问题
Long time listner, first time caller. I'm using Crystal Reports 2010.
I have daily trade information that I need to group together if the volume doesn't change. Here's what the data looks like.
Trade# BegDate EndDate Volume
1 1/1/2012 1/2/2012 500
1 1/2/2012 1/3/2012 500
1 1/3/2012 1/4/2012 1000
1 1/4/2012 1/5/2012 750
1 1/5/2012 1/6/2012 750
1 1/6/2012 1/7/2012 500
1 1/7/2012 1/8/2012 500
1 1/8/2012 1/9/2012 500
I need it to look like this.
Trade# DateRange Volume
1 1/1/2012 - 1/3/2012 500
1 1/3/2012 - 1/4/2012 1000
1 1/4/2012 - 1/6/2012 750
1 1/6/2012 - 1/9/2012 500
I need to group by the derived date ranges but I'm not sure how to accomplish this with Crystal. Any ideas??
回答1:
with w as (
select 1 id, to_date('1/1/2012', 'mm/dd/yyyy') start_date, to_date('1/2/2012', 'mm/dd/yyyy') end_date, 500 sales_volume from dual
union all
select 1, to_date('1/2/2012', 'mm/dd/yyyy'), to_date('1/3/2012', 'mm/dd/yyyy'), 500 from dual
union all
select 1, to_date('1/3/2012', 'mm/dd/yyyy'), to_date('1/4/2012', 'mm/dd/yyyy'), 1000 from dual
union all
select 1, to_date('1/4/2012', 'mm/dd/yyyy'), to_date('1/5/2012', 'mm/dd/yyyy'), 750 from dual
union all
select 1, to_date('1/5/2012', 'mm/dd/yyyy'), to_date('1/6/2012', 'mm/dd/yyyy'), 750 from dual
union all
select 1, to_date('1/6/2012', 'mm/dd/yyyy'), to_date('1/7/2012', 'mm/dd/yyyy'), 500 from dual
union all
select 1, to_date('1/7/2012', 'mm/dd/yyyy'), to_date('1/8/2012', 'mm/dd/yyyy'), 500 from dual
union all
select 1, to_date('1/8/2012', 'mm/dd/yyyy'), to_date('1/9/2012', 'mm/dd/yyyy'), 500 from dual
)
,t as (select sales_volume
,start_date
,end_date
,lag (sales_volume,1) over (order by start_date) prev_sales_volume
from w
order by start_date)
,u as (select *
from t
where nvl(prev_sales_volume,-1) != sales_volume
order by start_date)
select start_date
,nvl(lead (start_date,1) over (order by start_date),(select max(end_date) from w)) end_date
,sales_volume
from u
order by start_date
回答2:
I'd use the "X-2" suppression function to hide all but the last from each row, and Previous() and Next() to find the endpoints.
Group by BegDate; suppressing the details and group footer sections.
Create an {@UpdateCurrentBegDate} function somewhere in the group header
WhilePrintingRecords;
Global DateVar CurrentBegDate;
If PreviousIsNull({table.Volume}) or Previous({table.Volume}) <> {table.Volume}
Then ( //We found a new range
CurrentBegDate = {table.BegDate};
);
""; //Display nothing on screen
And a {@DisplayBegDate} function where you currently have BegDate being shown
EvaluateAfter({@UpdateCurrentBegDate});
Global DateVar CurrentBegDate;
Go to the Section Expert and click "X-2" next to the Suppress option (for the group header section). This is the formula for there
Not(NextIsNull({table.Volume}) or Next({table.Volume}) <> {table.Volume})
If you need to do totals or other calculations, you would do it in the {@UpdateCurrentBegDate} function (and you'd want to change the name to {@UpdateCurrentValues} or something similar). You can also create a new function that checks the Next() information if you only want to change things when the group changes -- using the default totalling functions will include the hidden values.
回答3:
with w as (
select 1 id, to_date('1/1/2012', 'mm/dd/yyyy') db, to_date('1/2/2012', 'mm/dd/yyyy') de, 500 s from dual
union all
select 1, to_date('1/2/2012', 'mm/dd/yyyy'), to_date('1/3/2012', 'mm/dd/yyyy'), 500 from dual
union all
select 1, to_date('1/3/2012', 'mm/dd/yyyy'), to_date('1/4/2012', 'mm/dd/yyyy'), 1000 from dual
union all
select 1, to_date('1/4/2012', 'mm/dd/yyyy'), to_date('1/5/2012', 'mm/dd/yyyy'), 750 from dual
union all
select 1, to_date('1/5/2012', 'mm/dd/yyyy'), to_date('1/6/2012', 'mm/dd/yyyy'), 750 from dual
union all
select 1, to_date('1/6/2012', 'mm/dd/yyyy'), to_date('1/7/2012', 'mm/dd/yyyy'), 500 from dual
union all
select 1, to_date('1/7/2012', 'mm/dd/yyyy'), to_date('1/8/2012', 'mm/dd/yyyy'), 500 from dual
union all
select 1, to_date('1/8/2012', 'mm/dd/yyyy'), to_date('1/9/2012', 'mm/dd/yyyy'), 510 from dual
)
select tmin.db, tmax.de, tmin.s
from
(
select
row_number() over (order by db) id,
db,
s
from
(
select
db,
s,
case
when ps is null
then 1
when ps != s
then row_number() over (order by db)
else 0 end num
from (
select
(db)
, (de)
, lag (s,1) over (ORDER BY db) ps
, s
from w
) t
) t1
where num != 0
) tmin,
(select
row_number() over (order by db) id,
de,
s
from
(
select
db,
de,
s,
case
when ps is null
then 1
when ps != s
then row_number() over (order by de desc)
else 0 end num
from (
select
db
,(de)
, lag (s,1) over (ORDER BY de desc) ps
, s
from w
order by db
) t
) t1
where num != 0) tmax
where tmin.id = tmax.id
回答4:
This is the most elegant solution I could come up with
WITH DirectTrades(tradeid, SourceDate, EndDate, Volume, Row, KillRow, Depth) AS
(
SELECT tradeid
BegDate AS SourceDate,
EndDate,
Volume,
ROW_NUMBER() over (partition by Table# order by BegDate) AS Row,
ROW_NUMBER() over (partition by Table# order by BegDate) AS KillRow,
0 AS Depth
FROM Trade
UNION ALL
SELECT t1.Tradeid
dt.SourceDate,
t1.EndDate,
t1.Volume,
dt.Row,
dt.Row + dt.Depth + 1,
dt.Depth + 1
FROM Trade AS t1
INNER JOIN
DirectTrades AS dt ON
t1.BegDate=dt.EndDate AND
t1.Volume=dt.Volume AND
t1.tradeid=dt.Tradeid
)
SELECT dt1.Tradeid
dt1.SourceDate,
dt1.EndDate,
dt1.Volume
FROM DirectTrades dt1
INNER JOIN
(
SELECT dt2.Row,
MAX(dt2.KillRow) AS KillRow
FROM DirectTrades dt2
WHERE dt2.Row NOT IN
(
SELECT dt3.KillRow
FROM DirectTrades dt3
WHERE dt3.Depth <> 0
)
GROUP BY dt2.Row
) dt4 ON dt1.Row=dt4.Row AND dt1.KillRow=dt4.KillRow
ORDER BY SourceDate
来源：https://stackoverflow.com/questions/13037749/crystal-reports-need-to-group-by-derived-date-range