Examples with FieldOperation io.cdap.cdap.etl.api.lineage.field.FieldOperation used on opensource projects

Example 16 with FieldOperation

use of io.cdap.cdap.etl.api.lineage.field.FieldOperation in project cdap by cdapio.

the class LineageOperationProcessorTest method testAnotherSimplePipeline.

@Test
public void testAnotherSimplePipeline() {
    // n1-->n2-->n3-->n4
    // n1 => read: file -> (offset, body)
    // n2 => parse: (body) -> (first_name, last_name) | n2
    // n3 => concat: (first_name, last_name) -> (name) | n
    // n4 => write: (offset, name) -> another_file
    Set<Connection> connections = new HashSet<>();
    connections.add(new Connection("n1", "n2"));
    connections.add(new Connection("n2", "n3"));
    connections.add(new Connection("n3", "n4"));
    Map<String, List<FieldOperation>> stageOperations = new HashMap<>();
    List<FieldOperation> fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldReadOperation("read", "some read", EndPoint.of("ns", "file1"), "offset", "body"));
    stageOperations.put("n1", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldTransformOperation("parse", "parsing body", Collections.singletonList("body"), "first_name", "last_name"));
    stageOperations.put("n2", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldTransformOperation("concat", "concatinating the fields", Arrays.asList("first_name", "last_name"), "name"));
    stageOperations.put("n3", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldWriteOperation("write_op", "writing data to file", EndPoint.of("myns", "another_file"), Arrays.asList("offset", "name")));
    stageOperations.put("n4", fieldOperations);
    LineageOperationsProcessor processor = new LineageOperationsProcessor(connections, stageOperations, Collections.emptySet());
    Set<Operation> processedOperations = processor.process();
    ReadOperation read = new ReadOperation("n1.read", "some read", EndPoint.of("ns", "file1"), "offset", "body");
    TransformOperation parse = new TransformOperation("n2.parse", "parsing body", Collections.singletonList(InputField.of("n1.read", "body")), "first_name", "last_name");
    TransformOperation concat = new TransformOperation("n3.concat", "concatinating the fields", Arrays.asList(InputField.of("n2.parse", "first_name"), InputField.of("n2.parse", "last_name")), "name");
    WriteOperation write = new WriteOperation("n4.write_op", "writing data to file", EndPoint.of("myns", "another_file"), Arrays.asList(InputField.of("n1.read", "offset"), InputField.of("n3.concat", "name")));
    List<Operation> expectedOperations = new ArrayList<>();
    expectedOperations.add(parse);
    expectedOperations.add(concat);
    expectedOperations.add(read);
    expectedOperations.add(write);
    Assert.assertEquals(new FieldLineageInfo(expectedOperations), new FieldLineageInfo(processedOperations));
}

Example 17 with FieldOperation

use of io.cdap.cdap.etl.api.lineage.field.FieldOperation in project cdap by cdapio.

the class LineageOperationProcessorTest method testSimpleJoinOperation.

@Test
public void testSimpleJoinOperation() {
    Set<Connection> connections = new HashSet<>();
    connections.add(new Connection("n1", "n3"));
    connections.add(new Connection("n2", "n3"));
    connections.add(new Connection("n3", "n4"));
    EndPoint cEndPoint = EndPoint.of("default", "customer");
    EndPoint pEndPoint = EndPoint.of("default", "purchase");
    EndPoint cpEndPoint = EndPoint.of("default", "customer_purchase");
    // customer -> (id)------------
    // |
    // JOIN  ------->(id, customer_id)
    // |
    // purchase -> (customer_id)---
    Map<String, List<FieldOperation>> stageOperations = new HashMap<>();
    stageOperations.put("n1", Collections.singletonList(new FieldReadOperation("ReadCustomer", "read description", cEndPoint, "id")));
    stageOperations.put("n2", Collections.singletonList(new FieldReadOperation("ReadPurchase", "read description", pEndPoint, "customer_id")));
    stageOperations.put("n3", Collections.singletonList(new FieldTransformOperation("Join", "Join Operation", Arrays.asList("n1.id", "n2.customer_id"), Arrays.asList("id", "customer_id"))));
    stageOperations.put("n4", Collections.singletonList(new FieldWriteOperation("Write", "write description", cpEndPoint, "id", "customer_id")));
    LineageOperationsProcessor processor = new LineageOperationsProcessor(connections, stageOperations, Collections.singleton("n3"));
    Set<Operation> expectedOperations = new HashSet<>();
    expectedOperations.add(new ReadOperation("n1.ReadCustomer", "read description", cEndPoint, "id"));
    expectedOperations.add(new ReadOperation("n2.ReadPurchase", "read description", pEndPoint, "customer_id"));
    expectedOperations.add(new TransformOperation("n3.Join", "Join Operation", Arrays.asList(InputField.of("n1.ReadCustomer", "id"), InputField.of("n2.ReadPurchase", "customer_id")), "id", "customer_id"));
    expectedOperations.add(new WriteOperation("n4.Write", "write description", cpEndPoint, Arrays.asList(InputField.of("n3.Join", "id"), InputField.of("n3.Join", "customer_id"))));
    Assert.assertEquals(expectedOperations, processor.process());
}

