具有2個返回值的SELECT子查詢

Question

我想從子查詢中選擇多個列。在這裏我的小例子：

一個功能返回兩個值：

CREATE OR REPLACE FUNCTION dummy_function(my_text text) 
RETURNS TABLE (id Integer, remark text) AS $$ 
BEGIN 
RETURN QUERY SELECT 42, upper(my_text); 
END; 
$$ LANGUAGE plpgsql; 

我不工作查詢：

SELECT 
    id, 
    city_name, 
    dummy_function(city_name) 
FROM 
    (SELECT 1 as id, 'Paris' as city_name 
    UNION ALL 
    SELECT 2 as id, 'Barcelona' as city_name 
    ) AS dummy_table 

我錯誤的結果：

id | city_name | dummy_function 
----+-----------+---------------- 
    1 | Paris  | (42,PARIS) 
    2 | Barcelona | (42,BARCELONA) 

但我想有這樣結果：

id | city_name | number | new_text 
----+-----------+--------------------- 
    1 | Paris  | 42 | PARIS 
    2 | Barcelona | 42 | BARCELONA 

你知道如何不運行函數兩次實現這一目標？

Answer 1

的FROM子句中使用函數返回行（或一組行的）：

SELECT 
    dummy_table.id, 
    city_name, 
    dummy_function.id, 
    remark 
FROM 
    (SELECT 1 as id, 'Paris' as city_name 
    UNION ALL 
    SELECT 2 as id, 'Barcelona' as city_name 
    ) AS dummy_table, 
    LATERAL dummy_function(city_name) 

id | city_name | id | remark 
----+-----------+----+----------- 
    1 | Paris  | 42 | PARIS 
    2 | Barcelona | 42 | BARCELONA 
(2 rows)  

每the documentation：

出現在

表功能也可以由前面關鍵字LATERAL，但對於功能來說，關鍵字是可選的;在任何情況下，函數的參數都可以包含對前面FROM項目提供的列的引用。

Answer 2

SELECT 
    dummy_table.id, 
    city_name, 
    df.id as number, 
    df.remark as new_text 
FROM 
    (SELECT 1 as id, 'Paris' as city_name 
    UNION ALL 
    SELECT 2 as id, 'Barcelona' as city_name 
    ) AS dummy_table, 
    dummy_function(city_name) df