[#2541] feat(spark-connector): support basic DDL and DML operations to iceberg catalog

catalogs/catalog-lakehouse-iceberg/src/main/java/com/datastrato/gravitino/catalog/lakehouse/iceberg/IcebergCatalogOperations.java

@@ -314,6 +314,11 @@ public boolean dropSchema(NameIdentifier ident, boolean cascade) throws NonEmpty
    */
   @Override
   public NameIdentifier[] listTables(Namespace namespace) throws NoSuchSchemaException {
+    NameIdentifier schemaIdent = NameIdentifier.of(namespace.levels());
+    if (!schemaExists(schemaIdent)) {
+      throw new NoSuchSchemaException("Schema (database) does not exist %s", namespace);
+    }
+
     try {
       ListTablesResponse listTablesResponse =
           icebergTableOps.listTable(IcebergTableOpsHelper.getIcebergNamespace(namespace));

integration-test/build.gradle.kts

@@ -30,7 +30,6 @@ dependencies {
   testImplementation(project(":server"))
   testImplementation(project(":server-common"))
   testImplementation(project(":spark-connector")) {
-    exclude("org.apache.iceberg")
     exclude("org.apache.hadoop", "hadoop-client-api")
     exclude("org.apache.hadoop", "hadoop-client-runtime")
   }

integration-test/src/test/java/com/datastrato/gravitino/integration/test/spark/SparkCommonIT.java

@@ -22,12 +22,12 @@
 import org.apache.spark.sql.connector.catalog.TableCatalog;
 import org.apache.spark.sql.types.DataType;
 import org.apache.spark.sql.types.DataTypes;
+import org.junit.jupiter.api.AfterAll;
 import org.junit.jupiter.api.Assertions;
 import org.junit.jupiter.api.BeforeAll;
 import org.junit.jupiter.api.BeforeEach;
 import org.junit.jupiter.api.Test;
 import org.junit.jupiter.api.condition.EnabledIf;
-import org.junit.platform.commons.util.StringUtils;
 
 public abstract class SparkCommonIT extends SparkEnvIT {
 
@@ -61,6 +61,8 @@ private static String getInsertWithPartitionSql(
   // Whether supports [CLUSTERED BY col_name3 SORTED BY col_name INTO num_buckets BUCKETS]
   protected abstract boolean supportsSparkSQLClusteredBy();
 
+  protected abstract boolean supportsPartition();
+
   // Use a custom database not the original default database because SparkIT couldn't read&write
   // data to tables in default database. The main reason is default database location is
   // determined by `hive.metastore.warehouse.dir` in hive-site.xml which is local HDFS address
@@ -79,6 +81,13 @@ void init() {
     sql("USE " + getDefaultDatabase());
   }
 
+  @AfterAll
+  void cleanUp() {
+    sql("USE " + getCatalogName());
+    getDatabases()
+        .forEach(database -> sql(String.format("DROP DATABASE IF EXISTS %s CASCADE", database)));
+  }
+
   @Test
   void testLoadCatalogs() {
     Set<String> catalogs = getCatalogs();
@@ -89,20 +98,20 @@ void testLoadCatalogs() {
   void testCreateAndLoadSchema() {
     String testDatabaseName = "t_create1";
     dropDatabaseIfExists(testDatabaseName);
-    sql("CREATE DATABASE " + testDatabaseName);
+    sql("CREATE DATABASE " + testDatabaseName + " WITH DBPROPERTIES (ID=001);");
     Map<String, String> databaseMeta = getDatabaseMetadata(testDatabaseName);
     Assertions.assertFalse(databaseMeta.containsKey("Comment"));
     Assertions.assertTrue(databaseMeta.containsKey("Location"));
     Assertions.assertEquals("datastrato", databaseMeta.get("Owner"));
     String properties = databaseMeta.get("Properties");
-    Assertions.assertTrue(StringUtils.isBlank(properties));
+    Assertions.assertTrue(properties.contains("(ID,001)"));
 
