Package de.taimos.pipeline.aws.utils

Class S3Utils

java.lang.Object

de.taimos.pipeline.aws.utils.S3Utils

public class S3Utils
extends Object

Method Summary

Modifier and Type	Method	Description
static boolean	doesObjectExist(software.amazon.awssdk.services.s3.S3Client s3Client, String bucket, String key)	v2 dropped v1's doesObjectExist convenience method, so this restores it.
static <T> T	joinTransfer(CompletableFuture<T> future)	Waits for a transfer, unwrapping the failure and staying interruptible.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

joinTransfer

public static <T> T joinTransfer(CompletableFuture<T> future)
                          throws InterruptedException

Waits for a transfer, unwrapping the failure and staying interruptible. Two reasons not to call join() directly. It wraps whatever went wrong in a CompletionException, where v1's waitForCompletion threw the AmazonS3Exception itself, so a Jenkinsfile could catch the S3 error and the log showed the service message rather than a wrapper. And join() is not interruptible: it records the interrupt and keeps waiting, so aborting a build would leave the transfer running - on an agent, against a workspace Jenkins believes is free. get() throws InterruptedException, which the callers already declare.

Throws:

InterruptedException

doesObjectExist

public static boolean doesObjectExist(software.amazon.awssdk.services.s3.S3Client s3Client,
 String bucket,
 String key)

v2 dropped v1's doesObjectExist convenience method, so this restores it. v1 implemented it as getObjectMetadata plus "treat any 404 as absent", which means a missing bucket answered false rather than throwing, and a 403 propagated. Catching NoSuchKeyException alone would not reproduce that - a missing bucket raises NoSuchBucketException instead - so the status code is what is matched on, exactly as v1 did.