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

Example 16 with FieldReadOperation

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

the class LineageOperationProcessorTest method testComplexMerge.

@Test
public void testComplexMerge() {
    // 
    // n1----n2---
    // |         |-------n6
    // |----n5---|
    // n3----n4---         |---n7----n8
    // 
    // 
    // n1: read: file1 -> offset,body
    // n2: parse: body -> name, address, zip
    // n3: read: file2 -> offset,body
    // n4: parse: body -> name, address, zip
    // n5: normalize: address -> address
    // n5: rename: address -> state_address
    // n6: write: offset, name, address -> file3
    // n7: rename: offset -> file_offset
    // n8: write: file_offset, name, address, zip -> file4
    Set<Connection> connections = new HashSet<>();
    connections.add(new Connection("n1", "n2"));
    connections.add(new Connection("n2", "n5"));
    connections.add(new Connection("n3", "n4"));
    connections.add(new Connection("n4", "n5"));
    connections.add(new Connection("n5", "n6"));
    connections.add(new Connection("n5", "n7"));
    connections.add(new Connection("n7", "n8"));
    EndPoint n1EndPoint = EndPoint.of("ns", "file1");
    EndPoint n3EndPoint = EndPoint.of("ns", "file2");
    EndPoint n6EndPoint = EndPoint.of("ns", "file3");
    EndPoint n8EndPoint = EndPoint.of("ns", "file4");
    Map<String, List<FieldOperation>> stageOperations = new HashMap<>();
    List<FieldOperation> fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldReadOperation("read", "reading file 1", n1EndPoint, "offset", "body"));
    stageOperations.put("n1", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldTransformOperation("parse", "parsing file 1", Collections.singletonList("body"), "name", "address", "zip"));
    stageOperations.put("n2", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldReadOperation("read", "reading file 2", n3EndPoint, "offset", "body"));
    stageOperations.put("n3", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldTransformOperation("parse", "parsing file 2", Collections.singletonList("body"), "name", "address", "zip"));
    stageOperations.put("n4", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldTransformOperation("normalize", "normalizing address", Collections.singletonList("address"), "address"));
    fieldOperations.add(new FieldTransformOperation("rename", "renaming address to state_address", Collections.singletonList("address"), "state_address"));
    stageOperations.put("n5", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldWriteOperation("write", "writing file 3", n6EndPoint, "offset", "name", "address"));
    stageOperations.put("n6", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldTransformOperation("rename", "renaming offset to file_offset", Collections.singletonList("offset"), "file_offset"));
    stageOperations.put("n7", fieldOperations);
    fieldOperations = new ArrayList<>();
    fieldOperations.add(new FieldWriteOperation("write", "writing file 4", n8EndPoint, "file_offset", "name", "address", "zip"));
    stageOperations.put("n8", fieldOperations);
    LineageOperationsProcessor processor = new LineageOperationsProcessor(connections, stageOperations, Collections.emptySet());
    Set<Operation> processedOperations = processor.process();
    Set<Operation> expectedOperations = new HashSet<>();
    ReadOperation read = new ReadOperation("n1.read", "reading file 1", n1EndPoint, "offset", "body");
    expectedOperations.add(read);
    TransformOperation parse = new TransformOperation("n2.parse", "parsing file 1", Collections.singletonList(InputField.of("n1.read", "body")), "name", "address", "zip");
    expectedOperations.add(parse);
    read = new ReadOperation("n3.read", "reading file 2", n3EndPoint, "offset", "body");
    expectedOperations.add(read);
    parse = new TransformOperation("n4.parse", "parsing file 2", Collections.singletonList(InputField.of("n3.read", "body")), "name", "address", "zip");
    expectedOperations.add(parse);
    TransformOperation merge1 = new TransformOperation("n2,n4.merge.offset", "Merged stages: n2,n4", ImmutableList.of(InputField.of("n1.read", "offset"), InputField.of("n3.read", "offset")), "offset");
    TransformOperation merge2 = new TransformOperation("n2,n4.merge.body", "Merged stages: n2,n4", ImmutableList.of(InputField.of("n1.read", "body"), InputField.of("n3.read", "body")), "body");
    TransformOperation merge3 = new TransformOperation("n2,n4.merge.address", "Merged stages: n2,n4", ImmutableList.of(InputField.of("n2.parse", "address"), InputField.of("n4.parse", "address")), "address");
    TransformOperation merge4 = new TransformOperation("n2,n4.merge.name", "Merged stages: n2,n4", ImmutableList.of(InputField.of("n2.parse", "name"), InputField.of("n4.parse", "name")), "name");
    TransformOperation merge5 = new TransformOperation("n2,n4.merge.zip", "Merged stages: n2,n4", ImmutableList.of(InputField.of("n2.parse", "zip"), InputField.of("n4.parse", "zip")), "zip");
    expectedOperations.add(merge1);
    expectedOperations.add(merge2);
    expectedOperations.add(merge3);
    expectedOperations.add(merge4);
    expectedOperations.add(merge5);
    TransformOperation normalize = new TransformOperation("n5.normalize", "normalizing address", Collections.singletonList(InputField.of("n2,n4.merge.address", "address")), "address");
    expectedOperations.add(normalize);
    TransformOperation rename = new TransformOperation("n5.rename", "renaming address to state_address", Collections.singletonList(InputField.of("n5.normalize", "address")), "state_address");
    expectedOperations.add(rename);
    WriteOperation write = new WriteOperation("n6.write", "writing file 3", n6EndPoint, InputField.of("n2,n4.merge.offset", "offset"), InputField.of("n2,n4.merge.name", "name"), InputField.of("n5.normalize", "address"));
    expectedOperations.add(write);
    rename = new TransformOperation("n7.rename", "renaming offset to file_offset", Collections.singletonList(InputField.of("n2,n4.merge.offset", "offset")), "file_offset");
    expectedOperations.add(rename);
    write = new WriteOperation("n8.write", "writing file 4", n8EndPoint, InputField.of("n7.rename", "file_offset"), InputField.of("n2,n4.merge.name", "name"), InputField.of("n5.normalize", "address"), InputField.of("n2,n4.merge.zip", "zip"));
    expectedOperations.add(write);
    Assert.assertEquals(expectedOperations, processedOperations);
}

