[redland-dev] [rapper] Use random names for blank nodes

[Nicholas Humfrey]

Dear F,


There isn't any functionality in the rapper command line tool for 
choosing different identifier generator.



However libraptor does have support for custom generation of blank node 
identifiers:

     raptor_world_set_generate_bnodeid_handler()

There is some documentation for this here:
http://librdf.org/raptor/api/raptor2-section-world.html#raptor-world-set-generate-bnodeid-handler

And the default implementation is defined here:
https://github.com/dajobe/raptor/blob/master/src/raptor_general.c#L283

librdf uses a seperate implementation, to ensure that every blank node 
in a world is unique:
https://github.com/dajobe/librdf/blob/master/src/rdf_raptor.c#L139
https://github.com/dajobe/librdf/blob/master/src/rdf_init.c#L532



The challenge with identifier generation is ensuring that they are 
unique.
I guess you could write some kind of hash function that hashes the 
integer identifier, but that wouldn't result in different bnode 
identifiers in different documents. Creating a random seed also wouldn't 
guarantee it, but would significantly reduce the risk.

The advantage of blank node identifiers like _:genid1, is that they 
don't pretend that there won't be conflicts between documents.


nick.



On 2017-04-21 01:56, F wrote:
> Hello,
> 
>     I'm trying to use rapper to convert a bunch of rdf/xml files to nt.
> Currently, it's assigning sequential names to blank nodes, like
> _:genid1, _:genid2, and so forth. I'd like to know if it's possible to
> use more randomized names instead (something like
> _:olret5ry67uyhgrt5zcx).
> 
> Thank you.
> 
> _______________________________________________
> redland-dev mailing list
> redland-dev at lists.librdf.org
> http://lists.librdf.org/mailman/listinfo/redland-dev