Ranking by partition in visual foxpro - sql

Question

I have the following table that looks like
+ --- + --- +
| AID | Tag |
+ --- + --- +
| 1 | 1 |
| 2 | 2 |
| 2 | 3 |
| 2 | 820 |
| 2 | 821 |
| 3 | 2 |
| 4 | 5 |
| 4 | 18 |
| 4 | 2744|
| 4 | 2745|
+ --- + --- +
When I write the following SQL Server 2008 code
select AID,
Tag,
RANK() over (partition by AID order by Tag asc) as rank
from My_Table
it produces the following results
+ --- + --- + ---- +
| AID | Tag | Rank |
+ --- + --- + ---- +
| 1 | 1 | 1 |
| 2 | 2 | 1 |
| 2 | 3 | 2 |
| 2 | 820 | 3 |
| 2 | 821 | 4 |
| 3 | 2 | 1 |
| 4 | 5 | 1 |
| 4 | 18 | 2 |
| 4 | 2744| 3 |
| 4 | 2745| 4 |
+ --- + --- + ---- +
which is exactly what I want.
Now, I want to write the same thing in Visual FoxPro 9 SQL. I tried it using recno() as demonstrated here; this numbers my records, but doesn't seem to support the ability to partition, and correlated subqueries don't seem to be supported in VFP 9 SQL. I know that I could do this with cursors and scans, but I don't want to do it that way. Any suggestions?

Answer 1

In VFP there is not a rank() function. However, you can achieve the same effect in a number of ways. One way is simple scan...endscan pass updating the ranking value as in the following example:
*** Sample Data
Create Cursor mytable ( AID Int, Tag Int)
Insert Into mytable Values (1,1 )
Insert Into mytable Values (2,2 )
Insert Into mytable Values (2,3 )
Insert Into mytable Values (2,820 )
Insert Into mytable Values (2,821 )
Insert Into mytable Values (3,2 )
Insert Into mytable Values (4,5 )
Insert Into mytable Values (4,18 )
Insert Into mytable Values (4,2744)
Insert Into mytable Values (4,2745)
*** Sample Data
Select AID, Tag, Cast(0 As Int) As rank ;
from mytable ;
order By AID, Tag ;
into Cursor crsRanked ;
readwrite
Scan
AID = AID
rcno = Recno()
Replace rank With Recno()-m.rcno+1 While AID = m.AID
Skip -1
Endscan
Locate
Browse
EDIT: Yesterday I overlooked how MS SQL server's RANK() function work, sorry. Here is one that work like MS SQL Server's Rank(), Dense_Rank(), Row_number():
Create Cursor mytable ( AID Int, Tag Int)
Insert Into mytable Values (1,1 )
Insert Into mytable Values (2,2 )
Insert Into mytable Values (2,3 )
Insert Into mytable Values (2,820 )
Insert Into mytable Values (2,821 )
Insert Into mytable Values (3,2 )
Insert Into mytable Values (4,5 )
Insert Into mytable Values (4,18 )
Insert Into mytable Values (4,18 )
Insert Into mytable Values (4,18 )
Insert Into mytable Values (4,2744)
Insert Into mytable Values (4,2745)
Select AID, Tag, ;
Cast(0 As Int) As rownum, ;
Cast(0 As Int) As rank, ;
Cast(0 As Int) As denserank ;
from mytable ;
order By AID, Tag ;
into Cursor crsRanked ;
readwrite
Local AID,rank,denserank,nextrank,rcno
Scan
AID = AID
rank = 0
nextrank = 0
denserank = 0
rcno = Recno()
Scan While m.AID = AID
Tag = Tag
rank = nextrank + 1
denserank = m.denserank + 1
Replace ;
rank With m.rank, ;
denserank With m.denserank, ;
rownum With Recno()-m.rcno+1 ;
While AID = m.AID And Tag = m.Tag
nextrank = m.nextrank + _Tally
Skip -1
Endscan
Skip -1
Endscan
Locate
Browse

Answer 2

I discovered the answer, for anyone who cares to know. The following SQL code is supported in Visual FoxPro 9.0 and will do what we want.
select t1.aid, ;
t1.tag, ;
count(*) as rank ;
from my_table t1 ;
inner join my_table t2 ;
on t2.aid = t1.aid ;
and t2.tag <= t1.tag ;
group by t1.aid, t1.tag
To see why, let's take a closer look at the inner join by leaving out the aggregate and including the tags from t2.
select t1.aid, ;
t1.tag, ;
t2.tag ;
from my_table t1 ;
inner join my_table t2 ;
on t2.aid = t1.aid ;
and t2.tag <= t1.tag ;
order by t1.aid, t1.tag
This code produces a table like
+ --- + ---- + ---- +
| AID | Tag1 | Tag2 |
+ --- + ---- + ---- +
| 1 | 1 | 1 |
| 2 | 2 | 2 |
| 2 | 3 | 2 |
| 2 | 3 | 3 |
| 2 | 820 | 2 |
| 2 | 820 | 3 |
| 2 | 820 | 820 |
| 2 | 821 | 2 |
| 2 | 821 | 3 |
| 2 | 821 | 820 |
| 2 | 821 | 821 |
| 3 | 2 | 2 |
| 4 | 5 | 5 |
| 4 | 18 | 5 |
| 4 | 18 | 18 |
| 4 | 2744 | 5 |
| 4 | 2744 | 18 |
| 4 | 2744 | 2744 |
| 4 | 2745 | 5 |
| 4 | 2745 | 18 |
| 4 | 2745 | 2744 |
| 4 | 2745 | 2745 |
+ --- + ---- + ---- +
We don't actually care about the data in Tag2, but now we can clearly see that the rank is the count of the Tag1 grouped by Aid and Tag1.