matches.elementAt(0);
- matches.removeElementAt(0);
+ SequenceI match = matches.remove(0);
best.add(match);
names.put(new SeqIdName(match.getName()), match);
int matchlen = match.getName().length();
while (matches.size() > 0)
{
// look through for a better one.
- SequenceI cand = (SequenceI) matches.elementAt(0);
- matches.remove(0);
+ SequenceI cand = matches.remove(0);
names.put(new SeqIdName(cand.getName()), cand);
int q, w, candlen = cand.getName().length();
// keep the one with an id 'closer' to the given seqnam string
- if ((q = Math.abs(matchlen - namlen)) > (w = Math.abs(candlen
- - namlen))
- && candlen > matchlen)
+ if ((q = Math.abs(matchlen - namlen)) > (w = Math
+ .abs(candlen - namlen)) && candlen > matchlen)
{
best.clear();
match = cand;
return null;
}
;
- return (SequenceI[]) best.toArray(new SequenceI[0]);
+ return best;
}
/**
*
* @param seqnam
* string to query Matcher with.
+ * @return a new array or (possibly) null
*/
public SequenceI[] findAllIdMatches(String seqnam)
{
SeqIdName nam = new SeqIdName(seqnam);
- return findAllIdMatches(nam);
+ List<SequenceI> m = findAllIdMatches(nam);
+ if (m != null)
+ {
+ return m.toArray(new SequenceI[m.size()]);
+ }
+ return null;
}
/**
* SeqIdName
* @return SequenceI[]
*/
- private SequenceI[] findAllIdMatches(
+ private List<SequenceI> findAllIdMatches(
jalview.analysis.SequenceIdMatcher.SeqIdName nam)
{
- Vector matches = new Vector();
+ ArrayList<SequenceI> matches = new ArrayList<SequenceI>();
while (names.containsKey(nam))
{
- matches.addElement(names.remove(nam));
+ matches.add(names.remove(nam));
}
- SequenceI[] r = pickbestMatches(nam, matches);
+ List<SequenceI> r = pickbestMatches(nam, matches);
return r;
}
- private class SeqIdName
+ class SeqIdName
{
String id;
{
if (s != null)
{
- id = new String(s);
+ id = s.toLowerCase(Locale.ROOT);
}
else
{
}
}
+ @Override
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());
}
+ @Override
public boolean equals(Object s)
{
+ if (s == null)
+ {
+ return false;
+ }
if (s instanceof SeqIdName)
{
- return this.equals((SeqIdName) s);
+ return this.stringequals(((SeqIdName) s).id);
}
else
{
if (s instanceof String)
{
- return this.equals((String) s);
+ return this.stringequals(((String) s).toLowerCase(Locale.ROOT));
}
}
* todo: (JBPNote) Set separator characters appropriately
*
* @param s
- * SeqIdName
* @return boolean
*/
- public boolean equals(SeqIdName s)
+ private boolean stringequals(String s)
{
- // TODO: JAL-732 patch for cases when name includes a list of IDs, and the
- // match contains one ID flanked
- if (id.length() > s.id.length())
+ if (id.length() > s.length())
{
- return id.startsWith(s.id) ? (WORD_SEP.indexOf(id.charAt(s.id
- .length())) > -1) : false;
+ return id.startsWith(s)
+ ? (WORD_SEP.indexOf(id.charAt(s.length())) > -1)
+ : false;
}
else
{
- return s.id.startsWith(id) ? (s.id.equals(id) ? true : (WORD_SEP
- .indexOf(s.id.charAt(id.length())) > -1)) : false;
+ return s.startsWith(id)
+ ? (s.equals(id) ? true
+ : (WORD_SEP.indexOf(s.charAt(id.length())) > -1))
+ : false;
}
}
- public boolean equals(String s)
+ /**
+ * toString method returns the wrapped sequence id. For debugging purposes
+ * only, behaviour not guaranteed not to change.
+ */
+ @Override
+ public String toString()
{
- 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;
- }
+ return id;
}
}
}
git://source.jalview.org / jalview.git / blobdiff