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

[osm] fix names with pipes in them

Browse Source
This commit is contained in:
Al
2016-10-17 02:32:25 -04:00
parent a796b41d90
commit d34faf42b8
1 changed files with 10 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
scripts/geodata/osm/formatter.py
View File

@@ -566,6 +566,9 @@ class OSMAddressFormatter(object):
elif six.u(',') in name: elif six.u(',') in name:
name = name.split(six.u(','), 1)[0] name = name.split(six.u(','), 1)[0]
if six.u('|') in name:
name = name.replace(six.u('|'), six.u(''))
for i in xrange(num_references if name_tag == 'name' else 1): for i in xrange(num_references if name_tag == 'name' else 1):
address_components = {component_name: name.strip()} address_components = {component_name: name.strip()}
self.components.add_admin_boundaries(address_components, osm_components, country, UNKNOWN_LANGUAGE, self.components.add_admin_boundaries(address_components, osm_components, country, UNKNOWN_LANGUAGE,
@@ -591,6 +594,9 @@ class OSMAddressFormatter(object):
elif six.u(',') in name: elif six.u(',') in name:
name = name.split(six.u(','), 1)[0] name = name.split(six.u(','), 1)[0]
if six.u('|') in name:
name = name.replace(six.u('|'), six.u(''))
n = min_references / 2 n = min_references / 2
if name_tag == 'name': if name_tag == 'name':
if is_default: if is_default:
@@ -622,6 +628,9 @@ class OSMAddressFormatter(object):
elif six.u(',') in name: elif six.u(',') in name:
name = name.split(six.u(','), 1)[0] name = name.split(six.u(','), 1)[0]
if six.u('|') in name:
name = name.replace(six.u('|'), six.u(''))
# Add half as many English records as the local language, every other language gets min_referenes / 2 # Add half as many English records as the local language, every other language gets min_referenes / 2
for i in xrange(num_references / 2 if language == ENGLISH else min_references / 2): for i in xrange(num_references / 2 if language == ENGLISH else min_references / 2):
address_components = {component_name: name.strip()} address_components = {component_name: name.strip()}