def test_case_sensitive(self):
mapping = Mapping([{"in": "A", "out": "b"}], case_sensitive=False)
mapping_case_sensitive = Mapping([{"in": "A", "out": "b"}])
transducer = Transducer(mapping)
transducer_case_sensitive = Transducer(mapping_case_sensitive)
self.assertEqual(transducer("a").output_string, "b")
self.assertEqual(transducer_case_sensitive("a").output_string, "a")
self.assertEqual(transducer("A").output_string, "b")
def test_reverse(self):
mapping = Mapping([{"in": "a", "out": "b"}])
mapping_reversed = Mapping([{"in": "a", "out": "b"}], reverse=True)
transducer = Transducer(mapping)
transducer_reversed = Transducer(mapping_reversed)
self.assertEqual(transducer("a").output_string, "b")
self.assertEqual(transducer("b").output_string, "b")
self.assertEqual(transducer_reversed("a").output_string, "a")
self.assertEqual(transducer_reversed("b").output_string, "a")
def test_escape_special(self):
mapping = Mapping([{"in": r"\d", "out": "digit"}])
mapping_escaped = Mapping([{"in": r"\d", "out": "b"}], escape_special=True)
transducer = Transducer(mapping)
transducer_escaped = Transducer(mapping_escaped)
self.assertEqual(transducer("1").output_string, "digit")
self.assertEqual(transducer(r"\d").output_string, r"\d")
self.assertEqual(transducer_escaped("1").output_string, "1")
self.assertEqual(transducer_escaped(r"\d").output_string, "b")
def test_abbreviations(self):
mapping = Mapping(
os.path.join(
os.path.dirname(public_data), "mappings", "abbreviation_config.yaml"
)
)
self.assertEqual(mapping.mapping[0]["in"], "i|u")
self.assertEqual(mapping.mapping[1]["in"], "a|e|i|o|u")
transducer = Transducer(mapping)
self.assertEqual(transducer("i").output_string, "1")
self.assertEqual(transducer("e").output_string, "2")
def test_minimal(self):
mapping = Mapping(
os.path.join(
os.path.dirname(public_data), "mappings", "minimal_config.yaml"
)
)
transducer = Transducer(mapping)
self.assertEqual(transducer("abb").output_string, "aaa")
self.assertEqual(transducer("a").output_string, "a")
self.assertFalse(mapping.wants_rules_sorted())
self.assertFalse(mapping.kwargs["case_sensitive"])
self.assertTrue(mapping.kwargs["escape_special"])
self.assertEqual(mapping.kwargs["norm_form"], "NFD")
self.assertTrue(mapping.kwargs["reverse"])
def test_rule_ordering_from_config(self):
"""
Same as test_minimal, but uses "rule-ordering" instead of "as-is" in the config.
"""
mapping = Mapping(
os.path.join(os.path.dirname(public_data), "mappings", "rule-ordering.yaml")
)
transducer = Transducer(mapping)
self.assertEqual(transducer("abb").output_string, "aaa")
self.assertEqual(transducer("a").output_string, "a")
self.assertTrue(mapping.wants_rules_sorted())
self.assertEqual(mapping.kwargs["rule_ordering"], "apply-longest-first")
self.assertFalse(mapping.kwargs["case_sensitive"])
self.assertTrue(mapping.kwargs["escape_special"])
self.assertEqual(mapping.kwargs["norm_form"], "NFD")
self.assertTrue(mapping.kwargs["reverse"])
def test_as_is(self):
"""
Test deprecated config: as_is.
"""
# explicitly set as_is=False
log_output = io.StringIO()
with redirect_stderr(log_output):
mapping_sorted = Mapping(
[{"in": "a", "out": "b"}, {"in": "aa", "out": "c"}], as_is=False
)
self.assertTrue(mapping_sorted.wants_rules_sorted())
self.assertIn(
"deprecated",
log_output.getvalue(),
"it should warn that the feature is deprecated",
)
self.assertIn(
"apply-longest-first",
log_output.getvalue(),
"it should show the equivalent rule_ordering setting",
)
# explicitly set as_is=True
log_output = io.StringIO()
with redirect_stderr(log_output):
mapping = Mapping(
[{"in": "a", "out": "b"}, {"in": "aa", "out": "c"}], as_is=True
)
self.assertFalse(mapping.wants_rules_sorted())
self.assertIn(
"deprecated",
log_output.getvalue(),
"it should warn that the feature is deprecated",
)
self.assertIn(
"as-written",
log_output.getvalue(),
"it should show the equivalent rule_ordering setting",
)
# test the default (rule_ordering="as-written")
mapping_as_is = Mapping([{"in": "a", "out": "b"}, {"in": "aa", "out": "c"}])
self.assertFalse(mapping.wants_rules_sorted())
# test the alternative (rule_ordering="apply-longest-first")
transducer = Transducer(mapping_sorted)
transducer_as_is = Transducer(mapping_as_is)
self.assertEqual(transducer("aa").output_string, "c")
self.assertEqual(transducer_as_is("aa").output_string, "bb")
def test_no_escape(self):
mapping = Mapping(
os.path.join(os.path.dirname(public_data), "mappings", "no_escape.csv")
)
transducer = Transducer(mapping)
self.assertEqual(transducer("?").output_string, "ʔ")