我想仅使用 select 语句（无包或函数）在 SQL 中实现以下结果。 我该怎么做？

Question

I want to achieve following results in SQL using just select statement (no package or function). How can I do it? C is a calculation column ie C=A*B and it is dependent on A(where type = X)

OP      Typ     A       B       C   
10      X       0.5     2       1
10      X       2       4       8
10      X       1.5     2       3
10      Y       12      2       24      <----If Typ = Y then A = (1+8+3)
20      X       2       3       6
20      X       3       1.5     4.5
20      Y       46.6    1       46.5    <----If Typ = Y then A = (1+8+3+24+6+4.5)
30      X       1       1       1
30      X       2       1.5     3
30      Y       96      0.5     48      <----If Typ = Y then A = (1+8+3+24+6+4.5+46.5+1+3)

If Typ = X then A will be from database
Data is sorted by Op,Type ascending

Answer 1

You need SUM() OVER () Analytic function with conditional for the column A, assuming you have an ID column(Opr_seq) mentioned as operation sequence :

SELECT Op, Typ,
       CASE WHEN Typ='Y' 
            THEN 
                 SUM(C) OVER (ORDER BY Opr_seq) - C
            ELSE
                 A
            END AS A,
       B, C    
  FROM t
 ORDER BY Opr_seq 

Demo

Answer 2

SQL statements where rows reference rows, which reference rows, can be solved with either a recursive common table expression, a CONNECT BY statement, or a MODEL expression. A CONNECT BY solution would probably be smaller, and a MODEL solution would probably be faster, but both would also probably be even more cryptic, so below is a solution using recursive common table expressions.

with results(op, typ, a, b, c, runningTotal, rownumber) as
(
    select op, typ, a, b, c,
        case when typ = 'X' then c else 0 end runningTotal,
        rownumber
    from
    (
        select op, typ, a, b,
            case when typ = 'X' then a*b else null end c,
            row_number() over (order by op, typ) rownumber
        from sample_data
    )
    where rownumber = 1

    union all

    select
        next_row.op,
        next_row.typ,
        case when next_row.typ = 'Y' then runningTotal else next_row.a end a,
        next_row.b,
        case when next_row.typ = 'Y' then runningTotal else next_row.a end * next_row.b c,
        runningTotal + case when next_row.typ = 'Y' then runningTotal else next_row.a end * next_row.b runningTotal,
        next_row.rownumber
    from results
    join
    (
        select op, typ, a, b,
            case when typ = 'X' then a*b else null end c,
            row_number() over (order by op, typ) rownumber
        from sample_data
    ) next_row
    on results.rownumber+1 = next_row.rownumber
)
select op, typ, A, B, C from results;

I used t his sample table:

-- drop table sample_data;
create table sample_data as
select 10 op, 'X' Typ, 0.5  A, 2   B from dual union all  
select 10 op, 'X' Typ, 2    A, 4   B from dual union all  
select 10 op, 'X' Typ, 1.5  A, 2   B from dual union all
select 10 op, 'Y' Typ, null A, 2   B from dual union all
select 20 op, 'X' Typ, 2    A, 3   B from dual union all
select 20 op, 'X' Typ, 3    A, 1.5 B from dual union all
select 20 op, 'Y' Typ, null A, 1   B from dual union all
select 30 op, 'X' Typ, 1    A, 1   B from dual union all
select 30 op, 'X' Typ, 2    A, 1.5 B from dual union all
select 30 op, 'Y' Typ, null A, 0.5 B from dual;

My results are slightly different than yours, but I think that's only because your sample has a few minor addition errors. Here's a SQL Fiddle .