JAL-2656 recognise Gzipped file input streams by their content, not their filename
authorJim Procter <jprocter@issues.jalview.org>
Tue, 28 Jul 2020 10:29:20 +0000 (11:29 +0100)
committerJim Procter <jprocter@issues.jalview.org>
Tue, 28 Jul 2020 10:29:20 +0000 (11:29 +0100)
src/jalview/io/FileParse.java

index 167314e..128672a 100755 (executable)
@@ -26,10 +26,10 @@ 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;
@@ -38,6 +38,7 @@ 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;
 
 
@@ -185,29 +186,64 @@ public class FileParse
     }
     if (!error)
     {
-      if (fileStr.toLowerCase().endsWith(".gz"))
+      try
       {
-        try
-        {
-          dataIn = getGzipReader(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);
+  }
 
   /**
+   * 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.