Example 18 with FieldOperation

use of io.cdap.cdap.etl.api.lineage.field.FieldOperation in project cdap by cdapio.

the class LineageOperationProcessorTest method testSimpleJoinWithAdditionalFields.

@Test
public void testSimpleJoinWithAdditionalFields() {
    Set<Connection> connections = new HashSet<>();
    connections.add(new Connection("n1", "n3"));
    connections.add(new Connection("n2", "n3"));
    connections.add(new Connection("n3", "n4"));
    EndPoint cEndPoint = EndPoint.of("default", "customer");
    EndPoint pEndPoint = EndPoint.of("default", "purchase");
    EndPoint cpEndPoint = EndPoint.of("default", "customer_purchase");
    // customer -> (id)------------
    // |
    // JOIN  ------->(id, customer_id)
    // |
    // purchase -> (customer_id)---
    Map<String, List<FieldOperation>> stageOperations = new HashMap<>();
    stageOperations.put("n1", Collections.singletonList(new FieldReadOperation("ReadCustomer", "read description", cEndPoint, "id", "name")));
    stageOperations.put("n2", Collections.singletonList(new FieldReadOperation("ReadPurchase", "read description", pEndPoint, "customer_id", "item")));
    List<FieldOperation> operationsFromJoin = new ArrayList<>();
    operationsFromJoin.add(new FieldTransformOperation("Join", "Join Operation", Arrays.asList("n1.id", "n2.customer_id"), Arrays.asList("id", "customer_id")));
    operationsFromJoin.add(new FieldTransformOperation("Identity name", "Identity Operation", Collections.singletonList("n1.name"), Collections.singletonList("name")));
    operationsFromJoin.add(new FieldTransformOperation("Identity item", "Identity Operation", Collections.singletonList("n2.item"), Collections.singletonList("item")));
    stageOperations.put("n3", operationsFromJoin);
    stageOperations.put("n4", Collections.singletonList(new FieldWriteOperation("Write", "write description", cpEndPoint, "id", "name", "customer_id", "item")));
    LineageOperationsProcessor processor = new LineageOperationsProcessor(connections, stageOperations, Collections.singleton("n3"));
    Set<Operation> expectedOperations = new HashSet<>();
    expectedOperations.add(new ReadOperation("n1.ReadCustomer", "read description", cEndPoint, "id", "name"));
    expectedOperations.add(new ReadOperation("n2.ReadPurchase", "read description", pEndPoint, "customer_id", "item"));
    expectedOperations.add(new TransformOperation("n3.Join", "Join Operation", Arrays.asList(InputField.of("n1.ReadCustomer", "id"), InputField.of("n2.ReadPurchase", "customer_id")), "id", "customer_id"));
    expectedOperations.add(new TransformOperation("n3.Identity name", "Identity Operation", Collections.singletonList(InputField.of("n1.ReadCustomer", "name")), "name"));
    expectedOperations.add(new TransformOperation("n3.Identity item", "Identity Operation", Collections.singletonList(InputField.of("n2.ReadPurchase", "item")), "item"));
    expectedOperations.add(new WriteOperation("n4.Write", "write description", cpEndPoint, Arrays.asList(InputField.of("n3.Join", "id"), InputField.of("n3.Identity name", "name"), InputField.of("n3.Join", "customer_id"), InputField.of("n3.Identity item", "item"))));
    Set<Operation> processedOperations = processor.process();
    Assert.assertEquals(expectedOperations, processedOperations);
}

Example 19 with FieldOperation

use of io.cdap.cdap.etl.api.lineage.field.FieldOperation in project cdap by cdapio.

the class LineageOperationProcessorTest method testSimpleJoinWithRenameJoinKeys.

