Search in sources :

Example 76 with StorageObject

use of com.google.api.services.storage.model.StorageObject in project druid by druid-io.

the class GoogleTimestampVersionedDataFinderTest method getLatestVersionTrailingSlashKeyPrefix.

@Test
public void getLatestVersionTrailingSlashKeyPrefix() {
    String bucket = "bucket";
    String keyPrefix = "prefix/dir/0/";
    // object for directory prefix/dir/0/
    final StorageObject storageObject1 = ObjectStorageIteratorTest.makeStorageObject(bucket, keyPrefix + "/", 0);
    storageObject1.setUpdated(new DateTime(System.currentTimeMillis()));
    final StorageObject storageObject2 = ObjectStorageIteratorTest.makeStorageObject(bucket, keyPrefix + "v1", 1);
    storageObject2.setUpdated(new DateTime(System.currentTimeMillis()));
    final StorageObject storageObject3 = ObjectStorageIteratorTest.makeStorageObject(bucket, keyPrefix + "v2", 1);
    storageObject3.setUpdated(new DateTime(System.currentTimeMillis() + 100));
    final StorageObject storageObject4 = ObjectStorageIteratorTest.makeStorageObject(bucket, keyPrefix + "other", 4);
    storageObject4.setUpdated(new DateTime(System.currentTimeMillis() + 100));
    final GoogleStorage storage = ObjectStorageIteratorTest.makeMockClient(ImmutableList.of(storageObject1, storageObject2, storageObject3, storageObject4));
    final GoogleTimestampVersionedDataFinder finder = new GoogleTimestampVersionedDataFinder(storage);
    Pattern pattern = Pattern.compile("v.*");
    URI latest = finder.getLatestVersion(URI.create(StringUtils.format("gs://%s/%s", bucket, keyPrefix)), pattern);
    URI expected = URI.create(StringUtils.format("gs://%s/%s", bucket, storageObject3.getName()));
    Assert.assertEquals(expected, latest);
}
Also used : Pattern(java.util.regex.Pattern) StorageObject(com.google.api.services.storage.model.StorageObject) URI(java.net.URI) DateTime(com.google.api.client.util.DateTime) Test(org.junit.Test)

Example 77 with StorageObject

use of com.google.api.services.storage.model.StorageObject in project druid by druid-io.

the class ObjectStorageIteratorTest method makeStorageObject.

static StorageObject makeStorageObject(final String bucket, final String key, final long size) {
    final StorageObject summary = new StorageObject();
    summary.setBucket(bucket);
    summary.setName(key);
    summary.setSize(BigInteger.valueOf(size));
    return summary;
}
Also used : StorageObject(com.google.api.services.storage.model.StorageObject)

Example 78 with StorageObject

use of com.google.api.services.storage.model.StorageObject in project beam by apache.

the class GcsUtil method expand.

/**
 * Expands a pattern into matched paths. The pattern path may contain globs, which are expanded in
 * the result. For patterns that only match a single object, we ensure that the object exists.
 */
public List<GcsPath> expand(GcsPath gcsPattern) throws IOException {
    Pattern p = null;
    String prefix = null;
    if (isWildcard(gcsPattern)) {
        // Part before the first wildcard character.
        prefix = getNonWildcardPrefix(gcsPattern.getObject());
        p = Pattern.compile(wildcardToRegexp(gcsPattern.getObject()));
    } else {
        // Not a wildcard.
        try {
            // Use a get request to fetch the metadata of the object, and ignore the return value.
            // The request has strong global consistency.
            getObject(gcsPattern);
            return ImmutableList.of(gcsPattern);
        } catch (FileNotFoundException e) {
            // If the path was not found, return an empty list.
            return ImmutableList.of();
        }
    }
    LOG.debug("matching files in bucket {}, prefix {} against pattern {}", gcsPattern.getBucket(), prefix, p.toString());
    String pageToken = null;
    List<GcsPath> results = new ArrayList<>();
    do {
        Objects objects = listObjects(gcsPattern.getBucket(), prefix, pageToken);
        if (objects.getItems() == null) {
            break;
        }
        // Filter objects based on the regex.
        for (StorageObject o : objects.getItems()) {
            String name = o.getName();
            // Skip directories, which end with a slash.
            if (p.matcher(name).matches() && !name.endsWith("/")) {
                LOG.debug("Matched object: {}", name);
                results.add(GcsPath.fromObject(o));
            }
        }
        pageToken = objects.getNextPageToken();
    } while (pageToken != null);
    return results;
}
Also used : Pattern(java.util.regex.Pattern) StorageObject(com.google.api.services.storage.model.StorageObject) FileNotFoundException(java.io.FileNotFoundException) GcsPath(org.apache.beam.sdk.extensions.gcp.util.gcsfs.GcsPath) ArrayList(java.util.ArrayList) Objects(com.google.api.services.storage.model.Objects)

Example 79 with StorageObject

use of com.google.api.services.storage.model.StorageObject in project beam by apache.

the class GcsFileSystemTest method createStorageObject.

