Question

从SQL Return every row even if Null values开始，我需要在按两列分组时选择与MAX（contractPrice）相关联的行数据。

我已经设置了sqlfiddle.com/#!6/170043/2，以证明问题是下面的代码返回重复项（如果重复最高价格，则返回第7行和第6行的sqlfiddle结果）。

    SELECT
        t1.MeasurableID,
        t1.EntityID,
        t1.ContractPrice,
        ID,
        UTCMatched,
        NumberOfContracts,
        CurrencyCode
    FROM
        contracts AS t1
    INNER JOIN (
        SELECT
            MeasurableID,
            EntityID,
            MAX (contractprice) AS MAXprice
        FROM
            contracts
        WHERE
            MeasurableID IN (2017, 2018)
        AND CurrencyCode IN ('CAD', 'BTC')
        GROUP BY
            MeasurableID,
            EntityID
    ) t2 ON t1.MeasurableID = t2.MeasurableID
    AND t1.EntityID = t2.EntityID
    AND t1.ContractPrice = t2.MAXprice
    WHERE
        t1.MeasurableID IN (2017, 2018)
    AND t1.CurrencyCode IN ('CAD', 'BTC')
    ORDER BY
        MeasurableID,
        EntityID,
        UTCMatched

Answer 1

有一些事情令人困惑或需要澄清：

您的代码为Sql-Server，但您向我们提供了MySql
派生表的使用似乎没有必要，事实上它使它有点混乱。

话虽如此，如果你想要MAX ContractPrice及其MAX ID，可以通过多种方式完成。还请注意我没有正确地聚合其余的列，这是不好的做法。只需将未聚合或分组的其余列包装在MAX（）中：

SELECT 
    t1.MeasurableID,
    t1.EntityID,
    MAX(t1.contractprice) AS MAXprice,
    MAX(t2.ID) as ID,
    MAX(t2.UTCMatched) as UTCMatched,
    MAX(t2.NumberOfContracts) AS NumberOfContracts,
    MAX(t2.CurrencyCode) AS CurrencyCode

FROM
    contracts t1
LEFT JOIN contracts t2 ON t1.MeasurableID = t2.MeasurableID
                          AND
                          t1.EntityID = t2.EntityID
                          AND
                          t1.ContractPrice = t2.ContractPrice

WHERE
    t1.MeasurableID IN (2017, 2018)
AND t1.CurrencyCode IN ('CAD', 'BTC')
GROUP BY
    t1.MeasurableID,
    t1.EntityID

另一种方法是，如果这实际上是SQL-Server，您可以使用ROW_NUMBER和派生表：

SELECT MeasurableID
       ,EntityID
       ,ContractPrice
       ,ID
       ,UTCMatched
       ,NumberOfContracts
       ,CurrencyCode
FROM
(SELECT *
        ,ROW_NUMBER() OVER(PARTITION BY EntityID ORDER BY ContractPrice, ID DESC ) rn
FROM Contracts t1) t1
WHERE t1.rn = 1

Answer 2

您的查询可以通过简化和“更正”：

   SELECT
        t1.MeasurableID,
        t1.EntityID,
        t1.ContractPrice,
        t1.ID,
        t1.UTCMatched,
        t1.NumberOfContracts,
        t1.CurrencyCode
    FROM contracts AS t1
    INNER JOIN contracts AS T2 ON (t1.MeasurableID = t2.MeasurableID AND MeasurableID IN (2017, 2018) AND CurrencyCode IN ('CAD', 'BTC'))
    ORDER BY
        t1.MeasurableID,
        t1.EntityID,
        t1.ContractPrice,
        t1.ID,
        t1.UTCMatched,
        t1.NumberOfContracts,
        t1.CurrencyCode
    HAVING t1.ContractPrice = MAX(T2.contractprice)

但我认为你仍然会有重复......我认为你的整个查询概念都是错误的。

Answer 3

P.S。我可以使用Row_Number（）消除重复项 - 按下面的代码：

        SELECT MeasurableID
             ,EntityID
             ,ContractPrice
             ,ID AS HighTradeID
             ,UTCMatched AS HighTradeUTC
             ,NumberOfContracts AS HighTradeNumberOfContracts
             ,CurrencyCode AS HighTradeCurrency
FROM
        (SELECT *
                        ,ROW_NUMBER() OVER(PARTITION BY MeasurableID, EntityID ORDER BY ContractPrice DESC, ID DESC ) RowNumber
        FROM Contracts t1
            WHERE
                MeasurableID IN (2017, 2018)
            AND CurrencyCode IN ('CAD', 'BTC')
        ) AS InnerSelect

WHERE InnerSelect.RowNumber = 1

SQL获取MAX（）的行值，按2列

3 个答案: