JAL-3615 used gzip endpoints for Pfam and Rfam

diff --git a/src/jalview/io/FileParse.java b/src/jalview/io/FileParse.java
index a5a4e36..dac8fe6 100755 (executable)
--- a/src/jalview/io/FileParse.java
+++ b/src/jalview/io/FileParse.java
@@ -40,6 +40,8 @@ import jalview.api.AlignmentViewPanel;
 import jalview.api.FeatureSettingsModelI;
 import jalview.util.MessageManager;
 import jalview.util.Platform;
+import jalview.ws.dbsources.Pfam;
+import jalview.ws.dbsources.Rfam;
 
 /**
  * implements a random access wrapper around a particular datasource, for
@@ -51,6 +53,8 @@ public class FileParse
 
   protected static final String TAB = "\t";
 
+  private static final String GZ_EXT = ".gz";
+
   /**
    * text specifying source of data. usually filename or url.
    */
@@ -64,6 +68,7 @@ public class FileParse
   {
     return bytes;
   }
+
   /**
    * a viewport associated with the current file operation. May be null. May
    * move to different object.
@@ -190,7 +195,7 @@ public class FileParse
     }
     if (!error)
     {
-      if (fileStr.toLowerCase().endsWith(".gz"))
+      if (fileStr.toLowerCase().endsWith(GZ_EXT))
       {
         try
         {
@@ -232,7 +237,7 @@ public class FileParse
     // GZIPInputStream code borrowed from Aquaria (soon to be open sourced) via
     // Kenny Sabir
     Exception e = null;
-    if (fileStr.toLowerCase().endsWith(".gz"))
+    if (isGzipped(fileStr))
     {
       try
       {
@@ -264,6 +269,27 @@ public class FileParse
   }
 
   /**
+   * Answers true if the filename (or URL) has a format which Jalview recognises
+   * as denoting gzipped content.
+   * <p>
+   * Currently this means having a ".gz" extension, or ending in "/gzipped" or
+   * "?gz=1" (used to retrieve gzipped from Pfam and Rfam respectively).
+   * 
+   * @param filename
+   * @return
+   */
+  protected static boolean isGzipped(String filename)
+  {
+    if (filename == null)
+    {
+      return false;
+    }
+    String lower = filename.toLowerCase();
+    return lower.endsWith(GZ_EXT) || lower.endsWith(Pfam.GZIPPED)
+            || lower.endsWith(Rfam.GZIPPED);
+  }
+
+  /**
    * sets the suffix string (if any) and returns remainder (if suffix was
    * detected)
    * 
@@ -345,7 +371,8 @@ public class FileParse
       {
         // this will be from JavaScript
         inFile = file;
-        dataIn = new BufferedReader(new InputStreamReader(new ByteArrayInputStream(bytes)));
+        dataIn = new BufferedReader(
+                new InputStreamReader(new ByteArrayInputStream(bytes)));
         dataName = fileStr;
       }
       else if (checkFileSource(fileStr))
@@ -453,8 +480,7 @@ public class FileParse
     {
       // pass up the reason why we have no source to read from
       throw new IOException(MessageManager.formatMessage(
-              "exception.failed_to_read_data_from_source",
-              new String[]
+              "exception.failed_to_read_data_from_source", new String[]
               { errormessage }));
     }
     error = false;

diff --git a/src/jalview/ws/dbsources/Pfam.java b/src/jalview/ws/dbsources/Pfam.java
index 8877c34..de19078 100644 (file)
--- a/src/jalview/ws/dbsources/Pfam.java
+++ b/src/jalview/ws/dbsources/Pfam.java
@@ -35,6 +35,11 @@ import com.stevesoft.pat.Regex;
  */
 abstract public class Pfam extends Xfam
 {
+  /*
+   * append to URLs to retrieve as a gzipped file
+   */
+  public static final String GZIPPED = "/gzipped";
+
   static final String PFAM_BASEURL_KEY = "PFAM_BASEURL";
 
   private static final String DEFAULT_PFAM_BASEURL = "https://pfam.xfam.org";

diff --git a/src/jalview/ws/dbsources/PfamFull.java b/src/jalview/ws/dbsources/PfamFull.java
index 0600427..d71892b 100644 (file)
--- a/src/jalview/ws/dbsources/PfamFull.java
+++ b/src/jalview/ws/dbsources/PfamFull.java
@@ -34,7 +34,7 @@ public class PfamFull extends Pfam
   @Override
   public String getURLSuffix()
   {
-    return "/alignment/full";
+    return "/alignment/full" + GZIPPED;
   }
 
