排除行與相關值

Question

比方說，我有一個表，看起來像這樣：

++++++++++++++++++++++++++++++++++++++++ 
    url  | fieldname | value 
++++++++++++++++++++++++++++++++++++++++ 
    a,b,c  | RATE   | 45 
---------------------------------------- 
    a,b,c  | PLAY   | 5 
---------------------------------------- 
    a,b,c  | DATE   | 2013-05-20 
---------------------------------------- 
    d,e,f  | PLAY   | 2 
---------------------------------------- 
    d,e,f  | DATE   | 2013-01-01 
++++++++++++++++++++++++++++++++++++++++ 

，如何排除與在url列一個特定的（不詳）值的所有行，如果行之一與url相同的值在fieldname列中具有特定（已知！）值？

舉個例子，我會怎麼做，如果我想排除fieldname與RATE行，也是「相關」行（那些在url列相同的值） - 這樣它會返回一個表這樣?:

++++++++++++++++++++++++++++++++++++++++ 
    url  | fieldname | value 
++++++++++++++++++++++++++++++++++++++++ 
    d,e,f  | PLAY   | 2 
---------------------------------------- 
    d,e,f  | DATE   | 2013-01-01 
++++++++++++++++++++++++++++++++++++++++ 

再次，要記住的是，url值不應該在查詢中指定。

Answer 1

這是「sets-within-sets」查詢的示例。我喜歡使用having子句進行聚合，因爲它是最一般的方法。

下面的查詢讓你感興趣的網址：

select url 
from t 
group by url 
having sum(case when fieldname = 'RATE' then 1 else 0 end) = 0; 

要獲取原始表中的所有行，你需要加入回去吧：

select t.* 
from t join 
    (select url 
     from t 
     group by url 
     having sum(case when fieldname = 'RATE' then 1 else 0 end) = 0 
    ) turl 
    on t.url = turl.url