# prep()

# Now, go through the triples in the query document. if there is a bnode, check 
# to see if it's in self.vars (which it will be). Go through each triple in f, 
# matching the arrangement of the variables. Once one is found, store the values 
# in self.vars next to the label and move the quad to self.results.

# dothetriple()

# You either match things, or store things, for every triple.
# For example, if the query is :a :b _:c ., you match a and b, and check to 
# see if c already has a value. if it does, then you match that too, if not, 
# you store something in there, until all of the variables are satisfied.

20:51:59 <AaronSw> So how do you do query? just running thru triples to see if 
they match?
20:52:36 <sbp> if only it were that simple!
20:52:50 <sbp> you have two stores, one for the query quads, and one for the 
file quads
20:53:03 <sbp> then you have a list of universal variables: with labels and 
value
20:53:26 <sbp> scrape all of the universal variables out of the query quads, 
and set them all to an empty value (which later gets filled in)
20:53:56 <sbp> then, for each quad in the query quads, gague which pattern of 
variables needs to be filled in, and go to the appropriate function
20:54:32 <sbp> match each triple in the file quads store against the pattern, 
and if you find a term that matches a variable, checjk to see if it's still 
empty in the vars list
20:54:51 <sbp> if so, put the value in there, and move the quad from the file 
store to a results store
20:55:18 <sbp> do so until all of the query quads are satisfied, output, and 
run through until there's nothing left to match
20:55:38 <sbp> that's the basic jist of it... there's a lot of weird hacking 
in between to glue it all together
20:55:47 <sbp> a lot of conversion function to save space and so on
20:56:08 <AaronSw> * AaronSw reads several times to try and make sense of it