Search in sources :

Example 26 with FileSet

use of co.cask.cdap.api.dataset.lib.FileSet in project cdap by caskdata.

the class FileSetTest method testRollbackWithNonEmptyDir.

@Test
public void testRollbackWithNonEmptyDir() throws IOException, TransactionFailureException, DatasetManagementException {
    FileSet fileSet1 = createFileset(testFileSetInstance1);
    Location outputDir = fileSet1.getOutputLocation();
    Assert.assertFalse(outputDir.exists());
    Assert.assertTrue(outputDir.mkdirs());
    Location outputFile = outputDir.append("outputFile");
    // this will create the outputFile
    outputFile.getOutputStream().close();
    Assert.assertTrue(outputFile.exists());
    Assert.assertTrue(outputDir.exists());
    ((FileSetDataset) fileSet1).onFailure();
    // both the output dir and file in it should still exist
    Assert.assertTrue(outputDir.exists());
    Assert.assertTrue(outputFile.exists());
}
Also used : FileSet(co.cask.cdap.api.dataset.lib.FileSet) FileSetDataset(co.cask.cdap.data2.dataset2.lib.file.FileSetDataset) Location(org.apache.twill.filesystem.Location) Test(org.junit.Test)

Example 27 with FileSet

use of co.cask.cdap.api.dataset.lib.FileSet in project cdap by caskdata.

the class FileSetTest method testPermissions.

@Test
public void testPermissions() throws Exception {
    String fsPermissions = "rwxrwx--x";
    String customPermissions = "rwx--x--x";
    String group = UserGroupInformation.getCurrentUser().getPrimaryGroupName();
    // create one file set with default permission so that the namespace data dir exists
    dsFrameworkUtil.createInstance("fileSet", OTHER_NAMESPACE.dataset("dummy"), DatasetProperties.EMPTY);
    // determine the default permissions of created directories (we want to test with different perms)
    String defaultPermissions = dsFrameworkUtil.getInjector().getInstance(NamespacedLocationFactory.class).get(OTHER_NAMESPACE).getPermissions();
    if (fsPermissions.equals(defaultPermissions)) {
        // swap the permissions so we can test with different file set permissions than the default
        customPermissions = "rwxrwx--x";
        fsPermissions = "rwx--x--x";
    }
    // create a dataset with configured permissions that are different from the default
    DatasetId datasetId = OTHER_NAMESPACE.dataset("testPermFS");
    dsFrameworkUtil.createInstance("fileSet", datasetId, FileSetProperties.builder().setBasePath("perm/test/path").setFilePermissions(fsPermissions).setFileGroup(group).build());
    FileSet fs = dsFrameworkUtil.getInstance(datasetId);
    // validate that the entire hierarchy of directories was created with the correct permissions
    Location base = fs.getBaseLocation();
    Assert.assertEquals(group, base.getGroup());
    Assert.assertEquals(fsPermissions, base.getPermissions());
    Location parent = Locations.getParent(base);
    Assert.assertNotNull(parent);
    Assert.assertEquals(group, parent.getGroup());
    Assert.assertEquals(fsPermissions, parent.getPermissions());
    parent = Locations.getParent(parent);
    Assert.assertNotNull(parent);
    Assert.assertEquals(group, parent.getGroup());
    Assert.assertEquals(fsPermissions, parent.getPermissions());
    Location nsRoot = Locations.getParent(parent);
    Assert.assertNotNull(nsRoot);
    Assert.assertNotEquals(fsPermissions, nsRoot.getPermissions());
    // create an empty file and validate it is created with the fileset's permissions
    Location child = base.append("a");
    Location grandchild = child.append("b");
    grandchild.getOutputStream().close();
    Assert.assertEquals(group, child.getGroup());
    Assert.assertEquals(group, grandchild.getGroup());
    Assert.assertEquals(fsPermissions, child.getPermissions());
    Assert.assertEquals(fsPermissions, grandchild.getPermissions());
    // create an empty file with custom permissions and validate them
    child = base.append("x");
    grandchild = child.append("y");
    grandchild.getOutputStream(customPermissions).close();
    Assert.assertEquals(group, child.getGroup());
    Assert.assertEquals(group, grandchild.getGroup());
    Assert.assertEquals(customPermissions, child.getPermissions());
    Assert.assertEquals(customPermissions, grandchild.getPermissions());
    // instantiate the dataset with custom permissions in the runtime arguments
    fs = dsFrameworkUtil.getInstance(datasetId, ImmutableMap.of(FileSetProperties.PROPERTY_FILES_PERMISSIONS, customPermissions));
    // create an empty file with custom permissions and validate them
    base = fs.getBaseLocation();
    child = base.append("p");
    grandchild = child.append("q");
    grandchild.getOutputStream().close();
    Assert.assertEquals(group, child.getGroup());
    Assert.assertEquals(group, grandchild.getGroup());
    Assert.assertEquals(customPermissions, child.getPermissions());
    Assert.assertEquals(customPermissions, grandchild.getPermissions());
}
Also used : FileSet(co.cask.cdap.api.dataset.lib.FileSet) DatasetId(co.cask.cdap.proto.id.DatasetId) Location(org.apache.twill.filesystem.Location) Test(org.junit.Test)

