有没有办法只显示第一个非零值，但如果所有值都为零，则在连接表的单独表中显示列的第一个实例

Question

这两个表相当大，我的 select 语句有更多我获得的值，但我认为我可以简化这些数据和查询，以便回答我的问题。

这是我的选择语句：

SELECT invoice.InvoiceNum, Layer, InvoiceItemNum 
FROM (INVOICE 
        left outer join InvoiceItem item 
            ON item.InvoiceNum = Invoice.InvoiceNum
    ) 
ORDER BY invoice.InvoiceNum

所以我有两张桌子。 Invoice 表和 InvoiceItem 表。它们由每个表中的 InvoiceNum 列连接起来，并显示 InvoiceNum 和 Layer 列

以下是此查询的结果：

InvoiceNum  | Layer | InvoiceItemNum
1           | 10    | 1
1           | 0     | 2
1           | 7     | 3 
1           | 0     | 4 
2           | 0     | 1
2           | 3     | 2 
3           | 0     | 1
3           | 0     | 2
3           | 0     | 3 
4           | 0     | 1
4           | 0     | 2
4           | 5     | 3

由于我的 InvoiceItem 表有多行可以分配给 1 个 InvoiceNum，这导致我的结果中出现重复的 InvoiceNum，这是我不希望的。

这是我试图获得的结果，仅列出发票表中的 1 个发票编号，其中第一种情况是来自 InvoiceItem 表的图层列的非零值，如果没有非零值，则列出第一个零。

尝试这样的事情：

InvoiceNum  | Layer | InvoiceItemNum
1           | 10    | 1
2           | 3     | 2 
3           | 0     | 1
4           | 5     | 3

我只是不确定如何执行此操作，或者鉴于它们位于两个不同的表上，这是否可能。

Answer 1

假设/理解：

• 虽然用 sybase 标记，但问题并没有区分 4x 不同的 Sybase RDBMS 产品（ASE、SQLAnywhere、IQ、Advantage），因此我将坚持使用通用 SQL 语法（即 4x 产品具有不同的SQL 方言；此外，ASE 不支持 CTE）
• 我不明白 OP 对 left (outer) join 的使用，因为提供的输出似乎没有表明 InvoiceItem 中存在任何“缺失”行
• 不清楚 Layer 和 InvoiceItemNum 列属于哪个表，因此我假设它们属于 InvoiceItem

猜测一组最小表定义和关联的 insert 语句：

create table Invoice
(InvoiceNum     int
)

create table InvoiceItem
(InvoiceNum     int
,InvoiceItemNum int
,Layer          int
)

insert Invoice select 1 union all select 2 union all select 3 union all select 4

insert InvoiceItem values (1,1,10)
insert InvoiceItem values (1,2,0)
insert InvoiceItem values (1,3,7)
insert InvoiceItem values (1,4,0)

insert InvoiceItem values (2,1,0)
insert InvoiceItem values (2,2,3)

insert InvoiceItem values (3,1,0)
insert InvoiceItem values (3,2,0)
insert InvoiceItem values (3,3,0)

insert InvoiceItem values (4,1,0)
insert InvoiceItem values (4,2,0)
insert InvoiceItem values (4,3,5)

生成OP当前输出的查询：

select  inv.InvoiceNum,
        item.Layer,
        item.InvoiceItemNum
from    Invoice inv
left                          -- superfluous in this case?
join    InvoiceItem item
on      inv.InvoiceNum = item.InvoiceNum
order by 1,3

 InvoiceNum  Layer       InvoiceItemNum
 ----------- ----------- --------------
           1          10              1
           1           0              2
           1           7              3
           1           0              4
           2           0              1
           2           3              2
           3           0              1
           3           0              2
           3           0              3
           4           0              1
           4           0              2
           4           5              3

生成 OP 所需输出的几个不同（复杂、混乱）的想法：

-- join based on Layer!=0; if no rows found then override NULLs
-- with Layer=0 and InvoiceItemNum=min(InvoiceItemNum) where Layer=0;
-- needs more work in case there are no matching rows in InvoiceItem ...
-- wrap case/then in a coalesce() and set to, what, 0?

