Redo XSD Datetime, Date, Time, Duration parser and serializers (#2929)

* New xsd_datetime module, with parsers and serializers for XSD_Duration, XSD_Date, XSD_DatetTime, XSD_Time, XSD_gYear, XSD_gYearMonth. Based on isoformat for Python <3.11, and builtin fromisoformat for Python 3.11+

* ruff fixes in test suite changes

* Fix a failing test

* Add missing exports to xsd_datetime

* Fix some version constraints to help CI tests pass

* Fix generating negative duartion strings. This fixes the broken doctest.

* Fix black formatting in xsd_datetime _again_.

* Add isodate back into the dockerfile requirements so that can still build correctly with RDFLib v7.0

* correctly calculate total years in Duration constructor.

* Fix some docstring generation errors

* For documentation-generation reasons, don't re-export builtin parsers as xsd parsers.

* Add ashleysommer to contributors list on the xsd_datetime module.

* Fix wording in xsd_datetime header.

---------

Co-authored-by: Nicholas Car <[email protected]>

Author: ashleysommer

devtools/constraints.min

@@ -1,11 +1,11 @@
 # This file selects minimum versions to ensure that the test suite passes on
 # these versions. The file's extension (`.min`) is chosen to evade Dependabot
 # which operates on `*.{txt,in}` files.
-isodate==0.6.0
+isodate==0.7.2; python_version < "3.11"
 pyparsing==2.1.0
 importlib-metadata==4.0.0
 berkeleydb==18.1.2
 networkx==2.0
-html5lib==1.0.1
+html5lib-modern==1.2.0
 lxml==4.3.0
 orjson==3.9.14

docker/latest/requirements.in

@@ -1,4 +1,6 @@
-# This file is used for building a docker image of hte latest rdflib release. It
+# This file is used for building a docker image of the latest rdflib release. It
 # will be updated by dependabot when new releases are made.
 rdflib==7.0.0
-html5lib
+html5lib-modern==1.2.0
+# isodate is required to allow the Dockerfile to build on with pre-RDFLib-7.1 releases.
+isodate==0.7.2

docker/latest/requirements.txt

@@ -6,12 +6,12 @@
 #
 html5lib-modern==1.2
     # via -r docker/latest/requirements.in
-isodate==0.6.1
+isodate==0.7.2; python_version < "3.11"
     # via rdflib
 pyparsing==3.0.9
     # via rdflib
 rdflib==7.0.0
     # via -r docker/latest/requirements.in
-six==1.16.0
-    # via
-    #   isodate
+# isodate is required to allow the Dockerfile to build on with pre-RDFLib-7.1 releases.
+isodate==0.7.2
+    # via -r docker/latest/requirements.in

docs/rdf_terms.rst

@@ -207,7 +207,8 @@ rdf:HTML               :class:`xml.dom.minidom.DocumentFragment`
 .. [#f1] plain literals map directly to value space
 
 .. [#f2] Date, time and datetime literals are mapped to Python
-         instances using the `isodate <http://pypi.python.org/pypi/isodate/>`_
+         instances using the RDFlib xsd_datetime module, that is based
+         on the `isodate <http://pypi.python.org/pypi/isodate/>`_
          package).
 
 .. [#f3] this is a bit dirty - by accident the ``html5lib`` parser

poetry.lock

@@ -39,7 +39,7 @@ rdfgraphisomorphism = 'rdflib.tools.graphisomorphism:main'
 
 [tool.poetry.dependencies]
 python = "^3.8.1"
-isodate = "^0.6.0"
+isodate = {version=">=0.7.2,<1.0.0", python = "<3.11"}
 pyparsing = ">=2.1.0,<4"
 berkeleydb = {version = "^18.1.0", optional = true}
 networkx = {version = ">=2,<4", optional = true}

pyproject.toml

@@ -21,7 +21,6 @@
 from typing import Any, Callable, Dict, NoReturn, Optional, Tuple, Union, overload
 from urllib.parse import quote
 
-import isodate
 from pyparsing import ParseResults
 
 from rdflib.namespace import RDF, XSD
@@ -47,6 +46,7 @@
     URIRef,
     Variable,
 )
+from rdflib.xsd_datetime import Duration, parse_datetime  # type: ignore[attr-defined]
 
 
 def Builtin_IRI(expr: Expr, ctx: FrozenBindings) -> URIRef:
@@ -521,8 +521,13 @@ def Builtin_TZ(e: Expr, ctx) -> Literal:
     if not d.tzinfo:
         return Literal("")
     n = d.tzinfo.tzname(d)
-    if n == "UTC":
+    if n is None:
+        n = ""
+    elif n == "UTC":
         n = "Z"
+    elif n.startswith("UTC"):
+        # Replace tzname like "UTC-05:00" with simply "-05:00" to match Jena tz fn
+        n = n[3:]
     return Literal(n)
 
 
@@ -687,7 +692,7 @@ def default_cast(e: Expr, ctx: FrozenBindings) -> Literal:  # type: ignore[retur
         if x.datatype and x.datatype not in (XSD.dateTime, XSD.string):
             raise SPARQLError("Cannot cast %r to XSD:dateTime" % x.datatype)
         try:
-            return Literal(isodate.parse_datetime(x), datatype=e.iri)
+            return Literal(parse_datetime(x), datatype=e.iri)
         except:  # noqa: E722
             raise SPARQLError("Cannot interpret '%r' as datetime" % x)
 
