Merge remote-tracking branch 'upstream/master'

[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..c229716f8e8353d8d8788670b767ea0c9cebc22f 100644 (file)
--- a/nominatim/tokenizer/sanitizers/clean_housenumbers.py
+++ b/nominatim/tokenizer/sanitizers/clean_housenumbers.py
@@ -5,20 +5,36 @@
 # 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')
+    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
+import re
 
 class _HousenumberSanitizer:
 
     def __init__(self, config):
-        self.kinds = config.get('filter-kind', ('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):
@@ -27,8 +43,12 @@ 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):
+                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)
@@ -47,10 +67,15 @@ class _HousenumberSanitizer:
                 yield from self._regularize(hnr)
 
 
-    def _regularize(self, hnr):
+    @staticmethod
+    def _regularize(hnr):
         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.
     """