Github user fszabo2 commented on a diff in the pull request:
https://github.com/apache/sqoop/pull/60#discussion_r238251774
--- Diff: src/java/org/apache/sqoop/hive/HiveTypes.java ---
@@ -37,16 +42,28 @@
private static final String HIVE_TYPE_STRING = "STRING";
private static final String HIVE_TYPE_BOOLEAN = "BOOLEAN";
private static final String HIVE_TYPE_BINARY = "BINARY";
+ private static final String HIVE_TYPE_DECIMAL = "DECIMAL";
public static final Log LOG =
LogFactory.getLog(HiveTypes.class.getName());
private HiveTypes() { }
+
+ public static String toHiveType(int sqlType, SqoopOptions options) {
+
+ if
(options.getConf().getBoolean(ConfigurationConstants.PROP_ENABLE_PARQUET_LOGICAL_TYPE_DECIMAL,
false)
+ && (sqlType == Types.NUMERIC || sqlType == Types.DECIMAL)){
+ return HIVE_TYPE_DECIMAL;
+ }
+ return toHiveType(sqlType);
+ }
+
+
/**
* Given JDBC SQL types coming from another database, what is the best
* mapping to a Hive-specific type?
*/
- public static String toHiveType(int sqlType) {
+ private static String toHiveType(int sqlType) {
--- End diff --
After taking another look at this file, it turns out that this method is
only called textfile import. Since my only intended use case is parquetfile, I
reverted this part of the change.
The implementation will be throwing a RuntimeException (for mapping error)
and log warnings in case of a parquet files, as you've suggested.
---