@@ -1085,7 +1090,7 @@ def dateTimeObjects(expr: Literal) -> Any:
 def isCompatibleDateTimeDatatype(  # type: ignore[return]
     obj1: Union[py_datetime.date, py_datetime.datetime],
     dt1: URIRef,
-    obj2: Union[isodate.Duration, py_datetime.timedelta],
+    obj2: Union[Duration, py_datetime.timedelta],
     dt2: URIRef,
 ) -> bool:
     """
@@ -1098,7 +1103,7 @@ def isCompatibleDateTimeDatatype(  # type: ignore[return]
             return True
         elif dt2 == XSD.dayTimeDuration or dt2 == XSD.Duration:
             # checking if the dayTimeDuration has no Time Component
-            # else it wont be compatible with Date Literal
+            # else it won't be compatible with Date Literal
             if "T" in str(obj2):
                 return False
             else:
@@ -1110,7 +1115,7 @@ def isCompatibleDateTimeDatatype(  # type: ignore[return]
         elif dt2 == XSD.dayTimeDuration or dt2 == XSD.Duration:
             # checking if the dayTimeDuration has no Date Component
             # (by checking if the format is "PT...." )
-            # else it wont be compatible with Time Literal
+            # else it won't be compatible with Time Literal
             if "T" == str(obj2)[1]:
                 return True
             else:
@@ -1139,7 +1144,7 @@ def calculateDuration(
 def calculateFinalDateTime(
     obj1: Union[py_datetime.date, py_datetime.datetime],
     dt1: URIRef,
-    obj2: Union[isodate.Duration, py_datetime.timedelta],
+    obj2: Union[Duration, py_datetime.timedelta],
     dt2: URIRef,
     operation: str,
 ) -> Literal:

rdflib/plugins/sparql/operators.py

@@ -19,8 +19,6 @@
     Union,
 )
 
-import isodate
-
 import rdflib.plugins.sparql
 from rdflib.graph import ConjunctiveGraph, Dataset, Graph
 from rdflib.namespace import NamespaceManager
@@ -302,7 +300,7 @@ def __init__(
     @property
     def now(self) -> datetime.datetime:
         if self._now is None:
-            self._now = datetime.datetime.now(isodate.tzinfo.UTC)
+            self._now = datetime.datetime.now(datetime.timezone.utc)
         return self._now
 
     def clone(

rdflib/plugins/sparql/sparql.py

@@ -38,7 +38,6 @@
     "Literal",
     "Variable",
 ]
-
 import logging
 import math
 import warnings
@@ -67,19 +66,22 @@
 from uuid import uuid4
 
 import html5lib
-from isodate import (
+
+import rdflib
+import rdflib.util
+from rdflib.compat import long_type
+
+from .xsd_datetime import (  # type: ignore[attr-defined]
     Duration,
     duration_isoformat,
-    parse_date,
     parse_datetime,
-    parse_duration,
     parse_time,
+    parse_xsd_date,
+    parse_xsd_duration,
+    parse_xsd_gyear,
+    parse_xsd_gyearmonth,
 )
 
-import rdflib
-import rdflib.util
-from rdflib.compat import long_type
-
 if TYPE_CHECKING:
     from .namespace import NamespaceManager
     from .paths import AlternativePath, InvPath, NegatedPath, Path, SequencePath
@@ -1424,7 +1426,7 @@ def eq(self, other: Any) -> bool:
             ):
                 return self.value == other
         # NOTE for type ignore: bool is a subclass of int so this won't ever run.
-        elif isinstance(other, bool):  # type: ignore[unreachable]
+        elif isinstance(other, bool):  # type: ignore[unreachable, unused-ignore]
             if self.datatype == _XSD_BOOLEAN:
                 return self.value == other
 
@@ -2030,13 +2032,13 @@ def _castPythonToLiteral(  # noqa: N802
 XSDToPython: Dict[Optional[str], Optional[Callable[[str], Any]]] = {
     None: None,  # plain literals map directly to value space
     URIRef(_XSD_PFX + "time"): parse_time,
-    URIRef(_XSD_PFX + "date"): parse_date,
-    URIRef(_XSD_PFX + "gYear"): parse_date,
-    URIRef(_XSD_PFX + "gYearMonth"): parse_date,
+    URIRef(_XSD_PFX + "date"): parse_xsd_date,
+    URIRef(_XSD_PFX + "gYear"): parse_xsd_gyear,
+    URIRef(_XSD_PFX + "gYearMonth"): parse_xsd_gyearmonth,
     URIRef(_XSD_PFX + "dateTime"): parse_datetime,
-    URIRef(_XSD_PFX + "duration"): parse_duration,
-    URIRef(_XSD_PFX + "dayTimeDuration"): parse_duration,
-    URIRef(_XSD_PFX + "yearMonthDuration"): parse_duration,
+    URIRef(_XSD_PFX + "duration"): parse_xsd_duration,
+    URIRef(_XSD_PFX + "dayTimeDuration"): parse_xsd_duration,
+    URIRef(_XSD_PFX + "yearMonthDuration"): parse_xsd_duration,
     URIRef(_XSD_PFX + "hexBinary"): _unhexlify,
     URIRef(_XSD_PFX + "string"): None,
     URIRef(_XSD_PFX + "normalizedString"): None,