Example 28 with FileSet

use of co.cask.cdap.api.dataset.lib.FileSet in project cdap by caskdata.

the class FileSetTest method testRollbackOfNonDirectoryOutput.

@Test
public void testRollbackOfNonDirectoryOutput() throws IOException, TransactionFailureException, DatasetManagementException {
    // test deletion of an output location, pointing to a non-directory file
    FileSet fileSet1 = createFileset(testFileSetInstance1);
    Location outputFile = fileSet1.getOutputLocation();
    Assert.assertFalse(outputFile.exists());
    outputFile.getOutputStream().close();
    Assert.assertTrue(outputFile.exists());
    ((FileSetDataset) fileSet1).onFailure();
    // the output file should still not be deleted
    Assert.assertTrue(outputFile.exists());
}
Also used : FileSet(co.cask.cdap.api.dataset.lib.FileSet) FileSetDataset(co.cask.cdap.data2.dataset2.lib.file.FileSetDataset) Location(org.apache.twill.filesystem.Location) Test(org.junit.Test)

Example 29 with FileSet

use of co.cask.cdap.api.dataset.lib.FileSet in project cdap by caskdata.

the class PipelineTest method testTextFileSourceAndMoveAction.

@Test
public void testTextFileSourceAndMoveAction() throws Exception {
    // create the pipeline config
    String moveFromName = "sourceTestMoveFrom";
    String inputName = "sourceTestInput";
    String outputName = "sourceTestOutput";
    Map<String, String> actionProperties = new HashMap<>();
    actionProperties.put(FilesetMoveAction.Conf.SOURCE_FILESET, "sourceTestMoveFrom");
    actionProperties.put(FilesetMoveAction.Conf.DEST_FILESET, inputName);
    ETLStage moveAction = new ETLStage("moveInput", new ETLPlugin(FilesetMoveAction.NAME, Action.PLUGIN_TYPE, actionProperties, null));
    Map<String, String> sourceProperties = new HashMap<>();
    sourceProperties.put(TextFileSetSource.Conf.FILESET_NAME, inputName);
    sourceProperties.put(TextFileSetSource.Conf.CREATE_IF_NOT_EXISTS, "true");
    sourceProperties.put(TextFileSetSource.Conf.DELETE_INPUT_ON_SUCCESS, "true");
    sourceProperties.put(TextFileSetSource.Conf.FILES, "${file}");
    ETLStage source = new ETLStage("source", new ETLPlugin(TextFileSetSource.NAME, BatchSource.PLUGIN_TYPE, sourceProperties, null));
    ETLStage sink = new ETLStage("sink", MockSink.getPlugin(outputName));
    ETLBatchConfig pipelineConfig = ETLBatchConfig.builder("* * * * *").addStage(source).addStage(sink).addStage(moveAction).addConnection(moveAction.getName(), source.getName()).addConnection(source.getName(), sink.getName()).build();
    // create the move from fileset
    addDatasetInstance(FileSet.class.getName(), moveFromName);
    // create the pipeline
    ApplicationId pipelineId = NamespaceId.DEFAULT.app("textSourceTestPipeline");
    ApplicationManager appManager = deployApplication(pipelineId, new AppRequest<>(APP_ARTIFACT, pipelineConfig));
    // write some files that will be moved to the input fileset
    DataSetManager<FileSet> moveFromManager = getDataset(moveFromName);
    // this file starts with '.' and should be ignored.
    Location invisibleFile = moveFromManager.get().getBaseLocation().append(".hidden");
    try (OutputStream outputStream = invisibleFile.getOutputStream()) {
        outputStream.write(Bytes.toBytes("this should not be read"));
    }
    // this file should be moved
    String line1 = "Hello World!";
    String line2 = "Good to meet you";
    String line3 = "My name is Hal";
    String inputText = line1 + "\n" + line2 + "\n" + line3;
    Location inputFile = moveFromManager.get().getBaseLocation().append("inputFile");
    try (OutputStream outputStream = inputFile.getOutputStream()) {
        outputStream.write(Bytes.toBytes(inputText));
    }
    // run the pipeline
    Map<String, String> runtimeArgs = new HashMap<>();
    // the ${file} macro will be substituted with "inputFile" for our pipeline run
    runtimeArgs.put("file", "inputFile");
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start(runtimeArgs);
    workflowManager.waitForFinish(4, TimeUnit.MINUTES);
    // check the pipeline output
    DataSetManager<Table> outputManager = getDataset(outputName);
    Set<StructuredRecord> outputRecords = new HashSet<>();
    outputRecords.addAll(MockSink.readOutput(outputManager));
    Set<StructuredRecord> expected = new HashSet<>();
    expected.add(StructuredRecord.builder(TextFileSetSource.OUTPUT_SCHEMA).set("position", (long) inputText.indexOf(line1)).set("text", line1).build());
    expected.add(StructuredRecord.builder(TextFileSetSource.OUTPUT_SCHEMA).set("position", (long) inputText.indexOf(line2)).set("text", line2).build());
    expected.add(StructuredRecord.builder(TextFileSetSource.OUTPUT_SCHEMA).set("position", (long) inputText.indexOf(line3)).set("text", line3).build());
    Assert.assertEquals(expected, outputRecords);
    // check that the input file does not exist in the moveFrom fileSet,
    // and was deleted by the source in the input fileSet
    Assert.assertFalse(moveFromManager.get().getBaseLocation().append("inputFile").exists());
    DataSetManager<FileSet> inputManager = getDataset(inputName);
    Assert.assertFalse(inputManager.get().getBaseLocation().append("inputFile").exists());
}
Also used : ApplicationManager(co.cask.cdap.test.ApplicationManager) Table(co.cask.cdap.api.dataset.table.Table) KeyValueTable(co.cask.cdap.api.dataset.lib.KeyValueTable) FileSet(co.cask.cdap.api.dataset.lib.FileSet) HashMap(java.util.HashMap) OutputStream(java.io.OutputStream) WorkflowManager(co.cask.cdap.test.WorkflowManager) ETLPlugin(co.cask.cdap.etl.proto.v2.ETLPlugin) StructuredRecord(co.cask.cdap.api.data.format.StructuredRecord) ETLBatchConfig(co.cask.cdap.etl.proto.v2.ETLBatchConfig) ETLStage(co.cask.cdap.etl.proto.v2.ETLStage) ApplicationId(co.cask.cdap.proto.id.ApplicationId) Location(org.apache.twill.filesystem.Location) HashSet(java.util.HashSet) Test(org.junit.Test)

