[GitHub] [iceberg] jackye1995 commented on a change in pull request #1633: AWS Glue catalog and table operations

[GitBox <gi...@apache.org>]

jackye1995 commented on a change in pull request #1633:
URL: https://github.com/apache/iceberg/pull/1633#discussion_r508129106



##########
File path: aws/src/main/java/org/apache/iceberg/aws/glue/GlueCatalog.java
##########
@@ -0,0 +1,312 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+package org.apache.iceberg.aws.glue;
+
+import java.io.Closeable;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
+import java.util.stream.Collectors;
+import org.apache.hadoop.conf.Configuration;
+import org.apache.iceberg.BaseMetastoreCatalog;
+import org.apache.iceberg.CatalogUtil;
+import org.apache.iceberg.TableMetadata;
+import org.apache.iceberg.TableOperations;
+import org.apache.iceberg.catalog.Namespace;
+import org.apache.iceberg.catalog.SupportsNamespaces;
+import org.apache.iceberg.catalog.TableIdentifier;
+import org.apache.iceberg.exceptions.NamespaceNotEmptyException;
+import org.apache.iceberg.exceptions.NoSuchNamespaceException;
+import org.apache.iceberg.relocated.com.google.common.base.Preconditions;
+import org.apache.iceberg.relocated.com.google.common.collect.Lists;
+import org.apache.iceberg.relocated.com.google.common.collect.Maps;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+import software.amazon.awssdk.services.glue.GlueClient;
+import software.amazon.awssdk.services.glue.model.CreateDatabaseRequest;
+import software.amazon.awssdk.services.glue.model.DeleteDatabaseRequest;
+import software.amazon.awssdk.services.glue.model.DeleteTableRequest;
+import software.amazon.awssdk.services.glue.model.EntityNotFoundException;
+import software.amazon.awssdk.services.glue.model.GetDatabaseRequest;
+import software.amazon.awssdk.services.glue.model.GetDatabaseResponse;
+import software.amazon.awssdk.services.glue.model.GetDatabasesRequest;
+import software.amazon.awssdk.services.glue.model.GetDatabasesResponse;
+import software.amazon.awssdk.services.glue.model.GetTablesRequest;
+import software.amazon.awssdk.services.glue.model.GetTablesResponse;
+import software.amazon.awssdk.services.glue.model.InvalidInputException;
+import software.amazon.awssdk.services.glue.model.UpdateDatabaseRequest;
+
+public class GlueCatalog extends BaseMetastoreCatalog implements Closeable, SupportsNamespaces {
+
+  private static final Logger LOG = LoggerFactory.getLogger(GlueCatalog.class);
+
+  private final Configuration conf;
+  private final GlueClient glue;
+  // Catalog ID is the AWS account ID. When it is null, it uses the account of the Glue client.
+  private final String catalogId;
+  private final String warehousePath;
+
+  /**
+   * constructor with default glue client
+   * @param conf Hadoop config, passed in for users of HadoopIO
+   * @param catalogId Glue catalog ID, which is the AWS account ID. When null, it uses the account of the Glue client.
+   * @param warehousePath warehouse path, such as s3://my-iceberg-bucket/metadata_prod
+   */
+  public GlueCatalog(Configuration conf, String catalogId, String warehousePath) {
+    this(conf, GlueClient.create(), catalogId, warehousePath);
+  }
+
+  /**
+   * constructor
+   * @param conf Hadoop config, passed in for users of HadoopIO
+   * @param glue Glue client
+   * @param catalogId Glue catalog ID, which is the AWS account ID. When null, it uses the account of the Glue client.
+   * @param warehousePath warehouse path, such as s3://my-iceberg-bucket/metadata_prod
+   */
+  public GlueCatalog(Configuration conf, GlueClient glue, String catalogId, String warehousePath) {
+    this.conf = conf;
+    this.glue = glue;
+    this.catalogId = catalogId;
+    this.warehousePath = warehousePath;
+    validateWarehousePath();
+  }
+
+  private void validateWarehousePath() {
+    Preconditions.checkArgument(warehousePath != null && warehousePath.length() > 0,
+        "Glue warehouse path must be set");
+    Preconditions.checkArgument(warehousePath.charAt(warehousePath.length() - 1) != '/',
+        "Glue warehouse path %s must not end with /", warehousePath);
+  }
+
+  @Override
+  public void close() throws IOException {
+    glue.close();
+  }
+
+  @Override
+  protected TableOperations newTableOps(TableIdentifier tableIdentifier) {
+    String dbName = IcebergToGlueConverter.toDatabaseName(tableIdentifier.namespace());
+    String tableName = tableIdentifier.name();
+    return new GlueTableOperations(conf, glue, catalogId, dbName, tableName);
+  }
+
+  /**
+   * Use the same format as Hive for default warehouse location
+   * @param tableIdentifier table id
+   * @return default warehouse path
+   */
+  @Override
+  protected String defaultWarehouseLocation(TableIdentifier tableIdentifier) {
+    // check if value is set in database
+    GetDatabaseResponse response = glue.getDatabase(GetDatabaseRequest.builder()
+        .name(IcebergToGlueConverter.toDatabaseName(tableIdentifier.namespace()))
+        .build());
+    String dbLocationUri = response.database().locationUri();
+    if (dbLocationUri != null) {
+      return String.format("%s/%s", dbLocationUri, tableIdentifier.name());
+    }
+
+    return String.format(
+        "%s/%s.db/%s",
+        warehousePath,
+        IcebergToGlueConverter.toDatabaseName(tableIdentifier.namespace()),
+        tableIdentifier.name());
+  }
+
+  @Override
+  public List<TableIdentifier> listTables(Namespace namespace) {
+    namespaceExists(namespace);
