我有两张代表账单的表格 票据清单和票据内容各有其相关数据，应在逻辑上“连接”以获得适当的结果。示例表列名称非常具有自我描述性。
DROP TABLE IF EXISTS my_list;
CREATE TABLE my_list (indexl int PRIMARY KEY, docnum int, mytime text, rebate double precision, status text);
INSERT INTO my_list
(indexl, docnum, mytime, rebate, status) VALUES
(10, 5, '01.01.2014 15:20:31', 0, ''),
(11, 6, '02.01.2014 11:10:11', 10, ''),
(12, 7, '02.01.2014 09:15:01', 0, ''),
(14, 8, '03.01.2014 12:12:49', 12, ''),
(17, 9, '04.01.2014 08:19:19', 10, ''),
(18, 10, '04.01.2014 10:10:10', 0, 'S'),
(19, 11, '04.01.2014 01:04:14', 0, 'B'),
(21, 12, '05.01.2014 02:49:14', 0, ''),
(22, 13, '12.01.2014 08:12:17', 0, '');
DROP TABLE IF EXISTS my_content;
CREATE TABLE my_content (indexc int PRIMARY KEY, docnum int, code int,
aname text, price double precision, qty double precision, secondtax double precision, meas text);
INSERT INTO my_content
(indexc, docnum, code, aname, price, qty, secondtax, meas) VALUES
(10, 5, 587, 'spaghetti', 75, 1, 0, 'kg'),
(15, 6, 3432, 'salt', 3, 2, 0, 'kg'),
(16, 12, 32, 'olive oil', 4, 1.5, 5, 'kg'),
(29, 7, 3432, 'salt', 3, 1, 0, 'kg'),
(17, 6, 449, 'sugar', 5, 2, 0, 'kg'),
(18, 7, 1582, 'dried eggs', 50, 1.2, 0, 'kg'),
(19, 8, 210, 'tomato', 80, 5.5, 0, 'kg'),
(20, 9, 211, 'mustard', 5, 3, 5, 'kg'),
(22, 10, 2014, 'clove', 1, 1, 0, 'kg'),
(23, 9, 8, 'oil', 120, 4, 0, 'lit'),
(24, 11, 816, 'laurel', 4, 1, 0, 'kg'),
(25, 8, 1582, 'dried eggs', 10, 0.2, 0, 'kg'),
(26, 12, 32, 'olive oil', 4, 1, 0, 'kg'),
(28, 13, 67, 'corned beef', 40, 0.5, 0, 'kg');
为了分析这些账单，我做了一个几乎不错的查询，但确信它可以写得更好，更短，更优雅，更合适。
SELECT s.code,
s.aname,
SUM( s.qty) AS sumused,
SUM( s.price * s.qty) AS bruttoprice,
SUM(( s.price/100 * l.rebate) * s.qty) AS sumrebate,
SUM((s.price - s.price/100 * l.rebate) * s.qty) AS clearprice,
SUM((s.price - s.price/100 * l.rebate) * s.qty/100 * s.secondtax) AS sumsecondtax,
SUM((s.price - s.price/100 * l.rebate) * s.qty - (s.price - s.price/100 * l.rebate) * s.qty/100 * s.secondtax) AS sumwithoutsecondtax
FROM my_content s, my_list l
WHERE s.docnum = l.docnum
AND NOT l.status='S'
AND l.mytime BETWEEN '02.01.2014 00:00:00' AND '05.01.2014 23:59:59'
GROUP BY s.code, s.aname, l.status
ORDER BY sumused
1）是否可以使用变量替换查询'（s.price / 100 * l.rebate）* s.qty'中的表达式，以便不必一直写入？理想的情况是，如果我可以写例如SUM（clearprice * s.secondtax）
2）如果票据在列表中有's'，则必须跳过内容中的行，而不是l.status ='S'。但是，如果状态为“B”，则表示必须将具有该文档的值减去（不添加）到SUM。也许更优雅的解决方案是将这些行中的数量乘以0-qty值 在显示查询状态'B'被忽略，因为我不知道如何应用它。 怎么做？
3）我实际上只需要通过s.code对结果进行分组，但是我无法从GROUP BY中删除s.aname和l.status，因为那时查询不想工作。在实际情况下，如果我更改某些代码的名称，它将单独显示不需要的内容 是否可以仅通过代码对结果进行分组，但是结果中会显示s.aname（比如说最后一个）？
我尽我所能创建示例表并以简单/即时方式查询 请通过建议或/和示例帮助解决具体问题 感谢。
修改 我借助'kordirko'解决了我的疑问......
