def test_raises_error_if_feature_is_not_separated_by_colon(self):
contents = ["1 qid:1 2:0.0 3=0.1 4:10.0"]
with self.assertRaisesRegex(
ParserError,
"failed to extract feature index and value from '3=0.1'"):
for _ in LibSVMRankingParser(contents, {}):
pass
def _generate_examples(self, path):
"""Yields examples."""
# Istella datasets seems to be encoded as latin1 and not utf-8, so we have
# to read the file contents as bytes and manually decode it as latin1.
with tf.io.gfile.GFile(path, "rb") as f:
lines = map(lambda bytes_line: bytes_line.decode("latin1"), f)
yield from LibSVMRankingParser(lines, _FEATURE_NAMES, _LABEL_NAME)
def test_raises_error_if_feature_value_is_malformatted(self):
contents = ["1 qid:1 2:0.0 3:0.00.1 4:10.0"]
with self.assertRaisesRegex(
ParserError,
"failed to extract feature index and value from '3:0.00.1'"):
for _ in LibSVMRankingParser(contents, {}):
pass
def test_raises_error_if_qid_is_missing(self):
contents = ["0 2:0.0 3:0.0"]
with self.assertRaisesRegex(
ParserError,
"line must contain a qid after the relevance label"):
for _ in LibSVMRankingParser(contents, {}):
pass
def test_parses_a_libsvm_ranking_dataset(self):
results = iter(
LibSVMRankingParser(_TEST_DATASET.split("\n"),
feature_names={
1: "bm25",
2: "tfidf",
3: "querylen",
4: "doclen",
5: "qualityscore"
}))
qid, features = next(results)
self.assertEqual(qid, "1")
self.assertAllEqual(features["label"], [3., 1., 0., 0.])
self.assertAllEqual(features["bm25"], [1., 0., 0., 0.])
self.assertAllEqual(features["tfidf"], [1., 0., 1., 0.])
self.assertAllEqual(features["querylen"], [0., 1., 0., 1.])
self.assertAllEqual(features["doclen"], [0.2, 0.1, 0.4, 0.3])
self.assertAllEqual(features["qualityscore"], [0., 1., 0., 0])
qid, features = next(results)
self.assertEqual(qid, "2")
self.assertAllEqual(features["label"], [0., 1., 0., 0., 1., 2.])
self.assertAllEqual(features["bm25"], [0., 1., 0., 0., 0., 1.])
self.assertAllEqual(features["tfidf"], [0., 0., 0., 0., 0., 1.])
self.assertAllEqual(features["querylen"], [1., 1., 1., 1., 1., 0.])
self.assertAllEqual(features["doclen"], [0.2, 0.4, 0.1, 0.2, 0.1, 0.3])
self.assertAllEqual(features["qualityscore"], [0., 0., 0., 0., 1., 0.])
qid, features = next(results)
self.assertEqual(qid, "3")
self.assertAllEqual(features["label"], [3., 0.])
self.assertAllEqual(features["bm25"], [1., 0.])
self.assertAllEqual(features["tfidf"], [0., 1.])
self.assertAllEqual(features["querylen"], [0., 1.])
self.assertAllEqual(features["doclen"], [0.4, 0.5])
self.assertAllEqual(features["qualityscore"], [1., 0.])
# Assert that the end of the file has been reached.
with self.assertRaises(StopIteration):
next(results)
def test_raises_error_if_label_is_not_a_number(self):
contents = ["abc qid:1 2:0.0 3:0.0"]
with self.assertRaisesRegex(
ParserError, "label 'abc' could not be converted to a float"):
for _ in LibSVMRankingParser(contents, {}):
pass
def test_raises_error_if_line_only_contains_label(self):
contents = ["1"]
with self.assertRaisesRegex(
ParserError, "could not extract label, qid and features"):
for _ in LibSVMRankingParser(contents, {}):
pass
def test_raises_error_with_offending_line(self):
contents = ["1 qid:1 2:1000.0", "2 qid:1 9:9.9", "malformatted line"]
with self.assertRaisesWithPredicateMatch(
ParserError, lambda error: error.line == "malformatted line"):
for _ in LibSVMRankingParser(contents, {}):
pass
def test_raises_error_with_offending_line_number(self):
contents = ["1 qid:1 2:0.0", "malformatted line", "1 qid:1 2:0.0"]
with self.assertRaisesWithPredicateMatch(
ParserError, lambda error: error.line_number == 2):
for _ in LibSVMRankingParser(contents, {}):
pass
def test_raises_error_if_qid_is_malformatted(self):
contents = ["1 qid: 2:0.0 3:0.0"]
with self.assertRaisesRegex(ParserError, "qid can not be empty"):
for _ in LibSVMRankingParser(contents, {}):
pass
def _generate_examples(self, path):
"""Yields examples."""
with tf.io.gfile.GFile(path, "r") as f:
yield from LibSVMRankingParser(f, _FEATURE_NAMES, _LABEL_NAME)
