Scramble the values of a column by replacing them with other values from the same domain

PHOTO EMBED

Sun Aug 07 2022 01:51:06 GMT+0000 (UTC)

Saved by @jmbenedetto #bash

//#########################################################################################//
/* -------------------------------------------------------------------

Name : Anon_Resampling

----------------------------------------------------------------------
Original Rule :	Replace with other values from the same domain:
1 - Table Name
2 - Field Name

-------------------------------------------------------------------*/

SUB Anon_Resampling (P_TABLENAME , P_FIELDNAME)


TRACE ##################################################;
TRACE ## Starting Function : Anon_Resampling  ##;
TRACE ## Anonymizing Field : $(P_FIELDNAME) #;
TRACE ##################################################;

//---------------------------------------//

[DistinctValues]:
Load Distinct 
[$(P_FIELDNAME)] as [OldDistinctValue],
RowNo() as [RowID],
Rand() as [Random]
Resident $(P_TABLENAME);

[AnonDistinctMapping]:
Mapping
Load
RowNo(),
[OldDistinctValue];
Load
[OldDistinctValue],
[Random]
Resident [DistinctValues]
Order By [Random];

[AnonDistinctValues]:
LOAD
*,
ApplyMap('AnonDistinctMapping',RowID,'Anon_Error') as [NewDistinctValue]
Resident DistinctValues;

Drop table DistinctValues;

[AnonMapping]:
Mapping
Load
[OldDistinctValue],
[NewDistinctValue]
Resident [AnonDistinctValues];

Drop table AnonDistinctValues;

[AnonValues]:
LOAD
*,
ApplyMap('AnonMapping',[$(P_FIELDNAME)],'Anon_Error') as [Anon_$(P_FIELDNAME)]
Resident $(P_TABLENAME);

Drop table $(P_TABLENAME);

Rename table AnonValues to $(P_TABLENAME);


END SUB

//#########################################################################################//
content_copyCOPY