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

[openaddresses] Replacing number sign in house number

Browse Source
This commit is contained in:
Al
2016-08-27 02:42:06 -04:00
parent caa6a405c5
commit 15f9817933
1 changed files with 3 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
scripts/geodata/openaddresses/formatter.py
View File

@@ -31,6 +31,7 @@ OPENADDRESS_FORMAT_DATA_FILENAME = 'openaddresses_formatted_addresses.tsv'
numeric_range_regex = re.compile('[\s]*\-[\s]*') numeric_range_regex = re.compile('[\s]*\-[\s]*')
null_regex = re.compile('^\s*(?:null|none)\s*$', re.I) null_regex = re.compile('^\s*(?:null|none)\s*$', re.I)
unknown_regex = re.compile('^\s*(?:unknown)\s*$', re.I) unknown_regex = re.compile('^\s*(?:unknown)\s*$', re.I)
number_sign_regex = re.compile('^#', re.UNICODE)
not_applicable_regex = re.compile('^\s*n\.?\s*/?\s*a\.?\s*$', re.I) not_applicable_regex = re.compile('^\s*n\.?\s*/?\s*a\.?\s*$', re.I)
@@ -96,7 +97,7 @@ class OpenAddressesFormatter(object):
house_number = int(house_number.strip()) house_number = int(house_number.strip())
return house_number > 0 return house_number > 0
except (ValueError, TypeError): except (ValueError, TypeError):
return house_number.strip() and is_numeric(house_number) and not all((c == '0' for c in house_number if c.isdigit())) return house_number.strip('# ') and is_numeric(house_number) and not all((c == '0' for c in house_number if c.isdigit()))
component_validators = { component_validators = {
AddressFormatter.HOUSE_NUMBER: validators.validate_house_number, AddressFormatter.HOUSE_NUMBER: validators.validate_house_number,
@@ -256,6 +257,7 @@ class OpenAddressesFormatter(object):
house_number = components.get(AddressFormatter.HOUSE_NUMBER, None) house_number = components.get(AddressFormatter.HOUSE_NUMBER, None)
if house_number: if house_number:
house_number = numeric_range_regex.replace(six.u('-'), house_number).strip() house_number = numeric_range_regex.replace(six.u('-'), house_number).strip()
house_number = number_sign_regex.replace(six.u(''), house_number)
house_number = self.cleanup_number(house_number) house_number = self.cleanup_number(house_number)
postcode = components.get(AddressFormatter.POSTCODE, None) postcode = components.get(AddressFormatter.POSTCODE, None)