tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[formatting] Constants for field names, a few options in format_address

Browse Source
This commit is contained in:
Al
2015-09-29 23:03:37 -04:00
parent 22e8178a97
commit 40cf247655
1 changed files with 50 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

93
scripts/geodata/address_formatting/formatter.py
View File

@@ -28,12 +28,6 @@ class AddressFormatter(object):
address_formatter.format_address('es', components) address_formatter.format_address('es', components)
''' '''
MINIMAL_COMPONENT_KEYS = [
('road', 'house_number'),
('road', 'house'),
('road', 'postcode')
]
whitespace_component_regex = re.compile('[\r\n]+[\s\r\n]*') whitespace_component_regex = re.compile('[\r\n]+[\s\r\n]*')
splitter = ' | ' splitter = ' | '
@@ -41,37 +35,54 @@ class AddressFormatter(object):
separator_tag = 'SEP' separator_tag = 'SEP'
field_separator_tag = 'FSEP' field_separator_tag = 'FSEP'
HOUSE = 'house'
HOUSE_NUMBER = 'house_number'
ROAD = 'road'
SUBURB = 'suburb'
CITY = 'city'
STATE = 'state'
POSTCODE = 'postcode'
COUNTRY = 'country'
aliases = OrderedDict([ aliases = OrderedDict([
('name', 'house'), ('name', HOUSE),
('addr:housename', 'house'), ('addr:housename', HOUSE),
('addr:housenumber', 'house_number'), ('addr:housenumber', HOUSE_NUMBER),
('addr:house_number', 'house_number'), ('addr:house_number', HOUSE_NUMBER),
('addr:street', 'road'), ('addr:street', ROAD),
('addr:city', 'city'), ('addr:city', CITY),
('addr:locality', 'city'), ('addr:locality', CITY),
('addr:municipality', 'city'), ('addr:municipality', CITY),
('addr:hamlet', 'village'), ('addr:hamlet', CITY),
('addr:suburb', 'suburb'), ('addr:suburb', SUBURB),
('addr:neighbourhood', 'suburb'), ('addr:neighbourhood', SUBURB),
('addr:neighborhood', 'suburb'), ('addr:neighborhood', SUBURB),
('addr:district', 'suburb'), ('addr:district', SUBURB),
('addr:state', 'state'), ('addr:state', STATE),
('addr:province', 'state'), ('addr:province', STATE),
('addr:region', 'state'), ('addr:region', STATE),
('addr:postal_code', 'postcode'), ('addr:postal_code', POSTCODE),
('addr:postcode', 'postcode'), ('addr:postcode', POSTCODE),
('addr:country', 'country'), ('addr:country', COUNTRY),
('street', 'road'), ('street', ROAD),
('street_name', 'road'), ('street_name', ROAD),
('residential', 'road'), ('residential', ROAD),
('hamlet', 'village'), ('hamlet', CITY),
('neighborhood', 'suburb'), ('neighborhood', SUBURB),
('neighbourhood', 'suburb'), ('neighbourhood', SUBURB),
('city_district', 'suburb'), ('city_district', SUBURB),
('state_code', 'state'), ('state_code', STATE),
('country_name', 'country'), ('country_name', COUNTRY),
('postal_code', POSTCODE),
('post_code', POSTCODE),
]) ])
MINIMAL_COMPONENT_KEYS = [
(ROAD, HOUSE_NUMBER),
(ROAD, HOUSE),
(ROAD, POSTCODE)
]
def __init__(self, scratch_dir='/tmp', splitter=None): def __init__(self, scratch_dir='/tmp', splitter=None):
if splitter is not None: if splitter is not None:
self.splitter = splitter self.splitter = splitter
@@ -213,20 +224,16 @@ class AddressFormatter(object):
return u' '.join(tokens[start:end]) return u' '.join(tokens[start:end])
def format_address(self, country, components, minimal_only=True, tag_components=True): def format_address(self, country, components, minimal_only=True, tag_components=True, replace_aliases=True):
template = self.config.get(country.upper()) template = self.config.get(country.upper())
if not template: if not template:
return None return None
template_text = template['address_template'] template_text = template['address_template']
self.replace_aliases(components) if replace_aliases:
self.replace_aliases(components)
if not self.minimal_components(components): if minimal_only and not self.minimal_components(components):
if minimal_only: return None
return None
if 'fallback_template' in template:
template_text = template['fallback_template']
else:
template_text = self.config['default']['fallback_template']
self.apply_replacements(template, components) self.apply_replacements(template, components)