fix: handling of rdf:HTML literals (#2490) · RDFLib/rdflib@588286b (original) (raw)
``
1
`+
from future import annotations
`
``
2
+
``
3
`+
import builtins
`
``
4
`+
import datetime
`
``
5
`+
import logging
`
``
6
`+
from decimal import Decimal
`
``
7
`+
from test.utils import affix_tuples
`
``
8
`+
from test.utils.literal import LiteralChecker, literal_idfn
`
``
9
`+
from test.utils.namespace import EGDC
`
``
10
`+
from test.utils.outcome import OutcomeChecker, OutcomePrimitive, OutcomePrimitives
`
``
11
`+
from typing import Any, Callable, Generator, Optional, Type, Union
`
``
12
+
1
13
`# NOTE: The config below enables strict mode for mypy.
`
2
14
`# mypy: no_ignore_errors
`
3
15
`# mypy: warn_unused_configs, disallow_any_generics
`
`@@ -7,14 +19,13 @@
`
7
19
`# mypy: no_implicit_optional, warn_redundant_casts, warn_unused_ignores
`
8
20
`# mypy: warn_return_any, no_implicit_reexport, strict_equality
`
9
21
``
10
``
`-
import datetime
`
11
``
`-
import logging
`
12
``
`-
from decimal import Decimal
`
13
``
`-
from test.utils import affix_tuples
`
14
``
`-
from test.utils.literal import LiteralChecker
`
15
``
`-
from test.utils.namespace import EGDC
`
16
``
`-
from test.utils.outcome import OutcomeChecker, OutcomePrimitive, OutcomePrimitives
`
17
``
`-
from typing import Any, Callable, Generator, Optional, Type, Union
`
``
22
+
``
23
`+
try:
`
``
24
`+
import html5lib as _ # noqa: F401
`
``
25
+
``
26
`+
_HAVE_HTML5LIB = True
`
``
27
`+
except ImportError:
`
``
28
`+
_HAVE_HTML5LIB = False
`
18
29
``
19
30
`import isodate
`
20
31
`import pytest
`
`@@ -915,6 +926,21 @@ def unlexify(s: str) -> str:
`
915
926
` )
`
916
927
``
917
928
``
``
929
`+
class _UnknownType:
`
``
930
`+
"""
`
``
931
`+
A class that is not known to rdflib, used to test the how
`
``
932
`+
rdflib.term.Literal handles unknown python types.
`
``
933
`+
"""
`
``
934
+
``
935
`+
def repr(self) -> str:
`
``
936
`+
return "_UnknownType()"
`
``
937
+
``
938
`+
def eq(self, __value: object) -> bool:
`
``
939
`+
if isinstance(__value, _UnknownType):
`
``
940
`+
return True
`
``
941
`+
return False
`
``
942
+
``
943
+
918
944
`@pytest.mark.parametrize(
`
919
945
` ["literal_maker", "outcome"],
`
920
946
` [
`
`@@ -951,7 +977,30 @@ def unlexify(s: str) -> str:
`
951
977
`lambda: Literal(Literal("blue sky", "en")),
`
952
978
`Literal("blue sky", "en"),
`
953
979
` ),
`
``
980
`+
(
`
``
981
`+
lambda: Literal("", datatype=RDF.HTML),
`
``
982
`+
LiteralChecker(
`
``
983
`+
..., None, RDF.HTML, True if _HAVE_HTML5LIB else None, ""
`
``
984
`+
),
`
``
985
`+
),
`
``
986
`+
(
`
``
987
`+
lambda: Literal("
`
``
988
`+
LiteralChecker(
`
``
989
`+
...,
`
``
990
`+
None,
`
``
991
`+
RDF.HTML,
`
``
992
`+
False if _HAVE_HTML5LIB else None,
`
``
993
`+
"
`
``
994
`+
),
`
``
995
`+
),
`
``
996
`+
(
`
``
997
`+
lambda: Literal(_UnknownType(), datatype=EGDC.UnknownType),
`
``
998
`+
LiteralChecker(
`
``
999
`+
_UnknownType(), None, EGDC.UnknownType, None, "_UnknownType()"
`
``
1000
`+
),
`
``
1001
`+
),
`
954
1002
` ],
`
``
1003
`+
ids=literal_idfn,
`
955
1004
`)
`
956
1005
`def test_literal_construction(
`
957
1006
`literal_maker: Callable[[], Literal],
`
`@@ -961,3 +1010,41 @@ def test_literal_construction(
`
961
1010
`with checker.context():
`
962
1011
`actual_outcome = literal_maker()
`
963
1012
`checker.check(actual_outcome)
`
``
1013
+
``
1014
+
``
1015
`+
@pytest.mark.parametrize(
`
``
1016
`+
["literal_maker", "normalize_literals", "outcome"],
`
``
1017
`+
[
`
``
1018
`+
(
`
``
1019
`+
lambda: Literal("001000", datatype=XSD.integer),
`
``
1020
`+
...,
`
``
1021
`+
LiteralChecker(1000, None, XSD.integer, False, "1000"),
`
``
1022
`+
),
`
``
1023
`+
(
`
``
1024
`+
lambda: Literal("001000", datatype=XSD.integer),
`
``
1025
`+
True,
`
``
1026
`+
LiteralChecker(1000, None, XSD.integer, False, "1000"),
`
``
1027
`+
),
`
``
1028
`+
(
`
``
1029
`+
lambda: Literal("001000", datatype=XSD.integer),
`
``
1030
`+
False,
`
``
1031
`+
LiteralChecker(1000, None, XSD.integer, False, "001000"),
`
``
1032
`+
),
`
``
1033
`+
],
`
``
1034
`+
ids=literal_idfn,
`
``
1035
`+
)
`
``
1036
`+
def test_global_normalize(
`
``
1037
`+
literal_maker: Callable[[], Literal],
`
``
1038
`+
normalize_literals: Union[builtins.ellipsis, bool],
`
``
1039
`+
outcome: OutcomePrimitives[Literal],
`
``
1040
`+
) -> None:
`
``
1041
`+
_normalize_literals = rdflib.NORMALIZE_LITERALS
`
``
1042
`+
try:
`
``
1043
`+
if normalize_literals is not ...:
`
``
1044
`+
rdflib.NORMALIZE_LITERALS = normalize_literals
`
``
1045
`+
checker = OutcomeChecker[Literal].from_primitives(outcome)
`
``
1046
`+
with checker.context():
`
``
1047
`+
actual_outcome = literal_maker()
`
``
1048
`+
checker.check(actual_outcome)
`
``
1049
`+
finally:
`
``
1050
`+
rdflib.NORMALIZE_LITERALS = _normalize_literals
`