如果时间在开始和结束时间之间，则按时间间隔聚合列

Question

我正在尝试对聚合表进行逆向工程以计算并发值。 表格样本：

start_time                end_time                  duration_s    some_id     total_val
2023-03-30 01:03:19.000    2023-03-30 01:03:58.699    39.699        A          3098.214
2023-03-30 00:41:50.000    2023-03-30 01:03:59.920    1329.92       A          1264.4286
2023-03-30 01:01:48.000    2023-03-30 01:04:00.759    132.759       B          2323.6356
2023-03-30 00:56:55.000    2023-03-30 01:04:00.946    425.946       B          1676.4208
2023-03-30 01:03:50.000    2023-03-30 01:04:02.711    12.711        A          1211.4553

我正在尝试创建一个按秒分组的连续图。 total_val 是整个持续时间 (end_time - start_time) 的成本，应该平均分配 (total_val/duration_s)。按 start_time 的简单分组不会说明事件何时结束。

我看过类似的questions但我是一个SQL n00b并且会很感激一些指示

这是有效但非常慢的代码（按分钟而不是秒聚合），

WITH temp AS (
        SELECT
            total_val / duration_s AS usage_per_sec,
            SEQUENCE(start_time, end_time, INTERVAL '1' MINUTE) exploded_times
        FROM tbl
    ),
    temp2 AS (
        SELECT
            exploded_times,
            TRANSFORM(exploded_times, x -> usage_per_sec * 60) usage_per_min
        FROM temp
    ),
    temp3 AS (
        SELECT
            ts_min,
            usage
        FROM temp2
        CROSS JOIN UNNEST(exploded_times, usage_per_min) AS t (ts_min, usage)
    )

SELECT
    ts_min,
    SUM(usage)
FROM temp3
GROUP BY
    ts_min

期望的输出将采用以下格式：ts 和 sum(total_val/duration)

2023-03-29 23:59:56.000   196.93959475378
2023-03-29 23:59:57.000   46216.598554771
2023-03-29 23:59:58.000   57332.587457153
2023-03-29 23:59:59.000   404240.20792671
2023-03-30 00:00:00.000   252365.2397336
2023-03-30 00:00:01.000   665889.58236006
2023-03-30 00:00:02.000   587338.31764963