Search in sources :

Example 71 with AmazonS3Client

use of com.talend.shaded.com.amazonaws.services.s3.AmazonS3Client in project nifi by apache.

the class AbstractS3Processor method createClient.

/**
 * Create client using credentials provider. This is the preferred way for creating clients
 */
@Override
protected AmazonS3Client createClient(final ProcessContext context, final AWSCredentialsProvider credentialsProvider, final ClientConfiguration config) {
    getLogger().info("Creating client with credentials provider");
    initializeSignerOverride(context, config);
    final AmazonS3Client s3 = new AmazonS3Client(credentialsProvider, config);
    initalizeEndpointOverride(context, s3);
    return s3;
}
Also used : AmazonS3Client(com.amazonaws.services.s3.AmazonS3Client)

Example 72 with AmazonS3Client

use of com.talend.shaded.com.amazonaws.services.s3.AmazonS3Client in project nifi by apache.

the class AbstractS3IT method oneTimeSetup.

@BeforeClass
public static void oneTimeSetup() {
    // Creates a client and bucket for this test
    final FileInputStream fis;
    try {
        fis = new FileInputStream(CREDENTIALS_FILE);
    } catch (FileNotFoundException e1) {
        fail("Could not open credentials file " + CREDENTIALS_FILE + ": " + e1.getLocalizedMessage());
        return;
    }
    try {
        final PropertiesCredentials credentials = new PropertiesCredentials(fis);
        client = new AmazonS3Client(credentials);
        if (client.doesBucketExist(BUCKET_NAME)) {
            fail("Bucket " + BUCKET_NAME + " exists. Choose a different bucket name to continue test");
        }
        CreateBucketRequest request = REGION.contains("east") ? // See https://github.com/boto/boto3/issues/125
        new CreateBucketRequest(BUCKET_NAME) : new CreateBucketRequest(BUCKET_NAME, REGION);
        client.createBucket(request);
    } catch (final AmazonS3Exception e) {
        fail("Can't create the key " + BUCKET_NAME + ": " + e.getLocalizedMessage());
    } catch (final IOException e) {
        fail("Caught IOException preparing tests: " + e.getLocalizedMessage());
    } finally {
        FileUtils.closeQuietly(fis);
    }
    if (!client.doesBucketExist(BUCKET_NAME)) {
        fail("Setup incomplete, tests will fail");
    }
}
Also used : AmazonS3Client(com.amazonaws.services.s3.AmazonS3Client) CreateBucketRequest(com.amazonaws.services.s3.model.CreateBucketRequest) FileNotFoundException(java.io.FileNotFoundException) PropertiesCredentials(com.amazonaws.auth.PropertiesCredentials) AmazonS3Exception(com.amazonaws.services.s3.model.AmazonS3Exception) IOException(java.io.IOException) FileInputStream(java.io.FileInputStream) BeforeClass(org.junit.BeforeClass)

Example 73 with AmazonS3Client

use of com.talend.shaded.com.amazonaws.services.s3.AmazonS3Client in project nifi by apache.

the class PutS3Object method onTrigger.

@Override
public void onTrigger(final ProcessContext context, final ProcessSession session) {
    FlowFile flowFile = session.get();
    if (flowFile == null) {
        return;
    }
    final long startNanos = System.nanoTime();
    final String bucket = context.getProperty(BUCKET).evaluateAttributeExpressions(flowFile).getValue();
    final String key = context.getProperty(KEY).evaluateAttributeExpressions(flowFile).getValue();
    final String cacheKey = getIdentifier() + "/" + bucket + "/" + key;
    final AmazonS3Client s3 = getClient();
    final FlowFile ff = flowFile;
    final Map<String, String> attributes = new HashMap<>();
    final String ffFilename = ff.getAttributes().get(CoreAttributes.FILENAME.key());
    attributes.put(S3_BUCKET_KEY, bucket);
    attributes.put(S3_OBJECT_KEY, key);
    final Long multipartThreshold = context.getProperty(MULTIPART_THRESHOLD).asDataSize(DataUnit.B).longValue();
    final Long multipartPartSize = context.getProperty(MULTIPART_PART_SIZE).asDataSize(DataUnit.B).longValue();
    final long now = System.currentTimeMillis();
    /*
         * If necessary, run age off for existing uploads in AWS S3 and local state
         */
    ageoffS3Uploads(context, s3, now);
    /*
         * Then
         */
    try {
        session.read(flowFile, new InputStreamCallback() {

