与包含值的 nn 表交叉连接(透视)
我有 3 个表:
TABLE MyColumn (
ColumnId INT NOT NULL,
Label VARCHAR(80) NOT NULL,
PRIMARY KEY (ColumnId)
)
TABLE MyPeriod (
PeriodId CHAR(6) NOT NULL, -- format yyyyMM
Label VARCHAR(80) NOT NULL,
PRIMARY KEY (PeriodId)
)
TABLE MyValue (
ColumnId INT NOT NULL,
PeriodId CHAR(6) NOT NULL,
Amount DECIMAL(8, 4) NOT NULL,
PRIMARY KEY (ColumnId, PeriodId),
FOREIGN KEY (ColumnId) REFERENCES MyColumn(ColumnId),
FOREIGN KEY (PeriodId) REFERENCES MyPeriod(PeriodId)
)
MyValue 的行仅在提供实际值时创建。
我希望以表格方式获得结果,如下所示:
Column | Month 1 | Month 2 | Month 4 | Month 5 |
Potatoes | 25.00 | 5.00 | 1.60 | NULL |
Apples | 2.00 | 1.50 | NULL | NULL |
我已经成功创建了交叉联接:
SELECT
MyColumn.Label AS [Column],
MyPeriod.Label AS [Period],
ISNULL(MyValue.Amount, 0) AS [Value]
FROM
MyColumn CROSS JOIN
MyPeriod LEFT OUTER JOIN
MyValue ON (MyValue.ColumnId = MyColumn.ColumnId AND MyValue.PeriodId = MyPeriod.PeriodId)
或者,在 linq 中:
from p in MyPeriods
from c in MyColumns
join v in MyValues on new { c.ColumnId, p.PeriodId } equals new { v.ColumnId, v.PeriodId } into values
from nv in values.DefaultIfEmpty()
select new {
Column = c.Label,
Period = p.Label,
Value = nv.Amount
}
并了解了如何在 linq 中创建数据透视表(此处 或这里) :
(假设 MyDatas 是一个包含上一个查询结果的视图) :
from c in MyDatas
group c by c.Column into line
select new
{
Column = line.Key,
Month1 = line.Where(l => l.Period == "Month 1").Sum(l => l.Value),
Month2 = line.Where(l => l.Period == "Month 2").Sum(l => l.Value),
Month3 = line.Where(l => l.Period == "Month 3").Sum(l => l.Value),
Month4 = line.Where(l => l.Period == "Month 4").Sum(l => l.Value)
}
但我想找到一种方法来创建一个结果集,如果可能的话,使用 Month1, ... 动态属性。
注意:导致 n+1 查询的解决方案:
from c in MyDatas
group c by c.Column into line
select new
{
Column = line.Key,
Months =
from l in line
group l by l.Period into period
select new
{
Period = period.Key,
Amount = period.Sum(l => l.Value)
}
}
I have 3 tables :
TABLE MyColumn (
ColumnId INT NOT NULL,
Label VARCHAR(80) NOT NULL,
PRIMARY KEY (ColumnId)
)
TABLE MyPeriod (
PeriodId CHAR(6) NOT NULL, -- format yyyyMM
Label VARCHAR(80) NOT NULL,
PRIMARY KEY (PeriodId)
)
TABLE MyValue (
ColumnId INT NOT NULL,
PeriodId CHAR(6) NOT NULL,
Amount DECIMAL(8, 4) NOT NULL,
PRIMARY KEY (ColumnId, PeriodId),
FOREIGN KEY (ColumnId) REFERENCES MyColumn(ColumnId),
FOREIGN KEY (PeriodId) REFERENCES MyPeriod(PeriodId)
)
MyValue's rows are only created when a real value is provided.
I want my results in a tabular way, as :
Column | Month 1 | Month 2 | Month 4 | Month 5 |
Potatoes | 25.00 | 5.00 | 1.60 | NULL |
Apples | 2.00 | 1.50 | NULL | NULL |
I have successfully created a cross-join :
SELECT
MyColumn.Label AS [Column],
MyPeriod.Label AS [Period],
ISNULL(MyValue.Amount, 0) AS [Value]
FROM
MyColumn CROSS JOIN
MyPeriod LEFT OUTER JOIN
MyValue ON (MyValue.ColumnId = MyColumn.ColumnId AND MyValue.PeriodId = MyPeriod.PeriodId)
Or, in linq :
from p in MyPeriods
from c in MyColumns
join v in MyValues on new { c.ColumnId, p.PeriodId } equals new { v.ColumnId, v.PeriodId } into values
from nv in values.DefaultIfEmpty()
select new {
Column = c.Label,
Period = p.Label,
Value = nv.Amount
}
And seen how to create a pivot in linq (here or here) :
(assuming MyDatas is a view with the result of the previous query) :
from c in MyDatas
group c by c.Column into line
select new
{
Column = line.Key,
Month1 = line.Where(l => l.Period == "Month 1").Sum(l => l.Value),
Month2 = line.Where(l => l.Period == "Month 2").Sum(l => l.Value),
Month3 = line.Where(l => l.Period == "Month 3").Sum(l => l.Value),
Month4 = line.Where(l => l.Period == "Month 4").Sum(l => l.Value)
}
But I want to find a way to create a resultset with, if possible, Month1, ... properties dynamic.
Note : A solution which results in a n+1 query :
from c in MyDatas
group c by c.Column into line
select new
{
Column = line.Key,
Months =
from l in line
group l by l.Period into period
select new
{
Period = period.Key,
Amount = period.Sum(l => l.Value)
}
}
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(1)
您是否有日历期间 (1-12) 或会计期间(1-12 或 1-13,具体取决于年份)?
您只会运行该报告一年吗?
如果可能的话,我会将PeriodId 分成“年份”和“年份”两个字段。 时期。 这将使选择特定年份变得更加容易。 查询会更简单。
Do you have calendar periods (1-12) or fiscal periods (1-12 or 1-13 depending on the year)?
Would you only run the report for one year?
If possible, I would break the PeriodId into two fields of Year & Period. This will make it easier to select for a specific year. The queries will be simpler.