     testDatabaseName = "t_create2";
     dropDatabaseIfExists(testDatabaseName);
     String testDatabaseLocation = "/tmp/" + testDatabaseName;
     sql(
         String.format(
-            "CREATE DATABASE %s COMMENT 'comment' LOCATION '%s'\n" + " WITH DBPROPERTIES (ID=001);",
+            "CREATE DATABASE %s COMMENT 'comment' LOCATION '%s'\n" + " WITH DBPROPERTIES (ID=002);",
             testDatabaseName, testDatabaseLocation));
     databaseMeta = getDatabaseMetadata(testDatabaseName);
     String comment = databaseMeta.get("Comment");
@@ -111,19 +120,22 @@ void testCreateAndLoadSchema() {
     // underlying catalog may change /tmp/t_create2 to file:/tmp/t_create2
     Assertions.assertTrue(databaseMeta.get("Location").contains(testDatabaseLocation));
     properties = databaseMeta.get("Properties");
-    Assertions.assertEquals("((ID,001))", properties);
+    Assertions.assertTrue(properties.contains("(ID,002)"));
   }
 
   @Test
   void testAlterSchema() {
     String testDatabaseName = "t_alter";
     dropDatabaseIfExists(testDatabaseName);
-    sql("CREATE DATABASE " + testDatabaseName);
+    sql("CREATE DATABASE " + testDatabaseName + " WITH DBPROPERTIES (ID=001);");
     Assertions.assertTrue(
-        StringUtils.isBlank(getDatabaseMetadata(testDatabaseName).get("Properties")));
+        getDatabaseMetadata(testDatabaseName).get("Properties").contains("(ID,001)"));
 
-    sql(String.format("ALTER DATABASE %s SET DBPROPERTIES ('ID'='001')", testDatabaseName));
-    Assertions.assertEquals("((ID,001))", getDatabaseMetadata(testDatabaseName).get("Properties"));
+    sql(String.format("ALTER DATABASE %s SET DBPROPERTIES ('ID'='002')", testDatabaseName));
+    Assertions.assertFalse(
+        getDatabaseMetadata(testDatabaseName).get("Properties").contains("(ID,001)"));
+    Assertions.assertTrue(
+        getDatabaseMetadata(testDatabaseName).get("Properties").contains("(ID,002)"));
 
     // Hive metastore doesn't support alter database location, therefore this test method
     // doesn't verify ALTER DATABASE database_name SET LOCATION 'new_location'.
@@ -334,9 +346,9 @@ void testAlterTableUpdateColumnType() {
     checkTableColumns(tableName, simpleTableColumns, getTableInfo(tableName));
 
     sql(String.format("ALTER TABLE %S ADD COLUMNS (col1 int)", tableName));
-    sql(String.format("ALTER TABLE %S CHANGE COLUMN col1 col1 string", tableName));
+    sql(String.format("ALTER TABLE %S CHANGE COLUMN col1 col1 bigint", tableName));
     ArrayList<SparkColumnInfo> updateColumns = new ArrayList<>(simpleTableColumns);
-    updateColumns.add(SparkColumnInfo.of("col1", DataTypes.StringType, null));
+    updateColumns.add(SparkColumnInfo.of("col1", DataTypes.LongType, null));
     checkTableColumns(tableName, updateColumns, getTableInfo(tableName));
   }
 
@@ -354,7 +366,7 @@ void testAlterTableRenameColumn() {
     sql(String.format("ALTER TABLE %S ADD COLUMNS (col1 int)", tableName));
     sql(
         String.format(
-            "ALTER TABLE %S RENAME COLUMN %S TO %S", tableName, oldColumnName, newColumnName));
+            "ALTER TABLE %s RENAME COLUMN %s TO %s", tableName, oldColumnName, newColumnName));
     ArrayList<SparkColumnInfo> renameColumns = new ArrayList<>(simpleTableColumns);
     renameColumns.add(SparkColumnInfo.of(newColumnName, DataTypes.IntegerType, null));
     checkTableColumns(tableName, renameColumns, getTableInfo(tableName));
@@ -373,7 +385,7 @@ void testUpdateColumnPosition() {
 
     sql(
         String.format(
-            "CREATE TABLE %s (id STRING COMMENT '', name STRING COMMENT '', age STRING COMMENT '') USING PARQUET",
+            "CREATE TABLE %s (id STRING COMMENT '', name STRING COMMENT '', age STRING COMMENT '')",
             tableName));
     checkTableColumns(tableName, simpleTableColumns, getTableInfo(tableName));
 
