Search in sources :

Example 11 with GcsPath

use of org.apache.beam.sdk.util.gcsfs.GcsPath 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.util.gcsfs.GcsPath) FileNotFoundException(java.io.FileNotFoundException) Matchers.anyString(org.mockito.Matchers.anyString) MatchResult(org.apache.beam.sdk.io.fs.MatchResult) Test(org.junit.Test)

Example 12 with GcsPath

use of org.apache.beam.sdk.util.gcsfs.GcsPath 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.util.gcsfs.GcsPath) BigInteger(java.math.BigInteger) Nullable(javax.annotation.Nullable)

Example 13 with GcsPath

use of org.apache.beam.sdk.util.gcsfs.GcsPath in project beam by apache.

the class GcsUtilTest method testRecursiveGlobExpansion.

@Test
public void testRecursiveGlobExpansion() throws IOException {
    GcsOptions pipelineOptions = gcsOptionsWithTestCredential();
    GcsUtil gcsUtil = pipelineOptions.getGcsUtil();
    Storage mockStorage = Mockito.mock(Storage.class);
    gcsUtil.setStorageClient(mockStorage);
    Storage.Objects mockStorageObjects = Mockito.mock(Storage.Objects.class);
    Storage.Objects.Get mockStorageGet = Mockito.mock(Storage.Objects.Get.class);
    Storage.Objects.List mockStorageList = Mockito.mock(Storage.Objects.List.class);
    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(new StorageObject().setBucket("testbucket").setName("test/directory/file1.txt"));
    items.add(new StorageObject().setBucket("testbucket").setName("test/directory/file2.txt"));
    items.add(new StorageObject().setBucket("testbucket").setName("test/directory/file3.txt"));
    items.add(new StorageObject().setBucket("testbucket").setName("test/directory/otherfile"));
    items.add(new StorageObject().setBucket("testbucket").setName("test/directory/anotherfile"));
    items.add(new StorageObject().setBucket("testbucket").setName("test/file4.txt"));
    modelObjects.setItems(items);
    when(mockStorage.objects()).thenReturn(mockStorageObjects);
    when(mockStorageObjects.get("testbucket", "test/directory/otherfile")).thenReturn(mockStorageGet);
    when(mockStorageObjects.list("testbucket")).thenReturn(mockStorageList);
    when(mockStorageGet.execute()).thenReturn(new StorageObject().setBucket("testbucket").setName("test/directory/otherfile"));
    when(mockStorageList.execute()).thenReturn(modelObjects);
    {
        GcsPath pattern = GcsPath.fromUri("gs://testbucket/test/**/*.txt");
        List<GcsPath> expectedFiles = ImmutableList.of(GcsPath.fromUri("gs://testbucket/test/directory/file1.txt"), GcsPath.fromUri("gs://testbucket/test/directory/file2.txt"), GcsPath.fromUri("gs://testbucket/test/directory/file3.txt"), GcsPath.fromUri("gs://testbucket/test/file4.txt"));
        assertThat(expectedFiles, contains(gcsUtil.expand(pattern).toArray()));
    }
}
Also used : Storage(com.google.api.services.storage.Storage) 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.util.gcsfs.GcsPath) List(java.util.List) ArrayList(java.util.ArrayList) ImmutableList(com.google.common.collect.ImmutableList) GcsOptions(org.apache.beam.sdk.extensions.gcp.options.GcsOptions) Test(org.junit.Test)

Example 14 with GcsPath

use of org.apache.beam.sdk.util.gcsfs.GcsPath in project beam by apache.

the class GcsUtilTest method testGlobExpansion.

