重构 - 找到一个函数来添加列到 postgres 查询

Question

我希望将以下查询添加为 Rails 应用程序中的数据库视图：

select
  recruiters.full_name as recruiter,
  advisors.advisor as agent,
  sum(case when policies.week_number=1 then policies.premium else 0 end) as wk1,
  sum(case when policies.week_number=2 then policies.premium else 0 end) as wk2,
  sum(case when policies.week_number=3 then policies.premium else 0 end) as wk3,
  sum(case when policies.week_number=4 then policies.premium else 0 end) as wk4,
  sum(policies.premium) as total
from policies
  join advisors on policies.advisor_id=advisors.id
  join recruiters on advisors.recruiter_id=recruiters.id
where policies.current_status='T' or policies.current_status='I'
group by recruiters.full_name, advisors.advisor

在 select 语句中，我目前每周添加一个 sum 语句来创建新列（在代码中创建重复）。这将导致一整年的结果如下：

sum(case when policies.week_number=1 then policies.premium else 0 end) as wk1,
sum(case when policies.week_number=2 then policies.premium else 0 end) as wk2,
  ....
  ....
sum(case when policies.week_number=52 then policies.premium else 0 end) as wk52,
sum(case when policies.week_number=53 then policies.premium else 0 end) as wk53

结果如下：

Week 1: 
RECRUITER | AGENT      | WK1      
----------|------------|----------
MC        | DM         | 523.8    
MC        | BO         | 0.0      

Week 2:
RECRUITER | AGENT      | WK1      | WK2     
----------|------------|----------|---------
MC        | DM         | 523.8    | -540.0  
MC        | BO         | 0.0      | 0.0   

....

Week 4:
RECRUITER | AGENT      | WK1      | WK2     | WK3      | WK4      | TOTAL
----------|------------|----------|---------|----------|----------|---------
MC        | DM         | 523.8    | -540.0  | 358.44   | 510.0    | 852.24
MC        | BO         | 0.0      | 0.0     | 1543.72  | 0.0      | 1543.72

我想用类似于以下逻辑的东西重构代码：

1. 在数据集中找到可用周数的 uniq 集：

   => [1,2,3,4]
   

2. 每个招聘人员/代理配对的每周总和政策。溢价。在 Rails 中，我会尝试这样的方法将 select 语句附加到查询中：

   table_columns = []
   [1,2,3,4].each do |i|
     table_columns << sum(case when policies.week_number=#{i} then policies.premium else 0 end) as wk#{i}
   end
   

例如，当policies.week_number=5 的数据可用时，将添加一年中所有53 周的“wk5”列等等。

是否可以在 Postgres 中创建一个函数来适应这样的重构？

仅供参考..我尝试使用交叉表，但我也在为分组招聘人员计算小计。

Answer 1

好吧，根据你的例子，我确实编写了一个 plpgsql 函数，它根据输入变量折射并动态构建一个 sql 查询 - 在这种情况下：请求周数。

该查询还有很大的改进空间，因此它更易于阅读，但真是太糟糕了……这只是一个快速烹饪。

DROP FUNCTION if exists stack_test(integer[]);

CREATE OR REPLACE FUNCTION stack_test(integer[])
  RETURNS text AS
$BODY$
declare 
rec record;
sql text;

select_field_sum text;
core_crostab_1 text;
core_crostab_2 text;
ct_field_list text ;

begin

select_field_sum ='0.0';
core_crostab_1   ='';
core_crostab_2   ='';
ct_field_list    ='';

FOR rec IN select unnest(( $1)) i  order by 1
   loop
   ct_field_list     =  ct_field_list ||',wk' || (rec.i)::text  || ' numeric'  ;
   select_field_sum  =  select_field_sum ||  '+coalesce(wk' || (rec.i)::text|| ',0)' ;   
   end loop;

core_crostab_1  = 'select advisor_id, week_number,sum(premium) premium from policies  where week_number = any ( array['|| array_to_string($1, ',') ||'] ) and policies.current_status=''''T'''' or policies.current_status=''''I'''' group by advisor_id,week_number order by 1,2';
core_crostab_2  = 'select unnest((array['|| array_to_string($1, ',') ||']  )) order by 1';

sql  = 'select *, ('||select_field_sum||') premiumsum  from crosstab('''||core_crostab_1||''', '''||core_crostab_2||''') as ct(advisor_id integer '||  ct_field_list ||') ';
sql = 'select recruiters.full_name as requiter,advisors.advisor ,q1.* from ( '||sql||')q1 left join advisors on(q1.advisor_id= advisors.id )  left join recruiters on (advisors.recruiter_id = recruiters.id) order by 1,2';

return  sql::text;
 end;
$BODY$
  LANGUAGE plpgsql volatile
  COST 100;