Improve error handling and adjust geom definition criteria

- Added a missing return statement to immediately exit the `lonlat_lookup` function when the `refs` column is unavailable, enhancing error resilience.
- Modified the minimum number of references required to classify geometries as areas or ways: now considering a sequence as an area if it's closed (first and last refs are the same) and consists of at least 4 points (previously 3), and as a way if it includes at least 2 points (previously 3). This adjustment aligns better with common geometric definitions, ensuring that areas have a more defined shape and ways are simplified.
- Replaced a TODO comment with a debug log statement for instances where geometries have less than the intended number of references, improving debuggability and future maintenance.

This change aims to enhance the accuracy of geometrical data processing and improve error handling for better stability and clarity in logs.

Author: Matin Mahmood

earth_osm/utils.py

@@ -38,6 +38,7 @@ def lonlat_lookup(df_way, primary_data):
     """
     if "refs" not in df_way.columns:
         logger.warning("refs column not found")
+        return []
 
     def look(ref):
         lonlat_row = list(map(lambda r: tuple(primary_data["Node"][str(r)]["lonlat"]), ref))
@@ -52,13 +53,12 @@ def way_or_area(df_way):
         raise KeyError("refs column not found")
 
     def check_closed(refs):
-        if (refs[0] == refs[-1]) and (len(refs) >= 3):
+        if (refs[0] == refs[-1]) and (len(refs) >= 4):
             return "area"
-        elif len(refs) >= 3:
+        elif len(refs) >= 2:
             return "way"
         else:
-            # TODO: improve error handling
-            # logger.debug(f"Way with less than 3 refs: {refs}")
+            logger.debug(f"Way with less than 2 refs: {refs}")
             return None
 
     type_list = df_way["refs"].apply(check_closed)