fix linting issues

[nominatim.git] / nominatim / tokenizer / sanitizers / clean_housenumbers.py

diff --git a/nominatim/tokenizer/sanitizers/clean_housenumbers.py b/nominatim/tokenizer/sanitizers/clean_housenumbers.py
index b65880c38e5179e167b8a2711e31b079e900d8ab..49f9b4f05c6c14c237e46c76d99be6711abeff23 100644 (file)
--- a/nominatim/tokenizer/sanitizers/clean_housenumbers.py
+++ b/nominatim/tokenizer/sanitizers/clean_housenumbers.py
@@ -5,19 +5,28 @@
 # Copyright (C) 2022 by the Nominatim developer community.
 # For a full list of authors see the git log.
 """
-Sanitizer that cleans and normalizes housenumbers.
+Sanitizer that preprocesses address tags for house numbers. The sanitizer
+allows to
+
+* define which tags are to be considered house numbers (see 'filter-kind')
+* split house number lists into individual numbers (see 'delimiters')
 
 Arguments:
     delimiters: Define the set of characters to be used for
-                splitting a list of housenumbers into parts. (default: ',;')
+                splitting a list of house numbers into parts. (default: ',;')
+    filter-kind: Define the address tags that are considered to be a
+                 house number. Either takes a single string or a list of strings,
+                 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')
 
 """
-from nominatim.tokenizer.sanitizers.helpers import create_split_regex
+from nominatim.tokenizer.sanitizers.helpers import create_split_regex, create_kind_filter
 
 class _HousenumberSanitizer:
 
     def __init__(self, config):
-        self.kinds = config.get('filter-kind', ('housenumber', ))
+        self.filter_kind = create_kind_filter(config, 'housenumber')
         self.split_regexp = create_split_regex(config)
 
 
@@ -27,8 +36,9 @@ class _HousenumberSanitizer:
 
         new_address = []
         for item in obj.address:
-            if item.kind in self.kinds:
-                new_address.extend(item.clone(kind='housenumber', name=n) for n in self.sanitize(item.name))
+            if self.filter_kind(item):
+                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)
@@ -47,7 +57,8 @@ class _HousenumberSanitizer:
                 yield from self._regularize(hnr)
 
 
-    def _regularize(self, hnr):
+    @staticmethod
+    def _regularize(hnr):
         yield hnr