@Test
public void testGlobExpansion() throws IOException {
    GcsOptions pipelineOptions = gcsOptionsWithTestCredential();
    GcsUtil gcsUtil = pipelineOptions.getGcsUtil();
    Storage mockStorage = Mockito.mock(Storage.class);
    gcsUtil.setStorageClient(mockStorage);
    Storage.Objects mockStorageObjects = Mockito.mock(Storage.Objects.class);
    Storage.Objects.Get mockStorageGet = Mockito.mock(Storage.Objects.Get.class);
    Storage.Objects.List mockStorageList = Mockito.mock(Storage.Objects.List.class);
    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(new StorageObject().setBucket("testbucket").setName("testdirectory/file1name"));
    items.add(new StorageObject().setBucket("testbucket").setName("testdirectory/file2name"));
    items.add(new StorageObject().setBucket("testbucket").setName("testdirectory/file3name"));
    items.add(new StorageObject().setBucket("testbucket").setName("testdirectory/otherfile"));
    items.add(new StorageObject().setBucket("testbucket").setName("testdirectory/anotherfile"));
    modelObjects.setItems(items);
    when(mockStorage.objects()).thenReturn(mockStorageObjects);
    when(mockStorageObjects.get("testbucket", "testdirectory/otherfile")).thenReturn(mockStorageGet);
    when(mockStorageObjects.list("testbucket")).thenReturn(mockStorageList);
    when(mockStorageGet.execute()).thenReturn(new StorageObject().setBucket("testbucket").setName("testdirectory/otherfile"));
    when(mockStorageList.execute()).thenReturn(modelObjects);
    // Test a single file.
    {
        GcsPath pattern = GcsPath.fromUri("gs://testbucket/testdirectory/otherfile");
        List<GcsPath> expectedFiles = ImmutableList.of(GcsPath.fromUri("gs://testbucket/testdirectory/otherfile"));
        assertThat(expectedFiles, contains(gcsUtil.expand(pattern).toArray()));
    }
    // Test patterns.
    {
        GcsPath pattern = GcsPath.fromUri("gs://testbucket/testdirectory/file*");
        List<GcsPath> expectedFiles = ImmutableList.of(GcsPath.fromUri("gs://testbucket/testdirectory/file1name"), GcsPath.fromUri("gs://testbucket/testdirectory/file2name"), GcsPath.fromUri("gs://testbucket/testdirectory/file3name"));
        assertThat(expectedFiles, contains(gcsUtil.expand(pattern).toArray()));
    }
    {
        GcsPath pattern = GcsPath.fromUri("gs://testbucket/testdirectory/file[1-3]*");
        List<GcsPath> expectedFiles = ImmutableList.of(GcsPath.fromUri("gs://testbucket/testdirectory/file1name"), GcsPath.fromUri("gs://testbucket/testdirectory/file2name"), GcsPath.fromUri("gs://testbucket/testdirectory/file3name"));
        assertThat(expectedFiles, contains(gcsUtil.expand(pattern).toArray()));
    }
    {
        GcsPath pattern = GcsPath.fromUri("gs://testbucket/testdirectory/file?name");
        List<GcsPath> expectedFiles = ImmutableList.of(GcsPath.fromUri("gs://testbucket/testdirectory/file1name"), GcsPath.fromUri("gs://testbucket/testdirectory/file2name"), GcsPath.fromUri("gs://testbucket/testdirectory/file3name"));
        assertThat(expectedFiles, contains(gcsUtil.expand(pattern).toArray()));
    }
    {
        GcsPath pattern = GcsPath.fromUri("gs://testbucket/test*ectory/fi*name");
        List<GcsPath> expectedFiles = ImmutableList.of(GcsPath.fromUri("gs://testbucket/testdirectory/file1name"), GcsPath.fromUri("gs://testbucket/testdirectory/file2name"), GcsPath.fromUri("gs://testbucket/testdirectory/file3name"));
        assertThat(expectedFiles, contains(gcsUtil.expand(pattern).toArray()));
    }
}
Also used : Storage(com.google.api.services.storage.Storage) 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.util.gcsfs.GcsPath) List(java.util.List) ArrayList(java.util.ArrayList) ImmutableList(com.google.common.collect.ImmutableList) GcsOptions(org.apache.beam.sdk.extensions.gcp.options.GcsOptions) Test(org.junit.Test)

Example 15 with GcsPath

use of org.apache.beam.sdk.util.gcsfs.GcsPath in project beam by apache.

the class GcsUtil method makeCopyBatches.

List<BatchRequest> makeCopyBatches(Iterable<String> srcFilenames, Iterable<String> destFilenames) throws IOException {
    List<String> srcList = Lists.newArrayList(srcFilenames);
    List<String> destList = Lists.newArrayList(destFilenames);
    checkArgument(srcList.size() == destList.size(), "Number of source files %s must equal number of destination files %s", srcList.size(), destList.size());
    List<BatchRequest> batches = new LinkedList<>();
    BatchRequest batch = createBatchRequest();
    for (int i = 0; i < srcList.size(); i++) {
        final GcsPath sourcePath = GcsPath.fromUri(srcList.get(i));
        final GcsPath destPath = GcsPath.fromUri(destList.get(i));
        enqueueCopy(sourcePath, destPath, batch);
        if (batch.size() >= MAX_REQUESTS_PER_BATCH) {
            batches.add(batch);
            batch = createBatchRequest();
        }
    }
    if (batch.size() > 0) {
        batches.add(batch);
    }
    return batches;
}
Also used : BatchRequest(com.google.api.client.googleapis.batch.BatchRequest) GcsPath(org.apache.beam.sdk.util.gcsfs.GcsPath) LinkedList(java.util.LinkedList)

Aggregations

GcsPath (org.apache.beam.sdk.util.gcsfs.GcsPath)19 ImmutableList (com.google.common.collect.ImmutableList)9 List (java.util.List)8 Objects (com.google.api.services.storage.model.Objects)7 StorageObject (com.google.api.services.storage.model.StorageObject)6 ArrayList (java.util.ArrayList)6 LinkedList (java.util.LinkedList)6 GcsUtil (org.apache.beam.sdk.util.GcsUtil)6 Test (org.junit.Test)6 InvocationOnMock (org.mockito.invocation.InvocationOnMock)5 Storage (com.google.api.services.storage.Storage)4 GcsOptions (org.apache.beam.sdk.extensions.gcp.options.GcsOptions)4 SeekableByteChannel (java.nio.channels.SeekableByteChannel)3 BatchRequest (com.google.api.client.googleapis.batch.BatchRequest)2 GoogleJsonResponseException (com.google.api.client.googleapis.json.GoogleJsonResponseException)2 FileNotFoundException (java.io.FileNotFoundException)2 DataflowPipelineOptions (org.apache.beam.runners.dataflow.options.DataflowPipelineOptions)2 TestCredential (org.apache.beam.sdk.extensions.gcp.auth.TestCredential)2 MatchResult (org.apache.beam.sdk.io.fs.MatchResult)2 TupleTagList (org.apache.beam.sdk.values.TupleTagList)2