fix(ingest): ignore pre-Web inception dates as website launch dates

P571 is the owner's inception (e.g. a newspaper founded in 1785), so dates
before 1991 are skipped for launch_date.

Refs #99

Author: Seungpyo1007

app/ingest/wikidata_fill.py

@@ -27,6 +27,7 @@
                 "P127": "owners"},
 }
 DATES = {"release_date", "launch_date"}
+WEB_EPOCH = "1991-01-01"
 PROPERTIES = {prop for mapping in MAPPINGS.values() for prop in mapping}
 
 
@@ -159,6 +160,9 @@ def fill(record: dict[str, Any], category: str, entity: dict[str, Any],
         candidates = values(entity, prop)
         if field in DATES:
             dates = [parsed for v in candidates if (parsed := calendar_date(v))]
+            if field == "launch_date":
+                # P571 is the owner's inception; before the Web it can't be a site launch.
+                dates = [d for d in dates if d >= WEB_EPOCH]
             if dates:
                 updated[field] = min(dates)
         elif field == "homepage_url":

tests/test_wikidata_fill.py

@@ -164,3 +164,10 @@ def handler(request):
         assert len(cache.fetch(f"Q{i}" for i in range(1, 51))) == 50
     assert sizes == [50, 25, 24]
     assert len(list(tmp_path.glob("*.json"))) == 50
+
+
+def test_pre_web_inception_is_not_a_website_launch():
+    entity = {"id": "Q1", "claims": {"P571": [claim(timestamp(time="+1785-01-01T00:00:00Z"))]}}
+    assert "launch_date" not in fill({}, "website", entity, {})
+    entity = {"id": "Q1", "claims": {"P577": [claim(timestamp(time="+1985-11-20T00:00:00Z"))]}}
+    assert fill({}, "software", entity, {})["release_date"] == "1985-11-20"