@Test
public void testSimpleJoinWithRenameJoinKeys() {
    Set<Connection> connections = new HashSet<>();
    connections.add(new Connection("n1", "n3"));
    connections.add(new Connection("n2", "n3"));
    connections.add(new Connection("n3", "n4"));
    EndPoint cEndPoint = EndPoint.of("default", "customer");
    EndPoint pEndPoint = EndPoint.of("default", "purchase");
    EndPoint cpEndPoint = EndPoint.of("default", "customer_purchase");
    // customer -> (id, name)------------
    // |
    // JOIN  ------->(id_from_customer, id_from_purchase, name, item)
    // |
    // purchase -> (customer_id, item)---
    Map<String, List<FieldOperation>> stageOperations = new HashMap<>();
    stageOperations.put("n1", Collections.singletonList(new FieldReadOperation("ReadCustomer", "read description", cEndPoint, "id", "name")));
    stageOperations.put("n2", Collections.singletonList(new FieldReadOperation("ReadPurchase", "read description", pEndPoint, "customer_id", "item")));
    List<FieldOperation> operationsFromJoin = new ArrayList<>();
    operationsFromJoin.add(new FieldTransformOperation("Join", "Join Operation", Arrays.asList("n1.id", "n2.customer_id"), Arrays.asList("id", "customer_id")));
    operationsFromJoin.add(new FieldTransformOperation("Rename id", "Rename id", Collections.singletonList("id"), "id_from_customer"));
    operationsFromJoin.add(new FieldTransformOperation("Rename customer_id", "Rename customer_id", Collections.singletonList("customer_id"), "id_from_purchase"));
    operationsFromJoin.add(new FieldTransformOperation("Identity name", "Identity Operation", Collections.singletonList("n1.name"), Collections.singletonList("name")));
    operationsFromJoin.add(new FieldTransformOperation("Identity item", "Identity Operation", Collections.singletonList("n2.item"), Collections.singletonList("item")));
    stageOperations.put("n3", operationsFromJoin);
    stageOperations.put("n4", Collections.singletonList(new FieldWriteOperation("Write", "write description", cpEndPoint, "id_from_customer", "id_from_purchase", "name", "item")));
    LineageOperationsProcessor processor = new LineageOperationsProcessor(connections, stageOperations, Collections.singleton("n3"));
    Set<Operation> processedOperations = processor.process();
    Set<Operation> expectedOperations = new HashSet<>();
    expectedOperations.add(new ReadOperation("n1.ReadCustomer", "read description", cEndPoint, "id", "name"));
    expectedOperations.add(new ReadOperation("n2.ReadPurchase", "read description", pEndPoint, "customer_id", "item"));
    expectedOperations.add(new TransformOperation("n3.Join", "Join Operation", Arrays.asList(InputField.of("n1.ReadCustomer", "id"), InputField.of("n2.ReadPurchase", "customer_id")), "id", "customer_id"));
    expectedOperations.add(new TransformOperation("n3.Rename id", "Rename id", Collections.singletonList(InputField.of("n3.Join", "id")), "id_from_customer"));
    expectedOperations.add(new TransformOperation("n3.Rename customer_id", "Rename customer_id", Collections.singletonList(InputField.of("n3.Join", "customer_id")), "id_from_purchase"));
    expectedOperations.add(new TransformOperation("n3.Identity name", "Identity Operation", Collections.singletonList(InputField.of("n1.ReadCustomer", "name")), "name"));
    expectedOperations.add(new TransformOperation("n3.Identity item", "Identity Operation", Collections.singletonList(InputField.of("n2.ReadPurchase", "item")), "item"));
    expectedOperations.add(new WriteOperation("n4.Write", "write description", cpEndPoint, Arrays.asList(InputField.of("n3.Rename id", "id_from_customer"), InputField.of("n3.Rename customer_id", "id_from_purchase"), InputField.of("n3.Identity name", "name"), InputField.of("n3.Identity item", "item"))));
    Assert.assertEquals(expectedOperations, processedOperations);
}

