fix: remove Literal.toPython date conversion for gYear/gYearMonth (#3115)

* fix: remove Literal.toPython casting for gYear and gYearMonth

Issue #3078 reports, that rdflib.Literal.toPython casting of xsd:gYear and
xsd:gYearMonth to datetime objects is not possible, as there is no
appropriate Python equivalence for those types.

The current implementation casts xsd:gYear and xsd:gYearMonth to
datetime objects assuming January 1st for xsd:gYear and the 1st day
of the given month for xsd:gYearMonth. This is plain wrong.

The change removes casting to datetime objects in
rdflib.Literal.toPython for xsd:gYear and xsd:gYearMonth.

Closes #3078 .

* test: adapt rdflib.Literal tests to gYear/gYearMonth toPython change

---------

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

Author: lu-pl

rdflib/term.py

@@ -69,8 +69,6 @@
     parse_time,
     parse_xsd_date,
     parse_xsd_duration,
-    parse_xsd_gyear,
-    parse_xsd_gyearmonth,
 )
 
 if TYPE_CHECKING:
@@ -2085,8 +2083,6 @@ def _castPythonToLiteral(  # noqa: N802
     None: None,  # plain literals map directly to value space
     URIRef(_XSD_PFX + "time"): parse_time,
     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_xsd_duration,
     URIRef(_XSD_PFX + "dayTimeDuration"): parse_xsd_duration,

rdflib/xsd_datetime.py

@@ -593,79 +593,6 @@ def parse_xsd_date(date_string: str):
     return parse_date(date_string if not minus else ("-" + date_string))
 
 
-def parse_xsd_gyear(gyear_string: str):
-    """
-    XSD gYear has more features than ISO8601 dates, specifically
-    XSD allows timezones on a gYear, that must be stripped off.
-    """
-    if gyear_string.endswith("Z") or gyear_string.endswith("z"):
-        gyear_string = gyear_string[:-1]
-    if gyear_string.startswith("-"):
-        gyear_string = gyear_string[1:]
-        minus = True
-    else:
-        minus = False
-    has_plus = gyear_string.rfind("+")
-    if has_plus > 0:
-        # Drop the +07:00 timezone part
-        gyear_string = gyear_string[:has_plus]
-    else:
-        split_parts = gyear_string.rsplit("-", 1)
-        if len(split_parts) > 1 and ":" in split_parts[-1]:
-            # Drop the -09:00 timezone part
-            gyear_string = split_parts[0]
-    if len(gyear_string) < 4:
-        raise ValueError("gYear string must be at least 4 numerals in length")
-    gyear_string = gyear_string.lstrip("0")  # strip all leading zeros
-    try:
-        y = int(gyear_string if not minus else ("-" + gyear_string))
-    except ValueError:
-        raise ValueError("gYear string must be a valid integer")
-    return date(y, 1, 1)
-
-
-def parse_xsd_gyearmonth(gym_string: str):
-    """
-    XSD gYearMonth has more features than ISO8601 dates, specifically
-    XSD allows timezones on a gYearMonth, that must be stripped off.
-    """
-    if gym_string.endswith("Z") or gym_string.endswith("z"):
-        gym_string = gym_string[:-1]
-    if gym_string.startswith("-"):
-        gym_string = gym_string[1:]
-        minus = True
-    else:
-        minus = False
-    has_plus = gym_string.rfind("+")
-    if has_plus > 0:
-        # Drop the +07:00 timezone part
-        gym_string = gym_string[:has_plus]
-    else:
-        split_parts = gym_string.rsplit("-", 1)
-        if len(split_parts) > 1 and ":" in split_parts[-1]:
-            # Drop the -09:00 timezone part
-            gym_string = split_parts[0]
-    year_month_parts = gym_string.split("-", 1)
-    if len(year_month_parts) < 2:
-        raise ValueError("XSD gYearMonth string must contain one dash")
-
-    if len(year_month_parts[0]) < 4:
-        raise ValueError("gYearMonth Year part must be at least 4 numerals in length")
-    elif len(year_month_parts[1]) < 2:
-        raise ValueError("gYearMonth Month part must be exactly 2 numerals in length")
-    year_string = year_month_parts[0].lstrip("0")  # strip all leading zeros
-    month_string = year_month_parts[1].lstrip("0")  # strip all leading zeros
-    try:
-        y = int(year_string if not minus else ("-" + year_string))
-    except ValueError:
-        raise ValueError("gYearMonth Year part must be a valid integer")
-    try:
-        m = int(month_string)
-    except ValueError:
-        raise ValueError("gYearMonth Month part must be a valid integer")
-    return date(y, m, 1)
-
-
 # Parse XSD Datetime is the same as ISO8601 Datetime
 # It uses datetime.fromisoformat for python 3.11 and above
 # or isodate.parse_datetime for older versions

test/test_literal/test_literal.py

@@ -855,21 +855,23 @@ def unlexify(s: str) -> str:
         ("0000-00-00", XSD.date, None),
         ("NOT A VALID HEX STRING", XSD.hexBinary, None),
         ("NOT A VALID BASE64 STRING", XSD.base64Binary, None),
+        # xsd:gYear and xsd:gYearMonth also do not get converted
+        ("1921-05", XSD.gYearMonth, None),
+        ("0001-01", XSD.gYearMonth, None),
+        ("0001-12", XSD.gYearMonth, None),
+        ("2002-01", XSD.gYearMonth, None),
+        ("9999-01", XSD.gYearMonth, None),
+        ("9999-12", XSD.gYearMonth, None),
+        ("1921", XSD.gYear, None),
+        ("2000", XSD.gYear, None),
+        ("0001", XSD.gYear, None),
+        ("9999", XSD.gYear, None),
+        ("1982", XSD.gYear, None),
+        ("2002", XSD.gYear, None),
+
         # these literals get converted to python types
         ("1921-05-01", XSD.date, datetime.date),
         ("1921-05-01T00:00:00", XSD.dateTime, datetime.datetime),
-        ("1921-05", XSD.gYearMonth, datetime.date),
-        ("0001-01", XSD.gYearMonth, datetime.date),
-        ("0001-12", XSD.gYearMonth, datetime.date),
-        ("2002-01", XSD.gYearMonth, datetime.date),
-        ("9999-01", XSD.gYearMonth, datetime.date),
-        ("9999-12", XSD.gYearMonth, datetime.date),
-        ("1921", XSD.gYear, datetime.date),
-        ("2000", XSD.gYear, datetime.date),
-        ("0001", XSD.gYear, datetime.date),
-        ("9999", XSD.gYear, datetime.date),
-        ("1982", XSD.gYear, datetime.date),
-        ("2002", XSD.gYear, datetime.date),
         ("1921-05-01T00:00:00+00:30", XSD.dateTime, datetime.datetime),
         ("1921-05-01T00:00:00-00:30", XSD.dateTime, datetime.datetime),
         ("true", XSD.boolean, bool),