JAL-2154 always compare canonicalised DBRefEntry source strings!

[jalview.git] / src / jalview / util / DBRefUtils.java

diff --git a/src/jalview/util/DBRefUtils.java b/src/jalview/util/DBRefUtils.java
index ed6d860..405f6e6 100755 (executable)
--- a/src/jalview/util/DBRefUtils.java
+++ b/src/jalview/util/DBRefUtils.java
@@ -31,6 +31,7 @@ import java.util.HashSet;
 import java.util.Hashtable;
 import java.util.List;
 import java.util.Map;
+import java.util.Set;
 
 import com.stevesoft.pat.Regex;
 
@@ -59,6 +60,18 @@ public class DBRefUtils
 
     canonicalSourceNameLookup.put("pdb", DBRefSource.PDB);
     canonicalSourceNameLookup.put("ensembl", DBRefSource.ENSEMBL);
+    // Ensembl Gn and Tr are for Ensembl genomic and transcript IDs as served
+    // from ENA.
+    canonicalSourceNameLookup.put("ensembl-tr", DBRefSource.ENSEMBL);
+    canonicalSourceNameLookup.put("ensembl-gn", DBRefSource.ENSEMBL);
+
+    // Make sure we have lowercase entries for all canonical string lookups
+    Set<String> keys = canonicalSourceNameLookup.keySet();
+    for (String k : keys)
+    {
+      canonicalSourceNameLookup.put(k.toLowerCase(),
+              canonicalSourceNameLookup.get(k));
+    }
 
     dasCoordinateSystemsLookup.put("pdbresnum", DBRefSource.PDB);
     dasCoordinateSystemsLookup.put("uniprot", DBRefSource.UNIPROT);
@@ -235,7 +248,8 @@ public class DBRefUtils
     public boolean matches(DBRefEntry refa, DBRefEntry refb)
     {
       if (refa.getSource() == null
-              || refb.getSource().equals(refa.getSource()))
+              || DBRefUtils.getCanonicalName(refb.getSource()).equals(
+                      DBRefUtils.getCanonicalName(refa.getSource())))
       {
         if (refa.getVersion() == null
                 || refb.getVersion().equals(refa.getVersion()))
@@ -266,7 +280,7 @@ public class DBRefUtils
     @Override
     public boolean matches(DBRefEntry refa, DBRefEntry refb)
     {
-      if (nullOrEqual(refa.getSource(), refb.getSource())
+      if (nullOrEqualSource(refa.getSource(), refb.getSource())
               && nullOrEqual(refa.getVersion(), refb.getVersion())
               && nullOrEqual(refa.getAccessionId(), refb.getAccessionId())
               && nullOrEqual(refa.getMap(), refb.getMap()))
@@ -288,7 +302,8 @@ public class DBRefUtils
     public boolean matches(DBRefEntry refa, DBRefEntry refb)
     {
       if (refa.getSource() != null && refb.getSource() != null
-              && refb.getSource().equals(refa.getSource()))
+              && DBRefUtils.getCanonicalName(refb.getSource()).equals(
+                      DBRefUtils.getCanonicalName(refa.getSource())))
       {
         // We dont care about version
         if (refa.getAccessionId() != null && refb.getAccessionId() != null
@@ -319,7 +334,8 @@ public class DBRefUtils
     public boolean matches(DBRefEntry refa, DBRefEntry refb)
     {
       if (refa.getSource() != null && refb.getSource() != null
-              && refb.getSource().equals(refa.getSource()))
+              && DBRefUtils.getCanonicalName(refb.getSource()).equals(
+                      DBRefUtils.getCanonicalName(refa.getSource())))
       {
         // We dont care about version
         if (refa.getAccessionId() != null && refb.getAccessionId() != null
@@ -355,7 +371,8 @@ public class DBRefUtils
     public boolean matches(DBRefEntry refa, DBRefEntry refb)
     {
       if (refa.getSource() != null && refb.getSource() != null
-              && refb.getSource().equals(refa.getSource()))
+              && DBRefUtils.getCanonicalName(refb.getSource()).equals(
+                      DBRefUtils.getCanonicalName(refa.getSource())))
       {
         // We dont care about version
         // if ((refa.getVersion()==null || refb.getVersion()==null)
@@ -394,7 +411,8 @@ public class DBRefUtils
     public boolean matches(DBRefEntry refa, DBRefEntry refb)
     {
       if (refa.getSource() != null && refb.getSource() != null
-              && refb.getSource().equals(refa.getSource()))
+              && DBRefUtils.getCanonicalName(refb.getSource()).equals(
+                      DBRefUtils.getCanonicalName(refa.getSource())))
       {
         // We dont care about version
 
@@ -521,7 +539,28 @@ public class DBRefUtils
     {
       return true;
     }
-    return (o1 == null ? o2.equals(o1) : o1.equals(o2));
+    return o1.equals(o2);
+  }
+
+  /**
+   * canonicalise source string before comparing. null is always wildcard
+   * 
+   * @param o1
+   *          - null or source string to compare
+   * @param o2
+   *          - null or source string to compare
+   * @return true if either o1 or o2 are null, or o1 equals o2 under
+   *         DBRefUtils.getCanonicalName
+   *         (o1).equals(DBRefUtils.getCanonicalName(o2))
+   */
+  public static boolean nullOrEqualSource(String o1, String o2)
+  {
+    if (o1 == null || o2 == null)
+    {
+      return true;
+    }
+    return DBRefUtils.getCanonicalName(o1).equals(
+            DBRefUtils.getCanonicalName(o2));
   }
 
   /**
@@ -546,7 +585,7 @@ public class DBRefUtils
 
   /**
    * Returns the (possibly empty) list of those supplied dbrefs which have the
-   * specified source databse
+   * specified source database, with a case-insensitive match of source name
    * 
    * @param dbRefs
    * @param source
@@ -560,7 +599,7 @@ public class DBRefUtils
     {
       for (DBRefEntry dbref : dbRefs)
       {
-        if (source.equals(dbref.getSource()))
+        if (source.equalsIgnoreCase(dbref.getSource()))
         {
           matches.add(dbref);
         }