Merge branch 'master' into dal

addons/hive-bridge/src/main/java/org/apache/hadoop/metadata/hive/bridge/HiveMetaStoreBridge.java

@@ -26,6 +26,7 @@ import org.apache.hadoop.hive.metastore.api.Index;
 import org.apache.hadoop.hive.metastore.api.Order;
 import org.apache.hadoop.hive.metastore.api.SerDeInfo;
 import org.apache.hadoop.hive.metastore.api.StorageDescriptor;
+import org.apache.hadoop.hive.metastore.api.hive_metastoreConstants;
 import org.apache.hadoop.hive.ql.metadata.Hive;
 import org.apache.hadoop.hive.ql.metadata.Partition;
 import org.apache.hadoop.hive.ql.metadata.Table;
@@ -263,17 +264,22 @@ public class HiveMetaStoreBridge {
             tableRef = new Referenceable(HiveDataTypes.HIVE_TABLE.getName());
             tableRef.set("name", hiveTable.getTableName());
             tableRef.set("owner", hiveTable.getOwner());
-            //todo fix
-            tableRef.set("createTime", hiveTable.getLastAccessTime());
+
+            tableRef.set("createTime", hiveTable.getMetadata().getProperty(hive_metastoreConstants.DDL_TIME));
             tableRef.set("lastAccessTime", hiveTable.getLastAccessTime());
             tableRef.set("retention", hiveTable.getRetention());
 
+            tableRef.set(HiveDataModelGenerator.COMMENT, hiveTable.getParameters().get(HiveDataModelGenerator.COMMENT));
+
             // add reference to the database
             tableRef.set("dbName", dbReference);
 
+            List<Referenceable> colList = getColumns(hiveTable.getCols());
+            tableRef.set("columns", colList);
+
             // add reference to the StorageDescriptor
             StorageDescriptor storageDesc = hiveTable.getSd();
-            Referenceable sdReferenceable = fillStorageDescStruct(storageDesc);
+            Referenceable sdReferenceable = fillStorageDescStruct(storageDesc, colList);
             tableRef.set("sd", sdReferenceable);
 
             // add reference to the Partition Keys
@@ -293,8 +299,6 @@ public class HiveMetaStoreBridge {
             tableRef.set("tableType", hiveTable.getTableType().name());
             tableRef.set("temporary", hiveTable.isTemporary());
 
-            List<Referenceable> colList = getColumns(hiveTable.getAllCols());
-            tableRef.set("columns", colList);
 
             tableRef = createInstance(tableRef);
         } else {
@@ -388,7 +392,7 @@ public class HiveMetaStoreBridge {
         indexRef.set("origTableName", index.getOrigTableName());
         indexRef.set("indexTableName", index.getIndexTableName());
 
-        Referenceable sdReferenceable = fillStorageDescStruct(index.getSd());
+        Referenceable sdReferenceable = fillStorageDescStruct(index.getSd(), null);
         indexRef.set("sd", sdReferenceable);
 
         indexRef.set("parameters", index.getParameters());
@@ -398,7 +402,7 @@ public class HiveMetaStoreBridge {
         createInstance(indexRef);
     }
 
-    private Referenceable fillStorageDescStruct(StorageDescriptor storageDesc) throws Exception {
+    private Referenceable fillStorageDescStruct(StorageDescriptor storageDesc, List<Referenceable> colList) throws Exception {
         LOG.debug("Filling storage descriptor information for " + storageDesc);
 
         Referenceable sdReferenceable = new Referenceable(HiveDataTypes.HIVE_STORAGEDESC.getName());
@@ -415,6 +419,8 @@ public class HiveMetaStoreBridge {
         serdeInfoStruct.set("parameters", serdeInfo.getParameters());
 
         sdReferenceable.set("serdeInfo", serdeInfoStruct);
+        sdReferenceable.set(HiveDataModelGenerator.STORAGE_NUM_BUCKETS, storageDesc.getNumBuckets());
+        sdReferenceable.set(HiveDataModelGenerator.STORAGE_IS_STORED_AS_SUB_DIRS, storageDesc.isStoredAsSubDirectories());
 
         // Will need to revisit this after we fix typesystem.
         /*
@@ -433,8 +439,15 @@ public class HiveMetaStoreBridge {
         }
         */
 
-        List<Referenceable> fieldsList = getColumns(storageDesc.getCols());
-        sdReferenceable.set("cols", fieldsList);
+        //Use the passed column list if not null, ex: use same references for table and SD
+        List<FieldSchema> columns = storageDesc.getCols();
+        if (columns != null && !columns.isEmpty()) {
+            if (colList != null) {
+                sdReferenceable.set("cols", colList);
+            } else {
+                sdReferenceable.set("cols", getColumns(columns));
+            }
+        }
 
