SAP DS - Split a Field at Separator into Multiple Rows

[former_member188458]

Salut mes amis!

Prelude :

I recently started my adventure in SAP DS ( 4.2) and found some great things in the tool .  The most powerful being able to add Python code to do any kind of transformation . I read from many SAP DS experts that one should avoid using Python UDT transformation but in my use case I wanted to keep it very generic and Python came as a big rescue to make the code more flexible . One of the use case I will be discussing here , not the exact use case but a portion of it where I used Python Code . Hope this helps someone who is trying to achieve the same .

There are other options as you will see in the link below , but again I found python more flexible and without any dependency on xsd etc as in below post https://wiki.scn.sap.com/wiki/display/EIM/How+to+split+a+comma+separated+String+into+multiple+rows+v...

Use Case : 

Input looked like below :

Row_Id    PartNo  Qty  Unit    RefDes

1              Part1     3     EA      Ref11,Ref12,Ref13

2              Part2     1     EA .    <NULL>

3              Part3     5     EA      Ref31,Ref32,Ref33,Ref34,Ref35

 

Expected Output :

 

Row_Id    PartNo  Qty  Unit    RefDes

1              Part1     3     EA      Ref11

1              Part1     3     EA      Ref12

1              Part1     3     EA      Ref13

2              Part2     1     EA      <NULL>

3              Part3     5     EA      Ref31

3              Part3     5     EA      Ref32

3              Part3     5     EA      Ref33

3              Part3     5     EA      Ref34

3              Part3     5     EA     Ref35

 

Implementation :

 

In order to achieve above, implemented a UDT transform in SAP DS  . In the Query transform feeding the UDT kept only 2 fields - Row_Id and RefDes . In the UDT Output , kept again 2 outputs Row_Id and RefDes , which will now have split Ref Des values . Finally joined the input data with output from this UDT with a left outer join on Row_Id

Heres a snapshot to get a better idea



 

UDT Definittion :

• Select Mode - Per Collection - As we ll be adding new rows to the input collection


• Launch Python Editor


• Added new field in I/O fields - For RefDes as the input Ref Des could be longer than 255 e.g. 8000 in my case and output will be 255(Restriction in UDT - cannot be more than 255)

Here are a few snapshots again to get a better idea:





 

Algorithm :

1. Get the Whole collection or total number of records in the Input


2. Loop through the number of records to do below


3. Get Field value for Ref Des


4. If the Ref Des value has separator in my case ',' , split and add each split value as a new entry to the collection with Row Id and split Ref Des


5. If the Ref Des value has  no separator in my case ',' , no need to split , so add as is as a new entry to the collection with Row Id and Ref Des value


6. Free memory using DatManager.DeleteDataRecord and del

 

Code :

 

print('Start')

newRecord = DataManager.NewDataRecord()

#gets the number of records
numRecords = Collection.Size()
num_rows = numRecords
#iterate over the collection
for recordNum in range(1, numRecords + 1):
#get a record
Collection.GetRecord(newRecord, recordNum)

#set a field on the record

rowID=newRecord.GetField(u'Row_Id')
refDes=newRecord.GetField(u'RefDesignator')

refDesSplitArr = refDes.split(",")

if len(refDesSplitArr) > 1:

for i in range(len(refDesSplitArr)):
num_rows=num_rows+i
newRecord2 = DataManager.NewDataRecord(num_rows)
newRecord2.SetField(u'Row_Id', unicode(rowID))
newRecord2.SetField(u'Ref_Des', unicode(refDesSplitArr[i]))
Collection.AddRecord(newRecord2)
else:
num_rows=num_rows+1
newRecord2 = DataManager.NewDataRecord(num_rows)
newRecord2.SetField(u'Row_Id', unicode(rowID))
newRecord2.SetField(u'Ref_Des', unicode(refDes))
Collection.AddRecord(newRecord2)

DataManager.DeleteDataRecord(newRecord)
del newRecord

print('End')