Search in sources :

Example 31 with EndPoint

use of io.cdap.cdap.api.lineage.field.EndPoint in project cdap by caskdata.

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);
}
Also used : ReadOperation(io.cdap.cdap.api.lineage.field.ReadOperation) FieldReadOperation(io.cdap.cdap.etl.api.lineage.field.FieldReadOperation) HashMap(java.util.HashMap) Connection(io.cdap.cdap.etl.proto.Connection) ArrayList(java.util.ArrayList) EndPoint(io.cdap.cdap.api.lineage.field.EndPoint) ReadOperation(io.cdap.cdap.api.lineage.field.ReadOperation) FieldOperation(io.cdap.cdap.etl.api.lineage.field.FieldOperation) FieldTransformOperation(io.cdap.cdap.etl.api.lineage.field.FieldTransformOperation) WriteOperation(io.cdap.cdap.api.lineage.field.WriteOperation) FieldWriteOperation(io.cdap.cdap.etl.api.lineage.field.FieldWriteOperation) TransformOperation(io.cdap.cdap.api.lineage.field.TransformOperation) FieldReadOperation(io.cdap.cdap.etl.api.lineage.field.FieldReadOperation) Operation(io.cdap.cdap.api.lineage.field.Operation) FieldTransformOperation(io.cdap.cdap.etl.api.lineage.field.FieldTransformOperation) TransformOperation(io.cdap.cdap.api.lineage.field.TransformOperation) FieldWriteOperation(io.cdap.cdap.etl.api.lineage.field.FieldWriteOperation) FieldReadOperation(io.cdap.cdap.etl.api.lineage.field.FieldReadOperation) WriteOperation(io.cdap.cdap.api.lineage.field.WriteOperation) FieldWriteOperation(io.cdap.cdap.etl.api.lineage.field.FieldWriteOperation) FieldTransformOperation(io.cdap.cdap.etl.api.lineage.field.FieldTransformOperation) ArrayList(java.util.ArrayList) List(java.util.List) ImmutableList(com.google.common.collect.ImmutableList) FieldOperation(io.cdap.cdap.etl.api.lineage.field.FieldOperation) HashSet(java.util.HashSet) Test(org.junit.Test)

Example 32 with EndPoint

use of io.cdap.cdap.api.lineage.field.EndPoint in project cdap by caskdata.

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);
}
Also used : ReadOperation(io.cdap.cdap.api.lineage.field.ReadOperation) FieldReadOperation(io.cdap.cdap.etl.api.lineage.field.FieldReadOperation) HashMap(java.util.HashMap) Connection(io.cdap.cdap.etl.proto.Connection) ArrayList(java.util.ArrayList) EndPoint(io.cdap.cdap.api.lineage.field.EndPoint) ReadOperation(io.cdap.cdap.api.lineage.field.ReadOperation) FieldOperation(io.cdap.cdap.etl.api.lineage.field.FieldOperation) FieldTransformOperation(io.cdap.cdap.etl.api.lineage.field.FieldTransformOperation) WriteOperation(io.cdap.cdap.api.lineage.field.WriteOperation) FieldWriteOperation(io.cdap.cdap.etl.api.lineage.field.FieldWriteOperation) TransformOperation(io.cdap.cdap.api.lineage.field.TransformOperation) FieldReadOperation(io.cdap.cdap.etl.api.lineage.field.FieldReadOperation) Operation(io.cdap.cdap.api.lineage.field.Operation) FieldTransformOperation(io.cdap.cdap.etl.api.lineage.field.FieldTransformOperation) TransformOperation(io.cdap.cdap.api.lineage.field.TransformOperation) FieldWriteOperation(io.cdap.cdap.etl.api.lineage.field.FieldWriteOperation) FieldReadOperation(io.cdap.cdap.etl.api.lineage.field.FieldReadOperation) WriteOperation(io.cdap.cdap.api.lineage.field.WriteOperation) FieldWriteOperation(io.cdap.cdap.etl.api.lineage.field.FieldWriteOperation) FieldTransformOperation(io.cdap.cdap.etl.api.lineage.field.FieldTransformOperation) ArrayList(java.util.ArrayList) List(java.util.List) ImmutableList(com.google.common.collect.ImmutableList) FieldOperation(io.cdap.cdap.etl.api.lineage.field.FieldOperation) HashSet(java.util.HashSet) Test(org.junit.Test)

Example 33 with EndPoint

use of io.cdap.cdap.api.lineage.field.EndPoint in project cdap by caskdata.

