From: Paweł Wroniszewski Date: Fri, 20 Oct 2023 10:39:47 +0000 (+0200) Subject: Revert country settings X-Git-Tag: v4.4.0~90^2 X-Git-Url: https://git.openstreetmap.org./nominatim.git/commitdiff_plain/2cae37ccdee1fc5bad88a492af91020ad4060069 Revert country settings --- diff --git a/settings/country_settings.yaml b/settings/country_settings.yaml index bda3b9e7..667684c6 100644 --- a/settings/country_settings.yaml +++ b/settings/country_settings.yaml @@ -4,7 +4,8 @@ ad: languages: ca names: !include country-names/ad.yaml postcode: - pattern: "ADddd" + pattern: "(ddd)" + output: AD\1 # United Arab Emirates (الإمارات العربية المتحدة) @@ -38,7 +39,8 @@ ai: languages: en names: !include country-names/ai.yaml postcode: - pattern: "AI-2640" + pattern: "2640" + output: AI-2640 # Albania (Shqipëria) @@ -963,7 +965,8 @@ ky: languages: en names: !include country-names/ky.yaml postcode: - pattern: "KYd-dddd" + pattern: "(d)-(dddd)" + output: KY\1-\2 # Kazakhstan (Қазақстан) @@ -999,7 +1002,7 @@ lc: languages: en names: !include country-names/lc.yaml postcode: - pattern: "LC(dd) ?(ddd)" + pattern: "(dd) ?(ddd)" output: LC\1 \2 @@ -1063,7 +1066,8 @@ lv: languages: lv names: !include country-names/lv.yaml postcode: - pattern: "LV-dddd" + pattern: "(dddd)" + output: LV-\1 # Libya (ليبيا) @@ -1098,7 +1102,8 @@ md: languages: ro, ru, uk names: !include country-names/md.yaml postcode: - pattern: "MD-dddd" + pattern: "(dddd)" + output: MD-\1 # Montenegro (Crna Gora / Црна Гора) @@ -1840,7 +1845,8 @@ vc: languages: en names: !include country-names/vc.yaml postcode: - pattern: "VCdddd" + pattern: "(dddd)" + output: VC\1 # Venezuela (Venezuela) @@ -1858,7 +1864,8 @@ vg: languages: en names: !include country-names/vg.yaml postcode: - pattern: "VGdddd" + pattern: "(dddd)" + output: VG\1 # Vietnam (Việt Nam) diff --git a/test/python/tokenizer/sanitizers/test_clean_postcodes.py b/test/python/tokenizer/sanitizers/test_clean_postcodes.py index 63d77202..14498cbc 100644 --- a/test/python/tokenizer/sanitizers/test_clean_postcodes.py +++ b/test/python/tokenizer/sanitizers/test_clean_postcodes.py @@ -89,23 +89,23 @@ def test_postcode_sweden_fail(sanitize, postcode): assert sanitize(country='se', postcode=postcode) == [] -@pytest.mark.parametrize("postcode", ('AD123', 'AD AD123')) +@pytest.mark.parametrize("postcode", ('AD123', '123', 'AD 123', 'AD-123')) def test_postcode_andorra_pass(sanitize, postcode): assert sanitize(country='ad', postcode=postcode) == [('postcode', 'AD123')] -@pytest.mark.parametrize("postcode", ('123', 'AD 123', 'AD-123', 'AD1234')) +@pytest.mark.parametrize("postcode", ('AD1234', 'AD AD123', 'XX123')) @pytest.mark.sanitizer_params(convert_to_address=False) def test_postcode_andorra_fail(sanitize, postcode): assert sanitize(country='ad', postcode=postcode) == [] -@pytest.mark.parametrize("postcode", ('AI-2640', 'AI AI-2640')) +@pytest.mark.parametrize("postcode", ('AI-2640', '2640', 'AI 2640')) def test_postcode_anguilla_pass(sanitize, postcode): assert sanitize(country='ai', postcode=postcode) == [('postcode', 'AI-2640')] -@pytest.mark.parametrize("postcode", ('2640', 'AI 2640', 'AI-2000', 'AI US-2640')) +@pytest.mark.parametrize("postcode", ('AI-2000', 'AI US-2640', 'AI AI-2640')) @pytest.mark.sanitizer_params(convert_to_address=False) def test_postcode_anguilla_fail(sanitize, postcode): assert sanitize(country='ai', postcode=postcode) == [] @@ -144,45 +144,45 @@ def test_postcode_jersey_fail(sanitize, postcode): assert sanitize(country='je', postcode=postcode) == [] -@pytest.mark.parametrize("postcode", ('KY1-1234', 'KY KY1-1234')) +@pytest.mark.parametrize("postcode", ('KY1-1234', '1-1234', 