Hi @funkyfuture That would be great! Looking forward to it

and six as a dependency would be okay?

yes, six is ok. Thanks!

hm, when i run the tests with Python 2 i get 15 fails. it hardly makes sense to refactor in that state.

@funkyfuture I am planning to streamline surf a bit and drop support for legacy versions of rdflib (2/3) and python (2.5) with the upcoming 1.1.9 version. There will be a couple of notable changes that will happen:

1. tests will be migrated to pytest from unittest
2. the surf.rdflib plugin will be integrated in surf directly (not installable as a separate plugin)

this will make testing easier, as for now one needs to install the surf.rdflib plugin in order to run some of the tests. I will spend the next couple of days bringing surf up to speed with these changes and will let you know when this is done.

in addition I am planning tox / travis support for integration tests on PRs. Stay tuned!

this is the first time I hear about dropbox' migration script. That will come in handy (at least for the plugin tests which I have not yet migrated). Thanks for the heads up!

the docker + tox helper script is most welcome and appreciated 👍

ok, I've migrated the tests to py.test and added travis integration (all tests pass for now).
To simplify SuRF's usage, the rdflib and sparql_protocol plugins are now embedded within.

@cosminbasca +1 for the new release, I'm looking forward to trying it out. I'm also very pleased to see that this project is still being actively developed (within the limits of one's time availability, of course ;-) )

A potential fix has been proposed in #18

Please fix this.

from surf import *

store = Store(  reader='rdflib',
            writer='rdflib',
            rdflib_store = 'IOMemory')

session = Session(store)

print('Load RDF data')
store.load_triples(source='http://www.w3.org/People/Berners-Lee/card.rdf')

Person = session.get_class(ns.FOAF['Person'])

all_persons = Person.all()

print('Found %d persons that Tim Berners-Lee knows'%(len(all_persons)))
for person in all_persons:
    print(person.foaf_name.first)

#create a person object
somebody = Person()
somebody_else = Person()

somebody.foaf_knows = somebody_else

in python3.7 produces:

Traceback (most recent call last):
  File "../SuRF_TEST.py", line 1, in <module>
    from surf import *
  File "/home/heumos/.local/lib/python3.7/site-packages/surf/__init__.py", line 43, in <module>
    from surf.query import Query, select, ask, describe, construct, OptionalGroup, \
  File "/home/heumos/.local/lib/python3.7/site-packages/surf/query/__init__.py", line 82, in <module>
    class Filter(unicode):
NameError: name 'unicode' is not defined

adding

patch pour python3

import sys
if sys.version_info.major >= 3:
unicode = str

to the
surf\query_init_.py
file, seems to solve the
name 'unicode' is not defined
error
But, another error appears in the namespace.py file
RuntimeError: dictionary changed size during iteration

Dear all,
I did this pull request a long time ago to help a friend who was using SurfRDF (I never used it) , and it seems this repository is stale. However, @franzist seems to have taken up the mantle and fixed what I did not in his own fork, much more up to date: https://github.com/franzlst/surfrdf/tree/master

I'd like to add that this might be the time where a new Pypi library should get registered, such as SurfingRDF, given how the developers here seem to have abandoned this repository (and this is life, I am definitely not complaining about the status of this repo, everyone can move on to something new)

(And honestly, from a personal point of view, I'd drop the support for 2.x Python)