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);
}
Also used : ReadOperation(io.cdap.cdap.api.lineage.field.ReadOperation) FieldReadOperation(io.cdap.cdap.etl.api.lineage.field.FieldReadOperation) HashMap(java.util.HashMap) Connection(io.cdap.cdap.etl.proto.Connection) ArrayList(java.util.ArrayList) EndPoint(io.cdap.cdap.api.lineage.field.EndPoint) ReadOperation(io.cdap.cdap.api.lineage.field.ReadOperation) FieldOperation(io.cdap.cdap.etl.api.lineage.field.FieldOperation) FieldTransformOperation(io.cdap.cdap.etl.api.lineage.field.FieldTransformOperation) WriteOperation(io.cdap.cdap.api.lineage.field.WriteOperation) FieldWriteOperation(io.cdap.cdap.etl.api.lineage.field.FieldWriteOperation) TransformOperation(io.cdap.cdap.api.lineage.field.TransformOperation) FieldReadOperation(io.cdap.cdap.etl.api.lineage.field.FieldReadOperation) Operation(io.cdap.cdap.api.lineage.field.Operation) FieldTransformOperation(io.cdap.cdap.etl.api.lineage.field.FieldTransformOperation) TransformOperation(io.cdap.cdap.api.lineage.field.TransformOperation) FieldWriteOperation(io.cdap.cdap.etl.api.lineage.field.FieldWriteOperation) FieldReadOperation(io.cdap.cdap.etl.api.lineage.field.FieldReadOperation) WriteOperation(io.cdap.cdap.api.lineage.field.WriteOperation) FieldWriteOperation(io.cdap.cdap.etl.api.lineage.field.FieldWriteOperation) FieldTransformOperation(io.cdap.cdap.etl.api.lineage.field.FieldTransformOperation) ArrayList(java.util.ArrayList) List(java.util.List) ImmutableList(com.google.common.collect.ImmutableList) FieldOperation(io.cdap.cdap.etl.api.lineage.field.FieldOperation) HashSet(java.util.HashSet) Test(org.junit.Test)

Aggregations

EndPoint (io.cdap.cdap.api.lineage.field.EndPoint)33 HashSet (java.util.HashSet)28 Test (org.junit.Test)26 ReadOperation (io.cdap.cdap.api.lineage.field.ReadOperation)24 TransformOperation (io.cdap.cdap.api.lineage.field.TransformOperation)24 WriteOperation (io.cdap.cdap.api.lineage.field.WriteOperation)24 Operation (io.cdap.cdap.api.lineage.field.Operation)23 ArrayList (java.util.ArrayList)19 HashMap (java.util.HashMap)14 List (java.util.List)11 ImmutableList (com.google.common.collect.ImmutableList)10 FieldOperation (io.cdap.cdap.etl.api.lineage.field.FieldOperation)10 FieldReadOperation (io.cdap.cdap.etl.api.lineage.field.FieldReadOperation)10 FieldTransformOperation (io.cdap.cdap.etl.api.lineage.field.FieldTransformOperation)10 FieldWriteOperation (io.cdap.cdap.etl.api.lineage.field.FieldWriteOperation)10 Connection (io.cdap.cdap.etl.proto.Connection)10 ImmutableSet (com.google.common.collect.ImmutableSet)9 Set (java.util.Set)9 EndPointField (io.cdap.cdap.data2.metadata.lineage.field.EndPointField)8 LinkedHashSet (java.util.LinkedHashSet)7