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

[openaddresses] Better handling of float postcodes

Browse Source
This commit is contained in:
Al
2016-08-26 20:10:30 -04:00
parent 0618e506c4
commit 0f9e8ee95d
1 changed files with 21 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
scripts/geodata/openaddresses/formatter.py
View File

@@ -117,6 +117,26 @@ class OpenAddressesFormatter(object):
return country_name
def cleanup_postcode(self, postcode):
postcode = postcode.strip()
try:
postcode_int = int(postcode)
except (ValueError, TypeError):
try:
postcode_float = float(postcode)
num_leading_zeros = 0
for c in postcode:
if c == six.u('0'):
num_leading_zeros += 1
else:
break
postcode = safe_decode(int(postcode_float))
if num_leading_zeros:
postcode = six.u('{}{}').format(six.u('0') * num_leading_zeros, postcode)
except (ValueError, TypeError):
pass
return postcode
def formatted_addresses(self, path, configs, tag_components=True):
abbreviate_street_prob = float(self.get_property('abbreviate_street_probability', *configs))
separate_street_prob = float(self.get_property('separate_street_probability', *configs) or 0.0)
@@ -208,15 +228,7 @@ class OpenAddressesFormatter(object):
postcode = components.get(AddressFormatter.POSTCODE, None)
if postcode:
postcode = postcode.strip()
try:
postcode = int(postcode)
except (ValueError, TypeError):
try:
postcode = float(postcode)
postcode = safe_decode(int(postcode))
except (ValueError, TypeError):
postcode = safe_decode(postcode)
postcode = self.cleanup_postcode(postcode)
if postcode_strip_non_digit_chars:
postcode = six.u('').join((c for c in postcode if c.isdigit()))