Merge branch 'alpha/origin_2022_JAL-3066_Jalview_212_slivka-integration' into spike...

[jalview.git] / src / jalview / ws / dbsources / Uniprot.java

diff --git a/src/jalview/ws/dbsources/Uniprot.java b/src/jalview/ws/dbsources/Uniprot.java
index 2fc5893..0775870 100644 (file)
--- a/src/jalview/ws/dbsources/Uniprot.java
+++ b/src/jalview/ws/dbsources/Uniprot.java
@@ -20,6 +20,7 @@
  */
 package jalview.ws.dbsources;
 
+import java.util.Locale;
 import jalview.bin.Cache;
 import jalview.datamodel.Alignment;
 import jalview.datamodel.AlignmentI;
@@ -32,7 +33,6 @@ import jalview.datamodel.SequenceI;
 import jalview.schemes.ResidueProperties;
 import jalview.util.StringUtils;
 import jalview.ws.seqfetcher.DbSourceProxyImpl;
-import jalview.xml.binding.embl.ROOT;
 import jalview.xml.binding.uniprot.DbReferenceType;
 import jalview.xml.binding.uniprot.Entry;
 import jalview.xml.binding.uniprot.FeatureType;
@@ -41,8 +41,8 @@ import jalview.xml.binding.uniprot.PositionType;
 import jalview.xml.binding.uniprot.PropertyType;
 
 import java.io.InputStream;
+import java.net.HttpURLConnection;
 import java.net.URL;
-import java.net.URLConnection;
 import java.util.ArrayList;
 import java.util.List;
 import java.util.Vector;
@@ -70,6 +70,7 @@ public class Uniprot extends DbSourceProxyImpl
   private static final String DEFAULT_UNIPROT_DOMAIN = "https://www.uniprot.org";
 
   private static final String BAR_DELIMITER = "|";
+  private static Regex ACCESSION_REGEX;
 
   /**
    * Constructor
@@ -103,7 +104,12 @@ public class Uniprot extends DbSourceProxyImpl
   @Override
   public Regex getAccessionValidator()
   {
-    return new Regex("([A-Z]+[0-9]+[A-Z0-9]+|[A-Z0-9]+_[A-Z0-9]+)");
+    if (ACCESSION_REGEX == null)
+    {
+      ACCESSION_REGEX = Platform
+              .newRegex("([A-Z]+[0-9]+[A-Z0-9]+|[A-Z0-9]+_[A-Z0-9]+)");
+    }
+    return ACCESSION_REGEX;
   }
 
   /*
@@ -139,7 +145,7 @@ public class Uniprot extends DbSourceProxyImpl
     startQuery();
     try
     {
-      queries = queries.toUpperCase().replaceAll(
+      queries = queries.toUpperCase(Locale.ROOT).replaceAll(
               "(UNIPROT\\|?|UNIPROT_|UNIREF\\d+_|UNIREF\\d+\\|?)", "");
       AlignmentI al = null;
 
@@ -147,10 +153,11 @@ public class Uniprot extends DbSourceProxyImpl
               + ".xml";
 
       URL url = new URL(downloadstring);
-      URLConnection urlconn = url.openConnection();
-      InputStream istr = urlconn.getInputStream();
-      List<Entry> entries = getUniprotEntries(istr);
-      if (entries != null)
+      HttpURLConnection urlconn = (HttpURLConnection)url.openConnection();
+      // anything other than 200 means we don't have data
+      // TODO: JAL-3882 reuse the EnsemblRestClient's fair 
+      // use/backoff logic to retry when the server tells us to go away
+      if (urlconn.getResponseCode() == 200)
       {
         List<SequenceI> seqs = new ArrayList<>();
         for (Entry entry : entries)
@@ -162,6 +169,7 @@ public class Uniprot extends DbSourceProxyImpl
 
       stopQuery();
       return al;
+      
     } catch (Exception e)
     {
       throw (e);
@@ -516,6 +524,11 @@ public class Uniprot extends DbSourceProxyImpl
     } catch (JAXBException | XMLStreamException
             | FactoryConfigurationError e)
     {
+      if (e instanceof javax.xml.bind.UnmarshalException && e.getCause()!=null && e.getCause() instanceof XMLStreamException && e.getCause().getMessage().contains("[row,col]:[1,1]"))
+      {
+        // trying to parse an empty stream
+        return null;
+      }
       e.printStackTrace();
     }
     return entries;