            @Override
            public void process(final InputStream rawIn) throws IOException {
                try (final InputStream in = new BufferedInputStream(rawIn)) {
                    final ObjectMetadata objectMetadata = new ObjectMetadata();
                    objectMetadata.setContentDisposition(ff.getAttribute(CoreAttributes.FILENAME.key()));
                    objectMetadata.setContentLength(ff.getSize());
                    final String contentType = context.getProperty(CONTENT_TYPE).evaluateAttributeExpressions(ff).getValue();
                    if (contentType != null) {
                        objectMetadata.setContentType(contentType);
                        attributes.put(S3_CONTENT_TYPE, contentType);
                    }
                    final String expirationRule = context.getProperty(EXPIRATION_RULE_ID).evaluateAttributeExpressions(ff).getValue();
                    if (expirationRule != null) {
                        objectMetadata.setExpirationTimeRuleId(expirationRule);
                    }
                    final Map<String, String> userMetadata = new HashMap<>();
                    for (final Map.Entry<PropertyDescriptor, String> entry : context.getProperties().entrySet()) {
                        if (entry.getKey().isDynamic()) {
                            final String value = context.getProperty(entry.getKey()).evaluateAttributeExpressions(ff).getValue();
                            userMetadata.put(entry.getKey().getName(), value);
                        }
                    }
                    final String serverSideEncryption = context.getProperty(SERVER_SIDE_ENCRYPTION).getValue();
                    if (!serverSideEncryption.equals(NO_SERVER_SIDE_ENCRYPTION)) {
                        objectMetadata.setSSEAlgorithm(serverSideEncryption);
                        attributes.put(S3_SSE_ALGORITHM, serverSideEncryption);
                    }
                    if (!userMetadata.isEmpty()) {
                        objectMetadata.setUserMetadata(userMetadata);
                    }
                    if (ff.getSize() <= multipartThreshold) {
                        // ----------------------------------------
                        // single part upload
                        // ----------------------------------------
                        final PutObjectRequest request = new PutObjectRequest(bucket, key, in, objectMetadata);
                        request.setStorageClass(StorageClass.valueOf(context.getProperty(STORAGE_CLASS).getValue()));
                        final AccessControlList acl = createACL(context, ff);
                        if (acl != null) {
                            request.setAccessControlList(acl);
                        }
                        final CannedAccessControlList cannedAcl = createCannedACL(context, ff);
                        if (cannedAcl != null) {
                            request.withCannedAcl(cannedAcl);
                        }
                        try {
                            final PutObjectResult result = s3.putObject(request);
                            if (result.getVersionId() != null) {
                                attributes.put(S3_VERSION_ATTR_KEY, result.getVersionId());
                            }
                            if (result.getETag() != null) {
                                attributes.put(S3_ETAG_ATTR_KEY, result.getETag());
                            }
                            if (result.getExpirationTime() != null) {
                                attributes.put(S3_EXPIRATION_ATTR_KEY, result.getExpirationTime().toString());
                            }
                            if (result.getMetadata().getRawMetadata().keySet().contains(S3_STORAGECLASS_META_KEY)) {
                                attributes.put(S3_STORAGECLASS_ATTR_KEY, result.getMetadata().getRawMetadataValue(S3_STORAGECLASS_META_KEY).toString());
                            }
                            if (userMetadata.size() > 0) {
                                StringBuilder userMetaBldr = new StringBuilder();
                                for (String userKey : userMetadata.keySet()) {
                                    userMetaBldr.append(userKey).append("=").append(userMetadata.get(userKey));
                                }
                                attributes.put(S3_USERMETA_ATTR_KEY, userMetaBldr.toString());
                            }
                            attributes.put(S3_API_METHOD_ATTR_KEY, S3_API_METHOD_PUTOBJECT);
                        } catch (AmazonClientException e) {
                            getLogger().info("Failure completing upload flowfile={} bucket={} key={} reason={}", new Object[] { ffFilename, bucket, key, e.getMessage() });
