Xapian discuss - Sep 2013 - pure data xapian search engine

Hi list,
???? I'm currently refactoring Pure Data's search function to use
Xapian.
Mostly it works well, without a lot of finessing.

The one sticking point is with indexing Pd "objects", which are
typically
letters, numbers, hyphens, underscores, and sometimes a trailing tilde,
with no whitespaces.
So "clip" is one object, and "clip~" is a different object.

I can index these terms just fine, and I'm currently using prefix XO to
make the clip~ object searchable using "object:clip~".? The problem
is that the query parser is removing the tilde, even if I set the stemming
strategy to $xapian::QueryParser_STEM_NONE (using the tcl bindings).

How do I tell the query parser to keep the tilde?? I suppose I could
change all instances of object~ to object_tilde in both the index and
the user input, but since they go into the index with no problem is there
a way to tell the parser to keep them for the search?

Thanks,
Jonathan

On Wed, Sep 18, 2013 at 02:53:48PM -0700, Jonathan Wilkes
wrote:
> The one sticking point is with indexing Pd "objects", which are
typically
> letters, numbers, hyphens, underscores, and sometimes a trailing tilde,
> with no whitespaces.
> So "clip" is one object, and "clip~" is a different
object.
> 
> I can index these terms just fine, and I'm currently using prefix XO to
> make the clip~ object searchable using "object:clip~".? The
problem
> is that the query parser is removing the tilde, even if I set the stemming
> strategy to $xapian::QueryParser_STEM_NONE (using the tcl bindings).
> 
> How do I tell the query parser to keep the tilde?? I suppose I could
> change all instances of object~ to object_tilde in both the index and
> the user input, but since they go into the index with no problem is there
> a way to tell the parser to keep them for the search?
I think this got answered on IRC, but in case not (and for the archives)
I'd suggest making "object" a boolean prefix.

Cheers,
    Olly