如何在 postgres 中返回过去 12 个月内每个月发生的所有记录的总和

Question

桌子看起来像这样

所以我得到了一个看起来像这样的查询

with date_ranges (range_name, range_dates) as
              ( values ('week_0', tstzrange ((now()::date-interval '6 days'),  now(), '[]'))
                      , ('week_1', tstzrange ((now()::date-interval '13 days'), (now()::date-interval '6 days')))
                      , ('week_2', tstzrange ((now()::date-interval '20 days'), (now()::date-interval '13 days')))
                      , ('week_3', tstzrange ((now()::date-interval '27 days'), (now()::date-interval '20 days')))
              )
                select range_name, range_dates, COALESCE(SUM("transactionTotal"), 0) total_amount 
        from date_ranges dr
        LEFT JOIN "MoneyTransactions" mt on (mt."entryType" = 'sold' AND mt."createdAt" <@ range_dates)
        group by range_name, range_dates
        order by range_name;

它返回过去四个星期每周的总和。 现在我试图返回相同的东西，但不是几周，而是过去 12 个月的总和。

我能够获得过去几个月的所需日期，例如

SELECT
  to_char(now() - series.n * '1 month'::interval, 'month') AS "month",
  tstzrange(
    date_trunc('month', now()) - series.n * '1 month'::interval,
    date_trunc('month', now()) - (series.n - 1) * '1 month'::interval,
    '[]'
  ) AS range_dates
FROM (
  SELECT generate_series(0, 11) AS n
) AS series

我试过了，显然它不起作用

with date_ranges (range_name, range_dates) as
              ( values (
                to_char(now() - series.n * '1 month'::interval, 'month') as "month",
                tstzrange(
                  date_trunc('month', now()) - series.n * '1 month'::interval,
                  date_trunc('month', now()) - (series.n - 1) * '1 month'::interval,
                  '[]'
              ))
              )
            select range_name, range_dates(0, 11), COALESCE(SUM("transactionTotal"), 0) total_amount 
        from date_ranges dr
        LEFT JOIN "MoneyTransactions" mt on (mt."entryType" = 'sold' AND mt."createdAt" <@ range_dates)
        group by range_name, range_dates
        order by range_name;

然而，我对 SQL 的有限知识阻止我将所有这些都放在一个查询中以获得所需的输出，看起来像这样

[
        {
            "range_name": "september",
            "range_dates": [
                {
                    "value": "2020-09-01T00:00:00.000Z",
                    "inclusive": true
                },
                {
                    "value": "2020-10-01T00:00:00.000Z",
                    "inclusive": false
                }
            ],
            "total_amount": "253050"
        },
        {
            "range_name": "august",
            "range_dates": [
                {
                    "value": "2020-08-01T00:00:00.000Z",
                    "inclusive": true
                },
                {
                    "value": "2020-09-01T00:00:00.000Z",
                    "inclusive": false
                }
            ],
            "total_amount": "55200"
        },
        ...
]

有人可以帮忙吗

Answer 1

这是获得 12 个月范围的更简洁的方法。 根据money_transactions的日期，您可以轻松地使用它们与create_date列进行比较。

to_char返回不同格式的 varchar，在我的示例中，我使用MM/YYYY ，它将返回 2019 年 10 月的10/2019 。您必须确保生成的系列日期和MoneyTransactions列createdAt日期格式匹配。 但是，您可以以任何方式调整它。

with date_range as 
(
   select
      to_char(months, 'MM/YYYY') as "month" -- returns month/year 10/19
   from
      generate_series( date_trunc('month', current_date - interval '11 months'), date_trunc('month', current_date), '1 month':: interval) as months 
)
select
   dr."month",
   COALESCE(SUM("transactionTotal"), 0) total_amount 
from
   date_range dr 
   left join
      "MoneyTransactions" mt 
      on mt."entryType" = 'sold' 
      and to_char(mt."createdAt", 'MM/YYYY') = dr."month" 
group by
   dr."month";

另外，SQL命名约定建议在标识符（表、列等）中使用snake_case和所有小写字母，因此money_transactions而不是MoneyTransactions 。