                            throw (e);
                        }
                    } else {
                        // ----------------------------------------
                        // multipart upload
                        // ----------------------------------------
                        // load or create persistent state
                        // ------------------------------------------------------------
                        MultipartState currentState;
                        try {
                            currentState = getLocalStateIfInS3(s3, bucket, cacheKey);
                            if (currentState != null) {
                                if (currentState.getPartETags().size() > 0) {
                                    final PartETag lastETag = currentState.getPartETags().get(currentState.getPartETags().size() - 1);
                                    getLogger().info("Resuming upload for flowfile='{}' bucket='{}' key='{}' " + "uploadID='{}' filePosition='{}' partSize='{}' storageClass='{}' " + "contentLength='{}' partsLoaded={} lastPart={}/{}", new Object[] { ffFilename, bucket, key, currentState.getUploadId(), currentState.getFilePosition(), currentState.getPartSize(), currentState.getStorageClass().toString(), currentState.getContentLength(), currentState.getPartETags().size(), Integer.toString(lastETag.getPartNumber()), lastETag.getETag() });
                                } else {
                                    getLogger().info("Resuming upload for flowfile='{}' bucket='{}' key='{}' " + "uploadID='{}' filePosition='{}' partSize='{}' storageClass='{}' " + "contentLength='{}' no partsLoaded", new Object[] { ffFilename, bucket, key, currentState.getUploadId(), currentState.getFilePosition(), currentState.getPartSize(), currentState.getStorageClass().toString(), currentState.getContentLength() });
                                }
                            } else {
                                currentState = new MultipartState();
                                currentState.setPartSize(multipartPartSize);
                                currentState.setStorageClass(StorageClass.valueOf(context.getProperty(STORAGE_CLASS).getValue()));
                                currentState.setContentLength(ff.getSize());
                                persistLocalState(cacheKey, currentState);
                                getLogger().info("Starting new upload for flowfile='{}' bucket='{}' key='{}'", new Object[] { ffFilename, bucket, key });
                            }
                        } catch (IOException e) {
                            getLogger().error("IOException initiating cache state while processing flow files: " + e.getMessage());
                            throw (e);
                        }
                        // ------------------------------------------------------------
                        if (currentState.getUploadId().isEmpty()) {
                            final InitiateMultipartUploadRequest initiateRequest = new InitiateMultipartUploadRequest(bucket, key, objectMetadata);
                            initiateRequest.setStorageClass(currentState.getStorageClass());
                            final AccessControlList acl = createACL(context, ff);
                            if (acl != null) {
                                initiateRequest.setAccessControlList(acl);
                            }
                            final CannedAccessControlList cannedAcl = createCannedACL(context, ff);
                            if (cannedAcl != null) {
                                initiateRequest.withCannedACL(cannedAcl);
                            }
                            try {
                                final InitiateMultipartUploadResult initiateResult = s3.initiateMultipartUpload(initiateRequest);
                                currentState.setUploadId(initiateResult.getUploadId());
                                currentState.getPartETags().clear();
                                try {
                                    persistLocalState(cacheKey, currentState);
                                } catch (Exception e) {
                                    getLogger().info("Exception saving cache state while processing flow file: " + e.getMessage());
                                    throw (new ProcessException("Exception saving cache state", e));
                                }
                                getLogger().info("Success initiating upload flowfile={} available={} position={} " + "length={} bucket={} key={} uploadId={}", new Object[] { ffFilename, in.available(), currentState.getFilePosition(), currentState.getContentLength(), bucket, key, currentState.getUploadId() });
