Search in sources :

Example 1 with ResponseHeaderOverrides

use of com.amazonaws.services.s3.model.ResponseHeaderOverrides in project Singularity by HubSpot.

the class S3LogResource method getS3LogsWithExecutorService.

// Fetching logs
private List<SingularityS3LogMetadata> getS3LogsWithExecutorService(S3Configuration s3Configuration, ListeningExecutorService executorService, Map<SingularityS3Service, Set<String>> servicesToPrefixes, int totalPrefixCount, final SingularityS3SearchRequest search, final ConcurrentHashMap<String, ContinuationToken> continuationTokens, final boolean paginated) throws InterruptedException, ExecutionException, TimeoutException {
    List<ListenableFuture<List<S3ObjectSummaryHolder>>> futures = Lists.newArrayListWithCapacity(totalPrefixCount);
    final AtomicInteger resultCount = new AtomicInteger();
    for (final Map.Entry<SingularityS3Service, Set<String>> entry : servicesToPrefixes.entrySet()) {
        final String s3Bucket = entry.getKey().getBucket();
        final String group = entry.getKey().getGroup();
        final AmazonS3 s3Client = entry.getKey().getS3Client();
        for (final String s3Prefix : entry.getValue()) {
            final String key = String.format(CONTINUATION_TOKEN_KEY_FORMAT, group, s3Bucket, s3Prefix);
            if (search.getContinuationTokens().containsKey(key) && search.getContinuationTokens().get(key).isLastPage()) {
                LOG.trace("No further content for prefix {} in bucket {}, skipping", s3Prefix, s3Bucket);
                continuationTokens.putIfAbsent(key, search.getContinuationTokens().get(key));
                continue;
            }
            futures.add(executorService.submit(new Callable<List<S3ObjectSummaryHolder>>() {

                @Override
                public List<S3ObjectSummaryHolder> call() throws Exception {
                    ListObjectsV2Request request = new ListObjectsV2Request().withBucketName(s3Bucket).withPrefix(s3Prefix);
                    if (paginated) {
                        Optional<ContinuationToken> token = Optional.absent();
                        if (search.getContinuationTokens().containsKey(key) && !Strings.isNullOrEmpty(search.getContinuationTokens().get(key).getValue())) {
                            request.setContinuationToken(search.getContinuationTokens().get(key).getValue());
                            token = Optional.of(search.getContinuationTokens().get(key));
                        }
                        int targetResultCount = search.getMaxPerPage().or(DEFAULT_TARGET_MAX_RESULTS);
                        request.setMaxKeys(targetResultCount);
                        if (resultCount.get() < targetResultCount) {
                            ListObjectsV2Result result = s3Client.listObjectsV2(request);
                            if (result.getObjectSummaries().isEmpty()) {
                                continuationTokens.putIfAbsent(key, new ContinuationToken(result.getNextContinuationToken(), true));
                                return Collections.emptyList();
                            } else {
                                boolean addToList = incrementIfLessThan(resultCount, result.getObjectSummaries().size(), targetResultCount);
                                if (addToList) {
                                    continuationTokens.putIfAbsent(key, new ContinuationToken(result.getNextContinuationToken(), !result.isTruncated()));
                                    List<S3ObjectSummaryHolder> objectSummaryHolders = new ArrayList<>();
                                    for (S3ObjectSummary objectSummary : result.getObjectSummaries()) {
                                        objectSummaryHolders.add(new S3ObjectSummaryHolder(group, objectSummary));
                                    }
                                    return objectSummaryHolders;
                                } else {
                                    continuationTokens.putIfAbsent(key, token.or(new ContinuationToken(null, false)));
                                    return Collections.emptyList();
                                }
                            }
                        } else {
                            continuationTokens.putIfAbsent(key, token.or(new ContinuationToken(null, false)));
                            return Collections.emptyList();
                        }
                    } else {
                        ListObjectsV2Result result = s3Client.listObjectsV2(request);
                        List<S3ObjectSummaryHolder> objectSummaryHolders = new ArrayList<>();
                        for (S3ObjectSummary objectSummary : result.getObjectSummaries()) {
                            objectSummaryHolders.add(new S3ObjectSummaryHolder(group, objectSummary));
                        }
                        return objectSummaryHolders;
                    }
                }
            }));
        }
    }
    final long start = System.currentTimeMillis();
    List<List<S3ObjectSummaryHolder>> results = Futures.allAsList(futures).get(s3Configuration.getWaitForS3ListSeconds(), TimeUnit.SECONDS);
    List<S3ObjectSummaryHolder> objects = Lists.newArrayListWithExpectedSize(results.size() * 2);
    for (List<S3ObjectSummaryHolder> s3ObjectSummaryHolders : results) {
        for (final S3ObjectSummaryHolder s3ObjectHolder : s3ObjectSummaryHolders) {
            objects.add(s3ObjectHolder);
        }
    }
    LOG.trace("Got {} objects from S3 after {}", objects.size(), JavaUtils.duration(start));
    List<ListenableFuture<SingularityS3LogMetadata>> logFutures = Lists.newArrayListWithCapacity(objects.size());
    final Date expireAt = new Date(System.currentTimeMillis() + s3Configuration.getExpireS3LinksAfterMillis());
    for (final S3ObjectSummaryHolder s3ObjectHolder : objects) {
        final S3ObjectSummary s3Object = s3ObjectHolder.getObjectSummary();
        final AmazonS3 s3Client = s3Services.getServiceByGroupAndBucketOrDefault(s3ObjectHolder.getGroup(), s3Object.getBucketName()).getS3Client();
        logFutures.add(executorService.submit(new Callable<SingularityS3LogMetadata>() {

