PostgreSQL函數如何返回數據集

背景： PostgreSQL里面沒有存儲過程，只有函數，其他數據庫里的這兩個對象在PG里都叫函數。 函數由函數頭，體和語言所組成，函數頭主要是函數的定義，變量的定義等，函數體主要是函數的實現，函數的語言是指該函數實現的方式，目前內置的有c,plpgsql,sql和internal,可以通過pg_language來查看當前DB支持的語言，也可以通過擴展來支持python等
函數返回值一般是類型，比如return int,varchar，返回結果集時就需要setof來表示。
一、數據准備

create table department(id int primary key, name text); create table employee(id int primary key, name text, salary int, departmentid int references department); insert into department values (1, 'Management'),(2, 'IT'),(3, 'BOSS'); insert into employee values (1, 'kenyon', 30000, 1); insert into employee values (2, 'francs', 50000, 1); insert into employee values (3, 'digoal', 60000, 2); insert into employee values (4, 'narutu', 120000, 3);

二、例子
1.sql一例

create or replace function f_get_employee() returns setof employee as $$ select * from employee; $$ language 'sql';

等同的另一個效果(Query)

create or replace function f_get_employee_query() returns setof employee as $$ begin return query select * from employee; end; $$ language plpgsql;

查詢圖解如下

postgres=# select * from f_get_employee(); id | name | salary | departmentid ----+--------+--------+-------------- 1 | kenyon | 30000 | 1 2 | francs | 50000 | 1 3 | digoal | 60000 | 2 4 | narutu | 120000 | 3 (4 rows)

查詢出來的函數還可以像普通的表一樣按條件查詢 ,但如果查詢的方式不一樣，則結果也不一樣，以下查詢方式將會得到類似數組的效果

postgres=# select f_get_employee(); f_get_employee --------------------- (1,kenyon,30000,1) (2,francs,50000,1) (3,digoal,60000,2) (4,narutu,120000,3) (4 rows)

因為返回的結果集類似一個表的數據集，PostgreSQL還支持對該函數執行結果進行條件判斷並過濾

postgres=# select * from f_get_employee() where id >3; id | name | salary | departmentid ----+--------+--------+-------------- 4 | narutu | 120000 | 3 (1 row)

上面的例子相對簡單，如果要返回不是表結構的數據集該怎么辦呢？看下面

2.返回指定結果集
a.用新建type來構造返回的結果集

--新建的type在有些圖形化工具界面中可能看不到，
要查找的話可以通過select * from pg_class where relkind='c'去查，c表示composite type

create type dept_salary as (departmentid int, totalsalary int); create or replace function f_dept_salary() returns setof dept_salary as $$ declare rec dept_salary%rowtype; begin for rec in select departmentid, sum(salary) as totalsalary from f_get_employee() group by departmentid loop return next rec; end loop; return; end; $$ language 'plpgsql';

b.用Out傳出的方式

create or replace function f_dept_salary_out(out o_dept text,out o_salary text) returns setof record as $$ declare v_rec record; begin for v_rec in select departmentid as dept_id, sum(salary) as total_salary from f_get_employee() group by departmentid loop o_dept:=v_rec.dept_id; o_salary:=v_rec.total_salary; return next; end loop; end; $$ language plpgsql;

執行結果:

postgres=# select * from f_dept_salary(); departmentid | totalsalary --------------+------------- 1 | 80000 3 | 120000 2 | 60000 (3 rows) postgres=# select * from f_dept_salary_out(); o_dept | o_salary --------+---------- 1 | 80000 3 | 120000 2 | 60000 (3 rows)

c.根據執行函數變量不同返回不同數據集

create or replace function f_get_rows(text) returns setof record as $$ declare rec record; begin for rec in EXECUTE 'select * from ' || $1 loop return next rec; end loop; return; end $$ language 'plpgsql';

執行結果：

postgres=# select * from f_get_rows('department') as dept(deptid int, deptname text); deptid | deptname --------+------------ 1 | Management 2 | IT 3 | BOSS (3 rows) postgres=# select * from f_get_rows('employee') as employee(employee_id int, employee_name text,employee_salary int,dept_id int); employee_id | employee_name | employee_salary | dept_id -------------+---------------+-----------------+--------- 1 | kenyon | 30000 | 1 2 | francs | 50000 | 1 3 | digoal | 60000 | 2 4 | narutu | 120000 | 3 (4 rows)

這樣同一個函數就可以返回不同的結果集了，很靈活。