         List<Struct> sortColsStruct = new ArrayList<>();
         for (Order sortcol : storageDesc.getSortCols()) {
@@ -472,7 +485,7 @@ public class HiveMetaStoreBridge {
             Referenceable colReferenceable = new Referenceable(HiveDataTypes.HIVE_COLUMN.getName());
             colReferenceable.set("name", fs.getName());
             colReferenceable.set("type", fs.getType());
-            colReferenceable.set("comment", fs.getComment());
+            colReferenceable.set(HiveDataModelGenerator.COMMENT, fs.getComment());
 
             colList.add(createInstance(colReferenceable));
         }

addons/hive-bridge/src/main/java/org/apache/hadoop/metadata/hive/model/HiveDataModelGenerator.java

@@ -55,6 +55,10 @@ public class HiveDataModelGenerator {
     private final Map<String, EnumTypeDefinition> enumTypeDefinitionMap;
     private final Map<String, StructTypeDefinition> structTypeDefinitionMap;
 
+    public static final String COMMENT = "comment";
+    public static final String STORAGE_NUM_BUCKETS = "numBuckets";
+    public static final String STORAGE_IS_STORED_AS_SUB_DIRS = "storedAsSubDirectories";
+
     public HiveDataModelGenerator() {
         classTypeDefinitions = new HashMap<>();
         enumTypeDefinitionMap = new HashMap<>();
@@ -237,7 +241,7 @@ public class HiveDataModelGenerator {
                         Multiplicity.OPTIONAL, false, null),
                 new AttributeDefinition("compressed", DataTypes.BOOLEAN_TYPE.getName(),
                         Multiplicity.REQUIRED, false, null),
-                new AttributeDefinition("numBuckets", DataTypes.INT_TYPE.getName(),
+                new AttributeDefinition(STORAGE_NUM_BUCKETS, DataTypes.INT_TYPE.getName(),
                         Multiplicity.OPTIONAL, false, null),
                 new AttributeDefinition("serdeInfo", HiveDataTypes.HIVE_SERDE.getName(),
                         Multiplicity.OPTIONAL, false, null),
@@ -251,7 +255,7 @@ public class HiveDataModelGenerator {
                         Multiplicity.OPTIONAL, false, null),
                 //new AttributeDefinition("skewedInfo", DefinedTypes.HIVE_SKEWEDINFO.getName(),
                 // Multiplicity.OPTIONAL, false, null),
-                new AttributeDefinition("storedAsSubDirectories", DataTypes.BOOLEAN_TYPE.getName(),
+                new AttributeDefinition(STORAGE_IS_STORED_AS_SUB_DIRS, DataTypes.BOOLEAN_TYPE.getName(),
                         Multiplicity.OPTIONAL, false, null),
         };
 
@@ -326,7 +330,7 @@ public class HiveDataModelGenerator {
                         Multiplicity.REQUIRED, false, null),
                 new AttributeDefinition("type", DataTypes.STRING_TYPE.getName(),
                         Multiplicity.REQUIRED, false, null),
-                new AttributeDefinition("comment", DataTypes.STRING_TYPE.getName(),
+                new AttributeDefinition(COMMENT, DataTypes.STRING_TYPE.getName(),
                         Multiplicity.OPTIONAL, false, null),
         };
         HierarchicalTypeDefinition<ClassType> definition =
@@ -377,6 +381,8 @@ public class HiveDataModelGenerator {
                         Multiplicity.OPTIONAL, false, null),
                 new AttributeDefinition("lastAccessTime", DataTypes.INT_TYPE.getName(),
                         Multiplicity.OPTIONAL, false, null),
+                new AttributeDefinition(COMMENT, DataTypes.STRING_TYPE.getName(),
+                        Multiplicity.OPTIONAL, false, null),
                 new AttributeDefinition("retention", DataTypes.INT_TYPE.getName(),
                         Multiplicity.OPTIONAL, false, null),
                 new AttributeDefinition("sd", HiveDataTypes.HIVE_STORAGEDESC.getName(),

addons/hive-bridge/src/test/java/org/apache/hadoop/metadata/hive/hook/HiveHookIT.java

@@ -21,12 +21,15 @@ package org.apache.hadoop.metadata.hive.hook;
 import org.apache.commons.lang.RandomStringUtils;
 import org.apache.hadoop.hive.conf.HiveConf;
 import org.apache.hadoop.hive.metastore.TableType;
+import org.apache.hadoop.hive.metastore.api.StorageDescriptor;
 import org.apache.hadoop.hive.ql.Driver;
 import org.apache.hadoop.hive.ql.session.SessionState;
 import org.apache.hadoop.metadata.MetadataServiceClient;
 import org.apache.hadoop.metadata.hive.bridge.HiveMetaStoreBridge;
+import org.apache.hadoop.metadata.hive.model.HiveDataModelGenerator;
 import org.apache.hadoop.metadata.hive.model.HiveDataTypes;
 import org.apache.hadoop.metadata.typesystem.Referenceable;
+import org.apache.hadoop.metadata.typesystem.persistence.Id;
 import org.apache.log4j.spi.LoggerFactory;
 import org.codehaus.jettison.json.JSONArray;
 import org.codehaus.jettison.json.JSONObject;
@@ -118,7 +121,7 @@ public class HiveHookIT {
 