Example 30 with FileSet

use of co.cask.cdap.api.dataset.lib.FileSet in project cdap by caskdata.

the class PipelineTest method testTextFileSinkAndDeletePostAction.

@Test
public void testTextFileSinkAndDeletePostAction() throws Exception {
    // create the pipeline config
    String inputName = "sinkTestInput";
    String outputName = "sinkTestOutput";
    String outputDirName = "users";
    ETLStage source = new ETLStage("source", MockSource.getPlugin(inputName));
    Map<String, String> sinkProperties = new HashMap<>();
    sinkProperties.put(TextFileSetSink.Conf.FILESET_NAME, outputName);
    sinkProperties.put(TextFileSetSink.Conf.FIELD_SEPARATOR, "|");
    sinkProperties.put(TextFileSetSink.Conf.OUTPUT_DIR, "${dir}");
    ETLStage sink = new ETLStage("sink", new ETLPlugin(TextFileSetSink.NAME, BatchSink.PLUGIN_TYPE, sinkProperties, null));
    Map<String, String> actionProperties = new HashMap<>();
    actionProperties.put(FilesetDeletePostAction.Conf.FILESET_NAME, outputName);
    // mapreduce writes multiple files to the output directory. Along with the actual output,
    // there are various .crc files that do not contain any of the output content.
    actionProperties.put(FilesetDeletePostAction.Conf.DELETE_REGEX, ".*\\.crc|_SUCCESS");
    actionProperties.put(FilesetDeletePostAction.Conf.DIRECTORY, outputDirName);
    ETLStage postAction = new ETLStage("cleanup", new ETLPlugin(FilesetDeletePostAction.NAME, PostAction.PLUGIN_TYPE, actionProperties, null));
    ETLBatchConfig pipelineConfig = ETLBatchConfig.builder("* * * * *").addStage(source).addStage(sink).addPostAction(postAction).addConnection(source.getName(), sink.getName()).build();
    // create the pipeline
    ApplicationId pipelineId = NamespaceId.DEFAULT.app("textSinkTestPipeline");
    ApplicationManager appManager = deployApplication(pipelineId, new AppRequest<>(APP_ARTIFACT, pipelineConfig));
    // write some data to the input fileset
    Schema inputSchema = Schema.recordOf("test", Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("item", Schema.of(Schema.Type.STRING)));
    Map<String, String> users = new HashMap<>();
    users.put("samuel", "wallet");
    users.put("dwayne", "rock");
    users.put("christopher", "cowbell");
    List<StructuredRecord> inputRecords = new ArrayList<>();
    for (Map.Entry<String, String> userEntry : users.entrySet()) {
        String name = userEntry.getKey();
        String item = userEntry.getValue();
        inputRecords.add(StructuredRecord.builder(inputSchema).set("name", name).set("item", item).build());
    }
    DataSetManager<Table> inputManager = getDataset(inputName);
    MockSource.writeInput(inputManager, inputRecords);
    // run the pipeline
    Map<String, String> runtimeArgs = new HashMap<>();
    // the ${dir} macro will be substituted with "users" for our pipeline run
