Merge pull request #2784 from lonvia/doscs-customizing-icu-tokenizer

[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 480217934e8689ecc49ffc8bb38289169893159e..417d68d2025777b944d1944371dea3d9268c0616 100644 (file)
--- a/nominatim/tokenizer/sanitizers/clean_housenumbers.py
+++ b/nominatim/tokenizer/sanitizers/clean_housenumbers.py
@@ -19,29 +19,37 @@ 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
+from typing import Callable, Iterator, List
+import re
+
+from nominatim.tokenizer.sanitizers.base import ProcessInfo
+from nominatim.data.place_name import PlaceName
+from nominatim.tokenizer.sanitizers.config import SanitizerConfig
 
 class _HousenumberSanitizer:
 
-    def __init__(self, config):
-        self.filter_kind = create_kind_filter(config, 'housenumber')
-        self.split_regexp = create_split_regex(config)
+    def __init__(self, config: SanitizerConfig) -> None:
+        self.filter_kind = config.get_filter_kind('housenumber')
+        self.split_regexp = config.get_delimiter()
 
-        nameregexps = config.get('is-a-name', [])
+        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: ProcessInfo) -> None:
         if not obj.address:
             return
 
-        new_address = []
+        new_address: List[PlaceName] = []
         for item in obj.address:
-            if self.filter_kind(item):
-                if self.treat_as_name(item.name):
+            if self.filter_kind(item.kind):
+                if self._treat_as_name(item.name):
                     obj.names.append(item.clone(kind='housenumber'))
                 else:
                     new_address.extend(item.clone(kind='housenumber', name=n)
@@ -53,7 +61,7 @@ class _HousenumberSanitizer:
         obj.address = new_address
 
 
-    def sanitize(self, value):
+    def sanitize(self, value: str) -> Iterator[str]:
         """ Extract housenumbers in a regularized format from an OSM value.
 
             The function works as a generator that yields all valid housenumbers
@@ -64,16 +72,15 @@ class _HousenumberSanitizer:
                 yield from self._regularize(hnr)
 
 
-    @staticmethod
-    def _regularize(hnr):
+    def _regularize(self, hnr: str) -> Iterator[str]:
         yield hnr
 
 
-    def _treat_as_name(self, housenumber):
+    def _treat_as_name(self, housenumber: str) -> bool:
         return any(r.fullmatch(housenumber) is not None for r in self.is_name_regexp)
 
 
-def create(config):
+def create(config: SanitizerConfig) -> Callable[[ProcessInfo], None]:
     """ Create a housenumber processing function.
     """