Example 17 with FieldReadOperation

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

the class LineageOperationProcessorTest method testSimpleJoinWithRenameOnAdditionalFields.

@Test
public void testSimpleJoinWithRenameOnAdditionalFields() {
    // customer -> (id, name)----------
    // |
    // JOIN  --->(id_from_customer, customer_id, name_from_customer, item_from_purchase)
    // |
    // purchase ->(customer_id, item)---
    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");
    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 name", "Rename name", Collections.singletonList("n1.name"), "name_from_customer"));
    operationsFromJoin.add(new FieldTransformOperation("Rename item", "Rename item", Collections.singletonList("n2.item"), "item_from_purchase"));
    stageOperations.put("n3", operationsFromJoin);
    stageOperations.put("n4", Collections.singletonList(new FieldWriteOperation("Write", "write description", cpEndPoint, "id_from_customer", "customer_id", "name_from_customer", "item_from_purchase")));
    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 name", "Rename name", Collections.singletonList(InputField.of("n1.ReadCustomer", "name")), "name_from_customer"));
    expectedOperations.add(new TransformOperation("n3.Rename item", "Rename item", Collections.singletonList(InputField.of("n2.ReadPurchase", "item")), "item_from_purchase"));
    expectedOperations.add(new WriteOperation("n4.Write", "write description", cpEndPoint, Arrays.asList(InputField.of("n3.Rename id", "id_from_customer"), InputField.of("n3.Join", "customer_id"), InputField.of("n3.Rename name", "name_from_customer"), InputField.of("n3.Rename item", "item_from_purchase"))));
    Assert.assertEquals(expectedOperations, processedOperations);
}

Example 18 with FieldReadOperation

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

the class LineageOperationProcessorTest method testJoinWith3Inputs.

