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

[formatting] removing the ability to insert city between house number and road in France from discussion in #27

Browse Source
This commit is contained in:
Al
2017-04-08 15:42:59 -04:00
parent 7f7aada32a
commit 4ecd6c23c6
2 changed files with 25 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
resources/formatting/global.yaml
View File

@@ -698,12 +698,6 @@ countries:
unit_before_level: unit_before_level:
before: level before: level
probability: 0.005 probability: 0.005
# libpostal issue #27
city:
city_before_road:
before: road
probability: 0.001
allow_between_house_number_and_road: true
# Andorra, uses same template as France # Andorra, uses same template as France
ad_ca: ad_ca:

54
scripts/geodata/address_formatting/formatter.py
View File

@@ -286,16 +286,14 @@ class AddressFormatter(object):
if k == 'conditional' or not v: if k == 'conditional' or not v:
continue continue
allow_between_house_number_and_road = v.get('allow_between_house_number_and_road', False)
if 'before' in v: if 'before' in v:
val = (self.BEFORE, v['before'], allow_between_house_number_and_road) val = (self.BEFORE, v['before'])
elif 'after' in v: elif 'after' in v:
val = (self.AFTER, v['after'], allow_between_house_number_and_road) val = (self.AFTER, v['after'])
elif 'last' in v: elif 'last' in v:
val = (self.LAST, None, False) val = (self.LAST, None)
elif 'first' in v: elif 'first' in v:
val = (self.FIRST, None, False) val = (self.FIRST, None)
else: else:
raise ValueError('Insertions must contain one of {{first, before, after, last}}. Value was: {}'.format(v)) raise ValueError('Insertions must contain one of {{first, before, after, last}}. Value was: {}'.format(v))
@@ -670,17 +668,17 @@ class AddressFormatter(object):
conditional_insertions = v conditional_insertions = v
break break
order, other, allow_between_house_number_and_road = None, None, False order, other = None, None
# Check the conditional probabilities first # Check the conditional probabilities first
if conditional_insertions is not None: if conditional_insertions is not None:
values, probs = conditional_insertions values, probs = conditional_insertions
order, other, allow_between_house_number_and_road = weighted_choice(values, probs) order, other = weighted_choice(values, probs)
# If there are no conditional probabilites or the "default" value was chosen, sample from the marginals # If there are no conditional probabilites or the "default" value was chosen, sample from the marginals
if other is None: if other is None:
values, probs = insertions values, probs = insertions
order, other, allow_between_house_number_and_road = weighted_choice(values, probs) order, other = weighted_choice(values, probs)
# Even though we may change the value of "other" below, use # Even though we may change the value of "other" below, use
# the original cache key because changes from here on are # the original cache key because changes from here on are
@@ -705,30 +703,28 @@ class AddressFormatter(object):
# house_number, unit, road, which we don't want. So effectively # house_number, unit, road, which we don't want. So effectively
# treat house_number and road as an atomic unit. # treat house_number and road as an atomic unit.
if not allow_between_house_number_and_road: if other == self.HOUSE_NUMBER and component != self.ROAD:
road_tag = self.tag_token(self.ROAD)
house_number_tag = other_token
if other == self.HOUSE_NUMBER and component != self.ROAD: if house_number_tag in template and road_tag in template:
road_tag = self.tag_token(self.ROAD) road_after_house_number = template.index(road_tag) > template.index(house_number_tag)
house_number_tag = other_token
if house_number_tag in template and road_tag in template: if road_after_house_number and order == self.AFTER:
road_after_house_number = template.index(road_tag) > template.index(house_number_tag) other = self.ROAD
elif not road_after_house_number and order == self.BEFORE:
other = self.ROAD
elif other == self.ROAD and component != self.HOUSE_NUMBER:
house_number_tag = self.tag_token(self.HOUSE_NUMBER)
road_tag = other_token
if road_after_house_number and order == self.AFTER: if house_number_tag in template and road_tag in template:
other = self.ROAD road_before_house_number = template.index(road_tag) < template.index(house_number_tag)
elif not road_after_house_number and order == self.BEFORE:
other = self.ROAD
elif other == self.ROAD and component != self.HOUSE_NUMBER:
house_number_tag = self.tag_token(self.HOUSE_NUMBER)
road_tag = other_token
if house_number_tag in template and road_tag in template: if road_before_house_number and order == self.AFTER:
road_before_house_number = template.index(road_tag) < template.index(house_number_tag) other = self.HOUSE_NUMBER
elif not road_before_house_number and order == self.BEFORE:
if road_before_house_number and order == self.AFTER: other = self.HOUSE_NUMBER
other = self.HOUSE_NUMBER
elif not road_before_house_number and order == self.BEFORE:
other = self.HOUSE_NUMBER
if order == self.BEFORE and other_token in template: if order == self.BEFORE and other_token in template:
template = self.insert_component(template, component, before=other) template = self.insert_component(template, component, before=other)