How to understand the field "groups" and the agg "GROUPING" in EnumerableAggregate - sql

Question

I am new to Calcite and I am using Calcite to convert a SQL query to an optimized plan, where I will translate the plan to a dataflow graph in an execution engine. One challenge is the translation of different RelNodes (e.g., Filter, Project, Aggregate, Calc, etc). I found a difficulty in understanding the EnumerableAggregate RelNode. Specifically, for the following example, where I defined a table T as
create table T (src int, dst int, label int, time int);
and wrote a toy query as
select count(distinct dst), sum(distinct label), count(*)
from T
where dst > 1
group by src
having src = 0;
I will obtain an optimized plan which contains two EnumerableAggregate RelNodes and here is the first EnumerableAggregate RelNode:
{
"id": "2",
"relOp": "org.apache.calcite.adapter.enumerable.EnumerableAggregate",
"group": [ 0, 1, 2 ],
"groups": [
[ 0, 1 ], [ 0, 2 ], [ 0 ]
],
"aggs": [
{
"agg": {
"name": "COUNT",
"kind": "COUNT",
"syntax": "FUNCTION_STAR"
},
"type": {
"type": "BIGINT",
"nullable": false
},
"distinct": false,
"operands": [],
"name": "EXPR$2"
},
{
"agg": {
"name": "GROUPING",
"kind": "GROUPING",
"syntax": "FUNCTION"
},
"type": {
"type": "BIGINT",
"nullable": false
},
"distinct": false,
"operands": [ 0, 1, 2 ],
"name": "$g"
}
]
}
I think I understand the reason why there are two Aggregate RelNodes. The reason is due to the use of distinct on dst in count and the use of distinct on label in sum in the query, where the optimizer wants to first group the data by (1) the group key src and (2) the two distinct columns (dst and label), in order to remove duplications. Then in the second Aggregate we calculate count and sum,
What I do not understand is how the first Aggregate processes the input data, what does the field groups (i.e., [0, 1], [0, 2] and [0]) do, what does the agg function GROUPING do and how many columns are there in the ouput of the first Aggregate.
For example, given the following input data: [[2,3,4,0], [2,3,4,1], [3,2,4,2], [3,2,4,3], [5,6,7,4], [5,6,7,5]], I think the data will be firstly divided into three groups: [[2,3,4,0], [2,3,4,1]] and [[3,2,4,2], [3,2,4,3]] and [[5,6,7,4], [5,6,7,5]]. But what is the next step?
Any help would be appreciated. Thanks!