@Test
public void testJoinWith3Inputs() {
    // customer -> (id, name)---------- |
    // |
    // purchase ->(customer_id, item)------> JOIN --->(id_from_customer, customer_id, address_id,
    // |                   name_from_customer, address)
    // |
    // address ->(address_id, address)--|
    Set<Connection> connections = new HashSet<>();
    connections.add(new Connection("n1", "n4"));
    connections.add(new Connection("n2", "n4"));
    connections.add(new Connection("n3", "n4"));
    connections.add(new Connection("n4", "n5"));
    EndPoint cEndPoint = EndPoint.of("default", "customer");
    EndPoint pEndPoint = EndPoint.of("default", "purchase");
    EndPoint aEndPoint = EndPoint.of("default", "address");
    EndPoint acpEndPoint = EndPoint.of("default", "customer_purchase_address");
    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")));
    stageOperations.put("n3", Collections.singletonList(new FieldReadOperation("ReadAddress", "read description", aEndPoint, "address_id", "address")));
    List<FieldOperation> operationsFromJoin = new ArrayList<>();
    operationsFromJoin.add(new FieldTransformOperation("Join", "Join Operation", Arrays.asList("n1.id", "n2.customer_id", "n3.address_id"), Arrays.asList("id", "customer_id", "address_id")));
    operationsFromJoin.add(new FieldTransformOperation("Rename id", "Rename Operation", Collections.singletonList("id"), Collections.singletonList("id_from_customer")));
    operationsFromJoin.add(new FieldTransformOperation("Rename customer.name", "Rename Operation", Collections.singletonList("n1.name"), Collections.singletonList("name_from_customer")));
    operationsFromJoin.add(new FieldTransformOperation("Identity address.address", "Identity Operation", Collections.singletonList("n3.address"), Collections.singletonList("address")));
    stageOperations.put("n4", operationsFromJoin);
    stageOperations.put("n5", Collections.singletonList(new FieldWriteOperation("Write", "Write Operation", acpEndPoint, "id_from_customer", "customer_id", "address_id", "name_from_customer", "address")));
    LineageOperationsProcessor processor = new LineageOperationsProcessor(connections, stageOperations, Collections.singleton("n4"));
    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 ReadOperation("n3.ReadAddress", "read description", aEndPoint, "address_id", "address"));
    expectedOperations.add(new TransformOperation("n4.Join", "Join Operation", Arrays.asList(InputField.of("n1.ReadCustomer", "id"), InputField.of("n2.ReadPurchase", "customer_id"), InputField.of("n3.ReadAddress", "address_id")), "id", "customer_id", "address_id"));
    expectedOperations.add(new TransformOperation("n4.Rename id", "Rename Operation", Collections.singletonList(InputField.of("n4.Join", "id")), "id_from_customer"));
    expectedOperations.add(new TransformOperation("n4.Rename customer.name", "Rename Operation", Collections.singletonList(InputField.of("n1.ReadCustomer", "name")), "name_from_customer"));
    expectedOperations.add(new TransformOperation("n4.Identity address.address", "Identity Operation", Collections.singletonList(InputField.of("n3.ReadAddress", "address")), "address"));
    expectedOperations.add(new WriteOperation("n5.Write", "Write Operation", acpEndPoint, Arrays.asList(InputField.of("n4.Rename id", "id_from_customer"), InputField.of("n4.Join", "customer_id"), InputField.of("n4.Join", "address_id"), InputField.of("n4.Rename customer.name", "name_from_customer"), InputField.of("n4.Identity address.address", "address"))));
    Assert.assertEquals(expectedOperations, processedOperations);
}

Example 19 with FieldReadOperation

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

the class StageOperationsValidatorTest method testInvalidInputs.