                                if (initiateResult.getUploadId() != null) {
                                    attributes.put(S3_UPLOAD_ID_ATTR_KEY, initiateResult.getUploadId());
                                }
                            } catch (AmazonClientException e) {
                                getLogger().info("Failure initiating upload flowfile={} bucket={} key={} reason={}", new Object[] { ffFilename, bucket, key, e.getMessage() });
                                throw (e);
                            }
                        } else {
                            if (currentState.getFilePosition() > 0) {
                                try {
                                    final long skipped = in.skip(currentState.getFilePosition());
                                    if (skipped != currentState.getFilePosition()) {
                                        getLogger().info("Failure skipping to resume upload flowfile={} " + "bucket={} key={} position={} skipped={}", new Object[] { ffFilename, bucket, key, currentState.getFilePosition(), skipped });
                                    }
                                } catch (Exception e) {
                                    getLogger().info("Failure skipping to resume upload flowfile={} bucket={} " + "key={} position={} reason={}", new Object[] { ffFilename, bucket, key, currentState.getFilePosition(), e.getMessage() });
                                    throw (new ProcessException(e));
                                }
                            }
                        }
                        // upload parts
                        // ------------------------------------------------------------
                        long thisPartSize;
                        for (int part = currentState.getPartETags().size() + 1; currentState.getFilePosition() < currentState.getContentLength(); part++) {
                            if (!PutS3Object.this.isScheduled()) {
                                throw new IOException(S3_PROCESS_UNSCHEDULED_MESSAGE + " flowfile=" + ffFilename + " part=" + part + " uploadId=" + currentState.getUploadId());
                            }
                            thisPartSize = Math.min(currentState.getPartSize(), (currentState.getContentLength() - currentState.getFilePosition()));
                            UploadPartRequest uploadRequest = new UploadPartRequest().withBucketName(bucket).withKey(key).withUploadId(currentState.getUploadId()).withInputStream(in).withPartNumber(part).withPartSize(thisPartSize);
                            try {
                                UploadPartResult uploadPartResult = s3.uploadPart(uploadRequest);
                                currentState.addPartETag(uploadPartResult.getPartETag());
                                currentState.setFilePosition(currentState.getFilePosition() + thisPartSize);
                                try {
                                    persistLocalState(cacheKey, currentState);
                                } catch (Exception e) {
                                    getLogger().info("Exception saving cache state processing flow file: " + e.getMessage());
                                }
                                getLogger().info("Success uploading part flowfile={} part={} available={} " + "etag={} uploadId={}", new Object[] { ffFilename, part, in.available(), uploadPartResult.getETag(), currentState.getUploadId() });
                            } catch (AmazonClientException e) {
                                getLogger().info("Failure uploading part flowfile={} part={} bucket={} key={} " + "reason={}", new Object[] { ffFilename, part, bucket, key, e.getMessage() });
                                throw (e);
                            }
                        }
                        // complete multipart upload
                        // ------------------------------------------------------------
                        CompleteMultipartUploadRequest completeRequest = new CompleteMultipartUploadRequest(bucket, key, currentState.getUploadId(), currentState.getPartETags());
                        try {
                            CompleteMultipartUploadResult completeResult = s3.completeMultipartUpload(completeRequest);
                            getLogger().info("Success completing upload flowfile={} etag={} uploadId={}", new Object[] { ffFilename, completeResult.getETag(), currentState.getUploadId() });
                            if (completeResult.getVersionId() != null) {
                                attributes.put(S3_VERSION_ATTR_KEY, completeResult.getVersionId());
