X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/1e5a8561c09f65e390ec51f2322919dd187bfbdf..baf1b955fb27e9b4e4c7c1146b87627e13be6ac1:/nominatim/tokenizer/sanitizers/clean_housenumbers.py?ds=sidebyside diff --git a/nominatim/tokenizer/sanitizers/clean_housenumbers.py b/nominatim/tokenizer/sanitizers/clean_housenumbers.py index 49f9b4f0..c229716f 100644 --- a/nominatim/tokenizer/sanitizers/clean_housenumbers.py +++ b/nominatim/tokenizer/sanitizers/clean_housenumbers.py @@ -19,15 +19,22 @@ Arguments: where each string is a regular expression. An address item is considered a house number if the 'kind' fully matches any of the given regular expressions. (default: 'housenumber') - + convert-to-name: Define house numbers that should be treated as a name + instead of a house number. Either takes a single string + or a list of strings, where each string is a regular + expression that must match the full house number value. """ -from nominatim.tokenizer.sanitizers.helpers import create_split_regex, create_kind_filter +import re class _HousenumberSanitizer: def __init__(self, config): - self.filter_kind = create_kind_filter(config, 'housenumber') - self.split_regexp = create_split_regex(config) + self.filter_kind = config.get_filter_kind('housenumber') + self.split_regexp = config.get_delimiter() + + nameregexps = config.get_string_list('convert-to-name', []) + self.is_name_regexp = [re.compile(r) for r in nameregexps] + def __call__(self, obj): @@ -37,8 +44,11 @@ class _HousenumberSanitizer: new_address = [] for item in obj.address: if self.filter_kind(item): - new_address.extend(item.clone(kind='housenumber', name=n) - for n in self.sanitize(item.name)) + if self._treat_as_name(item.name): + obj.names.append(item.clone(kind='housenumber')) + else: + new_address.extend(item.clone(kind='housenumber', name=n) + for n in self.sanitize(item.name)) else: # Don't touch other address items. new_address.append(item) @@ -62,6 +72,10 @@ class _HousenumberSanitizer: yield hnr + def _treat_as_name(self, housenumber): + return any(r.fullmatch(housenumber) is not None for r in self.is_name_regexp) + + def create(config): """ Create a housenumber processing function. """