如何生成月度报告，用零填充一个月中缺少的几天

Question

假设我在 QuestDB table 上有以下数据：

ts               | item
------------------------
2020-01-02T10:12 | a
2020-01-02T10:20 | b
2020-01-03T11:20 | a
2020-01-03T11:30 | b
2020-01-03T11:40 | c

我可以执行以下查询来计算特定月份按天采样的不同项目：

SELECT DAY(ts) as day, COUNT_DISTINCT(item) AS count_items
FROM my_table
WHERE ts IN '2020-01'
SAMPLE BY d FILL(0);

我得到以下数据：

day | count_items
------------------
2   | 2
3   | 3

我想要一个完整的月度报告而不是那个结果（即使是 table 没有数据并用零填充的日子）如下所示：

day | count_items
------------------
1   | 0
2   | 2
3   | 3
4   | 0
5   | 0
6   | 0
...
31  | 0

产生这种结果的最佳方法是什么？我一直在考虑将 JOIN 与 table 一起使用，它会生成特定月份的所有日期，但 QuestDB 没有实现 LEFT JOIN。

是否存在使用 QuestDB 生成此类报告的简单方法？

Answer 1

你可以用左连接

SELECT x as day, COALESCE(sb.count_items, 0) AS count_items
from long_sequence(31) ls
LEFT JOIN (
   SELECT DAY(ts) as day, COUNT_DISTINCT(item) AS count_items
   FROM my_table
   WHERE ts IN '2020-01'
   SAMPLE BY d
) AS sb ON sb.day = ls.x

Answer 2

我已经修改了 Alex 的答案，添加了一个 CAST(DAY(ts) AS LONG)（如果没有这个，左连接条件中会出现类型错误）并使用 DAYS_IN_MONTH 函数：

SELECT x as day, COALESCE(sb.count_items, 0) AS count_items
from long_sequence(DAYS_IN_MONTH('2020-01') ls
LEFT JOIN (
   SELECT CAST(DAY(ts) AS LONG) as day, COUNT_DISTINCT(item) AS count_items
   FROM my_table
   WHERE ts IN '2020-01'
   SAMPLE BY d
) AS sb ON sb.day = ls.x

如何生成月度报告，用零填充一个月中缺少的几天

How to generate a monthly report filling missing days of the month with zero

questdb