Question

I’m building something on Google App Engine in Java (JDO). I’m programmatically compressing a large byte[] with Deflater and then storing the compressed byte[] as in the blobstore. This works great:

 public class Functions {

public static byte[] compress(byte[] input) throws UnsupportedEncodingException, IOException, MessagingException
    {

        Deflater df = new Deflater();       //this function mainly generate the byte code
        df.setLevel(Deflater.BEST_COMPRESSION);
        df.setInput(input);

        ByteArrayOutputStream baos = new ByteArrayOutputStream(input.length);   //we write the generated byte code in this array
        df.finish();
        byte[] buff = new byte[1024];   //segment segment pop....segment set 1024
        while(!df.finished())
        {
            int count = df.deflate(buff);       //returns the generated code... index
            baos.write(buff, 0, count);     //write 4m 0 to count
        }
        baos.close();

        int baosLength = baos.toByteArray().length;
        int inputLength = input.length;
        //System.out.println("Original: "+inputLength);
        // System.out.println("Compressed: "+ baosLength);

        return baos.toByteArray();

    }

 public static byte[] decompress(byte[] input) throws UnsupportedEncodingException, IOException, DataFormatException
    {

        Inflater decompressor = new Inflater();
        decompressor.setInput(input);

        // Create an expandable byte array to hold the decompressed data
        ByteArrayOutputStream bos = new ByteArrayOutputStream(input.length);

        // Decompress the data
        byte[] buf = new byte[1024];
        while (!decompressor.finished()) {
            try {
                int count = decompressor.inflate(buf);
                bos.write(buf, 0, count);
            } catch (DataFormatException e) {
            }
        }
        try {
            bos.close();
        } catch (IOException e) {
        }

        // Get the decompressed data
        byte[] decompressedData = bos.toByteArray();

        return decompressedData;


    }

 public static BlobKey putInBlobStore(String contentType, byte[] filebytes) throws IOException {

        // Get a file service
          FileService fileService = FileServiceFactory.getFileService();


          AppEngineFile file = fileService.createNewBlobFile(contentType);

          // Open a channel to write to it
          boolean lock = true;
          FileWriteChannel writeChannel = fileService.openWriteChannel(file, lock);

          // This time we write to the channel using standard Java
          BufferedInputStream in = new BufferedInputStream(new ByteArrayInputStream(filebytes));
          byte[] buffer;
          int defaultBufferSize = 524288;
          if(filebytes.length > defaultBufferSize){
              buffer = new byte[defaultBufferSize]; // 0.5 MB buffers
          }
          else{
              buffer = new byte[filebytes.length]; // buffer the size of the data
          }

            int read;
            while( (read = in.read(buffer)) > 0 ){ //-1 means EndOfStream
                System.out.println(read);
                if(read < defaultBufferSize){
                    buffer = new byte[read];
                }
                ByteBuffer bb = ByteBuffer.wrap(buffer);
                writeChannel.write(bb);
            }
            writeChannel.closeFinally();

        return fileService.getBlobKey(file);
    }
}

Using the static compress() and putInBlobStore() functions in my Functions class, I can compress and store a byte[] like so:

BlobKey dataBlobKey =  Functions.putInBlobStore("MULTIPART_FORM_DATA", Functions.compress(orginalDataByteArray));

Pretty sweet. I’m really digging GAE.

But now, the problem:

I’m storing compressed HTML which I’d like to retrieve and decompress on the fly to display in an iframe within a JSP page. Compression is quick, but decompression takes FOREVER! Even when the compressed HTML is 15k, sometimes the decompression just dies.

Here’s my decompression approach:

 URL file = new URL("/blobserve?key=" + htmlBlobKey);
         URLConnection conn = file.openConnection();
         conn.setReadTimeout(30000);
         conn.setConnectTimeout(30000);
         InputStream inputStream = conn.getInputStream();
         byte[] data = IOUtils.toByteArray(inputStream);
         return new String(Functions.decompress(data));

Any thoughts on how best to take compressed HTML from the blobstore, decompress it and display it? Even if I need to pass it to a task queue and poll for completion while showing a progress bar – that would be fine. I truly don’t care, so long as it’s efficient and ultimately WORKS. I’d appreciate any guidance you can share with me here.

Thanks for your help.

Answer 1

I took Nick Johnson’s idea and read from the Blobstore directly vs serving the blobs. Now it’s lightning fast! Here’s the code:

try{
        ChainedBlobstoreInputStream inputStream = new ChainedBlobstoreInputStream(this.getHtmlBlobKey());
        //StringWriter writer = new StringWriter();
         byte[] data = IOUtils.toByteArray(inputStream);
         return new String(Functions.decompress(Encrypt.AESDecrypt(data)));
         //return new String(data);
    } 
    catch(Exception e){
            return "No HTML Version";
        }

I got the ChainedBlobstoreInputStream class from here:
Reading a BlobstoreInputStream >= 1MB in size