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

[osm] use CLDR country names with random probability

Browse Source
This commit is contained in:
Al
2016-07-28 02:37:12 -04:00
parent 095c808cea
commit 543048bc26
2 changed files with 17 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
resources/parser/data_sets/osm.yaml
View File

@@ -14,6 +14,7 @@ intersections:
places: places:
drop_postcode_probability: 0.6 drop_postcode_probability: 0.6
cldr_country_probability: 0.5
streets: streets:
abbreviate_probability: 0.3 abbreviate_probability: 0.3

16
scripts/geodata/osm/formatter.py
View File

@@ -459,6 +459,8 @@ class OSMAddressFormatter(object):
if num_references > 1000: if num_references > 1000:
num_references = 1000 num_references = 1000
cldr_country_prob = float(nested_get(self.config, ('places', 'cldr_country_probability'), default=0.0))
for name_tag in ('name', 'alt_name', 'loc_name', 'short_name', 'int_name'): for name_tag in ('name', 'alt_name', 'loc_name', 'short_name', 'int_name'):
if more_than_one_official_language: if more_than_one_official_language:
name = tags.get(name_tag) name = tags.get(name_tag)
@@ -477,6 +479,11 @@ class OSMAddressFormatter(object):
language_suffix=language_suffix, language_suffix=language_suffix,
drop_duplicate_city_names=False) drop_duplicate_city_names=False)
if random.random() < cldr_country_prob and AddressFormatter.COUNTRY in address_components:
address_country = self.cldr_country_name(country, language)
if address_country:
address_components[AddressFormatter.COUNTRY] = address_country
place_tags.append((address_components, None, True)) place_tags.append((address_components, None, True))
for language, is_default in local_languages: for language, is_default in local_languages:
@@ -501,6 +508,10 @@ class OSMAddressFormatter(object):
random_key=is_default, random_key=is_default,
language_suffix=language_suffix, language_suffix=language_suffix,
drop_duplicate_city_names=False) drop_duplicate_city_names=False)
if random.random() < cldr_country_prob and AddressFormatter.COUNTRY in address_components:
address_country = self.cldr_country_name(country, language)
if address_country:
address_components[AddressFormatter.COUNTRY] = address_country
place_tags.append((address_components, language, is_default)) place_tags.append((address_components, language, is_default))
@@ -526,6 +537,11 @@ class OSMAddressFormatter(object):
language_suffix=language_suffix, language_suffix=language_suffix,
drop_duplicate_city_names=False) drop_duplicate_city_names=False)
if random.random() < cldr_country_prob and AddressFormatter.COUNTRY in address_components:
address_country = self.cldr_country_name(country, language)
if address_country:
address_components[AddressFormatter.COUNTRY] = address_country
place_tags.append((address_components, language, False)) place_tags.append((address_components, language, False))
if postal_codes: if postal_codes: