From 12bc18f74b3cb77912aadc653958ee324f45ad18 Mon Sep 17 00:00:00 2001 From: Al Date: Sat, 28 Jan 2017 02:03:19 -0500 Subject: [PATCH] [openaddresses] fix Chinese house number validation --- resources/parser/data_sets/openaddresses.yaml | 5 +---- scripts/geodata/openaddresses/formatter.py | 6 ++++-- 2 files changed, 5 insertions(+), 6 deletions(-) diff --git a/resources/parser/data_sets/openaddresses.yaml b/resources/parser/data_sets/openaddresses.yaml index 6f5ea558..be706c95 100644 --- a/resources/parser/data_sets/openaddresses.yaml +++ b/resources/parser/data_sets/openaddresses.yaml @@ -789,10 +789,7 @@ countries: - filename: wuhan.csv add_osm_boundaries: true fields: - NUMBER: - component: house_number - STREET: - component: road + <<: *default_fields DISTIRCT: component: city_district cz: diff --git a/scripts/geodata/openaddresses/formatter.py b/scripts/geodata/openaddresses/formatter.py index beeb495c..46f202ef 100644 --- a/scripts/geodata/openaddresses/formatter.py +++ b/scripts/geodata/openaddresses/formatter.py @@ -150,7 +150,10 @@ class OpenAddressesFormatter(object): if not house_number: return False tokens = tokenize(house_number) - return all((c in token_types.NUMERIC_TOKEN_TYPES or t in (u'号', u'栋', u'附')) for t, c in tokens) + + if all((c in token_types.NUMERIC_TOKEN_TYPES or t in (u'号', u'栋', u'附')) for t, c in tokens): + return True + return cls.validate_house_number(house_number) component_validators = { AddressFormatter.HOUSE_NUMBER: validators.validate_house_number, @@ -158,7 +161,6 @@ class OpenAddressesFormatter(object): AddressFormatter.POSTCODE: validators.validate_postcode, } - language_validators = { SPANISH: { AddressFormatter.HOUSE_NUMBER: validators.validate_house_number_sin_numero,