     private String createTable(boolean partition) throws Exception {
         String tableName = tableName();
-        runCommand("create table " + tableName + "(id int, name string)" + (partition ? " partitioned by(dt string)"
+        runCommand("create table " + tableName + "(id int, name string) comment 'table comment' " + (partition ? " partitioned by(dt string)"
                 : ""));
         return tableName;
     }
@@ -127,18 +130,32 @@ public class HiveHookIT {
     public void testCreateTable() throws Exception {
         String tableName = tableName();
         String dbName = createDatabase();
-        runCommand("create table " + dbName + "." + tableName + "(id int, name string)");
+        String colName = "col" + random();
+        runCommand("create table " + dbName + "." + tableName + "(" + colName + " int, name string)");
         assertTableIsRegistered(dbName, tableName);
+        //there is only one instance of column registered
+        assertColumnIsRegistered(colName);
 
         tableName = createTable();
         String tableId = assertTableIsRegistered(DEFAULT_DB, tableName);
         Referenceable tableRef = dgiCLient.getEntity(tableId);
         Assert.assertEquals(tableRef.get("tableType"), TableType.MANAGED_TABLE.name());
+        Assert.assertEquals(tableRef.get(HiveDataModelGenerator.COMMENT), "table comment");
+        final Id sdId = (Id) tableRef.get("sd");
+        Referenceable sdRef = dgiCLient.getEntity(sdId.id);
+        Assert.assertEquals(sdRef.get(HiveDataModelGenerator.STORAGE_IS_STORED_AS_SUB_DIRS),false);
 
         //Create table where database doesn't exist, will create database instance as well
         assertDatabaseIsRegistered(DEFAULT_DB);
     }
 
+    private String assertColumnIsRegistered(String colName) throws Exception {
+        LOG.debug("Searching for column {}", colName);
+        String query = String.format("%s where name = '%s'", HiveDataTypes.HIVE_COLUMN.getName(), colName.toLowerCase());
+        return assertEntityIsRegistered(query, true);
+
+    }
+
     @Test
     public void testCTAS() throws Exception {
         String tableName = createTable();

repository/src/main/java/org/apache/hadoop/metadata/RepositoryMetadataModule.java

@@ -48,28 +48,28 @@ public class RepositoryMetadataModule extends com.google.inject.AbstractModule {
         ThrowingProviderBinder.create(binder())
                 .bind(GraphProvider.class, TitanGraph.class)
                 .to(TitanGraphProvider.class)
-                .in(Scopes.SINGLETON);
+                .asEagerSingleton();
 
         // allow for dynamic binding of the metadata repo & graph service
 
         // bind the MetadataRepositoryService interface to an implementation
-        bind(MetadataRepository.class).to(GraphBackedMetadataRepository.class);
+        bind(MetadataRepository.class).to(GraphBackedMetadataRepository.class).asEagerSingleton();
 
         // bind the ITypeStore interface to an implementation
-        bind(ITypeStore.class).to(GraphBackedTypeStore.class);
+        bind(ITypeStore.class).to(GraphBackedTypeStore.class).asEagerSingleton();
 
         // bind the GraphService interface to an implementation
         // bind(GraphService.class).to(graphServiceClass);
 
         // bind the MetadataService interface to an implementation
-        bind(MetadataService.class).to(DefaultMetadataService.class);
+        bind(MetadataService.class).to(DefaultMetadataService.class).asEagerSingleton();
 
         // bind the DiscoveryService interface to an implementation
-        bind(DiscoveryService.class).to(GraphBackedDiscoveryService.class);
+        bind(DiscoveryService.class).to(GraphBackedDiscoveryService.class).asEagerSingleton();
 
-        bind(SearchIndexer.class).to(GraphBackedSearchIndexer.class);
+        bind(SearchIndexer.class).to(GraphBackedSearchIndexer.class).asEagerSingleton();
 
-        bind(LineageService.class).to(HiveLineageService.class);
+        bind(LineageService.class).to(HiveLineageService.class).asEagerSingleton();
 
         MethodInterceptor interceptor = new GraphTransactionInterceptor();
         requestInjection(interceptor);

repository/src/main/java/org/apache/hadoop/metadata/services/DefaultMetadataService.java

@@ -89,10 +89,11 @@ public class DefaultMetadataService implements MetadataService {
     private void restoreTypeSystem() {
         LOG.info("Restoring type system from the store");
         try {
+            createSuperTypes();
+
             TypesDef typesDef = typeStore.restore();
             typeSystem.defineTypes(typesDef);
 
-            createSuperTypes();
         } catch (MetadataException e) {
             throw new RuntimeException(e);
         }

repository/src/test/java/org/apache/hadoop/metadata/discovery/HiveLineageServiceTest.java

@@ -57,11 +57,6 @@ import java.util.List;
 @Guice(modules = RepositoryMetadataModule.class)
 public class HiveLineageServiceTest {
 
-    static {
-        // this would override super types creation if not first thing
-        TypeSystem.getInstance().reset();
-    }
-
     @Inject
     private DefaultMetadataService metadataService;