diff options
Diffstat (limited to 'test/extractor')
-rw-r--r-- | test/extractor/text/test_metrics.py | 5 | ||||
-rw-r--r-- | test/extractor/text/test_summary.py | 10 |
2 files changed, 12 insertions, 3 deletions
diff --git a/test/extractor/text/test_metrics.py b/test/extractor/text/test_metrics.py index 9cc6a94..6d87889 100644 --- a/test/extractor/text/test_metrics.py +++ b/test/extractor/text/test_metrics.py @@ -55,8 +55,8 @@ class TestTextMetrics(unittest.TestCase): path = os.path.join(os.path.dirname(__file__), 'example-en.txt') # fetch document text = rdr(path) + # extracts all specified predicates triples = set(ext.extract(subject, text, principals)) - self.assertSetEqual({(s,p,o) for s,p,o in triples if p.uri != ns.bse.vocabulary_entropy}, { (subject, ext.schema.predicate(ns.bse.num_characters), 21997), (subject, ext.schema.predicate(ns.bse.num_paragraphs), 48), @@ -66,6 +66,9 @@ class TestTextMetrics(unittest.TestCase): entropy = {o for s,p,o in triples if p.uri == ns.bse.vocabulary_entropy} self.assertEqual(len(entropy), 1) self.assertAlmostEqual(list(entropy)[0], 8.830360505) + # skip unknown predicates + self.assertSetEqual(set(), set(ext.extract(subject, text, + {ext.schema.predicate(ns.bsfs.Predicate).child(ns.bse.unknown)}))) ## main ## diff --git a/test/extractor/text/test_summary.py b/test/extractor/text/test_summary.py index 78d3002..aee1ee2 100644 --- a/test/extractor/text/test_summary.py +++ b/test/extractor/text/test_summary.py @@ -36,14 +36,20 @@ class TestTextMetrics(unittest.TestCase): path = os.path.join(os.path.dirname(__file__), 'example-en.txt') # fetch document text = rdr(path) - + # empty input yields no triples + self.assertEqual(list(ext.extract(subject, [], principals)), []) + self.assertEqual(list(ext.extract(subject, [' '], principals)), []) + self.assertEqual(list(ext.extract(subject, [' ', ' ', ' '], principals)), []) + # creates a summary with warnings.catch_warnings(): warnings.simplefilter('ignore', category=FutureWarning) triples = list(ext.extract(subject, text, principals)) - self.assertEqual(triples, [ (subject, ext.schema.predicate(ns.bse.summary), 'Alice is tired of sitting by her sister on the bank')]) + # skip unknown predicates + self.assertSetEqual(set(), set(ext.extract(subject, text, + {ext.schema.predicate(ns.bsfs.Predicate).child(ns.bse.unknown)}))) ## main ## |