use of org.apache.beam.sdk.extensions.gcp.util.gcsfs.GcsPath in project beam by apache.
the class BatchStatefulParDoOverridesTest method buildPipelineOptions.
private static DataflowPipelineOptions buildPipelineOptions(String... args) throws IOException {
GcsUtil mockGcsUtil = mock(GcsUtil.class);
when(mockGcsUtil.expand(any(GcsPath.class))).then(invocation -> ImmutableList.of((GcsPath) invocation.getArguments()[0]));
when(mockGcsUtil.bucketAccessible(any(GcsPath.class))).thenReturn(true);
DataflowPipelineOptions options = PipelineOptionsFactory.fromArgs(args).as(DataflowPipelineOptions.class);
options.setRunner(DataflowRunner.class);
options.setGcpCredential(new TestCredential());
options.setJobName("some-job-name");
options.setProject("some-project");
options.setRegion("some-region");
options.setTempLocation(GcsPath.fromComponents("somebucket", "some/path").toString());
options.setFilesToStage(new ArrayList<>());
options.setGcsUtil(mockGcsUtil);
// Enable the FileSystems API to know about gs:// URIs in this test.
FileSystems.setDefaultPipelineOptions(options);
return options;
}
use of org.apache.beam.sdk.extensions.gcp.util.gcsfs.GcsPath in project beam by apache.
the class GcsUtil method makeRewriteOps.
LinkedList<RewriteOp> makeRewriteOps(Iterable<String> srcFilenames, Iterable<String> destFilenames, boolean deleteSource, boolean ignoreMissingSource, boolean ignoreExistingDest) 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());
LinkedList<RewriteOp> rewrites = Lists.newLinkedList();
for (int i = 0; i < srcList.size(); i++) {
final GcsPath sourcePath = GcsPath.fromUri(srcList.get(i));
final GcsPath destPath = GcsPath.fromUri(destList.get(i));
if (ignoreExistingDest && !sourcePath.getBucket().equals(destPath.getBucket())) {
throw new UnsupportedOperationException("Skipping dest existence is only supported within a bucket.");
}
rewrites.addLast(new RewriteOp(sourcePath, destPath, deleteSource, ignoreMissingSource));
}
return rewrites;
}
use of org.apache.beam.sdk.extensions.gcp.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 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;
}
use of org.apache.beam.sdk.extensions.gcp.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);
}
use of org.apache.beam.sdk.extensions.gcp.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()));
}
Aggregations