add support for external sanitizer modules

This commit is contained in:
Sarah Hoffmann
2022-07-25 16:10:19 +02:00
parent 7b7203c149
commit 6d41046b15
9 changed files with 142 additions and 94 deletions

View File

@@ -12,39 +12,45 @@ import pytest
from nominatim.tokenizer.place_sanitizer import PlaceSanitizer
from nominatim.data.place_info import PlaceInfo
def run_sanitizer_on(**kwargs):
place = PlaceInfo({'name': kwargs})
name, _ = PlaceSanitizer([{'step': 'strip-brace-terms'}]).process_names(place)
class TestStripBrace:
return sorted([(p.name, p.kind, p.suffix) for p in name])
@pytest.fixture(autouse=True)
def setup_country(self, def_config):
self.config = def_config
def run_sanitizer_on(self, **kwargs):
place = PlaceInfo({'name': kwargs})
name, _ = PlaceSanitizer([{'step': 'strip-brace-terms'}], self.config).process_names(place)
return sorted([(p.name, p.kind, p.suffix) for p in name])
def test_no_braces():
assert run_sanitizer_on(name='foo', ref='23') == [('23', 'ref', None),
('foo', 'name', None)]
def test_no_braces(self):
assert self.run_sanitizer_on(name='foo', ref='23') == [('23', 'ref', None),
('foo', 'name', None)]
def test_simple_braces():
assert run_sanitizer_on(name='Halle (Saale)', ref='3')\
== [('3', 'ref', None), ('Halle', 'name', None), ('Halle (Saale)', 'name', None)]
assert run_sanitizer_on(name='ack ( bar')\
== [('ack', 'name', None), ('ack ( bar', 'name', None)]
def test_simple_braces(self):
assert self.run_sanitizer_on(name='Halle (Saale)', ref='3')\
== [('3', 'ref', None), ('Halle', 'name', None), ('Halle (Saale)', 'name', None)]
assert self.run_sanitizer_on(name='ack ( bar')\
== [('ack', 'name', None), ('ack ( bar', 'name', None)]
def test_only_braces():
assert run_sanitizer_on(name='(maybe)') == [('(maybe)', 'name', None)]
def test_only_braces(self):
assert self.run_sanitizer_on(name='(maybe)') == [('(maybe)', 'name', None)]
def test_double_braces():
assert run_sanitizer_on(name='a((b))') == [('a', 'name', None),
('a((b))', 'name', None)]
assert run_sanitizer_on(name='a (b) (c)') == [('a', 'name', None),
('a (b) (c)', 'name', None)]
def test_double_braces(self):
assert self.run_sanitizer_on(name='a((b))') == [('a', 'name', None),
('a((b))', 'name', None)]
assert self.run_sanitizer_on(name='a (b) (c)') == [('a', 'name', None),
('a (b) (c)', 'name', None)]
def test_no_names():
def test_no_names(def_config):
place = PlaceInfo({'address': {'housenumber': '3'}})
name, address = PlaceSanitizer([{'step': 'strip-brace-terms'}]).process_names(place)
name, address = PlaceSanitizer([{'step': 'strip-brace-terms'}], def_config).process_names(place)
assert not name
assert len(address) == 1