Properly handle date values

Author: DavidMStraub

gedcom7/cast.py

@@ -142,6 +142,30 @@ def _cast_date(value: str) -> types.Date:
     )
 
 
+def _cast_date_approx(value: str) -> types.DateApprox:
+    """Cast a string to a DateApprox."""
+    match = _match(value, grammar.dateapprox, "DateApprox")
+    return types.DateApprox(
+        date=_cast_date(match.group("dateapprox")),
+        approx=match.group("qualifier"),
+    )
+
+
+def _cast_date_range(value: str) -> types.DateRange:
+    """Cast a string to a DateRange."""
+    match = _match(value, grammar.daterange, "DateRange")
+    res = {}
+    if match.group("between"):
+        res["start"] = _cast_date(match.group("between"))
+    if match.group("and"):
+        res["end"] = _cast_date(match.group("and"))
+    if match.group("after"):
+        res["start"] = _cast_date(match.group("after"))
+    if match.group("before"):
+        res["end"] = _cast_date(match.group("before"))
+    return types.DateRange(**res)
+
+
 def _cast_date_period(value: str) -> types.DatePeriod:
     """Cast a string to a DatePeriod."""
     match = _match(value, grammar.dateperiod, "DatePeriod")
@@ -178,14 +202,35 @@ def _cast_date_period(value: str) -> types.DatePeriod:
     return types.DatePeriod(**res)
 
 
+def _cast_date_value(value: str) -> types.DateValue:
+    """Cast a string to a DateValue.
+
+    A DateValue can be one of the following:
+    - A standard date: "1 OCT 2023"
+    - A dateperiod: "FROM 1 JAN 2023 TO 31 DEC 2023"
+    - A daterange: "BET 1 JAN 2023 AND 31 DEC 2023"
+    - A dateapprox: "ABT 1 OCT 2023"
+    """
+    dateapprox_match = re.fullmatch(grammar.dateapprox, value)
+    if dateapprox_match:
+        return _cast_date_approx(value)
+    daterange_match = re.fullmatch(grammar.daterange, value)
+    if daterange_match:
+        return _cast_date_range(value)
+    dateperiod_match = re.fullmatch(grammar.dateperiod, value)
+    if dateperiod_match:
+        return _cast_date_period(value)
+    return _cast_date(value)
+
+
 CAST_FUNCTIONS: dict[str, Callable[[str], types.DataType] | None] = {
     "Y|<NULL>": _cast_bool,
     "http://www.w3.org/2001/XMLSchema#Language": None,
     "http://www.w3.org/2001/XMLSchema#nonNegativeInteger": _cast_integer,
     "http://www.w3.org/2001/XMLSchema#string": None,
     "http://www.w3.org/ns/dcat#mediaType": _cast_mediatype,
     "https://gedcom.io/terms/v7/type-Age": _cast_age,
-    "https://gedcom.io/terms/v7/type-Date": _cast_date,
+    "https://gedcom.io/terms/v7/type-Date": _cast_date_value,
     "https://gedcom.io/terms/v7/type-Date#exact": _cast_date_exact,
     "https://gedcom.io/terms/v7/type-Date#period": _cast_date_period,
     "https://gedcom.io/terms/v7/type-Enum": _cast_enum,

gedcom7/types.py

@@ -126,7 +126,7 @@ class DateApprox:
 class DateRange:
     """Date range type."""
 
-    start: Date
+    start: Date | None = None
     end: Date | None = None
 
 

