Oracle 行轉列pivot 、列轉行unpivot 的Sql語句總結
多行轉字串
這個比較簡單,用||或concat函式可以實現
1. select concat(id,username) str from app_user
2.
3. select id||username str from app_user
字串轉多列
實際上就是拆分字串的問題,可以使用 substr、instr、regexp_substr函式方式
字串轉多行
使用union all函式等方式
首先讓我們來看看這個神奇的函式wm_concat(列名),該函式可以把列值以","號分隔起來,並顯示成一行,接下來上例子,看看這個神奇的函式如何應用準備測試
1. create table test(id number,name varchar2(20));
2.
3. insert into test values(1,'a');
4. insert into test values(1,'b');
5. insert into test values(1,'c');
6. insert into test values(2,'d');
7. insert into test values(2,'e');
效果1 : 行轉列,預設逗號隔開
1. select
效果2: 把結果裡的逗號替換成"|"
1. select replace(wm_concat(name),',','|') from test;
效果3: 按ID分組合並name
1. select id,wm_concat(name) name from test group by id;
sql語句等同於下面的sql語句
1. -------- 適用範圍:8i,9i,10g及以後版本( MAX + DECODE )
2. select
3. from (select id, name ,row_number() over(partition by id order by name) as rn from test) t group by id order by 1;
4.
5. -------- 適用範圍:8i,9i,10g及以後版本( ROW_NUMBER + LEAD )
6. select id, str from (select id,row_number() over(partition by id order by name) as rn,name || lead(',' || name, 1)
7. over(partition by id order by name) || lead(',' || name, 2) over(partition by id order by name) || lead(',' || name, 3)
8. over(partition by id order by name) as str from test) where rn = 1 order by 1;
9.
10. -------- 適用範圍:10g及以後版本( MODEL )
11. select id, substr(str, 2) str from test model return updated rows partition by(id) dimension by(row_number()
12. over(partition by id order by name) as rn) measures (cast(name as varchar2(20)) as str) rules upsert iterate(3)
13. until(presentv(str[iteration_number + 2], 1, 0)=0) (str[0] = str[0] || ',' || str[iteration_number + 1]) order by 1;
14.
15. -------- 適用範圍:8i,9i,10g及以後版本( MAX + DECODE )
16. select t.id id, max(substr(sys_connect_by_path(t.name, ','), 2)) str from (select id, name, row_number()
17. over(partition by id order by name) rn from test) t start with rn = 1 connect by rn = prior rn + 1 and id = prior id
18. group by t.id;
懶人擴充套件用法:
案例: 我要寫一個檢視,類似"create or replace view as select 欄位1,...欄位50 from tablename",基表有50多個欄位,要是靠手工寫太麻煩了,有沒有什麼簡便的方法? 當然有了,看我如果應用wm_concat來讓這個需求變簡單,假設我的APP_USER表中有(id,username,password,age)4個欄位。查詢結果如下
1. /** 這裡的表名預設區分大小寫 */
2. select 'create or replace view as select '|| wm_concat(column_name) || ' from APP_USER' sqlStr
3. from user_tab_columns where table_name='APP_USER';
利用系統表方式查詢
1. select * from user_tab_columns
在Oracle 11g中,Oracle 又增加了2個查詢:pivot(行轉列)和unpivot(列轉行)
參考:http://blog.csdn.NET/tianlesoftware/article/details/7060306、http://www.oracle.com/technetwork/cn/articles/11g-pivot-101924-zhs.html
pivot 列轉行
測試資料 (id,型別名稱,銷售數量),案例:根據水果的型別查詢出一條資料顯示出每種型別的銷售數量。
1. create table demo(id int,name varchar(20),nums int); ---- 建立表
2. insert into demo values(1, '蘋果', 1000);
3. insert into demo values(2, '蘋果', 2000);
4. insert into demo values(3, '蘋果', 4000);
5. insert into demo values(4, '橘子', 5000);
6. insert into demo values(5, '橘子', 3000);
7. insert into demo values(6, '葡萄', 3500);
8. insert into demo values(7, '芒果', 4200);
9. insert into demo values(8, '芒果', 5500);
分組查詢(當然這是不符合查詢一條資料的要求的)
1. select name, sum(nums) nums from demo group by name
行轉列查詢
1. select * from (select name, nums from demo) pivot (sum(nums) for name in ('蘋果' 蘋果, '橘子', '葡萄', '芒果'));
注意: pivot(聚合函式 for 列名 in(型別)),其中 in(‘’) 中可以指定別名,in中還可以指定子查詢,比如 select distinctcode from customers
當然也可以不使用pivot函式,等同於下列語句,只是程式碼比較長,容易理解
1. ------ 多項子查詢
2. select * from (select sum(nums) 蘋果 from demo where name='蘋果'),(select sum(nums) 橘子 from demo where name='橘子'),
3. (select sum(nums) 葡萄 from demo where name='葡萄'),(select sum(nums) 芒果 from demo where name='芒果');
4.
5. ------ decode 函式利用
6. select sum(decode(name,'蘋果',nums)) 蘋果, sum(decode(name,'橘子',nums)) 橘子,
7. sum(decode(name,'葡萄',nums)) 葡萄, sum(decode(name,'芒果',nums)) 芒果 from demo
unpivot 行轉列
顧名思義就是將多列轉換成1列中去案例:現在有一個水果表,記錄了4個季度的銷售數量,現在要將每種水果的每個季度的銷售情況用多行資料展示。
建立表和資料
1. create table Fruit(id int,name varchar(20), Q1 int, Q2 int, Q3 int, Q4 int);
2.
3. insert into Fruit values(1,'蘋果',1000,2000,3300,5000);
4. insert into Fruit values(2,'橘子',3000,3000,3200,1500);
5. insert into Fruit values(3,'香蕉',2500,3500,2200,2500);
6. insert into Fruit values(4,'葡萄',1500,2500,1200,3500);
7. select * from Fruit
列轉行查詢
1. select id , name, jidu, xiaoshou from Fruit unpivot (xiaoshou for jidu in (q1, q2, q3, q4) )
注意: unpivot沒有聚合函式,xiaoshou、jidu欄位也是臨時的變數
同樣不使用unpivot也可以實現同樣的效果,只是sql語句會很長,而且執行速度效率也沒有前者高
1. select id, name ,'Q1' jidu, (select q1 from fruit where id=f.id) xiaoshou from Fruit f
2. union
3. select id, name ,'Q2' jidu, (select q2 from fruit where id=f.id) xiaoshou from Fruit f
4. union
5. select id, name ,'Q3' jidu, (select q3 from fruit where id=f.id) xiaoshou from Fruit f
6. union
7. select id, name ,'Q4' jidu, (select q4 from fruit where id=f.id) xiaoshou from Fruit f