    runtimeArgs.put("dir", outputDirName);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start(runtimeArgs);
    workflowManager.waitForFinish(4, TimeUnit.MINUTES);
    // check the pipeline output
    DataSetManager<FileSet> outputManager = getDataset(outputName);
    FileSet output = outputManager.get();
    Location outputDir = output.getBaseLocation().append(outputDirName);
    Map<String, String> actual = new HashMap<>();
    for (Location outputFile : outputDir.list()) {
        if (outputFile.getName().endsWith(".crc") || "_SUCCESS".equals(outputFile.getName())) {
            Assert.fail("Post action did not delete file " + outputFile.getName());
        }
        try (BufferedReader reader = new BufferedReader(new InputStreamReader(outputFile.getInputStream()))) {
            String line;
            while ((line = reader.readLine()) != null) {
                String[] parts = line.split("\\|");
                actual.put(parts[0], parts[1]);
            }
        }
    }
    Assert.assertEquals(actual, users);
}
Also used : ApplicationManager(co.cask.cdap.test.ApplicationManager) Table(co.cask.cdap.api.dataset.table.Table) KeyValueTable(co.cask.cdap.api.dataset.lib.KeyValueTable) FileSet(co.cask.cdap.api.dataset.lib.FileSet) InputStreamReader(java.io.InputStreamReader) HashMap(java.util.HashMap) Schema(co.cask.cdap.api.data.schema.Schema) WorkflowManager(co.cask.cdap.test.WorkflowManager) ArrayList(java.util.ArrayList) ETLPlugin(co.cask.cdap.etl.proto.v2.ETLPlugin) StructuredRecord(co.cask.cdap.api.data.format.StructuredRecord) ETLBatchConfig(co.cask.cdap.etl.proto.v2.ETLBatchConfig) ETLStage(co.cask.cdap.etl.proto.v2.ETLStage) BufferedReader(java.io.BufferedReader) ApplicationId(co.cask.cdap.proto.id.ApplicationId) HashMap(java.util.HashMap) Map(java.util.Map) Location(org.apache.twill.filesystem.Location) Test(org.junit.Test)

Aggregations

FileSet (co.cask.cdap.api.dataset.lib.FileSet)40 Location (org.apache.twill.filesystem.Location)28 Test (org.junit.Test)19 PartitionedFileSet (co.cask.cdap.api.dataset.lib.PartitionedFileSet)12 HashMap (java.util.HashMap)11 KeyValueTable (co.cask.cdap.api.dataset.lib.KeyValueTable)8 TimePartitionedFileSet (co.cask.cdap.api.dataset.lib.TimePartitionedFileSet)8 ApplicationManager (co.cask.cdap.test.ApplicationManager)8 DatasetId (co.cask.cdap.proto.id.DatasetId)7 ColumnDesc (co.cask.cdap.proto.ColumnDesc)6 QueryResult (co.cask.cdap.proto.QueryResult)6 SparkManager (co.cask.cdap.test.SparkManager)5 WorkflowManager (co.cask.cdap.test.WorkflowManager)5 Table (co.cask.cdap.api.dataset.table.Table)4 ServiceManager (co.cask.cdap.test.ServiceManager)4 IOException (java.io.IOException)4 Map (java.util.Map)4 StructuredRecord (co.cask.cdap.api.data.format.StructuredRecord)3 PartitionedFileSetProperties (co.cask.cdap.api.dataset.lib.PartitionedFileSetProperties)3 FileSetDataset (co.cask.cdap.data2.dataset2.lib.file.FileSetDataset)3