Examples with GcsPath org.apache.beam.sdk.util.gcsfs.GcsPath used on opensource projects

Example 16 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()));
    }
}

Example 17 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;
}

Example 18 with GcsPath

use of org.apache.beam.sdk.util.gcsfs.GcsPath 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 LinkedList<>();
    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;
}

Example 19 with GcsPath

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

the class MinimalWordCountJava8Test method buildMockGcsUtil.

private GcsUtil buildMockGcsUtil() throws IOException {
    GcsUtil mockGcsUtil = Mockito.mock(GcsUtil.class);
    // Any request to open gets a new bogus channel
    Mockito.when(mockGcsUtil.open(Mockito.any(GcsPath.class))).then(new Answer<SeekableByteChannel>() {

        @Override
        public SeekableByteChannel answer(InvocationOnMock invocation) throws Throwable {
            return FileChannel.open(Files.createTempFile("channel-", ".tmp"), StandardOpenOption.CREATE, StandardOpenOption.DELETE_ON_CLOSE);
        }
    });
    // Any request for expansion returns a list containing the original GcsPath
    // This is required to pass validation that occurs in TextIO during apply()
    Mockito.when(mockGcsUtil.expand(Mockito.any(GcsPath.class))).then(new Answer<List<GcsPath>>() {

        @Override
        public List<GcsPath> answer(InvocationOnMock invocation) throws Throwable {
            return ImmutableList.of((GcsPath) invocation.getArguments()[0]);
        }
    });
    return mockGcsUtil;
}