@Test
public void testInvalidInputs() {
    FieldReadOperation read = new FieldReadOperation("read", "reading data", EndPoint.of("default", "file"), "offset", "body");
    StageOperationsValidator.Builder builder = new StageOperationsValidator.Builder(Collections.singletonList(read));
    builder.addStageOutputs(Arrays.asList("offset", "body"));
    StageOperationsValidator stageOperationsValidator = builder.build();
    stageOperationsValidator.validate();
    Assert.assertNull(stageOperationsValidator.getStageInvalids());
    FieldTransformOperation transform = new FieldTransformOperation("parse", "parsing data", Collections.singletonList("body"), Arrays.asList("name", "address", "zip"));
    builder = new StageOperationsValidator.Builder(Collections.singletonList(transform));
    builder.addStageInputs(Arrays.asList("offset", "body"));
    builder.addStageOutputs(Arrays.asList("name", "address", "zip"));
    stageOperationsValidator = builder.build();
    stageOperationsValidator.validate();
    Assert.assertNull(stageOperationsValidator.getStageInvalids());
    FieldWriteOperation write = new FieldWriteOperation("write", "writing data", EndPoint.of("default", "file2"), "name", "address", "zip");
    builder = new StageOperationsValidator.Builder(Collections.singletonList(write));
    builder.addStageInputs(Arrays.asList("address", "zip"));
    stageOperationsValidator = builder.build();
    stageOperationsValidator.validate();
    Assert.assertNotNull(stageOperationsValidator.getStageInvalids());
    InvalidFieldOperations invalidFieldOperations = stageOperationsValidator.getStageInvalids();
    Assert.assertEquals(1, invalidFieldOperations.getInvalidInputs().size());
    Map<String, List<String>> invalidInputs = new HashMap<>();
    invalidInputs.put("name", Collections.singletonList("write"));
    Assert.assertEquals(invalidInputs, invalidFieldOperations.getInvalidInputs());
    Assert.assertEquals(0, invalidFieldOperations.getInvalidOutputs().size());
    // name is provided by output of the operation previous to write
    List<FieldOperation> pipelineOperations = new ArrayList<>();
    pipelineOperations.add(new FieldTransformOperation("name_lookup", "generating name", Collections.singletonList("address"), "name"));
    pipelineOperations.add(new FieldWriteOperation("write", "writing data", EndPoint.of("default", "file2"), "name", "address", "zip"));
    builder = new StageOperationsValidator.Builder(pipelineOperations);
    builder.addStageInputs(Arrays.asList("address", "zip"));
    stageOperationsValidator = builder.build();
    stageOperationsValidator.validate();
    Assert.assertNull(stageOperationsValidator.getStageInvalids());
}

Example 20 with FieldReadOperation

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

the class StageOperationsValidatorTest method testUnusedValidOutputs.

@Test
public void testUnusedValidOutputs() {
    // the read operation has [offset, body] as output fields, the output schema only has [body], it
    // should still be valid
    FieldOperation read = new FieldReadOperation("read", "reading data", EndPoint.of("default", "file"), "offset", "body");
    StageOperationsValidator.Builder builder = new StageOperationsValidator.Builder(Collections.singletonList(read));
    builder.addStageOutputs(Collections.singleton("body"));
    StageOperationsValidator stageOperationsValidator = builder.build();
    stageOperationsValidator.validate();
    Assert.assertNull(stageOperationsValidator.getStageInvalids());
    // the transform has [body] -> [name, address, zip], the output schema has [name, address], it is still valid
    // since the field is generated by the input schema fields
    FieldOperation transform = new FieldTransformOperation("parse", "parsing data", Collections.singletonList("body"), Arrays.asList("name", "address", "zip"));
    builder = new StageOperationsValidator.Builder(Collections.singletonList(transform));
    builder.addStageInputs(Collections.singleton("body"));
    builder.addStageOutputs(Arrays.asList("name", "address"));
    stageOperationsValidator = builder.build();
    stageOperationsValidator.validate();
    Assert.assertNull(stageOperationsValidator.getStageInvalids());
    // simulate joiner, the operations are [d1.k1, d2.k2] -> [k1, k2], [d1.val1] -> [val1], [d2.val2] -> [val2],
    // the input schema are [d1.k1, d2.k2, d1.val1, d1.val2], output schemas are [val1, val2], this is still valid
    // since both d1.k1, d2.k2 are in the input schema though k1, k2 are not in output schema.
    FieldTransformOperation transform1 = new FieldTransformOperation("transform1", "description1", ImmutableList.of("d1.k1", "d2.k2"), ImmutableList.of("k1", "k2"));
    FieldTransformOperation transform2 = new FieldTransformOperation("transform2", "description2", ImmutableList.of("d1.val1"), ImmutableList.of("val1"));
    FieldTransformOperation transform3 = new FieldTransformOperation("transform3", "description3", ImmutableList.of("d2.val2"), ImmutableList.of("val2"));
    builder = new StageOperationsValidator.Builder(ImmutableList.of(transform1, transform2, transform3));
    builder.addStageInputs(ImmutableList.of("d1.k1", "d2.k2", "d1.val1", "d2.val2"));
    builder.addStageOutputs(ImmutableList.of("val1", "val2"));
    StageOperationsValidator validator = builder.build();
    validator.validate();
    Assert.assertNull(validator.getStageInvalids());
}