def test_trace_conversion(self):
traceroutes = [['N0', 'N5', 'N4', 'N6'], ['N5', 'N5', 'N4', 'N5'],
['N5', 'N2', 'N4']]
traceroutes_check = [[0, 5, 4, 6], [5, 5, 4, 5], [5, 2, 4]]
trace_id = vft.trace_in_vertex_id(self.sample_graph, traceroutes)
self.assertEqual(traceroutes_check, trace_id)
trace_back = vft.trace_in_vertex_name(self.sample_graph, trace_id)
self.assertEqual(traceroutes, trace_back)
def test_trace_conversion_error(self):
fake_trace = [
['FAKE', 'FAKE1'],
]
fake_trace_id = [
[2, 4, 55],
]
with self.assertRaises(ValueError):
_ = vft.trace_in_vertex_id(self.sample_graph, fake_trace)
with self.assertRaises(IndexError):
_ = vft.trace_in_vertex_name(self.sample_graph, fake_trace_id)
def purify(g, meta, filters):
results = dict()
traceroutes = [x[helpers.TRACE] for x in meta]
if 'cc' in filters:
results['cc'] = g.transitivity_undirected(mode=igraph.TRANSITIVITY_ZERO)
if 'ad' in filters:
results['ad'] = g.average_path_length(directed=False, unconn=True)
if 'nc' in filters:
results['nc'] = g.vcount()
if 'ec' in filters:
results['ec'] = g.ecount()
if 'rc' in filters:
k = 20
scores = g.degree()
indices = range(g.vcount())
indices.sort(key=scores.__getitem__)
e_k = [x for x in g.es
if g.degree(x.source) >= k and g.degree(x.target) >= 50]
e_k2 = float(2 * len(e_k))
n_k = float(len([x for x in g.vs if g.degree(x) >= k]))
fi_k = e_k2 / (n_k * (n_k - 1))
results['rc'] = fi_k
# remove traces with unknown nodes
before_caida = len(traceroutes)
traceroutes = vft.trace_in_vertex_id(g, traceroutes)
if 'tc' in filters:
results['tc'] = len(traceroutes)
if 'tl' in filters:
results['tl'] = np.mean([len(x) for x in traceroutes])
if 'tml' in filters:
results['tml'] = max([len(x) for x in traceroutes])
results['tml_sentence'] = vft.trace_in_vertex_name(g, [x for x in traceroutes if len(x) == results['tml']])[0]
if 'tsl' in filters:
results['tsl'] = min([len(x) for x in traceroutes])
results['tsl_sentence'] = vft.trace_in_vertex_name(g, [x for x in traceroutes if len(x) == results['tsl']])[0]
if 'rt' in filters:
results['rt'] = before_caida - len(traceroutes)
if 'vf_prelabeled' in filters:
results['vf_prelabeled'] = len([x for x in meta if x[helpers.IS_VF_PRELABELED] == 1])
if 'vf_degree' in filters:
results['vf_degree'] = len([x for x in meta if x[helpers.IS_VF_DEGREE] == 1])
if 'vf_closeness' in filters:
results['vf_closeness'] = len([x for x in meta if x[helpers.IS_VF_CLOSENESS] == 1])
if 'random_walk_vf_closeness' in filters:
results['random_walk_vf_closeness'] = len([x for x in meta if x[helpers.RANDOM_WALK_VF_CLOSENESS_ROUTE] == 1])
if 'lp_soft_prelabeled' in filters:
results['lp_soft_prelabeled'] = len([x for x in meta if x[helpers.IS_LP_SOFT_PRELABELED] == 1])
if 'lp_hard_prelabeled' in filters:
results['lp_hard_prelabeled'] = len([x for x in meta if x[helpers.IS_LP_HARD_PRELABELED] == 1])
if 'lp_soft_degree' in filters:
results['lp_soft_degree'] = len([x for x in meta if x[helpers.IS_LP_SOFT_DEGREE] == 1])
if 'lp_hard_degree' in filters:
results['lp_hard_degree'] = len([x for x in meta if x[helpers.IS_LP_HARD_DEGREE] == 1])
if 'lp_soft_closeness' in filters:
results['lp_soft_closeness'] = len([x for x in meta if x[helpers.IS_LP_SOFT_CLOSENESS] == 1])
if 'lp_hard_closeness' in filters:
results['lp_hard_closeness'] = len([x for x in meta if x[helpers.IS_LP_HARD_CLOSENESS] == 1])
if 'pred' in filters:
# SH prediction
sh_pred = len([x for x in meta if x[helpers.SH_LEN] == x[helpers.TRACE_LEN]])
# only VF with 1 extra hop
ppvf_pred = len([x for x in meta if x[helpers.TRACE_LEN] <= x[helpers.SH_LEN] + 1 and x[helpers.IS_VF_DEGREE] == 1])
# SH or VF with one extra hop
smart_pred = len([x for x in meta if x[helpers.SH_LEN] == x[helpers.TRACE_LEN] or (x[helpers.TRACE_LEN] <= x[helpers.SH_LEN] + 1 and x[helpers.IS_VF_DEGREE] == 1)])
# Brute force prediction
all_pred = len([x for x in meta if x[helpers.TRACE_LEN] <= x[helpers.SH_LEN] + 1])
results['sh_pred'] = sh_pred
results['ppvf_pred'] = ppvf_pred
results['smart_pred'] = smart_pred
results['all_pred'] = all_pred
return results