   /*

diff --git a/src/jalview/ws/dbsources/PfamSeed.java b/src/jalview/ws/dbsources/PfamSeed.java
index dff8a17..f64d07f 100644 (file)
--- a/src/jalview/ws/dbsources/PfamSeed.java
+++ b/src/jalview/ws/dbsources/PfamSeed.java
@@ -36,7 +36,7 @@ public class PfamSeed extends Pfam
   @Override
   public String getURLSuffix()
   {
-    return "/alignment/seed";
+    return "/alignment/seed" + GZIPPED;
   }
 
   /*

diff --git a/src/jalview/ws/dbsources/Rfam.java b/src/jalview/ws/dbsources/Rfam.java
index 1d9d99a..dba3e31 100644 (file)
--- a/src/jalview/ws/dbsources/Rfam.java
+++ b/src/jalview/ws/dbsources/Rfam.java
@@ -36,6 +36,11 @@ abstract public class Rfam extends Xfam
 
   private static final String DEFAULT_RFAM_BASEURL = "https://rfam.xfam.org";
 
+  /*
+   * append to URLs to retrieve as a gzipped file
+   */
+  public static final String GZIPPED = "?gz=1&download=1";
+
   @Override
   protected String getURLPrefix()
   {

diff --git a/src/jalview/ws/dbsources/RfamFull.java b/src/jalview/ws/dbsources/RfamFull.java
index d815336..396511c 100644 (file)
--- a/src/jalview/ws/dbsources/RfamFull.java
+++ b/src/jalview/ws/dbsources/RfamFull.java
@@ -36,7 +36,7 @@ public class RfamFull extends Rfam
   @Override
   public String getURLSuffix()
   {
-    return "/alignment/full";
+    return "/alignment/full" + GZIPPED;
   }
 
   /*

diff --git a/src/jalview/ws/dbsources/RfamSeed.java b/src/jalview/ws/dbsources/RfamSeed.java
index a74e829..eaa574b 100644 (file)
--- a/src/jalview/ws/dbsources/RfamSeed.java
+++ b/src/jalview/ws/dbsources/RfamSeed.java
@@ -36,8 +36,7 @@ public class RfamSeed extends Rfam
   @Override
   public String getURLSuffix()
   {
-    // to download gzipped file add '?gzip=1'
-    return "/alignment/stockholm";
+    return "/alignment/stockholm" + GZIPPED;
   }
 
   /*

diff --git a/src/jalview/ws/dbsources/Xfam.java b/src/jalview/ws/dbsources/Xfam.java
index b83f558..f0cb14b 100644 (file)
--- a/src/jalview/ws/dbsources/Xfam.java
+++ b/src/jalview/ws/dbsources/Xfam.java
@@ -36,7 +36,6 @@ import jalview.ws.seqfetcher.DbSourceProxyImpl;
  */
 public abstract class Xfam extends DbSourceProxyImpl
 {
-
   public Xfam()
   {
     super();

diff --git a/test/jalview/io/FileParseTest.java b/test/jalview/io/FileParseTest.java
new file mode 100644 (file)
index 0000000..7eb2c9c
--- /dev/null
+++ b/test/jalview/io/FileParseTest.java
@@ -0,0 +1,34 @@
+package jalview.io;
+
+import static org.testng.Assert.assertFalse;
+import static org.testng.Assert.assertTrue;
+
+import org.testng.annotations.Test;
+
+public class FileParseTest
+{
+  @Test(groups = "Functional")
+  public void setIsGzipped()
+  {
+    assertFalse(FileParse.isGzipped(null));
+    assertFalse(FileParse.isGzipped("foobar"));
+    assertFalse(FileParse.isGzipped(".gz.foobar"));
+
+    assertTrue(FileParse.isGzipped("abc.gz"));
+    assertTrue(FileParse.isGzipped("abc.GZ"));
+    assertTrue(FileParse.isGzipped(".gz"));
+    assertFalse(FileParse.isGzipped("abc/gz"));
+    assertFalse(FileParse.isGzipped("gz"));
+    
+    assertTrue(FileParse.isGzipped("http:/xy.com/abc/gzipped"));
+    assertTrue(FileParse.isGzipped("abc/gzipped"));
+    assertTrue(FileParse.isGzipped("abc/GZIPPED"));
+    assertTrue(FileParse.isGzipped("/gzipped"));
+    assertFalse(FileParse.isGzipped("gzipped"));
+    
+    assertTrue(FileParse.isGzipped("http:/xy.com/abc?gz=1"));
+    assertTrue(FileParse.isGzipped("http:/xy.com/abc?GZ=1"));
+    // currently only recognised if the last token on the URL
+    assertFalse(FileParse.isGzipped("http:/xy.com/abc?gz=1&content-type=text/xml"));
+  }
+}