[DRILL-4062] SUM function does not work against interval data type from csv or json files

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: Functions - Drill
• Labels: interval
• Assignee: Unassigned
• Reporter: Krystal
• Created: 10/Nov/15 20:03
• Updated: 10/Nov/15 20:03

Description

I have a csv table with the following interval year data:
select columns[1] from `interval_data.csv`;
---------

EXPR$0

---------

P12Y

P152Y

P-152Y

P5Y

P10Y

P0Y

P-50Y

P12Y

P152Y

---------

I got an exception when trying to use the SUM function against the data:
select sum(cast(columns[1] as interval year)) from `interval_data.csv`;
Error: VALIDATION ERROR: From line 1, column 8 to line 1, column 45: Cannot apply 'SUM' to arguments of type 'SUM(<INTERVAL YEAR>)'. Supported form(s): 'SUM(<NUMERIC>)'

Same error is encountered for same data from JSON file.
This works for interval data from parquet file:
./parquet-schema ../interval_data.parquet
message root {
optional int32 c_row;
optional fixed_len_byte_array(12) c_interval_year;
optional fixed_len_byte_array(12) c_interval_month;
optional fixed_len_byte_array(12) c_interval_day;
optional fixed_len_byte_array(12) c_interval_hour;
optional fixed_len_byte_array(12) c_interval_minute;
optional fixed_len_byte_array(12) c_interval_second;
optional binary c_interval (UTF8);
optional int32 c_time;
optional int32 c_date;
optional int64 c_timestamp;
}

select sum(c_interval_year) from `interval_data.parquet`;
---------

EXPR$0

---------

P1692M

---------