                            }
                            if (completeResult.getETag() != null) {
                                attributes.put(S3_ETAG_ATTR_KEY, completeResult.getETag());
                            }
                            if (completeResult.getExpirationTime() != null) {
                                attributes.put(S3_EXPIRATION_ATTR_KEY, completeResult.getExpirationTime().toString());
                            }
                            if (currentState.getStorageClass() != null) {
                                attributes.put(S3_STORAGECLASS_ATTR_KEY, currentState.getStorageClass().toString());
                            }
                            if (userMetadata.size() > 0) {
                                StringBuilder userMetaBldr = new StringBuilder();
                                for (String userKey : userMetadata.keySet()) {
                                    userMetaBldr.append(userKey).append("=").append(userMetadata.get(userKey));
                                }
                                attributes.put(S3_USERMETA_ATTR_KEY, userMetaBldr.toString());
                            }
                            attributes.put(S3_API_METHOD_ATTR_KEY, S3_API_METHOD_MULTIPARTUPLOAD);
                        } catch (AmazonClientException e) {
                            getLogger().info("Failure completing upload flowfile={} bucket={} key={} reason={}", new Object[] { ffFilename, bucket, key, e.getMessage() });
                            throw (e);
                        }
                    }
                }
            }
        });
        if (!attributes.isEmpty()) {
            flowFile = session.putAllAttributes(flowFile, attributes);
        }
        session.transfer(flowFile, REL_SUCCESS);
        final String url = s3.getResourceUrl(bucket, key);
        final long millis = TimeUnit.NANOSECONDS.toMillis(System.nanoTime() - startNanos);
        session.getProvenanceReporter().send(flowFile, url, millis);
        getLogger().info("Successfully put {} to Amazon S3 in {} milliseconds", new Object[] { ff, millis });
        try {
            removeLocalState(cacheKey);
        } catch (IOException e) {
            getLogger().info("Error trying to delete key {} from cache: {}", new Object[] { cacheKey, e.getMessage() });
        }
    } catch (final ProcessException | AmazonClientException pe) {
        if (pe.getMessage().contains(S3_PROCESS_UNSCHEDULED_MESSAGE)) {
            getLogger().info(pe.getMessage());
            session.rollback();
        } else {
            getLogger().error("Failed to put {} to Amazon S3 due to {}", new Object[] { flowFile, pe });
            flowFile = session.penalize(flowFile);
            session.transfer(flowFile, REL_FAILURE);
        }
    }
}
Also used : CannedAccessControlList(com.amazonaws.services.s3.model.CannedAccessControlList) AccessControlList(com.amazonaws.services.s3.model.AccessControlList) InitiateMultipartUploadResult(com.amazonaws.services.s3.model.InitiateMultipartUploadResult) HashMap(java.util.HashMap) AmazonClientException(com.amazonaws.AmazonClientException) CompleteMultipartUploadResult(com.amazonaws.services.s3.model.CompleteMultipartUploadResult) PartETag(com.amazonaws.services.s3.model.PartETag) UploadPartResult(com.amazonaws.services.s3.model.UploadPartResult) Entry(java.util.Map.Entry) BufferedInputStream(java.io.BufferedInputStream) PutObjectRequest(com.amazonaws.services.s3.model.PutObjectRequest) FlowFile(org.apache.nifi.flowfile.FlowFile) PutObjectResult(com.amazonaws.services.s3.model.PutObjectResult) BufferedInputStream(java.io.BufferedInputStream) FileInputStream(java.io.FileInputStream) InputStream(java.io.InputStream) InitiateMultipartUploadRequest(com.amazonaws.services.s3.model.InitiateMultipartUploadRequest) UploadPartRequest(com.amazonaws.services.s3.model.UploadPartRequest) IOException(java.io.IOException) CannedAccessControlList(com.amazonaws.services.s3.model.CannedAccessControlList) AmazonClientException(com.amazonaws.AmazonClientException) ProcessException(org.apache.nifi.processor.exception.ProcessException) AmazonS3Exception(com.amazonaws.services.s3.model.AmazonS3Exception) IOException(java.io.IOException) AmazonS3Client(com.amazonaws.services.s3.AmazonS3Client) ProcessException(org.apache.nifi.processor.exception.ProcessException) AtomicLong(java.util.concurrent.atomic.AtomicLong) InputStreamCallback(org.apache.nifi.processor.io.InputStreamCallback) ObjectMetadata(com.amazonaws.services.s3.model.ObjectMetadata) Map(java.util.Map) HashMap(java.util.HashMap) CompleteMultipartUploadRequest(com.amazonaws.services.s3.model.CompleteMultipartUploadRequest)