Example 20 with FieldOperation

use of io.cdap.cdap.etl.api.lineage.field.FieldOperation in project cdap by cdapio.

the class LineageOperationProcessorTest method testDirectMerge.

@Test
public void testDirectMerge() {
    // n1--------->n3
    // |
    // n2--------->n4
    // n1 => pRead: personFile -> (offset, body)
    // n2 => hRead: hrFile -> (offset, body)
    // n1.n2.merge => n1.n2.merge: (pRead.offset, pRead.body, hRead.offset, hRead.body) -> (offset, body)
    // n3 => write1: (n1.n2.merge.offset, n1.n2.merge.body) -> testStore
    // n4 => write1: (n1.n2.merge.offset, n1.n2.merge.body) -> prodStore
    Set<Connection> connections = new HashSet<>();
    connections.add(new Connection("n1", "n3"));
    connections.add(new Connection("n1", "n4"));
    connections.add(new Connection("n2", "n3"));
    connections.add(new Connection("n2", "n4"));
    EndPoint pEndPoint = EndPoint.of("ns", "personFile");
    EndPoint hEndPoint = EndPoint.of("ns", "hrFile");
    EndPoint testEndPoint = EndPoint.of("ns", "testStore");
    EndPoint prodEndPoint = EndPoint.of("ns", "prodStore");
    Map<String, List<FieldOperation>> stageOperations = new HashMap<>();
    List<FieldOperation> fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldReadOperation("pRead", "Reading from person file", pEndPoint, "offset", "body"));
    stageOperations.put("n1", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldReadOperation("hRead", "Reading from hr file", hEndPoint, "offset", "body"));
    stageOperations.put("n2", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldWriteOperation("write1", "Writing to test store", testEndPoint, "offset", "body"));
    stageOperations.put("n3", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldWriteOperation("write2", "Writing to prod store", prodEndPoint, "offset", "body"));
    stageOperations.put("n4", fieldOperations);
    LineageOperationsProcessor processor = new LineageOperationsProcessor(connections, stageOperations, Collections.emptySet());
    Set<Operation> processedOperations = processor.process();
    Set<Operation> expectedOperations = new HashSet<>();
    ReadOperation pRead = new ReadOperation("n1.pRead", "Reading from person file", pEndPoint, "offset", "body");
    expectedOperations.add(pRead);
    ReadOperation hRead = new ReadOperation("n2.hRead", "Reading from hr file", hEndPoint, "offset", "body");
    expectedOperations.add(hRead);
    // implicit merge should be added by app
    TransformOperation merge1 = new TransformOperation("n1,n2.merge.offset", "Merged stages: n1,n2", Arrays.asList(InputField.of("n1.pRead", "offset"), InputField.of("n2.hRead", "offset")), "offset");
    TransformOperation merge2 = new TransformOperation("n1,n2.merge.body", "Merged stages: n1,n2", Arrays.asList(InputField.of("n1.pRead", "body"), InputField.of("n2.hRead", "body")), "body");
    expectedOperations.add(merge1);
    expectedOperations.add(merge2);
    WriteOperation write1 = new WriteOperation("n3.write1", "Writing to test store", testEndPoint, Arrays.asList(InputField.of("n1,n2.merge.offset", "offset"), InputField.of("n1,n2.merge.body", "body")));
    expectedOperations.add(write1);
    WriteOperation write2 = new WriteOperation("n4.write2", "Writing to prod store", prodEndPoint, Arrays.asList(InputField.of("n1,n2.merge.offset", "offset"), InputField.of("n1,n2.merge.body", "body")));
    expectedOperations.add(write2);
    Assert.assertEquals(new FieldLineageInfo(expectedOperations), new FieldLineageInfo(processedOperations));
}