[openaddresses] adding clear_cache method to clear the LRU cache for point-in-polygon indices and using it in OpenAddresses import since it heavily reuses polygons and only for the current file

scripts/geodata/openaddresses/formatter.py

@@ -255,6 +255,9 @@ class OpenAddressesFormatter(object):
         latitude_index = headers.index('LAT')
         longitude_index = headers.index('LON')
 
+        # Clear cached polygons
+        self.components.osm_admin_rtree.clear_cache()
+
         for row in reader:
             try:
                 latitude = float(row[latitude_index])

scripts/geodata/polygons/index.py

@@ -1,4 +1,5 @@
 import fiona
+import gc
 import geohash
 import os
 import rtree
@@ -86,6 +87,12 @@ class PolygonIndex(object):
     def setup(self):
         pass
 
+    def clear_cache(self, garbage_collect=True):
+        if self.persistent_polygons and self.cache_size > 0:
+            self.polygons.clear()
+            if garbage_collect:
+                gc.collect()
+
     def simplify_polygon(self, poly, simplify_tolerance=None, preserve_topology=None):
         if simplify_tolerance is None:
             simplify_tolerance = self.simplify_tolerance