cmathiesen commented on a change in pull request #1103: URL: https://github.com/apache/iceberg/pull/1103#discussion_r438776518
########## File path: mr/src/main/java/org/apache/iceberg/mr/mapred/TableResolver.java ########## @@ -0,0 +1,121 @@ +/* + * 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.mr.mapred; + +import java.io.IOException; +import java.net.URI; +import java.net.URISyntaxException; +import java.util.Properties; +import org.apache.commons.lang3.StringUtils; +import org.apache.hadoop.conf.Configuration; +import org.apache.hadoop.mapred.JobConf; +import org.apache.iceberg.Table; +import org.apache.iceberg.catalog.TableIdentifier; +import org.apache.iceberg.hadoop.HadoopCatalog; +import org.apache.iceberg.hadoop.HadoopTables; +import org.apache.iceberg.mr.InputFormatConfig; + +final class TableResolver { + + private TableResolver() { + } + + static Table resolveTableFromJob(JobConf conf) throws IOException { + Properties properties = new Properties(); + properties.setProperty(InputFormatConfig.CATALOG_NAME, extractProperty(conf, InputFormatConfig.CATALOG_NAME)); + if (conf.get(InputFormatConfig.CATALOG_NAME).equals(InputFormatConfig.HADOOP_CATALOG)) { + properties.setProperty(InputFormatConfig.SNAPSHOT_TABLE, conf.get(InputFormatConfig.SNAPSHOT_TABLE, "true")); + } + properties.setProperty(InputFormatConfig.TABLE_LOCATION, extractProperty(conf, InputFormatConfig.TABLE_LOCATION)); + properties.setProperty(InputFormatConfig.TABLE_NAME, extractProperty(conf, InputFormatConfig.TABLE_NAME)); + return resolveTableFromConfiguration(conf, properties); + } + + static Table resolveTableFromConfiguration(Configuration conf, Properties properties) throws IOException { + String catalogName = properties.getProperty(InputFormatConfig.CATALOG_NAME); + URI tableLocation = pathAsURI(properties.getProperty(InputFormatConfig.TABLE_LOCATION)); + if (catalogName == null) { + throw new IllegalArgumentException("Catalog property: 'iceberg.catalog' not set in JobConf"); + } + switch (catalogName) { + case InputFormatConfig.HADOOP_TABLES: + HadoopTables tables = new HadoopTables(conf); + return tables.load(tableLocation.getPath()); + case InputFormatConfig.HADOOP_CATALOG: + String tableName = properties.getProperty(InputFormatConfig.TABLE_NAME); + TableIdentifier id = TableIdentifier.parse(tableName); + if (tableName.endsWith(InputFormatConfig.SNAPSHOT_TABLE_SUFFIX)) { + if (!Boolean.parseBoolean(properties.getProperty(InputFormatConfig.SNAPSHOT_TABLE, + Boolean.TRUE.toString()))) { + String tablePath = id.toString().replaceAll("\\.", "/"); + URI warehouseLocation = pathAsURI(tableLocation.getPath().replaceAll(tablePath, "")); + HadoopCatalog catalog = new HadoopCatalog(conf, warehouseLocation.getPath()); + return catalog.loadTable(id); + } else { + return resolveMetadataTable(conf, tableLocation.getPath(), tableName); + } + } else { + URI warehouseLocation = pathAsURI(extractWarehousePath(tableLocation.getPath(), tableName)); Review comment: We had initially implemented the IF using a method similar to what you're describing: we required a user to add an extra table property called `iceberg.warehouse.location` I think and used that to resolve the table with HadoopCatalog. The `LOCATION` property of your Hive table has to be unique for each table (which is why we always have a property for a full table path), so we couldn't put a path to a common warehouse there. We initially tried using the same location for multiple tables and relied on the table name/ID to resolve the table correctly in Iceberg but that caused major confusion in Hive when trying to join tables :') So originally we used an extra property for the warehouse location, but we realised we could extract this location from the full table path and save the user an extra configuration property. We don't have any strong preference either way, so I'm happy to change this as you've suggested Although thinking about this raises another point - I've just realised/tested that you could create an Iceberg table using HadoopCatalog but if you had the full path you can use HadoopTables to load the table again instead... And with Hive we can always expect to have the full path to the table because the location needs to be unique, so we could simplify here and only use HadoopTables for this IF? @massdosage ---------------------------------------------------------------- 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: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
