SQL -

Question

我目前使用下面的查詢來連接兩個表一起簡化聯合查詢：

select distinct book_id, clean_page_content as page_content, 'clean' as status, reg_date 
from book_db_clean_pages t 
where book_id = 'sarafi_book_5' 
and reg_date = '2015-08-26' 

UNION all 

select distinct book_id, dirty_page_content as page_content, 'dirty' as status, reg_date 
from book_db_dirty_pages e 
where book_id = 'sarafi_book_5' 
and reg_date = '2015-08-26' 

然而，當我想尋找一個不同的book_id我需要改變兩個分句（一個在上面，一個低於UNION）。同樣的原則適用於reg_date。所以我想知道是否有可能簡化我的查詢只是使用可能的命令，如加入或其他事情。原則上，我要尋找一個解決方案像這樣：

select distinct t.book_id, (t.clean_page_content or e.dirty_page_content) as page_content, 
     ('clean' or 'dirty') as status, t.reg_date 
from book_db_clean_pages t, book_db_dirty_pages e 
where t.book_id = e.book_id and t.reg_date = e.reg_date 
and t.book_id = 'sarafi_book_5' 
and t.reg_date = '2015-08-26' 

Answer 1

把你的工會成派生表：

select * 
from (
    select ... 
    from book_db_clean_pages t 
    -- no where clause 

    union all 

    select ... 
    from book_db_dirty_pages e 
    where t.book_id = e.book_id and t.reg_date = e.reg_date 
    -- no where clause for book_id and reg_date 

) t 
where book_id = 'sarafi_book_5' 
and reg_date = '2015-08-26'; 

所有現代DBMS將是足夠的智慧來推動的條件爲派生表，所以當你這樣做的時候不應該有性能損失。

---

順便說一句：如果book_id是主鍵（或唯一的列）的兩個表中，施加在內部選擇distinct不會移除任何東西。

Answer 2

你不能簡化查詢，因爲你已經寫了它。

你可以寫：

select distinct b.* 
from ((select book_id, clean_page_content as page_content, 'clean' as status, reg_date 
     from book_db_clean_pages t 
    ) union all 
     (select book_id, dirty_page_content as page_content, 'dirty' as status, reg_date 
     from book_db_dirty_pages e 
    ) 
    ) b 
where book_id = 'sarafi_book_5' and reg_date = '2015-08-26'; 

你沒有提到你正在使用的數據庫。如果子查詢可以使用索引，那麼您的初始查詢可能會更有效率。