src/jalview/ext/ensembl/SeqFetcher.java

   1 package jalview.ext.ensembl;
   2
   3 import jalview.io.FileParse;
   4
   5 import java.io.BufferedReader;
   6 import java.io.DataOutputStream;
   7 import java.io.IOException;
   8 import java.io.InputStream;
   9 import java.io.InputStreamReader;
  10 import java.net.HttpURLConnection;
  11 import java.net.URL;
  12 import java.net.URLConnection;
  13 import java.util.ArrayList;
  14 import java.util.List;
  15
  16 import org.apache.http.NameValuePair;
  17 import org.apache.http.message.BasicNameValuePair;
  18
  19 public class SeqFetcher
  20 {
  21   private final static String ENSEMBL_REST = "rest.ensembl.org";
  22
  23   private static final String SEQUENCE_ID_URL = "http://" + ENSEMBL_REST + "/sequence/id";
  24
  25   private static final String PING_URL = "http://" + ENSEMBL_REST + "/info/ping";
  26
  27   private final static long RETEST_INTERVAL = 10000L; // 10 seconds
  28
  29   private static boolean ensemblRestAvailable = false;
  30
  31   private static long lastCheck = -1;
  32
  33   /**
  34    * Rechecks if Ensembl is responding, unless the last check was successful and
  35    * the retest interval has not yet elapsed. Returns true if Ensembl is up,
  36    * else false.
  37    *
  38    * @return
  39    */
  40   public boolean isEnsemblAvailable()
  41   {
  42     long now = System.currentTimeMillis();
  43     boolean retest = now - lastCheck > RETEST_INTERVAL;
  44     if (ensemblRestAvailable && !retest)
  45     {
  46       return true;
  47     }
  48     ensemblRestAvailable = checkEnsembl();
  49     lastCheck = now;
  50     return ensemblRestAvailable;
  51   }
  52
  53   /**
  54    * Tries to connect to Ensembl's REST 'ping' endpoint, and returns true if
  55    * successful, else false
  56    *
  57    * @return
  58    */
  59   private boolean checkEnsembl()
  60   {
  61     try
  62     {
  63       URL ping = new URL(PING_URL);
  64       HttpURLConnection conn = (HttpURLConnection) ping.openConnection();
  65       int rc = conn.getResponseCode();
  66       conn.disconnect();
  67       if (rc >= 200 && rc < 300)
  68       {
  69         return true;
  70       }
  71     } catch (Throwable t)
  72     {
  73       System.err.println("Error connecting to " + PING_URL + ": "
  74               + t.getMessage());
  75     }
  76     return false;
  77   }
  78
  79   public SeqFetcher()
  80   {
  81   }
  82
  83   public enum EnsemblSeqType
  84   {
  85     GENOMIC("genomic"), CDS("cds"), TRANSCRIPT("cds"), PROTEIN("protein"), CDNA(
  86             "cdna");
  87
  88     private String type;
  89
  90     EnsemblSeqType(String t)
  91     {
  92       type = t;
  93     }
  94
  95     public String getType()
  96     {
  97       return type;
  98     }
  99   }
 100
 101   /**
 102    * Returns a list of additional URL query parameters to specify the desired
 103    * sequence type (genomic/cds/protein etc), and data format Fasta
 104    *
 105    * @param type
 106    */
 107   public List<NameValuePair> getAdditionalParameters(EnsemblSeqType type)
 108   {
 109     List<NameValuePair> params = new ArrayList<NameValuePair>();
 110     params.add(new BasicNameValuePair("type", type.getType()));
 111     params.add(new BasicNameValuePair("content-type", "text/x-fasta"));
 112     return params;
 113   }
 114
 115   /**
 116    * return a reader to a Fasta response from the Ensembl sequence endpoint
 117    *
 118    * @param returnType
 119    * @param ids
 120    * @return
 121    * @throws IOException
 122    */
 123   public FileParse getSequenceReader(EnsemblSeqType returnType,
 124           List<String> ids) throws IOException
 125   {
 126     // see http://rest.ensembl.org/documentation/info/sequence_id
 127
 128     String urlstring = SEQUENCE_ID_URL;
 129     List<NameValuePair> vals = getAdditionalParameters(returnType);
 130     boolean first = true;
 131     for (NameValuePair nvp : vals)
 132     {
 133       urlstring += first ? "?" : "&";
 134       first = false;
 135       urlstring += nvp.getName() + "=" + nvp.getValue();
 136     }
 137
 138     URL url = new URL(urlstring);
 139
 140     URLConnection connection = url.openConnection();
 141     HttpURLConnection httpConnection = (HttpURLConnection) connection;
 142
 143     httpConnection.setRequestMethod("POST");
 144     httpConnection.setRequestProperty("Content-Type", "application/json");
 145     httpConnection.setRequestProperty("Accept", "text/x-fasta");
 146     byte[] thepostbody;
 147     {
 148       StringBuilder postBody = new StringBuilder();
 149       postBody.append("{\"ids\":[");
 150       first = true;
 151       for (String id : ids)
 152       {
 153         if (!first)
 154         {
 155           postBody.append(",");
 156         }
 157         first = false;
 158         postBody.append("\"");
 159         postBody.append(id.trim());
 160         postBody.append("\"");
 161       }
 162       postBody.append("]}");
 163       thepostbody = postBody.toString().getBytes();
 164     }
 165     httpConnection.setRequestProperty("Content-Length",
 166             Integer.toString(thepostbody.length));
 167     httpConnection.setUseCaches(false);
 168     httpConnection.setDoInput(true);
 169     httpConnection.setDoOutput(true);
 170
 171     DataOutputStream wr = new DataOutputStream(
 172             httpConnection.getOutputStream());
 173     wr.write(thepostbody);
 174     wr.flush();
 175     wr.close();
 176
 177     InputStream response = connection.getInputStream();
 178     int responseCode = httpConnection.getResponseCode();
 179
 180     if (responseCode != 200)
 181     {
 182       throw new RuntimeException(
 183               "Response code was not 200. Detected response was "
 184                       + responseCode);
 185     }
 186
 187     BufferedReader reader = null;
 188     reader = new BufferedReader(new InputStreamReader(response, "UTF-8"));
 189     FileParse fp = new FileParse(reader, url.toString(), "HTTP_POST");
 190     return fp;
 191   }
 192
 193 }