sanitizer: move helpers into a configuration class

test/python/tokenizer/sanitizers/test_clean_housenumbers.py

@@ -42,3 +42,16 @@ def test_housenumber_lists(sanitize, number):
 def test_filter_kind(sanitize):
     assert sanitize(housenumber='34', number='4', badnumber='65') == \
             [('badnumber', '65'), ('housenumber', '34'), ('housenumber', '4')]
+
+
+@pytest.mark.parametrize('number', ('6523', 'n/a', '4'))
+def test_convert_to_name_converted(number):
+    sanitizer_args = {'step': 'clean-housenumbers',
+                      'convert-to-name': (r'\d+', 'n/a')}
+
+    place = PlaceInfo({'address': {'housenumber': number}})
+    names, address = PlaceSanitizer([sanitizer_args]).process_names(place)
+
+    assert ('housenumber', number) in set((p.kind, p.name) for p in names)
+    assert 'housenumber' not in set(p.kind for p in address)
+

test/python/tokenizer/sanitizers/test_config.py

@@ -5,17 +5,17 @@
 # Copyright (C) 2022 by the Nominatim developer community.
 # For a full list of authors see the git log.
 """
-Tests for sanitizer helper functions.
+Tests for sanitizer configuration helper functions.
 """
 import pytest
 
 from nominatim.errors import UsageError
 from nominatim.tokenizer.place_sanitizer import PlaceName
-import nominatim.tokenizer.sanitizers.helpers as helpers
+from nominatim.tokenizer.sanitizers.config import SanitizerConfig
 
 @pytest.mark.parametrize('inp', ('fg34', 'f\\f', 'morning [glory]', '56.78'))
 def test_create_split_regex_no_params_unsplit(inp):
-    regex = helpers.create_split_regex({})
+    regex = SanitizerConfig().get_delimiter()
 
     assert list(regex.split(inp)) == [inp]
 
@@ -26,14 +26,14 @@ def test_create_split_regex_no_params_unsplit(inp):
                                       ('1,  3  ,5', ['1', '3', '5'])
                                      ])
 def test_create_split_regex_no_params_split(inp, outp):
-    regex = helpers.create_split_regex({})
+    regex = SanitizerConfig().get_delimiter()
 
     assert list(regex.split(inp)) == outp
 
 
 @pytest.mark.parametrize('delimiter', ['.', '\\', '[]', '   ', '/.*+'])
 def test_create_split_regex_custom(delimiter):
-    regex = helpers.create_split_regex({'delimiters': delimiter})
+    regex = SanitizerConfig({'delimiters': delimiter}).get_delimiter()
 
     assert list(regex.split(f'out{delimiter}house')) == ['out', 'house']
     assert list(regex.split('out,house')) == ['out,house']
@@ -41,39 +41,39 @@ def test_create_split_regex_custom(delimiter):
 
 def test_create_split_regex_empty_delimiter():
     with pytest.raises(UsageError):
-        regex = helpers.create_split_regex({'delimiters': ''})
+        regex = SanitizerConfig({'delimiters': ''}).get_delimiter()
 
 
 @pytest.mark.parametrize('inp', ('name', 'name:de', 'na\\me', '.*'))
 def test_create_kind_filter_no_params(inp):
-    filt = helpers.create_kind_filter({})
+    filt = SanitizerConfig().get_filter_kind()
 
     assert filt(PlaceName('something', inp, ''))
 
 
 @pytest.mark.parametrize('kind', ('de', 'name:de', 'ende'))
 def test_create_kind_filter_custom_regex_positive(kind):
-    filt = helpers.create_kind_filter({'filter-kind': '.*de'})
+    filt = SanitizerConfig({'filter-kind': '.*de'}).get_filter_kind()
 
     assert filt(PlaceName('something', kind, ''))
 
 
 @pytest.mark.parametrize('kind', ('de ', '123', '', 'bedece'))
 def test_create_kind_filter_custom_regex_negative(kind):
-    filt = helpers.create_kind_filter({'filter-kind': '.*de'})
+    filt = SanitizerConfig({'filter-kind': '.*de'}).get_filter_kind()
 
     assert not filt(PlaceName('something', kind, ''))
 
 
 @pytest.mark.parametrize('kind', ('name', 'fr', 'name:fr', 'frfr', '34'))
 def test_create_kind_filter_many_positive(kind):
-    filt = helpers.create_kind_filter({'filter-kind': ['.*fr', 'name', r'\d+']})
+    filt = SanitizerConfig({'filter-kind': ['.*fr', 'name', r'\d+']}).get_filter_kind()
 
     assert filt(PlaceName('something', kind, ''))
 
 
 @pytest.mark.parametrize('kind', ('name:de', 'fridge', 'a34', '.*', '\\'))
 def test_create_kind_filter_many_negative(kind):
-    filt = helpers.create_kind_filter({'filter-kind': ['.*fr', 'name', r'\d+']})
+    filt = SanitizerConfig({'filter-kind': ['.*fr', 'name', r'\d+']}).get_filter_kind()
 
     assert not filt(PlaceName('something', kind, ''))