如何在子查询中填写日期？

Question

我有一张买东西的桌子：

player_id | date       | registration_date | price
   pl1    | 2019-01-21 | 2019-01-20        |  20
   pl1    | 2019-01-23 | 2019-01-20        |  10
   pl1    | 2019-01-24 | 2019-01-20        |  15

在“日期”上使用 groupArray 和在“价格”上使用 arrayCumSum 并使用 ArrayJoin 计算后，我得到了每天累积总和的表格：

player_id | date       | registration_date | sum_price
   pl1    | 2019-01-21 | 2019-01-20        |  20
   pl1    | 2019-01-23 | 2019-01-20        |  30
   pl1    | 2019-01-24 | 2019-01-20        |  45

但我需要添加从注册到今天的缺失日期（今天是'2019-01-25'）：

player_id | date       | registration_date | sum_price
   pl1    | 2019-01-20 | 2019-01-20        |  0
   pl1    | 2019-01-21 | 2019-01-20        |  20
   pl1    | 2019-01-22 | 2019-01-20        |  20
   pl1    | 2019-01-23 | 2019-01-20        |  30
   pl1    | 2019-01-24 | 2019-01-20        |  45
   pl1    | 2019-01-25 | 2019-01-20        |  45

我该怎么做？

Answer 1

试试这个：

SELECT player_id, result.1 as date, registrationDate as registration_date, result.2 as sum_price
FROM
(
    SELECT
        player_id,
        groupArray((date, price)) AS purchases,
        min(registration_date) AS registrationDate,
        arrayMap(x -> registrationDate + x, range(toUInt32(toDate('2019-01-25') - registrationDate + 1))) dates,
        arrayFilter(x -> arrayFirstIndex(p -> p.1 = x, purchases) = 0, dates) AS missed_dates,
        arrayMap(x -> (x, 0), missed_dates) AS dummy_purchases,
        arraySort(x -> x.1, arrayConcat(purchases, dummy_purchases)) all_purchases,
        arrayCumSum(x -> x.2, all_purchases) cum_prices,
        arrayMap(index -> (all_purchases[index].1, cum_prices[index]), arrayEnumerate(all_purchases)) flat_result,
        arrayJoin(flat_result) result
    FROM test.purchases01
    GROUP BY player_id
)

/* result
┌─player_id─┬───────date─┬─registration_date─┬─sum_price─┐
│ pl1       │ 2019-01-20 │        2019-01-20 │         0 │
│ pl1       │ 2019-01-21 │        2019-01-20 │        20 │
│ pl1       │ 2019-01-22 │        2019-01-20 │        20 │
│ pl1       │ 2019-01-23 │        2019-01-20 │        30 │
│ pl1       │ 2019-01-24 │        2019-01-20 │        45 │
│ pl1       │ 2019-01-25 │        2019-01-20 │        45 │
└───────────┴────────────┴───────────────────┴───────────┘

*/


/* Prepare test data */

CREATE TABLE test.purchases01
(
    `player_id` String,
    `date` Date,
    `registration_date` Date,
    `price` int
)
ENGINE = Memory;

INSERT INTO test.purchases01
VALUES  ('pl1', '2019-01-21', '2019-01-20', 20), 
        ('pl1', '2019-01-23', '2019-01-20', 10), 
        ('pl1', '2019-01-24', '2019-01-20', 15);