select  inv.InvoiceNum,
        coalesce(item1.Layer,0) as "Layer",
        case    when item1.InvoiceItemNum is NULL
                then (select min(InvoiceItemNum) from InvoiceItem item3 where item3.InvoiceNum = inv.InvoiceNum)
                else item1.InvoiceItemNum
        end as "InvoiceItemNum"

from    Invoice inv
left
join    InvoiceItem item1
on      inv.InvoiceNum = item1.InvoiceNum
and     item1.Layer != 0
and     not exists(select       1
                   from         InvoiceItem item2
                   where        item2.InvoiceNum = item1.InvoiceNum
                   and          item2.Layer != 0
                   and          item2.InvoiceItemNum < item1.InvoiceItemNum)
order by 1

-- OR

-- perform a mutually exclusive UNION of Layer!=0 and Layer=0 queries
-- with Layer!=0 having prcedence

select  inv.InvoiceNum,
        item1.Layer,
        item1.InvoiceItemNum

from    Invoice inv
--left  ??? needs work if this is really an outer join ???
join    InvoiceItem item1
on      inv.InvoiceNum = item1.InvoiceNum

and     (
         (      item1.Layer != 0
          and   not exists(select 1
                           from   InvoiceItem item2
                           where  item2.InvoiceNum = item1.InvoiceNum
                           and    item2.Layer != 0
                           and    item2.InvoiceItemNum < item1.InvoiceItemNum)
         )

         or

         (      item1.Layer = 0
          and   not exists(select 1
                           from   InvoiceItem item3
                           where  item3.InvoiceNum = item1.InvoiceNum
                           and    item3.Layer != 0)
          and   not exists(select 1
                           from   InvoiceItem item4
                           where  item4.InvoiceNum = item1.InvoiceNum
                           and    item4.Layer = 0
                           and    item4.InvoiceItemNum < item1.InvoiceItemNum)
         )
        )
order by 1

这两者都会生成：

InvoiceNum  Layer       InvoiceItemNum
 ----------- ----------- --------------
           1          10              1
           2           3              2
           3           0              1
           4           5              3

注释：

• 不确定输出应该是什么，因为它（对我来说）没有出现，OP已经证明了需要 left （外部）join
• 在 (Sybase)SAP ASE 16.0 中测试的所有查询

Answer 2

这个问题有点棘手：

在 Postgres 中尝试一下：

with cte as (
select 
inv.invoicenum,sum(layer::int) "sum_layer"
from invoice inv inner join invoiceitem item on item.invoicenum=inv.invoicenum
group by 1
)
,
cte1 as (

select distinct on (inv.invoicenum) inv.invoicenum,layer, InvoiceItemNum

from invoice inv inner join invoiceitem item on item.invoicenum=inv.invoicenum
where inv.invoicenum in (select invoicenum from cte where sum_layer=0)
order by inv.invoicenum, InvoiceItemNum
),
cte2 as (
select 
distinct on (inv.invoicenum) inv.invoicenum, layer , InvoiceItemNum
from invoice inv inner join invoiceitem item on item.invoicenum=inv.invoicenum
where inv.invoicenum in (select invoicenum from cte where sum_layer>0) and layer::int>0
order by inv.invoicenum, InvoiceItemNum
)
(
select * from cte1
union all
select * from cte2
)
order by 1

演示

在 MySQL 8 中：

试试这个：

with cte as (
select 
inv.invoicenum,sum(layer) "sum_layer"
from invoice inv inner join invoiceitem item on item.invoicenum=inv.invoicenum
group by 1
)
,
cte1 as (
select * from (
select inv.invoicenum, layer, InvoiceItemNum
from invoice inv inner join invoiceitem item on item.invoicenum=inv.invoicenum
where inv.invoicenum in (select invoicenum from cte where sum_layer=0)
order by inv.invoicenum, InvoiceItemNum
) c
group by invoicenum
),
cte2 as (
select * from (
select inv.invoicenum, layer, InvoiceItemNum
from invoice inv inner join invoiceitem item on item.invoicenum=inv.invoicenum
where inv.invoicenum in (select invoicenum from cte where sum_layer>0) and layer>0
order by inv.invoicenum, InvoiceItemNum ) c
group by invoicenum
)
(
select * from cte1
union all
select * from cte2
)