'KY 1-1234')) def test_postcode_cayman_islands_pass(sanitize, postcode): assert sanitize(country='ky', postcode=postcode) == [('postcode', 'KY1-1234')] -@pytest.mark.parametrize("postcode", ('1-1234', 'KY-1234', 'KZ1-1234', 'KY1 1234', 'KY 1-1234', 'KY1-123')) +@pytest.mark.parametrize("postcode", ('KY-1234', 'KZ1-1234', 'KY1 1234', 'KY1-123', 'KY KY1-1234')) @pytest.mark.sanitizer_params(convert_to_address=False) def test_postcode_cayman_islands_fail(sanitize, postcode): assert sanitize(country='ky', postcode=postcode) == [] -@pytest.mark.parametrize("postcode", ('LC11 222', 'LC LC11 222', 'LC LC11 222')) +@pytest.mark.parametrize("postcode", ('LC11 222', '11 222', '11222', 'LC 11 222')) def test_postcode_saint_lucia_pass(sanitize, postcode): assert sanitize(country='lc', postcode=postcode) == [('postcode', 'LC11 222')] -@pytest.mark.parametrize("postcode", ('11 222', '11222', 'LC 11 222')) +@pytest.mark.parametrize("postcode", ('11 2222', 'LC LC11 222')) @pytest.mark.sanitizer_params(convert_to_address=False) def test_postcode_saint_lucia_fail(sanitize, postcode): assert sanitize(country='lc', postcode=postcode) == [] -@pytest.mark.parametrize("postcode", ('LV-1111', 'LV LV-1111')) +@pytest.mark.parametrize("postcode", ('LV-1111', '1111', 'LV 1111', 'LV1111',)) def test_postcode_latvia_pass(sanitize, postcode): assert sanitize(country='lv', postcode=postcode) == [('postcode', 'LV-1111')] -@pytest.mark.parametrize("postcode", ('1111', 'LV 1111', 'LV1111', 'LV LV 1111')) +@pytest.mark.parametrize("postcode", ('111', '11111', 'LV LV-1111')) @pytest.mark.sanitizer_params(convert_to_address=False) def test_postcode_latvia_fail(sanitize, postcode): assert sanitize(country='lv', postcode=postcode) == [] -@pytest.mark.parametrize("postcode", ('MD-1111', 'MD MD-1111')) +@pytest.mark.parametrize("postcode", ('MD-1111', '1111', 'MD 1111', 'MD1111')) def test_postcode_moldova_pass(sanitize, postcode): assert sanitize(country='md', postcode=postcode) == [('postcode', 'MD-1111')] -@pytest.mark.parametrize("postcode", ('1111', 'MD 1111', 'MD1111')) +@pytest.mark.parametrize("postcode", ("MD MD-1111", "MD MD1111", "MD MD 1111")) @pytest.mark.sanitizer_params(convert_to_address=False) def test_postcode_moldova_fail(sanitize, postcode): assert sanitize(country='md', postcode=postcode) == [] @@ -204,23 +204,23 @@ def test_postcode_malta_fail(sanitize, postcode): assert sanitize(country='mt', postcode=postcode) == [] -@pytest.mark.parametrize("postcode", ('VC1111', 'VC VC1111')) +@pytest.mark.parametrize("postcode", ('VC1111', '1111', 'VC-1111', 'VC 1111')) def test_postcode_saint_vincent_pass(sanitize, postcode): assert sanitize(country='vc', postcode=postcode) == [('postcode', 'VC1111')] -@pytest.mark.parametrize("postcode", ('1111', 'VC-1111', 'VC 1111', 'VC11')) +@pytest.mark.parametrize("postcode", ('VC11', 'VC VC1111')) @pytest.mark.sanitizer_params(convert_to_address=False) def test_postcode_saint_vincent_fail(sanitize, postcode): assert sanitize(country='vc', postcode=postcode) == [] -@pytest.mark.parametrize("postcode", ('VG1111', 'VG VG1111')) +@pytest.mark.parametrize("postcode", ('VG1111', '1111', 'VG 1111', 'VG-1111')) def test_postcode_virgin_islands_pass(sanitize, postcode): assert sanitize(country='vg', postcode=postcode) == [('postcode', 'VG1111')] -@pytest.mark.parametrize("postcode", ('1111', 'VG 1111', 'VG-1111')) +@pytest.mark.parametrize("postcode", ('111', '11111', 'VG VG1111')) @pytest.mark.sanitizer_params(convert_to_address=False) def test_postcode_virgin_islands_fail(sanitize, postcode): assert sanitize(country='vg', postcode=postcode) == []