Document personal data YAML formats

validate_yaml.py

@@ -36,6 +36,23 @@ def check_currency(item: agenda.types.StrDict) -> None:
     sys.exit(-1)
 
 
+def check_country_code(
+    item: agenda.types.StrDict, source: str, required: bool = True
+) -> None:
+    """Throw error if country code is missing or invalid."""
+    country = item.get("country")
+    if country is None:
+        if not required:
+            return
+        pprint(item)
+        print(f"{source} missing country")
+        sys.exit(-1)
+    if not isinstance(country, str) or not agenda.get_country(country):
+        pprint(item)
+        print(f"{source} has invalid country {country!r}")
+        sys.exit(-1)
+
+
 def get_coords(item: agenda.types.StrDict) -> LatLon | None:
     """Return latitude/longitude tuple when present."""
     if "latitude" in item and "longitude" in item:
@@ -267,6 +284,8 @@ def check_conferences() -> None:
             print(f"currency {conf.currency!r} not in {currencies!r}")
             sys.exit(-1)
 
+        check_country_code(conf_data, "conference", required=False)
+
         current_start = normalize_datetime(conf_data["start"])
         if prev_start and current_start < prev_start:
             assert prev_conf_data is not None
@@ -290,6 +309,11 @@ def check_events() -> None:
     last_year = today - timedelta(days=365)
     next_year = today + timedelta(days=2 * 365)
 
+    filepath = os.path.join(data_dir, "events.yaml")
+    events_data = yaml.safe_load(open(filepath, "r"))
+    for event in events_data:
+        check_country_code(event, "event", required=False)
+
     events = agenda.events_yaml.read(data_dir, last_year, next_year)
     print(len(events), "events")
 
@@ -314,6 +338,7 @@ def check_accommodation() -> None:
     for stay in accommodation_list:
         try:
             assert all(field in stay for field in required_fields)
+            check_country_code(stay, "accommodation")
             check_coordinates(stay)
         except AssertionError:
             pprint(stay)
@@ -347,8 +372,7 @@ def check_airports() -> None:
     )
     print(len(airports), "airports")
     for airport in airports.values():
-        assert "country" in airport
-        assert agenda.get_country(airport["country"])
+        check_country_code(airport, "airport")
 
 
 def check_stations() -> None:
@@ -356,8 +380,31 @@ def check_stations() -> None:
     stations = agenda.travel.parse_yaml("stations", data_dir)
     print(len(stations), "stations")
     for station in stations:
-        assert "country" in station
-        assert agenda.get_country(station["country"])
+        check_country_code(station, "station")
+
+
+def check_ferry_terminals() -> None:
+    """Check ferry terminals."""
+    terminals = agenda.travel.parse_yaml("ferry_terminals", data_dir)
+    print(len(terminals), "ferry terminals")
+    for terminal in terminals:
+        check_country_code(terminal, "ferry terminal")
+
+
+def check_bus_stops() -> None:
+    """Check bus stops."""
+    stops = agenda.travel.parse_yaml("bus_stops", data_dir)
+    print(len(stops), "bus stops")
+    for stop in stops:
+        check_country_code(stop, "bus stop")
+
+
+def check_coach_stations() -> None:
+    """Check coach stations."""
+    stations = agenda.travel.parse_yaml("coach_stations", data_dir)
+    print(len(stations), "coach stations")
+    for station in stations:
+        check_country_code(station, "coach station")
 
 
 def check_ferries() -> None:
@@ -415,8 +462,8 @@ def check_buses() -> None:
 
 def check_airlines() -> list[agenda.types.StrDict]:
     """Check airlines."""
-    airlines = typing.cast(
-        list[agenda.types.StrDict], agenda.travel.parse_yaml("airlines", data_dir)
+    airlines: list[agenda.types.StrDict] = agenda.travel.parse_yaml(
+        "airlines", data_dir
     )
     print(len(airlines), "airlines")
     for airline in airlines:
@@ -450,6 +497,9 @@ def check() -> None:
     check_accommodation()
     check_airports()
     check_stations()
+    check_ferry_terminals()
+    check_bus_stops()
+    check_coach_stations()
 
 
 if __name__ == "__main__":