+    // should be safe to list all before returning the list, instead of dynamically load the list.
+    String nextToken = null;
+    List<TableIdentifier> results = new ArrayList<>();
+    while (true) {
+      GetTablesResponse response = glue.getTables(GetTablesRequest.builder()
+          .catalogId(catalogId)
+          .databaseName(IcebergToGlueConverter.toDatabaseName(namespace))
+          .nextToken(nextToken)
+          .build());
+      if (response.hasTableList()) {
+        results.addAll(response.tableList().stream()
+            .map(GlueToIcebergConverter::toTableId)
+            .collect(Collectors.toList()));
+        nextToken = response.nextToken();
+      }
+      if (nextToken == null) {
+        break;
+      }
+    }
+    return results;
+  }
+
+  @Override
+  public boolean dropTable(TableIdentifier identifier, boolean purge) {
+    try {
+      dropTable(newTableOps(identifier), identifier, purge);
+      return true;
+    } catch (Exception e) {
+      return false;
+    }
+  }
+
+  private TableMetadata dropTable(TableOperations ops, TableIdentifier identifier, boolean purge) {
+    TableMetadata lastMetadata = null;
+    if (ops.current() != null) {
+      lastMetadata = ops.current();
+    }
+
+    glue.deleteTable(DeleteTableRequest.builder()
+        .catalogId(catalogId)
+        .databaseName(IcebergToGlueConverter.toDatabaseName(identifier.namespace()))
+        .name(identifier.name())
+        .build());
+
+    if (purge && lastMetadata != null) {
+      CatalogUtil.dropTableData(ops.io(), lastMetadata);
+    }
+
+    LOG.info("Dropped table: {}", identifier);
+    return lastMetadata;
+  }
+
+  /**
+   * Rename table in Glue is a drop table and create table.
+   * @param from identifier of the table to rename
+   * @param to new table name
+   */
+  @Override
+  public void renameTable(TableIdentifier from, TableIdentifier to) {
+    TableOperations ops = newTableOps(from);
+    // keep metadata
+    TableMetadata tableMetadata = dropTable(ops, from, false);
+    // use the same info to create the new table
+    createTable(to, tableMetadata.schema(), tableMetadata.spec(), tableMetadata.properties());
+  }
+
+  @Override
+  public void createNamespace(Namespace namespace, Map<String, String> metadata) {
+    glue.createDatabase(CreateDatabaseRequest.builder()
+        .catalogId(catalogId)
+        .databaseInput(IcebergToGlueConverter.toDatabaseInput(namespace, metadata))
+        .build());
+  }
+
+  @Override
+  public List<Namespace> listNamespaces(Namespace namespace) throws NoSuchNamespaceException {
+    if (!namespace.isEmpty()) {
+      // if it is not a list all op, just check if the namespace exists and return itself.
+      if (namespaceExists(namespace)) {
+        return Lists.newArrayList(namespace);
+      }
+      throw new NoSuchNamespaceException(
+          "Glue does not support nested namespace, cannot list namespaces under %s", namespace);
+    }
+    // should be safe to list all before returning the list, instead of dynamically load the list.
+    String nextToken = null;
+    List<Namespace> results = new ArrayList<>();
+    while (true) {
+      GetDatabasesResponse response = glue.getDatabases(GetDatabasesRequest.builder()
+          .catalogId(catalogId)
+          .nextToken(nextToken)
+          .build());
+      if (response.hasDatabaseList()) {
+        results.addAll(response.databaseList().stream()
+            .map(GlueToIcebergConverter::toNamespace)
+            .collect(Collectors.toList()));
+        nextToken = response.nextToken();
+      }
+      if (nextToken == null) {
+        break;
+      }
+    }
+    return results;
+  }
+
+  @Override
+  public Map<String, String> loadNamespaceMetadata(Namespace namespace) throws NoSuchNamespaceException {
+    String databaseName = IcebergToGlueConverter.toDatabaseName(namespace);
+    try {
+      GetDatabaseResponse response = glue.getDatabase(GetDatabaseRequest.builder()
+          .catalogId(catalogId)
+          .name(databaseName)
+          .build());
+      return response.database().parameters();
+    } catch (InvalidInputException e) {
+      throw new NoSuchNamespaceException("invalid input for namespace %s, error message: %s",
+          namespace, e.getMessage());
+    } catch (EntityNotFoundException e) {
+      throw new NoSuchNamespaceException("fail to find Glue database for namespace %s, error message: %s",
+          databaseName, e.getMessage());
+    }
+  }
+
+  /**
+   * After completing this operation, you no longer have access to the tables
+   * (and all table versions and partitions that might belong to the tables)
+   * and the user-defined functions in the deleted database.
+   * AWS Glue deletes these "orphaned" resources asynchronously in a timely manner, at the discretion of the service.
+   * To ensure the immediate deletion of all related resources, before calling DeleteDatabase, use DeleteTableVersion
+   * or BatchDeleteTableVersion, DeletePartition or BatchDeletePartition, DeleteUserDefinedFunction,
+   * and DeleteTable or BatchDeleteTable, to delete any resources that belong to the database.
+   * @param namespace a namespace. {@link Namespace}
+   * @return always true
+   * @throws NamespaceNotEmptyException never thrown

Review comment:
       Good point, this is definitely a different behavior between Glue and Hive. I can do a ListTable before dropping.




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org