move postcode matcher in a separate file

[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 9777a7fcdcc6dc74103f727644cb8da5f81beaa9..c229716f8e8353d8d8788670b767ea0c9cebc22f 100644 (file)
--- a/nominatim/tokenizer/sanitizers/clean_housenumbers.py
+++ b/nominatim/tokenizer/sanitizers/clean_housenumbers.py
@@ -5,7 +5,11 @@
 # Copyright (C) 2022 by the Nominatim developer community.
 # For a full list of authors see the git log.
 """
 # Copyright (C) 2022 by the Nominatim developer community.
 # For a full list of authors see the git log.
 """

-Sanitizer that cleans and normalizes house numbers.
+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
 
 Arguments:
     delimiters: Define the set of characters to be used for


@@ -15,15 +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')
                  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):
 
 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):
 
 
     def __call__(self, obj):


@@ -33,7 +44,11 @@ class _HousenumberSanitizer:
         new_address = []
         for item in obj.address:
             if self.filter_kind(item):
         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)
             else:
                 # Don't touch other address items.
                 new_address.append(item)


@@ -52,10 +67,15 @@ class _HousenumberSanitizer:
                 yield from self._regularize(hnr)
 
 
                 yield from self._regularize(hnr)
 
 

-    def _regularize(self, hnr):
+    @staticmethod
+    def _regularize(hnr):

         yield 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.
     """
 def create(config):
     """ Create a housenumber processing function.
     """