            @Override
            public SingularityS3LogMetadata call() throws Exception {
                Optional<Long> maybeStartTime = Optional.absent();
                Optional<Long> maybeEndTime = Optional.absent();
                if (!search.isExcludeMetadata()) {
                    GetObjectMetadataRequest metadataRequest = new GetObjectMetadataRequest(s3Object.getBucketName(), s3Object.getKey());
                    Map<String, String> objectMetadata = s3Client.getObjectMetadata(metadataRequest).getUserMetadata();
                    maybeStartTime = getMetadataAsLong(objectMetadata, SingularityS3Log.LOG_START_S3_ATTR);
                    maybeEndTime = getMetadataAsLong(objectMetadata, SingularityS3Log.LOG_END_S3_ATTR);
                }
                if (search.isListOnly()) {
                    return new SingularityS3LogMetadata(s3Object.getKey(), s3Object.getLastModified().getTime(), s3Object.getSize(), maybeStartTime, maybeEndTime);
                } else {
                    GeneratePresignedUrlRequest getUrlRequest = new GeneratePresignedUrlRequest(s3Object.getBucketName(), s3Object.getKey()).withMethod(HttpMethod.GET).withExpiration(expireAt);
                    String getUrl = s3Client.generatePresignedUrl(getUrlRequest).toString();
                    ResponseHeaderOverrides downloadHeaders = new ResponseHeaderOverrides();
                    downloadHeaders.setContentDisposition(CONTENT_DISPOSITION_DOWNLOAD_HEADER);
                    downloadHeaders.setContentEncoding(CONTENT_ENCODING_DOWNLOAD_HEADER);
                    GeneratePresignedUrlRequest downloadUrlRequest = new GeneratePresignedUrlRequest(s3Object.getBucketName(), s3Object.getKey()).withMethod(HttpMethod.GET).withExpiration(expireAt).withResponseHeaders(downloadHeaders);
                    String downloadUrl = s3Client.generatePresignedUrl(downloadUrlRequest).toString();
                    return new SingularityS3Log(getUrl, s3Object.getKey(), s3Object.getLastModified().getTime(), s3Object.getSize(), downloadUrl, maybeStartTime, maybeEndTime);
                }
            }
        }));
    }
    return Futures.allAsList(logFutures).get(s3Configuration.getWaitForS3LinksSeconds(), TimeUnit.SECONDS);
}
Also used : AmazonS3(com.amazonaws.services.s3.AmazonS3) Set(java.util.Set) HashSet(java.util.HashSet) ContinuationToken(com.hubspot.singularity.api.ContinuationToken) ListObjectsV2Result(com.amazonaws.services.s3.model.ListObjectsV2Result) ArrayList(java.util.ArrayList) Callable(java.util.concurrent.Callable) S3ObjectSummaryHolder(com.hubspot.singularity.helpers.S3ObjectSummaryHolder) GetObjectMetadataRequest(com.amazonaws.services.s3.model.GetObjectMetadataRequest) ListObjectsV2Request(com.amazonaws.services.s3.model.ListObjectsV2Request) List(java.util.List) ArrayList(java.util.ArrayList) SingularityS3Service(com.hubspot.singularity.helpers.SingularityS3Service) SingularityS3LogMetadata(com.hubspot.singularity.SingularityS3LogMetadata) S3ObjectSummary(com.amazonaws.services.s3.model.S3ObjectSummary) Date(java.util.Date) GeneratePresignedUrlRequest(com.amazonaws.services.s3.model.GeneratePresignedUrlRequest) SingularityS3Log(com.hubspot.singularity.SingularityS3Log) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) ListenableFuture(com.google.common.util.concurrent.ListenableFuture) ResponseHeaderOverrides(com.amazonaws.services.s3.model.ResponseHeaderOverrides) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap) HashMap(java.util.HashMap)

