JAL-2656 recognise Gzipped file input streams by their content, not their filename
[jalview.git] / src / jalview / io / FileParse.java
index 7117d0f..128672a 100755 (executable)
@@ -26,19 +26,22 @@ import jalview.api.AlignmentViewPanel;
 import jalview.api.FeatureSettingsModelI;
 import jalview.util.MessageManager;
 
+import java.io.BufferedInputStream;
 import java.io.BufferedReader;
 import java.io.File;
 import java.io.FileInputStream;
-import java.io.FileReader;
 import java.io.IOException;
 import java.io.InputStream;
 import java.io.InputStreamReader;
 import java.io.Reader;
 import java.io.StringReader;
+import java.net.HttpURLConnection;
 import java.net.MalformedURLException;
 import java.net.URL;
+import java.net.URLConnection;
 import java.util.zip.GZIPInputStream;
 
+
 /**
  * implements a random access wrapper around a particular datasource, for
  * passing to identifyFile and AlignFile objects.
@@ -56,6 +59,7 @@ public class FileParse
 
   public File inFile = null;
 
+
   /**
    * a viewport associated with the current file operation. May be null. May
    * move to different object.
@@ -182,29 +186,73 @@ public class FileParse
     }
     if (!error)
     {
-      if (fileStr.toLowerCase().endsWith(".gz"))
+      try
       {
-        try
-        {
-          dataIn = tryAsGzipSource(new FileInputStream(fileStr));
-          dataName = fileStr;
-          return error;
-        } catch (Exception x)
-        {
-          warningMessage = "Failed  to resolve as a GZ stream ("
-                  + x.getMessage() + ")";
-          // x.printStackTrace();
-        }
-        ;
+        dataIn = checkForGzipStream(new FileInputStream(fileStr));
+        dataName = fileStr;
+      } catch (Exception x)
+      {
+        warningMessage = "Failed to resolve " + fileStr
+                + " as a data source. (" + x.getMessage() + ")";
+        // x.printStackTrace();
+        error = true;
       }
-
-      dataIn = new BufferedReader(new FileReader(fileStr));
-      dataName = fileStr;
+      ;
     }
     return error;
   }
+  
+  /**
+   * Recognise the 2-byte magic header for gzip streams
+   * 
+   * https://recalll.co/ask/v/topic/java-How-to-check-if-InputStream-is-Gzipped/555aadd62bd27354438b90f6
+   * 
+   * @param bytes - at least two bytes 
+   * @return 
+   */
+  private static boolean isGzipStream(byte[] bytes) {
+    int head = ((int) bytes[0] & 0xff) | ((bytes[1] << 8) & 0xff00);
+    return (GZIPInputStream.GZIP_MAGIC == head);
+  }
 
-  private BufferedReader tryAsGzipSource(InputStream inputStream)
+  /**
+   * Returns a Reader for the given input after wrapping it in a buffered input
+   * stream, and then checking if it needs to be wrapped by a GZipInputStream
+   * 
+   * @param input
+   * @return
+   */
+  private BufferedReader checkForGzipStream(InputStream input) throws Exception {
+
+    // NB: stackoverflow https://stackoverflow.com/questions/4818468/how-to-check-if-inputstream-is-gzipped
+    // could use a PushBackInputStream rather than a BufferedInputStream
+    
+    BufferedInputStream bufinput;
+    if (!input.markSupported()) {
+       bufinput= new BufferedInputStream(input,16);
+       input = bufinput;
+    }
+    input.mark(4);
+    byte[] bytes=input.readNBytes(2);
+    input.reset();
+    if (bytes.length==2 && isGzipStream(bytes)) {
+      return getGzipReader(input);
+    }
+    // return a buffered reader for the stream.
+    InputStreamReader isReader= new InputStreamReader(input);
+    BufferedReader toReadFrom=new BufferedReader(isReader);
+    return toReadFrom;
+  }
+  /**
+   * Returns a {@code BufferedReader} which wraps the input stream with a
+   * GZIPInputStream. Throws a {@code ZipException} if a GZIP format error
+   * occurs or the compression method used is unsupported.
+   * 
+   * @param inputStream
+   * @return
+   * @throws Exception
+   */
+  private BufferedReader getGzipReader(InputStream inputStream)
           throws Exception
   {
     BufferedReader inData = new BufferedReader(
@@ -215,44 +263,58 @@ public class FileParse
     return inData;
   }
 
-  private boolean checkURLSource(String fileStr)
+  /**
+   * Tries to read from the given URL. If successful, saves a reader to the
+   * response in field {@code dataIn}, otherwise (on exception, or HTTP response
+   * status not 200), throws an exception.
+   * <p>
+   * If the response status includes
+   * 
+   * <pre>
+   * Content-Type : application/x-gzip
+   * </pre>
+   * 
+   * then tries to read as gzipped content.
+   * 
+   * @param urlStr
+   * @throws IOException
+   * @throws MalformedURLException
+   */
+  private void checkURLSource(String urlStr)
           throws IOException, MalformedURLException
   {
     errormessage = "URL NOT FOUND";
-    URL url = new URL(fileStr);
-    //
-    // GZIPInputStream code borrowed from Aquaria (soon to be open sourced) via
-    // Kenny Sabir
+    URL url = new URL(urlStr);
+    HttpURLConnection conn = (HttpURLConnection) url.openConnection();
+    int rc = conn.getResponseCode();
+    if (rc != HttpURLConnection.HTTP_OK)
+    {
+      throw new IOException(
+              "Response status from " + urlStr + " was " + rc);
+    }
+    String encoding = conn.getContentEncoding();
+    String contentType = conn.getContentType();
+    boolean isgzipped = "application/x-gzip".equalsIgnoreCase(contentType)
+            || "gzip".equals(encoding);
     Exception e = null;
-    if (fileStr.toLowerCase().endsWith(".gz"))
+    InputStream inputStream = conn.getInputStream();
+    if (isgzipped)
     {
       try
       {
-        InputStream inputStream = url.openStream();
-        dataIn = tryAsGzipSource(inputStream);
-        dataName = fileStr;
-        return false;
-      } catch (Exception ex)
-      {
-        e = ex;
-      }
-    }
-
-    try
-    {
-      dataIn = new BufferedReader(new InputStreamReader(url.openStream()));
-    } catch (IOException q)
-    {
-      if (e != null)
+        dataIn = getGzipReader(inputStream);
+        dataName = urlStr;
+      } catch (Exception e1)
       {
         throw new IOException(MessageManager
                 .getString("exception.failed_to_resolve_gzip_stream"), e);
       }
-      throw q;
+      return;
     }
-    // record URL as name of datasource.
-    dataName = fileStr;
-    return false;
+
+    dataIn = new BufferedReader(new InputStreamReader(inputStream));
+    dataName = urlStr;
+    return;
   }
 
   /**