Github user jackylk commented on a diff in the pull request:
https://github.com/apache/carbondata/pull/2823#discussion_r227621527
--- Diff:
integration/spark-datasource/src/main/spark2.1andspark2.2/org/apache/spark/sql/CarbonVectorProxy.java
---
@@ -150,127 +140,189 @@ public void reset() {
columnarBatch.reset();
}
- public void putRowToColumnBatch(int rowId, Object value, int offset) {
- org.apache.spark.sql.types.DataType t = dataType(offset);
- if (null == value) {
- putNull(rowId, offset);
- } else {
- if (t == org.apache.spark.sql.types.DataTypes.BooleanType) {
- putBoolean(rowId, (boolean) value, offset);
- } else if (t == org.apache.spark.sql.types.DataTypes.ByteType)
{
- putByte(rowId, (byte) value, offset);
- } else if (t ==
org.apache.spark.sql.types.DataTypes.ShortType) {
- putShort(rowId, (short) value, offset);
- } else if (t ==
org.apache.spark.sql.types.DataTypes.IntegerType) {
- putInt(rowId, (int) value, offset);
- } else if (t == org.apache.spark.sql.types.DataTypes.LongType)
{
- putLong(rowId, (long) value, offset);
- } else if (t ==
org.apache.spark.sql.types.DataTypes.FloatType) {
- putFloat(rowId, (float) value, offset);
- } else if (t ==
org.apache.spark.sql.types.DataTypes.DoubleType) {
- putDouble(rowId, (double) value, offset);
- } else if (t ==
org.apache.spark.sql.types.DataTypes.StringType) {
- UTF8String v = (UTF8String) value;
- putByteArray(rowId, v.getBytes(), offset);
- } else if (t instanceof
org.apache.spark.sql.types.DecimalType) {
- DecimalType dt = (DecimalType) t;
- Decimal d = Decimal.fromDecimal(value);
- if (dt.precision() <= Decimal.MAX_INT_DIGITS()) {
- putInt(rowId, (int) d.toUnscaledLong(), offset);
- } else if (dt.precision() <= Decimal.MAX_LONG_DIGITS()) {
- putLong(rowId, d.toUnscaledLong(), offset);
- } else {
- final BigInteger integer =
d.toJavaBigDecimal().unscaledValue();
- byte[] bytes = integer.toByteArray();
- putByteArray(rowId, bytes, 0, bytes.length, offset);
+
+ public static class ColumnVectorProxy {
+
+ private ColumnVector vector;
+
+ public ColumnVectorProxy(ColumnarBatch columnarBatch, int ordinal)
{
+ this.vector = columnarBatch.column(ordinal);
+ }
+
+ public void putRowToColumnBatch(int rowId, Object value, int
offset) {
+ org.apache.spark.sql.types.DataType t = dataType(offset);
--- End diff --
It seems the offset param is not used in dataType, and please change the
function name of dataType
---