SELECT code,
MAX(aname) AS aname,
SUM(newqty) AS sumused,
SUM(price * newqty) AS bruttoprice,
SUM(crebate * newqty) AS sumrebate,
SUM(cprice * newqty) AS clearprice,
SUM(cprice * newqty/100 * secondtax) AS sumsecondtax,
SUM(cprice * newqty - cprice * newqty/100 * secondtax) AS sumwithoutsecondtax
FROM (
SELECT s.*, l.*,
s.price/100 * l.rebate AS crebate,
s.price - s.price/100 * l.rebate AS cprice,
CASE WHEN l.status = 'B' THEN 0 - s.qty ELSE s.qty END AS newqty
FROM my_content s, my_list l
WHERE s.docnum = l.docnum
AND NOT l.status='S'
AND l.mytime BETWEEN '02.01.2014 00:00:00' AND '05.01.2014 23:59:59'
) AS someAliasWhichhavetobehere
GROUP BY code
ORDER BY sumused;
一切似乎都没关系我只有结果值-0但我认为这不会导致进一步计算出现问题。如果我能怎样摆脱它？
答案 0 :(得分：2)
问题1
是的，可以使用子查询。
在子查询中计算变量some_variable，并以这种方式在外部查询中使用它：
SELECT code,
aname,
SUM( qty) AS sumused,
SUM( price * qty) AS bruttoprice,
SUM(( some_variable) * qty) AS sumrebate,
SUM((price - some_variable) * qty) AS clearprice,
SUM((price - some_variable) * qty/100 * secondtax) AS sumsecondtax,
SUM((price - some_variable) * qty - (price - some_variable) * qty/100 * secondtax) AS sumwithoutsecondtax
FROM (
SELECT s.*, l.*,
s.price/100 * l.rebate some_variable
FROM my_content s, my_list l
WHERE s.docnum = l.docnum
AND NOT l.status='S'
AND l.mytime BETWEEN '02.01.2014 00:00:00' AND '05.01.2014 23:59:59'
) as Alias
GROUP BY code, aname, status
ORDER BY sumused ;
请查看此处的第一个查询 - ＆gt; demo
问题2
请提供更多详情。目前尚不清楚哪些值必须减去。您的意思是price，或qty或整个表达，还是其他什么？
一般来说，这可以使用CASE ... WHEN .. THEN表达式来完成，例如：
SELECT ....
SUM ( CASE WHEN status = 'B'
THEN - price * qty
ELSE price * qty
END
) As column_alias,
....
FROM ....
或者也许是这样:(当status ='B'然后乘以-1，否则乘以1）：
SUM ( price * qty * CASE WHEN status = 'B' THEN -1 ELSE 1 END )
问题3
我会使用MAX或MIN函数，这是最简单的方法。他们检索一个随机名称（最大或最小）。
SELECT code,
max( aname ) aname,
SUM( qty) AS sumused,
SUM( price * qty) AS bruttoprice,
SUM(( some_variable) * qty) AS sumrebate,
SUM((price - some_variable) * qty) AS clearprice,
SUM((price - some_variable) * qty/100 * secondtax) AS sumsecondtax,
SUM((price - some_variable) * qty - (price - some_variable) * qty/100 * secondtax) AS sumwithoutsecondtax
FROM (
SELECT s.*, l.*,
s.price/100 * l.rebate some_variable
FROM my_content s, my_list l
WHERE s.docnum = l.docnum
AND NOT l.status='S'
AND l.mytime BETWEEN '02.01.2014 00:00:00' AND '05.01.2014 23:59:59'
) as Alias
GROUP BY code, status
ORDER BY sumused ;
请查看此处的第二个查询 - ＆gt; demo
问题4 - 它是什么：s.*和l.*
基本上这是来自Oracle SQL的我（坏？）习惯:)
每个人都知道这种语法：
SELECT * FROM sometable
*表示：给我表中的所有列。
假设我们想要获取所有列，但我们还想在结果集中显示计算值，显而易见的方法是：
SELECT *,
price * qty AS amount
FROM table
这在PostGeSQL中运行得很好，但不幸的是在Oracle中这会产生语法错误 Oracle强制我们使用表名或别名：
SELECT table.*,
price * qty AS amount
FROM table;
SELECT t.*,
price * qty AS amount
FROM table t
使用连接时相似 - 这项工作非常好：
SELECT *
FROM table1 JOIN table2 ON ....
但这会在Oracle中出现语法错误：
SELECT *,
price * qty AS amount
FROM table1 JOIN table2 ON ....
我们必须在这里使用表名或别名：
SELECT table1.*,
table2.*,
price * qty AS amount
FROM table1 JOIN table2 ON ....
SELECT t1.*,
t2.*,
price * qty AS amount
FROM table1 t1 JOIN table2 AS t2 ON ....
幸运的是，这符合ANSI SQL语法，并且应该适用于所有数据库：）
问题5 为什么将子查询结果命名为AS Alias
因为PostgeSQL需要为子查询指定一个名称（别名） 类似的MySql也需要别名：
SELECT ....
FROM(
subquery
) some_name
... or ....
SELECT ....
FROM(
subquery
) AS some_name
只有在Oracle中，子查询不会重新确定别名（但可能是），在Oracle中，这样可以正常工作：
SELECT ....
FROM (
subquery ....
)