def setUp(self):
corpus = ParallelGraphCorpus(
inf="../exp/corpora/news/pgc/ma/2006-11/news-2006-11-aligned-part-00.pgc")
self.graph_pair = corpus[0]
descriptor = Descriptor(cornet_sim)
self.feat_extr = Extractor(
descriptor,
node_selector=select_visible_node)
def test_align_count(self):
descriptor = Descriptor(term_align)
feat_extr = Extractor(
descriptor,
node_selector=select_visible_node)
graph_pair = self.corpus[0]
instances = feat_extr.extract(graph_pair)
self.dump(graph_pair, instances)
# check for a couple of "interesting" instances
#instance: 0
#source: 0: top: Posters Partij voor de Dieren verwijderd
#target: 0: top: Zeeland verwijdert posters Partij voor de Dieren
#align_inside_count: 6
#source_align_outside_count: 0
#target_align_outside_count: 0
#source_align_none_count: 0
#target_align_none_count: 1
self.assertEqual(instances[0]["align_inside_count"], 6)
self.assertEqual(instances[0]["source_align_outside_count"], 0)
self.assertEqual(instances[0]["target_align_outside_count"], 0)
self.assertEqual(instances[0]["source_align_none_count"], 0)
self.assertEqual(instances[0]["target_align_none_count"], 1)
#instance: 2
#source: 0: top: Posters Partij voor de Dieren verwijderd
#target: 2: name: Zeeland
#align_inside_count: 0
#source_align_outside_count: 6
#target_align_outside_count: 0
#source_align_none_count: 0
#target_align_none_count: 1
self.assertEqual(instances[2]["align_inside_count"], 0)
self.assertEqual(instances[2]["source_align_outside_count"], 6)
self.assertEqual(instances[2]["target_align_outside_count"], 0)
self.assertEqual(instances[2]["source_align_none_count"], 0)
self.assertEqual(instances[2]["target_align_none_count"], 1)
#instance: 25
#source: 2: noun: Posters
#target: 3: verb: verwijdert
#align_inside_count: 0
#source_align_outside_count: 1
#target_align_outside_count: 1
#source_align_none_count: 0
#target_align_none_count: 01
self.assertEqual(instances[25]["align_inside_count"], 0)
self.assertEqual(instances[25]["source_align_outside_count"], 1)
self.assertEqual(instances[25]["target_align_outside_count"], 1)
self.assertEqual(instances[25]["source_align_none_count"], 0)
self.assertEqual(instances[25]["target_align_none_count"], 0)
def test_align_count(self):
descriptor = Descriptor(term_align)
feat_extr = Extractor(descriptor, node_selector=select_visible_node)
graph_pair = self.corpus[0]
instances = feat_extr.extract(graph_pair)
self.dump(graph_pair, instances)
# check for a couple of "interesting" instances
#instance: 0
#source: 0: top: Posters Partij voor de Dieren verwijderd
#target: 0: top: Zeeland verwijdert posters Partij voor de Dieren
#align_inside_count: 6
#source_align_outside_count: 0
#target_align_outside_count: 0
#source_align_none_count: 0
#target_align_none_count: 1
self.assertEqual(instances[0]["align_inside_count"], 6)
self.assertEqual(instances[0]["source_align_outside_count"], 0)
self.assertEqual(instances[0]["target_align_outside_count"], 0)
self.assertEqual(instances[0]["source_align_none_count"], 0)
self.assertEqual(instances[0]["target_align_none_count"], 1)
#instance: 2
#source: 0: top: Posters Partij voor de Dieren verwijderd
#target: 2: name: Zeeland
#align_inside_count: 0
#source_align_outside_count: 6
#target_align_outside_count: 0
#source_align_none_count: 0
#target_align_none_count: 1
self.assertEqual(instances[2]["align_inside_count"], 0)
self.assertEqual(instances[2]["source_align_outside_count"], 6)
self.assertEqual(instances[2]["target_align_outside_count"], 0)
self.assertEqual(instances[2]["source_align_none_count"], 0)
self.assertEqual(instances[2]["target_align_none_count"], 1)
#instance: 25
#source: 2: noun: Posters
#target: 3: verb: verwijdert
#align_inside_count: 0
#source_align_outside_count: 1
#target_align_outside_count: 1
#source_align_none_count: 0
#target_align_none_count: 01
self.assertEqual(instances[25]["align_inside_count"], 0)
self.assertEqual(instances[25]["source_align_outside_count"], 1)
self.assertEqual(instances[25]["target_align_outside_count"], 1)
self.assertEqual(instances[25]["source_align_none_count"], 0)
self.assertEqual(instances[25]["target_align_none_count"], 0)
def __init__(self,
descriptor=None,
extractor=None,
classifier=None,
matcher=None):
self.descriptor = descriptor or Descriptor()
self.extractor = extractor or Extractor(self.descriptor)
self.classifier = classifier or Classifier()
self.matcher = matcher or Matcher(no_rel=self.descriptor.no_rel)
class TestCornetFeatures(unittest.TestCase):
def setUp(self):
corpus = ParallelGraphCorpus(
inf="../exp/corpora/news/pgc/ma/2006-11/news-2006-11-aligned-part-00.pgc")
self.graph_pair = corpus[0]
descriptor = Descriptor(cornet_sim)
self.feat_extr = Extractor(
descriptor,
node_selector=select_visible_node)
def test_cornet_server(self):
create_cornet_server_proxy(SERVER)
instances = self.feat_extr.extract(self.graph_pair)
print instances
# FIXME add asserts
def test_cornet_load(self):
print "\n(Loading Cornetto database - may take a long time...)"
load_cornet(CDB_LU_FNAME, CDB_SYN_FNAME)
instances = self.feat_extr.extract(self.graph_pair)
print instances
def test_init_2(self):
"""
adding a new feature
"""
def ff_new_feat(**kwargs):
return 1
def pp_hook1(): pass
f1 = Feat( ff_new_feat , "i", pp_graph_hooks=[pp_hook1]),
fd = Descriptor(f1)
e = Extractor(fd)
print
e.descriptor.pprint()
def test_init_1(self):
feat_extr = Extractor(Descriptor())
print
feat_extr.descriptor.pprint()
def extractor(self):
try:
return self._extractor
except AttributeError:
self._extractor = Extractor(self.descriptor, self.node_selector)
return self._extractor
