[
https://issues.apache.org/activemq/browse/CAMEL-2551?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=58266#action_58266
]
Sergey Zolotaryov commented on CAMEL-2551:
------------------------------------------
Hi Claus
We use Camel for a downstream distribution product and one of the requirements
is streaming of large amounts of data. The use case is bulk output of data - I
put a db cursor into an Iterable, forward to a home made velocity component
(out of the box velocity component writes to a String, which is not acceptable
for us) - and it puts PipedInputStream as message body. A separate thread feeds
the PipedOutputStream to velocity engine. Then when this stream is passed on to
file component we get InputStream closed exception, because of the bug
mentioned. For now I have to wrap PipedInputStream with a buffering stream, but
if you include the fix, we will remove the workaround.
> File component does not correctly handle PipedInputStream in message body.
> --------------------------------------------------------------------------
>
> Key: CAMEL-2551
> URL: https://issues.apache.org/activemq/browse/CAMEL-2551
> Project: Apache Camel
> Issue Type: Bug
> Components: camel-core
> Affects Versions: 2.2.0
> Environment: does not matter
> Reporter: Sergey Zolotaryov
> Assignee: Willem Jiang
> Attachments: patch.txt
>
>
> Streams that do not have their contents length at immediate disposal, like
> PipedInputStream, are not processed correctly by the file component.
> \\
> {code}
> private void writeFileByStream(InputStream in, File target) throws
> IOException {
> FileChannel out = null;
> try {
> out = prepareOutputFileChannel(target, out);
> if (LOG.isTraceEnabled()) {
> LOG.trace("Using InputStream to transfer from: " + in + " to:
> " + out);
> }
> int size = endpoint.getBufferSize();
> byte[] buffer = new byte[size];
> ByteBuffer byteBuffer = ByteBuffer.wrap(buffer);
> while (true) {
> int count = in.read(buffer);
> if (count <= 0) {
> break;
> } else if (count < size) {
> byteBuffer = ByteBuffer.wrap(buffer, 0, count);
> out.write(byteBuffer);
> break;
> } else {
> out.write(byteBuffer);
> byteBuffer.clear();
> }
> }
> } finally {
> ObjectHelper.close(in, target.getName(), LOG);
> ObjectHelper.close(out, target.getName(), LOG);
> }
> }
> {code}
> The code
> {code}
> } else if (count < size) {
> byteBuffer = ByteBuffer.wrap(buffer, 0, count);
> out.write(byteBuffer);
> break;
> } else {
> {code}
> does not take into account that bytes read can be less than the size of the
> buffer passed into the InputStream.read method and stream can still have more
> content. The only indication that EOF was reached is -1 returned from the
> read method according to Java API.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
