Question

我有一个查询，从2个子查询和左连接中选择。问题是左连接没有正确识别子查询，我不知道为什么。它无法识别dt.timestamp

SELECT dt.TIMESTAMP
    ,count(*)
    ,tc.total_cells - count(*)
FROM (
    SELECT DISTINCT TIMESTAMP
    FROM (
        SELECT to_char(trunc(TIMESTAMP, 'hh24'), 'dd/mm/yyyy hh24:mi') TIMESTAMP
        FROM t_date_time
        WHERE trunc(TIMESTAMP, 'dd') = to_date('16-mar-17', 'dd-mon-yy')
        ORDER BY TIMESTAMP DESC
        )
    ORDER BY to_date(TIMESTAMP, 'dd/mm/yyyy hh24:mi')
    ) dt,
  (
        SELECT count(DISTINCT cell) total_cells
        FROM t_ca_processed
        WHERE trunc(TIMESTAMP, 'dd') = trunc(to_date('16-mar-17', 'dd-mon-yy'), 'dd')
        ) tc
LEFT JOIN t_ca_processed p ON dt.timestamp = to_char(p.TIMESTAMP, 'dd/mm/yyyy hh24:mi')
GROUP BY dt.TIMESTAMP
ORDER BY to_date(dt.TIMESTAMP, 'dd/mm/yyyy hh24:mi')

Answer 1

**

SELECT dt.TIMESTAMP
    ,count(*)
    ,tc.total_cells - count(*)
FROM (
    SELECT DISTINCT TIMESTAMP
    FROM (
        SELECT to_char(trunc(TIMESTAMP, 'hh24'), 'dd/mm/yyyy hh24:mi') TIMESTAMP
        FROM t_date_time
        WHERE trunc(TIMESTAMP, 'dd') = to_date('16-mar-17', 'dd-mon-yy')
        ORDER BY TIMESTAMP DESC
        )
    ORDER BY to_date(TIMESTAMP, 'dd/mm/yyyy hh24:mi')
    ) dt
    left join 
  (
        SELECT count(DISTINCT cell) total_cells
        FROM t_ca_processed
        WHERE trunc(TIMESTAMP, 'dd') = trunc(to_date('16-mar-17', 'dd-mon-yy'), 'dd')
        ) tc
ON dt.timestamp = --(tc.something)--to_char(p.TIMESTAMP, 'dd/mm/yyyy hh24:mi')
GROUP BY dt.TIMESTAMP
ORDER BY to_date(dt.TIMESTAMP, 'dd/mm/yyyy hh24:mi')

**

并且您需要在TC查询中添加TIMESTAMP列，然后我们才能匹配DT和TC

Answer 2

编译器不喜欢混合连接符号（vs vs left join）

一种方法是使用公用表表达式（CTE）来实现交叉连接数据，然后在左连接中使用它。

WITH TC AS (
    SELECT DISTINCT TIMESTAMP
    FROM (
        SELECT to_char(trunc(TIMESTAMP, 'hh24'), 'dd/mm/yyyy hh24:mi') TIMESTAMP
        FROM t_date_time
        WHERE trunc(TIMESTAMP, 'dd') = to_date('16-mar-17', 'dd-mon-yy')
        ORDER BY TIMESTAMP DESC
        )
    ORDER BY to_date(TIMESTAMP, 'dd/mm/yyyy hh24:mi')
    ) dt,
  (
        SELECT count(DISTINCT cell) total_cells
        FROM t_ca_processed
        WHERE trunc(TIMESTAMP, 'dd') = trunc(to_date('16-mar-17', 'dd-mon-yy'), 'dd')
        )
SELECT dt.TIMESTAMP
    ,count(*)
    ,tc.total_cells - count(*)
FROM  tc
LEFT JOIN t_ca_processed p ON dt.timestamp = to_char(p.TIMESTAMP, 'dd/mm/yyyy hh24:mi')
GROUP BY dt.TIMESTAMP
ORDER BY to_date(dt.TIMESTAMP, 'dd/mm/yyyy hh24:mi')

sql从多个表中选择加入不起作用

2 个答案: