JAL-2582 recognise ENS…P peptide IDs and make two getParent requests to get gene ID.

[jalview.git] / src / jalview / ext / ensembl / EnsemblRestClient.java

diff --git a/src/jalview/ext/ensembl/EnsemblRestClient.java b/src/jalview/ext/ensembl/EnsemblRestClient.java
index b7b1985..2437588 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblRestClient.java
+++ b/src/jalview/ext/ensembl/EnsemblRestClient.java
@@ -20,6 +20,7 @@
  */
 package jalview.ext.ensembl;
 
+import jalview.io.DataSourceType;
 import jalview.io.FileParse;
 import jalview.util.StringUtils;
 
@@ -50,15 +51,19 @@ import com.stevesoft.pat.Regex;
  */
 abstract class EnsemblRestClient extends EnsemblSequenceFetcher
 {
+  private static final int DEFAULT_READ_TIMEOUT = 5 * 60 * 1000; // 5 minutes
+
+  private static final int CONNECT_TIMEOUT_MS = 10 * 1000; // 10 seconds
+
   /*
    * update these constants when Jalview has been checked / updated for
-   * changes to Ensembl REST API
+   * changes to Ensembl REST API (ref JAL-2105)
    * @see https://github.com/Ensembl/ensembl-rest/wiki/Change-log
    * @see http://rest.ensembl.org/info/rest?content-type=application/json
    */
-  private static final String LATEST_ENSEMBLGENOMES_REST_VERSION = "4.6";
+  private static final String LATEST_ENSEMBLGENOMES_REST_VERSION = "5.0";
 
-  private static final String LATEST_ENSEMBL_REST_VERSION = "4.7";
+  private static final String LATEST_ENSEMBL_REST_VERSION = "5.0";
 
   private static final String REST_CHANGE_LOG = "https://github.com/Ensembl/ensembl-rest/wiki/Change-log";
 
@@ -71,6 +76,9 @@ abstract class EnsemblRestClient extends EnsemblSequenceFetcher
 
   private final static long VERSION_RETEST_INTERVAL = 1000L * 3600; // 1 hr
 
+  private static final Regex PROTEIN_REGEX = new Regex(
+          "(ENS)([A-Z]{3}|)P[0-9]{11}$");
+
   private static final Regex TRANSCRIPT_REGEX = new Regex(
           "(ENS)([A-Z]{3}|)T[0-9]{11}$");
 
@@ -120,6 +128,18 @@ abstract class EnsemblRestClient extends EnsemblSequenceFetcher
 
   /**
    * Answers true if the query matches the regular expression pattern for an
+   * Ensembl protein stable identifier
+   * 
+   * @param query
+   * @return
+   */
+  public boolean isProteinIdentifier(String query)
+  {
+    return query == null ? false : PROTEIN_REGEX.search(query);
+  }
+
+  /**
+   * Answers true if the query matches the regular expression pattern for an
    * Ensembl gene stable identifier
    * 
    * @param query
@@ -186,7 +206,7 @@ abstract class EnsemblRestClient extends EnsemblSequenceFetcher
    */
   private boolean checkEnsembl()
   {
-    HttpURLConnection conn = null;
+    BufferedReader br = null;
     try
     {
       // note this format works for both ensembl and ensemblgenomes
@@ -196,8 +216,9 @@ abstract class EnsemblRestClient extends EnsemblSequenceFetcher
 
       /*
        * expect {"ping":1} if ok
+       * if ping takes more than 2 seconds to respond, treat as if unavailable
        */
-      BufferedReader br = getHttpResponse(ping, null);
+      br = getHttpResponse(ping, null, 2 * 1000);
       JSONParser jp = new JSONParser();
       JSONObject val = (JSONObject) jp.parse(br);
       String pingString = val.get("ping").toString();
@@ -208,9 +229,15 @@ abstract class EnsemblRestClient extends EnsemblSequenceFetcher
               + t.getMessage());
     } finally
     {
-      if (conn != null)
+      if (br != null)
       {
-        conn.disconnect();
+        try
+        {
+          br.close();
+        } catch (IOException e)
+        {
+          // ignore
+        }
       }
     }
     return false;
@@ -234,22 +261,39 @@ abstract class EnsemblRestClient extends EnsemblSequenceFetcher
       // request failed
       return null;
     }
-    FileParse fp = new FileParse(reader, url.toString(), "HTTP_POST");
+    FileParse fp = new FileParse(reader, url.toString(), DataSourceType.URL);
     return fp;
   }
 
   /**
+   * Gets a reader to the HTTP response, using the default read timeout of 5
+   * minutes
+   * 
+   * @param url
+   * @param ids
+   * @return
+   * @throws IOException
+   */
+  protected BufferedReader getHttpResponse(URL url, List<String> ids)
+          throws IOException
+  {
+    return getHttpResponse(url, ids, DEFAULT_READ_TIMEOUT);
+  }
+
+  /**
    * Writes the HTTP request and gets the response as a reader.
    * 
    * @param url
    * @param ids
    *          written as Json POST body if more than one
+   * @param readTimeout
+   *          in milliseconds
    * @return
    * @throws IOException
    *           if response code was not 200, or other I/O error
    */
-  protected BufferedReader getHttpResponse(URL url, List<String> ids)
-          throws IOException
+  protected BufferedReader getHttpResponse(URL url, List<String> ids,
+          int readTimeout) throws IOException
   {
     // long now = System.currentTimeMillis();
     HttpURLConnection connection = (HttpURLConnection) url.openConnection();
@@ -269,6 +313,9 @@ abstract class EnsemblRestClient extends EnsemblSequenceFetcher
     connection.setDoInput(true);
     connection.setDoOutput(multipleIds);
 
+    connection.setConnectTimeout(CONNECT_TIMEOUT_MS);
+    connection.setReadTimeout(readTimeout);
+
     if (multipleIds)
     {
       writePostBody(connection, ids);
@@ -481,9 +528,11 @@ abstract class EnsemblRestClient extends EnsemblSequenceFetcher
       boolean laterVersion = StringUtils.compareVersions(version, expected) == 1;
       if (laterVersion)
       {
-        System.err.println(String.format(
-                "Expected %s REST version %s but found %s, see %s",
-                getDbSource(), expected, version, REST_CHANGE_LOG));
+        System.err
+                .println(String
+                        .format("EnsemblRestClient expected %s REST version %s but found %s, see %s",
+                                getDbSource(), expected, version,
+                                REST_CHANGE_LOG));
       }
       info.restVersion = version;
     } catch (Throwable t)