private StorageObject createStorageObject(String gcsFilename, long fileSize) {
    GcsPath gcsPath = GcsPath.fromUri(gcsFilename);
    // Google APIs will use null for empty files.
    @Nullable BigInteger size = (fileSize == 0) ? null : BigInteger.valueOf(fileSize);
    return new StorageObject().setBucket(gcsPath.getBucket()).setName(gcsPath.getObject()).setSize(size);
}
Also used : StorageObject(com.google.api.services.storage.model.StorageObject) GcsPath(org.apache.beam.sdk.extensions.gcp.util.gcsfs.GcsPath) BigInteger(java.math.BigInteger) Nullable(org.checkerframework.checker.nullness.qual.Nullable)

Example 80 with StorageObject

use of com.google.api.services.storage.model.StorageObject in project beam by apache.

the class GcsFileSystemTest method testMatch.

@Test
public void testMatch() throws Exception {
    Objects modelObjects = new Objects();
    List<StorageObject> items = new ArrayList<>();
    // A directory
    items.add(new StorageObject().setBucket("testbucket").setName("testdirectory/"));
    // Files within the directory
    items.add(createStorageObject("gs://testbucket/testdirectory/file1name", 1L));
    items.add(createStorageObject("gs://testbucket/testdirectory/file2name", 2L));
    items.add(createStorageObject("gs://testbucket/testdirectory/file3name", 3L));
    items.add(createStorageObject("gs://testbucket/testdirectory/file4name", 4L));
    items.add(createStorageObject("gs://testbucket/testdirectory/otherfile", 5L));
    items.add(createStorageObject("gs://testbucket/testdirectory/anotherfile", 6L));
    modelObjects.setItems(items);
    when(mockGcsUtil.listObjects(eq("testbucket"), anyString(), isNull(String.class))).thenReturn(modelObjects);
    List<GcsPath> gcsPaths = ImmutableList.of(GcsPath.fromUri("gs://testbucket/testdirectory/non-exist-file"), GcsPath.fromUri("gs://testbucket/testdirectory/otherfile"));
    when(mockGcsUtil.getObjects(eq(gcsPaths))).thenReturn(ImmutableList.of(StorageObjectOrIOException.create(new FileNotFoundException()), StorageObjectOrIOException.create(createStorageObject("gs://testbucket/testdirectory/otherfile", 4L))));
    List<String> specs = ImmutableList.of("gs://testbucket/testdirectory/file[1-3]*", "gs://testbucket/testdirectory/non-exist-file", "gs://testbucket/testdirectory/otherfile");
    List<MatchResult> matchResults = gcsFileSystem.match(specs);
    assertEquals(3, matchResults.size());
    assertEquals(Status.OK, matchResults.get(0).status());
    assertThat(ImmutableList.of("gs://testbucket/testdirectory/file1name", "gs://testbucket/testdirectory/file2name", "gs://testbucket/testdirectory/file3name"), contains(toFilenames(matchResults.get(0)).toArray()));
    assertEquals(Status.NOT_FOUND, matchResults.get(1).status());
    assertEquals(Status.OK, matchResults.get(2).status());
    assertThat(ImmutableList.of("gs://testbucket/testdirectory/otherfile"), contains(toFilenames(matchResults.get(2)).toArray()));
}
Also used : StorageObject(com.google.api.services.storage.model.StorageObject) Objects(com.google.api.services.storage.model.Objects) ArrayList(java.util.ArrayList) GcsPath(org.apache.beam.sdk.extensions.gcp.util.gcsfs.GcsPath) FileNotFoundException(java.io.FileNotFoundException) ArgumentMatchers.anyString(org.mockito.ArgumentMatchers.anyString) MatchResult(org.apache.beam.sdk.io.fs.MatchResult) Test(org.junit.Test)

Aggregations

StorageObject (com.google.api.services.storage.model.StorageObject)81 Test (org.junit.Test)33 Objects (com.google.api.services.storage.model.Objects)21 IOException (java.io.IOException)20 ArrayList (java.util.ArrayList)16 Storage (com.google.api.services.storage.Storage)12 List (java.util.List)9 URI (java.net.URI)8 RetryHelperException (com.google.cloud.RetryHelper.RetryHelperException)7 GcsOptions (org.apache.beam.sdk.extensions.gcp.options.GcsOptions)7 GcsPath (org.apache.beam.sdk.extensions.gcp.util.gcsfs.GcsPath)7 GoogleJsonError (com.google.api.client.googleapis.json.GoogleJsonError)6 Nullable (javax.annotation.Nullable)6 File (java.io.File)5 Pattern (java.util.regex.Pattern)5 DateTime (com.google.api.client.util.DateTime)4 BlobSourceOption (com.google.cloud.storage.Storage.BlobSourceOption)4 BlobTargetOption (com.google.cloud.storage.Storage.BlobTargetOption)4 BlobWriteOption (com.google.cloud.storage.Storage.BlobWriteOption)4 BucketSourceOption (com.google.cloud.storage.Storage.BucketSourceOption)4