Hi
The file consumer will store the file as a GenericFile which basically
is a pointer to java.io.File.
So you ought to be fine to use FileInputStream or InputStream as your
bean parameter. And even close it after use.
As its a bean parameter then Camel dont replace the existing message
body with that, and the splitter should be able to work with the
original message body as-is.
It would be a different story if the message body was an InputStream
that was only readable once.
For that you can use stream caching from Camel.
On Fri, Dec 15, 2017 at 4:48 PM, Burkard Stephan
<stephan.burk...@visana.ch> wrote:
> Hi Camel users
>
> I have a route that consumes big XML files (100 to 300 MB), splits them into
> chunks and processes the chunks.
>
> from(pollingFileConsumer)
> .routeId("bigFileConsumer")
> // here I would like to make a very basic check if the file looks as
> expected
> .split().tokenizeXML("XmlElementNameToSplitWith").streaming()
> // Continue to work with file chunk from splitter
> .bean(whatever)
>
>
> Before the splitter, I would like to check if the file is in expected format.
> To avoid reading the whole big file (memory), I would like to read the first
> 2000 chars (no matter if XML is invalid) and look up some required specifics
> (for example with regex) like the namespace.
>
> What is best practice to do this?
>
> I could of course implement a bean and call it. But the message body is
> "java.io.File".
> => When I convert it into a String (implicitly by annotate the method with
> "@Body String body" this converts the whole big file.
> => When I read the File through a stream, I would probably have to reset the
> Stream so that Camel can read the File from beginning.
>
> Thanks for suggestions
> Stephan
>
>
--
Claus Ibsen
-----------------
http://davsclaus.com @davsclaus
Camel in Action 2: https://www.manning.com/ibsen2
