Postgresql postgres-根据条件计算累积和

我有一个名为asset_reporting的DB表，如下所示

asset_id      asset_type       entered_at             exited_at         site_id
  100           Cage       2019-05-02 10:00:00    2019-05-04 10:00:00     25
  102           Cage       2019-05-03 10:00:00         null               25
  103         Container    2019-05-01 10:00:00    2019-05-03 10:00:00     25

我需要计算资产在给定时间范围内在特定站点的运行总时间。假设我看到的时间范围是2019-05-01到2019-05-05，我需要的输出低于日

     day       asset_type        count
 2019-05-01      Cage              0
 2019-05-01    Container           1
 2019-05-02      Cage              1
 2019-05-02    Container           1
 2019-05-03      Cage              2
 2019-05-03    Container           1
 2019-05-04      Cage              2
 2019-05-04    Container           0
 2019-05-05      Cage              1
 2019-05-05    Container           0

到目前为止，我得到了下面的问题。我无法计算出如何保持资产在站点的总天数的运行总数

select date(ar.entered_at) as day, asset_type, count(*) from asset_reporting ar
group by day,asset_type

下面是上述查询的输出。请注意，它与上面的预期输出不符

     day       asset_type        count
 2019-05-01      Cage              0
 2019-05-01    Container           1
 2019-05-02      Cage              1
 2019-05-02    Container           0
 2019-05-03      Cage              1
 2019-05-03    Container           0
 2019-05-04      Cage              0
 2019-05-04    Container           0
 2019-05-05      Cage              0
 2019-05-05    Container           0

---

我们可以使用日历表方法解决此问题：

WITH dates AS (
    SELECT '2019-05-01'::date AS day UNION ALL
    SELECT '2019-05-02'::date UNION ALL
    SELECT '2019-05-03'::date UNION ALL
    SELECT '2019-05-04'::date UNION ALL
    SELECT '2019-05-05'::date
)

SELECT
    d.day,
    at.asset_type,
    COUNT(ar.asset_type) AS count
FROM dates d
CROSS JOIN (SELECT DISTINCT asset_type FROM asset_reporting) at
LEFT JOIN asset_reporting ar
    ON at.asset_type = ar.asset_type AND
       (d.day >= ar.entered_at::date OR ar.entered_at IS NULL) AND
       (d.day <= ar.exited_at::date OR ar.exited_at IS NULL)
GROUP BY
    d.day,
    at.asset_type
ORDER BY
    d.day,
    at.asset_type;

---

我们可以使用日历表方法解决此问题：

WITH dates AS (
    SELECT '2019-05-01'::date AS day UNION ALL
    SELECT '2019-05-02'::date UNION ALL
    SELECT '2019-05-03'::date UNION ALL
    SELECT '2019-05-04'::date UNION ALL
    SELECT '2019-05-05'::date
)

SELECT
    d.day,
    at.asset_type,
    COUNT(ar.asset_type) AS count
FROM dates d
CROSS JOIN (SELECT DISTINCT asset_type FROM asset_reporting) at
LEFT JOIN asset_reporting ar
    ON at.asset_type = ar.asset_type AND
       (d.day >= ar.entered_at::date OR ar.entered_at IS NULL) AND
       (d.day <= ar.exited_at::date OR ar.exited_at IS NULL)
GROUP BY
    d.day,
    at.asset_type
ORDER BY
    d.day,
    at.asset_type;