Is Bison the right tool for swish-e?

[Bill Moseley]

I have no experience with bison or yacc, but I'm wondering if I should
change that.  I've read a few yacc/bison tutorials, but haven't written
any code.  It's all still a bit fuzzy.

I'm the maintainer of swish-e, a search engine for small collections of
files.  Currently the query parser is hand-written C and is in bad need
of a rewrite.  Unfortunately, I don't have a lot of free time, so I
wanted to ask here if bison would work for us before starting out.

My main concern is one irregularity in our query syntax.  Swish-e indexes
things by "metaname", which really means an index is made up of a number
of fields or columns.  So, searches are really limited to a given
metaname:

    title=bison  # limit searches to the "title" metaname
    description=(foo OR bar)  # similar (case of the operator is not important)

Swish-e has a default metaname where things get indexed if not specified
like above.  It's called "swishdefault".  So a plain query like:

    foo

is the same as

    swishdefault=foo

So these two queries are the same:

    title=foo description=(foo OR bar) OR (baz AND woo)

    title=foo AND description=(foo OR bar) OR \ 
        (swishdefault=baz AND swishdescription=woo)

We also want to be able to use the standard + and - operators:

    foo OR bar NOT baz
    foo OR bar -baz  # same thing

We also have phrases by using double-quotes and the boolean operators
can be turned into search words by placing them in a phrase:

    foo or bar      # search for docs with either word
    foo "or" bar    # search for docs with all three words
    "foo or bar"    # find the phrase of all three words

Does this all seem like something bison can do?  I assume so, but I just
wanted to run it by experienced users before I start.

Thanks,




-- 
Bill Moseley
address@hidden