Claus,
Please forgive all the @ in here as that was the only way I could post this
to get around the spam filter.
Just to get this working for me so I can start refactoring my code to accept
and process individual items from the unmarshaling of the multi-line beanio
beans I hacked this bit of code. I don't want to use it but it let's me
proceed until I can get something like this to work with the Camel file
endpoint.
As you can see I'm injecting a seda queue in here to drop the individual
items on and a directory w@tcher to w@tch for incoming files. My unit test
copies a file over into the directory and I can see the results. I have to
put the exclusive lock mechanics in there or the bean reader would try to
read the file before it was truly available. But I don't want to replicate
all the mechanics that are already baked in and hardened in the Camel File
endpoint.
Doing it this way though changes this from a pull mechanism to a
push/publish mechanism that I prefer.
I have a couple of questions about how I might proceed with this. I'm fine
with reading the Beanio stream/definition from my blueprint bundle with
getResourceAsStream and foregoing the use of the Camel marsahaler itself.
If I do that, what mecahnics would I use to get a notification from the File
endpoint that a new file has been dropped into the directory. Just a
standard Camel Processor?
You'd mentioned an iterator/reader object and noticed some in the BeanIO
code itself but I don't think I saw anything in Camel itself.
I did try just chunking the raw text but the problem with that is if I read
in a chunk of 100 records the 100th might well fall in the middle of a
record.
Thanks for any insights.
@EndpointInject(uri = "seda:process")
private ProducerTemplate process;
Logger logger = Logger.getLogger(AHSBatchReader.class);
private StreamFactory factory;
private w@tchService w@tcher;
Path dir;
public void init() throws IOException {
logger.info("Starting stream factory...");
// create a StreamFactory
factory = StreamFactory.newInstance();
// load the mapping file...switch to use getResourceAsStream
from bundle.
factory.load("src/main/resources/dataformats/beanio-ahs-in.xml");
w@tcher = FileSystems.getDefault().neww@tchService();
dir = FileSystems.getDefault().getPath("/test/in");
dir.register(w@tcher, ENTRY_CREATE, ENTRY_DELETE, ENTRY_MODIFY);
new Thread(this).start();
}
public void run() {
while (true) {
w@tchKey key;
try {
key = w...@tcher.take();
for (w@tchEvent<?> event : key.pollEvents()) {
w...@tchevent.kind kind = event.kind();
w@tchEvent<Path> ev = cast(event);
Path name = ev.context();
Path child = dir.resolve(name);
// print out event
System.out.format("%s: %s\n",
event.kind().name(), child);
if (event.kind() == ENTRY_MODIFY) {
File file = child.toFile();
boolean exclusivelyLocked =
false;
//Get an exclusive lock on the file if we can. If we can't then the system
is still using and it we delay start of beanio reading.
while (!exclusivelyLocked) {
try {
FileChannel
channel = new RandomAccessFile(file, "rw").getChannel();
FileLock lock =
null;
try {
lock =
channel.tryLock();
System.out.println(lock);
System.out.println(lock.isShared());
System.out.println(lock.isValid());
exclusivelyLocked = lock.isValid() && !lock.isShared();
} catch
(Exception e) {
// File
is open by someone else
Thread.sleep(500);
}
//Release them now so that Beanio can grab them.
lock.close();
channel.close();
} catch (Exception e) {
}
}
//Start processing the beanio stream.
execute(file);
}
}
} catch (InterruptedException x) {
}
}
}
@SuppressWarnings("unchecked")
static <T> w@tchEvent<T> cast(w@tchEvent<?> event) {
return (w@tchEvent<T>) event;
}
private void execute(File file) {
BeanReader in = factory.createReader("myInputStream",
file.getAbsoluteFile());
Object o;
while ((o = in.read()) != null) {
//Send these to the SEDA queue
process.sendBody(o);
}
in.close();
file.delete();
}
}
--
View this message in context:
http://camel.465427.n5.nabble.com/Beanio-stream-tp5780468p5780717.html
Sent from the Camel - Users mailing list archive at Nabble.com.