test/test_cast.py

@@ -336,3 +336,181 @@ def test_cast_date_period():
     )
     with pytest.raises(ValueError):
         cast._cast_date_period("11 JAN 2022")
+
+
+def test_cast_date_range():
+    assert cast._cast_date_range("BET 1 JAN 2020 AND 31 DEC 2020") == types.DateRange(
+        start=types.Date(
+            calendar=None,
+            day=1,
+            month="JAN",
+            year=2020,
+            epoch=None,
+        ),
+        end=types.Date(
+            calendar=None,
+            day=31,
+            month="DEC",
+            year=2020,
+            epoch=None,
+        ),
+    )
+    assert cast._cast_date_range("AFT 1 JAN 2020") == types.DateRange(
+        start=types.Date(
+            calendar=None,
+            day=1,
+            month="JAN",
+            year=2020,
+            epoch=None,
+        ),
+        end=None,
+    )
+
+    assert cast._cast_date_range("BEF 31 DEC 2020") == types.DateRange(
+        start=None,
+        end=types.Date(
+            calendar=None,
+            day=31,
+            month="DEC",
+            year=2020,
+            epoch=None,
+        ),
+    )
+
+    assert cast._cast_date_range(
+        "BET JULIAN 1 FEB 1600 BCE AND 31 MAR 1500"
+    ) == types.DateRange(
+        start=types.Date(
+            calendar="JULIAN",
+            day=1,
+            month="FEB",
+            year=1600,
+            epoch="BCE",
+        ),
+        end=types.Date(
+            calendar=None,
+            day=31,
+            month="MAR",
+            year=1500,
+            epoch=None,
+        ),
+    )
+
+    with pytest.raises(ValueError):
+        cast._cast_date_range("FROM 1 JAN 2020 TO 31 DEC 2020")
+
+    with pytest.raises(ValueError):
+        cast._cast_date_range("11 JAN 2022")
+
+
+def test_cast_date_approx():
+    assert cast._cast_date_approx("ABT 11 JAN 2022") == types.DateApprox(
+        date=types.Date(
+            calendar=None,
+            day=11,
+            month="JAN",
+            year=2022,
+            epoch=None,
+        ),
+        approx="ABT",
+    )
+
+    assert cast._cast_date_approx("EST JULIAN 5 MAY 1755 BCE") == types.DateApprox(
+        date=types.Date(
+            calendar="JULIAN",
+            day=5,
+            month="MAY",
+            year=1755,
+            epoch="BCE",
+        ),
+        approx="EST",
+    )
+
+    assert cast._cast_date_approx("CAL 1 JAN 2021") == types.DateApprox(
+        date=types.Date(
+            calendar=None,
+            day=1,
+            month="JAN",
+            year=2021,
+            epoch=None,
+        ),
+        approx="CAL",
+    )
+
+    with pytest.raises(ValueError):
+        cast._cast_date_approx("APPROXIMATELY 11 JAN 2022")
+
+    with pytest.raises(ValueError):
+        cast._cast_date_approx("11 JAN 2022")
+
+
+def test_cast_date_value():
+    # Test with standard date
+    assert isinstance(cast._cast_date_value("11 JAN 2022"), types.Date)
+    assert cast._cast_date_value("11 JAN 2022") == types.Date(
+        calendar=None,
+        day=11,
+        month="JAN",
+        year=2022,
+        epoch=None,
+    )
+
+    # Test with date approximation
+    assert isinstance(cast._cast_date_value("ABT 11 JAN 2022"), types.DateApprox)
+    assert cast._cast_date_value("ABT 11 JAN 2022") == types.DateApprox(
+        date=types.Date(
+            calendar=None,
+            day=11,
+            month="JAN",
+            year=2022,
+            epoch=None,
+        ),
+        approx="ABT",
+    )
+
+    # Test with date range
+    assert isinstance(
+        cast._cast_date_value("BET 1 JAN 2020 AND 31 DEC 2020"), types.DateRange
+    )
+    assert cast._cast_date_value("BET 1 JAN 2020 AND 31 DEC 2020") == types.DateRange(
+        start=types.Date(
+            calendar=None,
+            day=1,
+            month="JAN",
+            year=2020,
+            epoch=None,
+        ),
+        end=types.Date(
+            calendar=None,
+            day=31,
+            month="DEC",
+            year=2020,
+            epoch=None,
+        ),
+    )
+
+    # Test with date period
+    assert isinstance(
+        cast._cast_date_value("FROM 1 JAN 2020 TO 31 DEC 2020"), types.DatePeriod
+    )
+    assert cast._cast_date_value("FROM 1 JAN 2020 TO 31 DEC 2020") == types.DatePeriod(
+        from_=types.Date(
+            calendar=None,
+            day=1,
+            month="JAN",
+            year=2020,
+            epoch=None,
+        ),
+        to=types.Date(
+            calendar=None,
+            day=31,
+            month="DEC",
+            year=2020,
+            epoch=None,
+        ),
+    )
+
+    # Test with invalid format
+    with pytest.raises(ValueError):
+        cast._cast_date_value("2022-01-11")
+