MySQL离开加入子查询分组

Question

我有一个要求，我需要将数据分组到相等数量的 ob 行中。由于 mysql 没有 rownum() 我正在模拟这种行为：

SET @row:=6; 
SELECT MAX(agg.timestamp) AS timestamp, MAX(agg.value) AS value, COUNT(agg.value) AS count 
FROM 
    (
        SELECT timestamp, value, @row:=@row+1 AS row 
        FROM data 
        WHERE channel_id=52 AND timestamp >= 0 ORDER BY timestamp
    ) AS agg 
GROUP BY row div 8
ORDER BY timestamp ASC;

注意：根据Can grouped expressions be used with variable assignments?，此查询可能不是 100% 正确，但它确实有效。

另一个要求是计算分组集之间的行差异。我一直在寻找使用子查询连接同一个表的解决方案：

SET @row:=6; 
SELECT MAX(agg.timestamp) AS timestamp, MAX(agg.value) AS value, COUNT(agg.value) AS count 
FROM 
    (
        SELECT timestamp, value, @row:=@row+1 AS row 
        FROM data 
        WHERE channel_id=52 AND timestamp >= 0 ORDER BY timestamp
    ) AS agg 
LEFT JOIN data AS prev
    ON prev.channel_id = agg.channel_id
    AND prev.timestamp = (
        SELECT MAX(timestamp) 
        FROM data
        WHERE data.channel_id = agg.channel_id
            AND data.timestamp < MIN(agg.timestamp)
    )
GROUP BY row div 8
ORDER BY timestamp ASC;

不幸的是，错误：

Error Code: 1054. Unknown column 'agg.channel_id' in 'on clause'

知道如何编写这个查询吗？

Answer 1

您从未从您的 sbuquery 中选择 channel_id，因此它不会返回到父查询，因此是不可见的。试试

SELECT MAX(agg.timestamp) AS timestamp, MAX(agg.value) AS value, COUNT(agg.value) AS count 
FROM 
    (
        SELECT timestamp, value, @row:=@row+1 AS row, channel_id
                                                    ^^^^^^^^^^^^-- need this
        FROM data 

由于 MySQL 只查看和使用您从该子查询显式返回的字段，并且不会“深入”查询底层的表，因此您需要选择/返回您将使用父查询的所有字段.

Answer 2

这个版本怎么样：

SELECT MAX(agg.timestamp) AS timestamp, MAX(agg.value) AS value, COUNT(agg.value) AS count, COALESCE(prev.timestamp, 0) AS prev_timestamp
FROM (SELECT d.*, @row:=@row+1 AS row 
      FROM data d CROSS JOIN
           (select @row := 6) vars
      WHERE channel_id = 52 AND timestamp >= 0 ORDER BY timestamp
     )  agg LEFT JOIN
    data prev
    ON prev.channel_id = agg.channel_id AND
       prev.timestamp = (SELECT MAX(timestamp) 
                         FROM data
                         WHERE data.channel_id = agg.channel_id AND
                               data.timestamp < agg.timestamp
                        )
GROUP BY row div 8
ORDER BY timestamp ASC;

这包括子查询中的所有列。并将变量初始化放在同一个查询中。