@@ -456,12 +468,13 @@ void testComplexType() {
   }
 
   @Test
+  @EnabledIf("supportsPartition")
   void testCreateDatasourceFormatPartitionTable() {
     String tableName = "datasource_partition_table";
 
     dropTableIfExists(tableName);
     String createTableSQL = getCreateSimpleTableString(tableName);
-    createTableSQL = createTableSQL + "USING PARQUET PARTITIONED BY (name, age)";
+    createTableSQL = createTableSQL + " USING PARQUET PARTITIONED BY (name, age)";
     sql(createTableSQL);
     SparkTableInfo tableInfo = getTableInfo(tableName);
     SparkTableInfoChecker checker =
@@ -558,6 +571,7 @@ void testInsertTableAsSelect() {
   }
 
   @Test
+  @EnabledIf("supportsPartition")
   void testInsertDatasourceFormatPartitionTableAsSelect() {
     String tableName = "insert_select_partition_table";
     String newTableName = "new_" + tableName;

integration-test/src/test/java/com/datastrato/gravitino/integration/test/spark/SparkEnvIT.java

@@ -12,6 +12,7 @@
 import com.datastrato.gravitino.integration.test.container.HiveContainer;
 import com.datastrato.gravitino.integration.test.util.spark.SparkUtilIT;
 import com.datastrato.gravitino.spark.connector.GravitinoSparkConfig;
+import com.datastrato.gravitino.spark.connector.iceberg.IcebergPropertiesConstants;
 import com.datastrato.gravitino.spark.connector.plugin.GravitinoSparkPlugin;
 import com.google.common.collect.Maps;
 import java.io.IOException;
@@ -37,6 +38,7 @@ public abstract class SparkEnvIT extends SparkUtilIT {
   private SparkSession sparkSession;
   private String hiveMetastoreUri = "thrift://127.0.0.1:9083";
   private String gravitinoUri = "http://127.0.0.1:8090";
+  private String warehouse;
 
   protected abstract String getCatalogName();
 
@@ -79,8 +81,18 @@ private void initMetalakeAndCatalogs() {
     client.createMetalake(NameIdentifier.of(metalakeName), "", Collections.emptyMap());
     GravitinoMetalake metalake = client.loadMetalake(NameIdentifier.of(metalakeName));
     Map<String, String> properties = Maps.newHashMap();
-    properties.put(GravitinoSparkConfig.GRAVITINO_HIVE_METASTORE_URI, hiveMetastoreUri);
-
+    switch (getProvider()) {
+      case "hive":
+        properties.put(GravitinoSparkConfig.GRAVITINO_HIVE_METASTORE_URI, hiveMetastoreUri);
+        break;
+      case "lakehouse-iceberg":
+        properties.put(IcebergPropertiesConstants.LAKEHOUSE_ICEBERG_CATALOG_BACKEND, "hive");
+        properties.put(IcebergPropertiesConstants.LAKEHOUSE_ICEBERG_CATALOG_WAREHOUSE, warehouse);
+        properties.put(IcebergPropertiesConstants.LAKEHOUSE_ICEBERG_CATALOG_URI, hiveMetastoreUri);
+        break;
+      default:
+        throw new IllegalArgumentException("Unsupported provider: " + getProvider());
+    }
     metalake.createCatalog(
         NameIdentifier.of(metalakeName, getCatalogName()),
         Catalog.Type.RELATIONAL,
@@ -102,6 +114,11 @@ private void initHiveEnv() {
             "thrift://%s:%d",
             containerSuite.getHiveContainer().getContainerIpAddress(),
             HiveContainer.HIVE_METASTORE_PORT);
+    warehouse =
+        String.format(
+            "hdfs://%s:%d/user/hive/warehouse",
+            containerSuite.getHiveContainer().getContainerIpAddress(),
+            HiveContainer.HDFS_DEFAULTFS_PORT);
   }
 
   private void initHdfsFileSystem() {
@@ -129,12 +146,7 @@ private void initSparkEnv() {
             .config(GravitinoSparkConfig.GRAVITINO_URI, gravitinoUri)
             .config(GravitinoSparkConfig.GRAVITINO_METALAKE, metalakeName)
             .config("hive.exec.dynamic.partition.mode", "nonstrict")
-            .config(
-                "spark.sql.warehouse.dir",
-                String.format(
-                    "hdfs://%s:%d/user/hive/warehouse",
-                    containerSuite.getHiveContainer().getContainerIpAddress(),
-                    HiveContainer.HDFS_DEFAULTFS_PORT))
+            .config("spark.sql.warehouse.dir", warehouse)
             .enableHiveSupport()
             .getOrCreate();
   }

integration-test/src/test/java/com/datastrato/gravitino/integration/test/spark/hive/SparkHiveCatalogIT.java

@@ -40,6 +40,11 @@ protected boolean supportsSparkSQLClusteredBy() {
     return true;
   }
 
+  @Override
+  protected boolean supportsPartition() {
+    return true;
+  }
+
   @Test
   public void testCreateHiveFormatPartitionTable() {
     String tableName = "hive_partition_table";

integration-test/src/test/java/com/datastrato/gravitino/integration/test/spark/iceberg/SparkIcebergCatalogIT.java

@@ -0,0 +1,34 @@
+/*
+ * Copyright 2024 Datastrato Pvt Ltd.
+ * This software is licensed under the Apache License version 2.
+ */
+package com.datastrato.gravitino.integration.test.spark.iceberg;
+
+import com.datastrato.gravitino.integration.test.spark.SparkCommonIT;
+import org.junit.jupiter.api.Tag;
+import org.junit.jupiter.api.TestInstance;
+
+@Tag("gravitino-docker-it")
+@TestInstance(TestInstance.Lifecycle.PER_CLASS)
+public class SparkIcebergCatalogIT extends SparkCommonIT {
+
+  @Override
+  protected String getCatalogName() {
+    return "iceberg";
+  }
+
+  @Override
+  protected String getProvider() {
+    return "lakehouse-iceberg";
+  }
+
+  @Override
+  protected boolean supportsSparkSQLClusteredBy() {
+    return false;
+  }
+
+  @Override
+  protected boolean supportsPartition() {
+    return false;
+  }
+}

spark-connector/src/main/java/com/datastrato/gravitino/spark/connector/GravitinoCatalogAdaptorFactory.java

@@ -6,6 +6,7 @@
 package com.datastrato.gravitino.spark.connector;
 
 import com.datastrato.gravitino.spark.connector.hive.HiveAdaptor;
+import com.datastrato.gravitino.spark.connector.iceberg.IcebergAdaptor;
 import java.util.Locale;
 
 /**
@@ -17,6 +18,8 @@ public static GravitinoCatalogAdaptor createGravitinoAdaptor(String provider) {
     switch (provider.toLowerCase(Locale.ROOT)) {
       case "hive":
         return new HiveAdaptor();
+      case "lakehouse-iceberg":
+        return new IcebergAdaptor();
       default:
         throw new RuntimeException(String.format("Provider:%s is not supported yet", provider));
     }

spark-connector/src/main/java/com/datastrato/gravitino/spark/connector/catalog/GravitinoCatalogManager.java

@@ -58,6 +58,7 @@ public void close() {
     Preconditions.checkState(!isClosed, "Gravitino Catalog is already closed");
     isClosed = true;
     gravitinoClient.close();
+    gravitinoCatalogManager = null;
   }
 
   public Catalog getGravitinoCatalogInfo(String name) {