Allow any Hijri year (drop 3+ digit year constraint and year-month-day)

Author: rlskoeser

src/undate/converters/calendars/hijri/hijri.lark

@@ -1,19 +1,17 @@
 %import common.WS
 %ignore WS
 
-hijri_date: year | month year | day month year | year month | year month day
-// NOTE: ISMI sample dates are year month day
-// if we can assume years are 3 digits minimum, we can support year month day AND we can use faster LALR parser
+// only support day month year format for now
+// parser requires numeric day and year to be distinguished based on order
+hijri_date: day month year | month year | year
 
 // TODO: handle date ranges?
 
 // TODO: add support for qualifiers?
 // PGP dates use qualifiers like "first decade of" (for beginning of month)
 // "first third of", seasons (can look for more examples)
 
-// TODO: is there a minimum year we need to support?
-// if we can assume 3+ digit years we can distinguish between days and years,
-year: /\d{3,}/
+year: /\d+/
 
 // months
 month: month_1

src/undate/converters/calendars/hijri/parser.py

@@ -5,5 +5,5 @@
 grammar_path = pathlib.Path(__file__).parent / "hijri.lark"
 
 with open(grammar_path) as grammar:
-    # NOTE: LALR parser is faster but requires assumption of 3+ digit years
-    hijri_parser = Lark(grammar.read(), start="hijri_date", strict=True, parser="lalr")
+    # NOTE: LALR parser is faster but can't be used to ambiguity between years and dates
+    hijri_parser = Lark(grammar.read(), start="hijri_date", strict=True)

tests/test_converters/test_calendars/test_hijri/test_hijri_parser.py

@@ -39,10 +39,11 @@
     "7 Jumādā I 1243",
     "29 Muḥarram 1243",
     "30 Muḥarram 1243",
-    # year month, if we can assume 3+ digit years
-    "901 Rabīʿ I",
-    # year month day
-    "901 Rabīʿ I 12",
+    "Rabīʿ I 901",
+    "12 Rabīʿ I 901",
+    # two and 1 digit years
+    "12 Rabīʿ I 90",
+    "12 Rabīʿ I 9",
 ]
 
 
@@ -63,6 +64,9 @@ def test_should_parse(date_string):
     "Foo 383",
     # wrong format
     "2024-10-02",
+    # year month day not supported
+    "901 Rabīʿ I",
+    "901 Rabīʿ I 12",
 ]