Example 2 with ResponseHeaderOverrides

use of com.amazonaws.services.s3.model.ResponseHeaderOverrides in project dataverse by IQSS.

the class S3AccessIO method generateTemporaryS3Url.

public String generateTemporaryS3Url() throws IOException {
    // A. 1 hour by default seems like an OK number. Making it configurable seems like a good idea too. -- L.A.
    if (s3 == null) {
        throw new IOException("ERROR: s3 not initialised. ");
    }
    if (dvObject instanceof DataFile) {
        key = getMainFileKey();
        java.util.Date expiration = new java.util.Date();
        long msec = expiration.getTime();
        msec += 1000 * getUrlExpirationMinutes();
        expiration.setTime(msec);
        GeneratePresignedUrlRequest generatePresignedUrlRequest = new GeneratePresignedUrlRequest(bucketName, key);
        // Default.
        generatePresignedUrlRequest.setMethod(HttpMethod.GET);
        generatePresignedUrlRequest.setExpiration(expiration);
        ResponseHeaderOverrides responseHeaders = new ResponseHeaderOverrides();
        // responseHeaders.setContentDisposition("attachment; filename="+this.getDataFile().getDisplayName());
        // Encode the file name explicitly specifying the encoding as UTF-8:
        // (otherwise S3 may not like non-ASCII characters!)
        // Most browsers are happy with just "filename="+URLEncoder.encode(this.getDataFile().getDisplayName(), "UTF-8")
        // in the header. But Firefox appears to require that "UTF8" is
        // specified explicitly, as below:
        responseHeaders.setContentDisposition("attachment; filename*=UTF-8''" + URLEncoder.encode(this.getDataFile().getDisplayName(), "UTF-8"));
        // - without it, download will work, but Firefox will leave the special
        // characters in the file name encoded. For example, the file name
        // will look like "1976%E2%80%932016.txt" instead of "1976–2016.txt",
        // where the dash is the "long dash", represented by a 3-byte UTF8
        // character "\xE2\x80\x93"
        responseHeaders.setContentType(this.getDataFile().getContentType());
        generatePresignedUrlRequest.setResponseHeaders(responseHeaders);
        URL s = s3.generatePresignedUrl(generatePresignedUrlRequest);
        return s.toString();
    } else if (dvObject instanceof Dataset) {
        throw new IOException("Data Access: GenerateTemporaryS3Url: Invalid DvObject type : Dataset");
    } else if (dvObject instanceof Dataverse) {
        throw new IOException("Data Access: Invalid DvObject type : Dataverse");
    } else {
        throw new IOException("Data Access: Invalid DvObject type");
    }
}
Also used : DataFile(edu.harvard.iq.dataverse.DataFile) GeneratePresignedUrlRequest(com.amazonaws.services.s3.model.GeneratePresignedUrlRequest) Date(java.util.Date) Dataset(edu.harvard.iq.dataverse.Dataset) IOException(java.io.IOException) ResponseHeaderOverrides(com.amazonaws.services.s3.model.ResponseHeaderOverrides) Dataverse(edu.harvard.iq.dataverse.Dataverse) Date(java.util.Date) URL(java.net.URL)

Aggregations

GeneratePresignedUrlRequest (com.amazonaws.services.s3.model.GeneratePresignedUrlRequest)2 ResponseHeaderOverrides (com.amazonaws.services.s3.model.ResponseHeaderOverrides)2 Date (java.util.Date)2 AmazonS3 (com.amazonaws.services.s3.AmazonS3)1 GetObjectMetadataRequest (com.amazonaws.services.s3.model.GetObjectMetadataRequest)1 ListObjectsV2Request (com.amazonaws.services.s3.model.ListObjectsV2Request)1 ListObjectsV2Result (com.amazonaws.services.s3.model.ListObjectsV2Result)1 S3ObjectSummary (com.amazonaws.services.s3.model.S3ObjectSummary)1 ImmutableMap (com.google.common.collect.ImmutableMap)1 ListenableFuture (com.google.common.util.concurrent.ListenableFuture)1 SingularityS3Log (com.hubspot.singularity.SingularityS3Log)1 SingularityS3LogMetadata (com.hubspot.singularity.SingularityS3LogMetadata)1 ContinuationToken (com.hubspot.singularity.api.ContinuationToken)1 S3ObjectSummaryHolder (com.hubspot.singularity.helpers.S3ObjectSummaryHolder)1 SingularityS3Service (com.hubspot.singularity.helpers.SingularityS3Service)1 DataFile (edu.harvard.iq.dataverse.DataFile)1 Dataset (edu.harvard.iq.dataverse.Dataset)1 Dataverse (edu.harvard.iq.dataverse.Dataverse)1 IOException (java.io.IOException)1 URL (java.net.URL)1