Example 74 with AmazonS3Client

use of com.talend.shaded.com.amazonaws.services.s3.AmazonS3Client in project nifi by apache.

the class TestAWSCredentials method setUp.

@Before
public void setUp() {
    mockAwsProcessor = new AbstractAWSCredentialsProviderProcessor<AmazonS3Client>() {

        protected List<PropertyDescriptor> getSupportedPropertyDescriptors() {
            return Arrays.asList(AbstractAWSCredentialsProviderProcessor.AWS_CREDENTIALS_PROVIDER_SERVICE, AbstractAWSProcessor.CREDENTIALS_FILE, AbstractAWSProcessor.ACCESS_KEY, AbstractAWSProcessor.SECRET_KEY, AbstractAWSProcessor.TIMEOUT);
        }

        @Override
        protected AmazonS3Client createClient(final ProcessContext context, final AWSCredentials credentials, final ClientConfiguration config) {
            awsCredentials = credentials;
            clientConfiguration = config;
            final AmazonS3Client s3 = new AmazonS3Client(credentials, config);
            return s3;
        }

        @Override
        protected AmazonS3Client createClient(final ProcessContext context, final AWSCredentialsProvider credentialsProvider, final ClientConfiguration config) {
            awsCredentialsProvider = credentialsProvider;
            clientConfiguration = config;
            final AmazonS3Client s3 = new AmazonS3Client(credentialsProvider, config);
            return s3;
        }

        @Override
        public void onTrigger(final ProcessContext context, final ProcessSession session) {
        }
    };
    runner = TestRunners.newTestRunner(mockAwsProcessor);
}
Also used : ProcessSession(org.apache.nifi.processor.ProcessSession) AmazonS3Client(com.amazonaws.services.s3.AmazonS3Client) List(java.util.List) BasicAWSCredentials(com.amazonaws.auth.BasicAWSCredentials) AnonymousAWSCredentials(com.amazonaws.auth.AnonymousAWSCredentials) AWSCredentials(com.amazonaws.auth.AWSCredentials) ProcessContext(org.apache.nifi.processor.ProcessContext) ClientConfiguration(com.amazonaws.ClientConfiguration) AWSCredentialsProvider(com.amazonaws.auth.AWSCredentialsProvider) Before(org.junit.Before)

Example 75 with AmazonS3Client

use of com.talend.shaded.com.amazonaws.services.s3.AmazonS3Client in project apex-malhar by apache.

the class S3Reconciler method setup.

@Override
public void setup(Context.OperatorContext context) {
    s3client = new AmazonS3Client(new BasicAWSCredentials(accessKey, secretKey));
    if (region != null) {
        s3client.setRegion(Region.getRegion(Regions.fromName(region)));
    }
    filePath = context.getValue(DAG.APPLICATION_PATH);
    try {
        fs = FileSystem.newInstance(new Path(filePath).toUri(), new Configuration());
    } catch (IOException e) {
        logger.error("Unable to create FileSystem: {}", e.getMessage());
    }
    super.setup(context);
}
Also used : Path(org.apache.hadoop.fs.Path) AmazonS3Client(com.amazonaws.services.s3.AmazonS3Client) Configuration(org.apache.hadoop.conf.Configuration) IOException(java.io.IOException) BasicAWSCredentials(com.amazonaws.auth.BasicAWSCredentials)

Aggregations

AmazonS3Client (com.amazonaws.services.s3.AmazonS3Client)107 Test (org.junit.Test)23 BasicAWSCredentials (com.amazonaws.auth.BasicAWSCredentials)20 AmazonClientException (com.amazonaws.AmazonClientException)16 ClientConfiguration (com.amazonaws.ClientConfiguration)15 ArrayList (java.util.ArrayList)13 HashMap (java.util.HashMap)13 AmazonS3 (com.amazonaws.services.s3.AmazonS3)12 File (java.io.File)12 InvocationOnMock (org.mockito.invocation.InvocationOnMock)12 PutObjectResult (com.amazonaws.services.s3.model.PutObjectResult)11 UploadPartRequest (com.amazonaws.services.s3.model.UploadPartRequest)11 AWSCredentials (com.amazonaws.auth.AWSCredentials)10 AWSCredentialsProvider (com.amazonaws.auth.AWSCredentialsProvider)10 ObjectMetadata (com.amazonaws.services.s3.model.ObjectMetadata)10 AmazonServiceException (com.amazonaws.AmazonServiceException)9 AmazonS3Exception (com.amazonaws.services.s3.model.AmazonS3Exception)9 InternalEvent (com.nextdoor.bender.InternalEvent)9 TestContext (com.nextdoor.bender.aws.TestContext)9 IOException (java.io.IOException)9