From 4ab60cd4fcb2796b3e3b4205b4a9e579eec4f870 Mon Sep 17 00:00:00 2001 From: Al Date: Tue, 2 Aug 2016 03:13:05 -0400 Subject: [PATCH] [osm] Remove boundary names with trailing commas --- scripts/geodata/addresses/components.py | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/scripts/geodata/addresses/components.py b/scripts/geodata/addresses/components.py index 59d640b9..804fa28d 100644 --- a/scripts/geodata/addresses/components.py +++ b/scripts/geodata/addresses/components.py @@ -982,6 +982,19 @@ class AddressComponents(object): if value.isdigit(): address_components.pop(component) + def cleanup_boundary_names(self, address_components): + ''' + Boundary name cleanup + --------------------- + + Cleanup things like addr:city=Rockport, + ''' + for component in list(address_components): + if component not in self.BOUNDARY_COMPONENTS: + continue + + address_components[component] = address_components[component].strip(six.u(', ')) + def prune_duplicate_names(self, address_components): ''' Name deduping @@ -1199,6 +1212,7 @@ class AddressComponents(object): self.cleanup_house_number(address_components) self.remove_numeric_boundary_names(address_components) + self.cleanup_boundary_names(address_components) self.add_house_number_phrase(address_components, language, country=country) self.add_postcode_phrase(address_components, language, country=country)