Support Dutch BSN in numeric columns

[fbascheper]

Anonimatron version: 1.9.3

Currently anonimatron generates an Exception if a Dutch BSN is configured for a numeric column in a database table, like so:

Exception in thread "main" java.lang.RuntimeException: java.lang.ClassCastException: java.math.BigDecimal incompatible with java.lang.String
        at com.rolfje.anonimatron.jdbc.JdbcAnonymizerService.processTableColumns(JdbcAnonymizerService.java:239)
        at com.rolfje.anonimatron.jdbc.JdbcAnonymizerService.anonymizeTableInPlace(JdbcAnonymizerService.java:155)
        at com.rolfje.anonimatron.jdbc.JdbcAnonymizerService.anonymize(JdbcAnonymizerService.java:94)
        at com.rolfje.anonimatron.Anonimatron.anonymize(Anonimatron.java:99)
        at com.rolfje.anonimatron.Anonimatron.main(Anonimatron.java:67)
Caused by: java.lang.ClassCastException: java.math.BigDecimal incompatible with java.lang.String
        at com.rolfje.anonimatron.synonyms.StringSynonym.setFrom(StringSynonym.java:43)
        at com.rolfje.anonimatron.anonymizer.DutchBSNAnononymizer.anonymize(DutchBSNAnononymizer.java:34)
        at com.rolfje.anonimatron.anonymizer.AnonymizerService.anonymize(AnonymizerService.java:97)
        at com.rolfje.anonimatron.jdbc.JdbcAnonymizerService$2.processColumn(JdbcAnonymizerService.java:140)
        at com.rolfje.anonimatron.jdbc.JdbcAnonymizerService.processTableColumns(JdbcAnonymizerService.java:210)
        ... 4 more

It would be nice to support numeric columns as well, because a Dutch BSN is a valid number.

[realrolfje]

I would strongly advise against storing a BSN in a numeric column. It may cause the same impracticalities as storing a bank account number in a numeric column, which caused interesting problems in some systems when IBAN was introduced.

Just because something has digits, does not make it a scalar number. There is no "bigger" or "smaller" BSN, and they can not be added or subtracted. In that sense, it is not a number. Particularly if it is an identifier for which the syntax rules are external to your system.

Nothing against making this technically possible, but it makes me wonder... why?

[fbascheper]

Point taken and yes, I agree with all the statements above. The point is that more than a few legacy systems have been implemented this way, i.e. by storing social security numbers in a numeric column.

[realrolfje]

True. I've seen systems like that. I usually push a bit to get that fixed. But technically speaking, yes it should be possible to anonymize number-like data based on numeric fields in a database.

[fbascheper]

That's exactly my point, because I believe it's not the responsibility of the anonymisation tool to determine what column definition is used to store BSN in an application's database.

[realrolfje]

Merged into develop, part of next release.