GEDCOM specification compliance: DatePeriod and Date (#3)

* fix dateperiod pattern

* change date parsing to accept empty day; add test

* chore: cleanup

* cleanup dateperiod; special case epoch

* align with updated GEDCOM7 definition

---------

Co-authored-by: Dr. Georg Ostertag <[email protected]>

Author: geostag

gedcom7/grammar.py

@@ -21,19 +21,19 @@
 ucletter = '[A-Z]'
 tagchar = f'({ucletter}|[0-9]|{underscore})'
 exttag = f'{underscore}({tagchar})+'
-calendar = f'(GREGORIAN|JULIAN|FRENCH_R|HEBREW|{exttag})'
+daterestrict = 'FROM|TO|BET|AND|BEF|AFT|ABT|CAL|EST'
+calendar = f'(?!{daterestrict})(GREGORIAN|JULIAN|FRENCH_R|HEBREW|{exttag})'
 day = f'{integer}'
 stdtag = f'{ucletter}({tagchar})*'
-month = f'({stdtag}|{exttag})'
+month = f'(?!{daterestrict})({stdtag}|{exttag})'
 year = f'{integer}'
-epoch = f'(BCE|{exttag})'
+epoch = f'(?!{daterestrict})(BCE|{exttag})'
 date = f'({calendar}{d})?(({day}{d})?{month}{d})?{year}({d}{epoch})?'
 date_capture = f'((?P<calendar>{calendar}){d})?(((?P<day>{day}){d})?(?P<month>{month}){d})?(?P<year>{year})({d}(?P<epoch>{epoch}))?'
 dateapprox = f'(?P<qualifier>ABT|CAL|EST){d}(?P<dateapprox>{date})'
 dateexact = f'(?P<day>{day}){d}(?P<month>{month}){d}(?P<year>{year})'
 dateperiod = f'((TO{d}(?P<todate1>{date}))?|FROM{d}(?P<fromdate>{date})({d}TO{d}(?P<todate2>{date}))?)'
 daterange = f'(BET{d}(?P<between>{date}){d}AND{d}(?P<and>{date})|AFT{d}(?P<after>{date})|BEF{d}(?P<before>{date}))'
-daterestrict = '(FROM|TO|BET|AND|BEF|AFT|ABT|CAL|EST|BCE)'
 datevalue = f'({date}|{dateperiod}|{daterange}|{dateapprox})?'
 tag = f'({stdtag}|{exttag})'
 enum = f'{tag}'

gedcom7/types.py

@@ -234,7 +234,7 @@ def parse(self) -> Dict[str, Union[str, int]]:
         match = self.match.groupdict()
         return {
             "calendar": match["calendar"],
-            "day": int(match["day"]),
+            "day": int(match["day"]) if match["day"] else None,
             "month": match["month"],
             "year": int(match["year"]),
             "epoch": match["epoch"],

test/test_types.py

@@ -126,6 +126,27 @@ def test_date():
         "year": 2022,
         "epoch": "BCE",
     }
+    assert T.DateValue("11 JAN 2022").parse() == {
+        "calendar": None,
+        "day": 11,
+        "month": "JAN",
+        "year": 2022,
+        "epoch": None,
+    }
+    assert T.DateValue("JAN 2022").parse() == {
+        "calendar": None,
+        "day": None,
+        "month": "JAN",
+        "year": 2022,
+        "epoch": None,
+    }
+    assert T.DateValue("2022").parse() == {
+        "calendar": None,
+        "day": None,
+        "month": None,
+        "year": 2022,
+        "epoch": None,
+    }
 
 
 def test_dateperiod():
@@ -149,6 +170,22 @@ def test_dateperiod():
             "epoch": None,
         },
     }
+    assert T.DatePeriod("FROM 2021 TO 2022").parse() == {
+        "from": {
+            "calendar": None,
+            "year": 2021,
+            "month": None,
+            "day": None,
+            "epoch": None,
+        },
+        "to": {
+            "calendar": None,
+            "year": 2022,
+            "month": None,
+            "day": None,
+            "epoch": None,
+        },
+    }
     assert T.DatePeriod("FROM 1 JAN 2021 TO 11 JAN 2022").parse() == {
         "from": {
             "calendar": None,