在MySQL

Question

「無效使用組功能的」使用MySQL，我的數據庫看起來是這樣的：

+-----------------------+--------------------+------+--------+ 
|  moviesInfo  |     |  |  | 
+-----------------------+--------------------+------+--------+ 
| title     | director   | year | oscars | 
+-----------------------+--------------------+------+--------+ 
| BoBoiBoy: The Movie | Nizam Razak  | 2016 | 0  | 
| Inception    | Christopher Nolan | 2010 | 4  | 
| Interstellar   | Christopher Nolan | 2014 | 1  | 
| Munna Bhai M.B.B.S. | Rajkumar Hirani | 2003 | 0  | 
| My Dear Brother  | Ertem Egilmez  | 1973 | 0  | 
| Rocky     | John G. Avildsen | 1976 | 3  | 
| The Nights of Cabiria | Federico Fellini | 1957 | 1  | 
| The Sixth Sense  | M. Night Shyamalan | 1999 | 6  | 
| Tokyo Story   | Yasujirô Ozu  | 1953 | 0  | 
| Yojimbo    | Akira Kurosawa  | 1961 | 1  | 
+-----------------------+--------------------+------+--------+ 


select director from moviesInfo 
where MAX(year) > 2000 and SUM(oscars) > 2 
group by director; 

通過上述查詢我試圖讓誰後來有一部電影在2000年，有更多的導演比2 oscars，但我得到一個錯誤

Invalid use of group function 

仍然是MySQL的新手我不完全理解這一點。

預期輸出將是

+-------------------+ 
|  director  | 
+-------------------+ 
| Christopher Nolan | 
+-------------------+ 

Answer 1

的GROUP BY aggregate functions不能WHERE子句中使用。他們應該停留在HAVING子句中，在GROUP BY之後，因爲它們在分組生成的值集上運行。

您的查詢應該是：

SELECT director 
FROM moviesInfo 
GROUP BY director 
HAVING MAX(year) > 2000 AND SUM(oscars) > 2 

更新

在OP的要求，這，大概，怎麼上面的SQL查詢被執行：

1. 的GROUP BY條款告知服務器對匹配的行進行分組（所有行爲 ，因爲沒有WHERE子句可對其進行過濾）。所述組如下：

   +-----------------------+--------------------+------+--------+ 
   | title     | director   | year | oscars | 
   +-----------------------+--------------------+------+--------+ 
   | BoBoiBoy: The Movie | Nizam Razak  | 2016 | 0  | 
   +-----------------------+--------------------+------+--------+ 
   | Inception    | Christopher Nolan | 2010 | 4  | 
   | Interstellar   | Christopher Nolan | 2014 | 1  | 
   +-----------------------+--------------------+------+--------+ 
   | Munna Bhai M.B.B.S. | Rajkumar Hirani | 2003 | 0  | 
   +-----------------------+--------------------+------+--------+ 
   | My Dear Brother  | Ertem Egilmez  | 1973 | 0  | 
   +-----------------------+--------------------+------+--------+ 
   | Rocky     | John G. Avildsen | 1976 | 3  | 
   +-----------------------+--------------------+------+--------+ 
   | The Nights of Cabiria | Federico Fellini | 1957 | 1  | 
   +-----------------------+--------------------+------+--------+ 
   | The Sixth Sense  | M. Night Shyamalan | 1999 | 6  | 
   +-----------------------+--------------------+------+--------+ 
   | Tokyo Story   | Yasujirô Ozu  | 1953 | 0  | 
   +-----------------------+--------------------+------+--------+ 
   | Yojimbo    | Akira Kurosawa  | 1961 | 1  | 
   +-----------------------+--------------------+------+--------+ 
   

2. HAVING的子句使得計算MAX(year)和SUM(oscars)爲每個組。該數據現在看起來像：

   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | title     | director   | year | oscars | MAX(year) | SUM(oscars) | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | BoBoiBoy: The Movie | Nizam Razak  | 2016 | 0  | 2016  | 0   | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | Inception    | Christopher Nolan | 2010 | 4  | 2014  | 5   | 
   | Interstellar   | Christopher Nolan | 2014 | 1  |   |    | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | Munna Bhai M.B.B.S. | Rajkumar Hirani | 2003 | 0  | 2003  | 0   | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | My Dear Brother  | Ertem Egilmez  | 1973 | 0  | 1973  | 0   | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | Rocky     | John G. Avildsen | 1976 | 3  | 1976  | 3   | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | The Nights of Cabiria | Federico Fellini | 1957 | 1  | 1957  | 1   | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | The Sixth Sense  | M. Night Shyamalan | 1999 | 6  | 1999  | 6   | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | Tokyo Story   | Yasujirô Ozu  | 1953 | 0  | 1953  | 0   | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | Yojimbo    | Akira Kurosawa  | 1961 | 1  | 1961  | 1   | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   

3. 接下來，HAVING的條件進行評估和數據集的樣子：

   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | title     | director   | year | oscars | MAX(year) | SUM(oscars) | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   | Inception    | Christopher Nolan | 2010 | 4  | 2014  | 5   | 
   | Interstellar   | Christopher Nolan | 2014 | 1  |   |    | 
   +-----------------------+--------------------+------+--------+-----------+-------------+ 
   

4. 現在SELECT子句用於生成結果集你期望。

請注意，上面的解釋是GROUP BY查詢背後的理論模型。在現實世界中，MySQL（和其他RDBMS）以不同的順序混合並運行上述步驟。它結合了WHERE和HAVING子句（如果可能），一次處理一行，並嘗試儘可能多地過濾儘可能多的行，以便儘快完成處理。

備註

的SELECT子句上的GROUP BY查詢can contain only expressions在於：

1. 也存在於GROUP BY子句中;
   例如，director;
2. 是GROUP BY aggregate functions;
   例如SUM(oscars),COUNT(*),AVG(year)等。沒有要求它們存在於其他條款中;
3. 在功能上依賴於存在於GROUP BY子句中的列;例如，
   例如，如果GROUP BY子句包含表的PK列，那麼SELECT子句中允許該表的任何列，因爲組中存在的所有行將具有該表的所有列的相同值。

讓我們此言的列title，year和oscars不能成爲SELECT子句中，因爲不符合上述任何條件。將它們添加到SELECT子句中會生成無效的SQL。原因很明顯：第二組中有title有兩個不同的值。哪一個可以選擇？

在版本5.7.5之前，MySQL用於接受這種無效的SQL查詢，但保留其對return indeterminate results的權利。這意味着如果對不同計算機上的相同數據集執行查詢（或從備份中恢復數據之後），結果可能會更改。

從版本5.7.5開始，MySQL拒絕無效的GROUP BY查詢。