X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/31412e06740727695c5d9512e0cd59c0dd683322..2aba0ad4bc0a924337a2541e983d61a26cf67ce6:/test/python/api/search/test_token_assignment.py diff --git a/test/python/api/search/test_token_assignment.py b/test/python/api/search/test_token_assignment.py index 6d4156e5..fff8d471 100644 --- a/test/python/api/search/test_token_assignment.py +++ b/test/python/api/search/test_token_assignment.py @@ -9,7 +9,7 @@ Test for creation of token assignments from tokenized queries. """ import pytest -from nominatim_api.search.query import QueryStruct, Phrase, PhraseType, TokenRange, Token +from nominatim_api.search.query import QueryStruct, Phrase, TokenRange, Token import nominatim_api.search.query as qmod from nominatim_api.search.token_assignment import yield_token_assignments, TokenAssignment, PENALTY_TOKENCHANGE @@ -25,7 +25,7 @@ def make_query(*args): for btype, ptype, _ in args[1:]: q.add_node(btype, ptype) - q.add_node(qmod.BREAK_END, PhraseType.NONE) + q.add_node(qmod.BREAK_END, qmod.PHRASE_ANY) for start, t in enumerate(args): for end, ttype in t[2]: @@ -44,14 +44,14 @@ def check_assignments(actual, *expected): def test_query_with_missing_tokens(): - q = QueryStruct([Phrase(PhraseType.NONE, '')]) - q.add_node(qmod.BREAK_END, PhraseType.NONE) + q = QueryStruct([Phrase(qmod.PHRASE_ANY, '')]) + q.add_node(qmod.BREAK_END, qmod.PHRASE_ANY) assert list(yield_token_assignments(q)) == [] def test_one_word_query(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL), (1, qmod.TOKEN_WORD), (1, qmod.TOKEN_HOUSENUMBER)])) @@ -61,7 +61,7 @@ def test_one_word_query(): def test_single_postcode(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_POSTCODE)])) res = list(yield_token_assignments(q)) @@ -69,7 +69,7 @@ def test_single_postcode(): def test_single_country_name(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_COUNTRY)])) res = list(yield_token_assignments(q)) @@ -77,7 +77,7 @@ def test_single_country_name(): def test_single_word_poi_search(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_NEAR_ITEM), (1, qmod.TOKEN_QUALIFIER)])) @@ -87,9 +87,9 @@ def test_single_word_poi_search(): @pytest.mark.parametrize('btype', [qmod.BREAK_WORD, qmod.BREAK_PART, qmod.BREAK_TOKEN]) def test_multiple_simple_words(btype): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (btype, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)]), - (btype, PhraseType.NONE, [(3, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (btype, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)]), + (btype, qmod.PHRASE_ANY, [(3, qmod.TOKEN_PARTIAL)])) penalty = PENALTY_TOKENCHANGE[btype] @@ -107,8 +107,8 @@ def test_multiple_simple_words(btype): def test_multiple_words_respect_phrase_break(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q), TokenAssignment(name=TokenRange(0, 1), @@ -118,8 +118,8 @@ def test_multiple_words_respect_phrase_break(): def test_housenumber_and_street(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_HOUSENUMBER)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_HOUSENUMBER)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q), TokenAssignment(name=TokenRange(1, 2), @@ -129,8 +129,8 @@ def test_housenumber_and_street(): def test_housenumber_and_street_backwards(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(2, qmod.TOKEN_HOUSENUMBER)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(2, qmod.TOKEN_HOUSENUMBER)])) check_assignments(yield_token_assignments(q), TokenAssignment(name=TokenRange(0, 1), @@ -140,10 +140,10 @@ def test_housenumber_and_street_backwards(): def test_housenumber_and_postcode(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_HOUSENUMBER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(3, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(4, qmod.TOKEN_POSTCODE)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_HOUSENUMBER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(3, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(4, qmod.TOKEN_POSTCODE)])) check_assignments(yield_token_assignments(q), TokenAssignment(penalty=pytest.approx(0.3), @@ -157,10 +157,10 @@ def test_housenumber_and_postcode(): postcode=TokenRange(3, 4))) def test_postcode_and_housenumber(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_POSTCODE)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(3, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(4, qmod.TOKEN_HOUSENUMBER)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_POSTCODE)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(3, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(4, qmod.TOKEN_HOUSENUMBER)])) check_assignments(yield_token_assignments(q), TokenAssignment(penalty=pytest.approx(0.3), @@ -175,10 +175,10 @@ def test_postcode_and_housenumber(): def test_country_housenumber_postcode(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_COUNTRY)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(3, qmod.TOKEN_HOUSENUMBER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(4, qmod.TOKEN_POSTCODE)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_COUNTRY)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(3, qmod.TOKEN_HOUSENUMBER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(4, qmod.TOKEN_POSTCODE)])) check_assignments(yield_token_assignments(q)) @@ -186,27 +186,27 @@ def test_country_housenumber_postcode(): @pytest.mark.parametrize('ttype', [qmod.TOKEN_POSTCODE, qmod.TOKEN_COUNTRY, qmod.TOKEN_NEAR_ITEM, qmod.TOKEN_QUALIFIER]) def test_housenumber_with_only_special_terms(ttype): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_HOUSENUMBER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, ttype)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_HOUSENUMBER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, ttype)])) check_assignments(yield_token_assignments(q)) @pytest.mark.parametrize('ttype', [qmod.TOKEN_POSTCODE, qmod.TOKEN_HOUSENUMBER, qmod.TOKEN_COUNTRY]) def test_multiple_special_tokens(ttype): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, ttype)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(3, ttype)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, ttype)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(3, ttype)])) check_assignments(yield_token_assignments(q)) def test_housenumber_many_phrases(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(3, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(4, qmod.TOKEN_HOUSENUMBER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(5, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(3, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(4, qmod.TOKEN_HOUSENUMBER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(5, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q), TokenAssignment(penalty=0.1, @@ -221,8 +221,8 @@ def test_housenumber_many_phrases(): def test_country_at_beginning(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_COUNTRY)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_COUNTRY)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q), TokenAssignment(penalty=0.1, name=TokenRange(1, 2), @@ -230,8 +230,8 @@ def test_country_at_beginning(): def test_country_at_end(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_COUNTRY)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_COUNTRY)])) check_assignments(yield_token_assignments(q), TokenAssignment(penalty=0.1, name=TokenRange(0, 1), @@ -239,16 +239,16 @@ def test_country_at_end(): def test_country_in_middle(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_COUNTRY)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(3, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_COUNTRY)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(3, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q)) def test_postcode_with_designation(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_POSTCODE)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_POSTCODE)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q), TokenAssignment(penalty=0.1, name=TokenRange(1, 2), @@ -258,8 +258,8 @@ def test_postcode_with_designation(): def test_postcode_with_designation_backwards(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(2, qmod.TOKEN_POSTCODE)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(2, qmod.TOKEN_POSTCODE)])) check_assignments(yield_token_assignments(q), TokenAssignment(name=TokenRange(0, 1), @@ -269,8 +269,8 @@ def test_postcode_with_designation_backwards(): def test_near_item_at_beginning(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_NEAR_ITEM)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_NEAR_ITEM)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q), TokenAssignment(penalty=0.1, name=TokenRange(1, 2), @@ -278,8 +278,8 @@ def test_near_item_at_beginning(): def test_near_item_at_end(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_NEAR_ITEM)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_NEAR_ITEM)])) check_assignments(yield_token_assignments(q), TokenAssignment(penalty=0.1, name=TokenRange(0, 1), @@ -287,17 +287,17 @@ def test_near_item_at_end(): def test_near_item_in_middle(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_NEAR_ITEM)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(3, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_NEAR_ITEM)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(3, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q)) def test_qualifier_at_beginning(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_QUALIFIER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(3, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_QUALIFIER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(3, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q), @@ -309,11 +309,11 @@ def test_qualifier_at_beginning(): def test_qualifier_after_name(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(3, qmod.TOKEN_QUALIFIER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(4, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(5, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(3, qmod.TOKEN_QUALIFIER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(4, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(5, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q), @@ -326,27 +326,27 @@ def test_qualifier_after_name(): def test_qualifier_before_housenumber(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_QUALIFIER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_HOUSENUMBER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(3, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_QUALIFIER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_HOUSENUMBER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(3, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q)) def test_qualifier_after_housenumber(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_HOUSENUMBER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(2, qmod.TOKEN_QUALIFIER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(3, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_HOUSENUMBER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(2, qmod.TOKEN_QUALIFIER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(3, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q)) def test_qualifier_in_middle_of_phrase(): - q = make_query((qmod.BREAK_START, PhraseType.NONE, [(1, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(2, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(3, qmod.TOKEN_QUALIFIER)]), - (qmod.BREAK_WORD, PhraseType.NONE, [(4, qmod.TOKEN_PARTIAL)]), - (qmod.BREAK_PHRASE, PhraseType.NONE, [(5, qmod.TOKEN_PARTIAL)])) + q = make_query((qmod.BREAK_START, qmod.PHRASE_ANY, [(1, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(2, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(3, qmod.TOKEN_QUALIFIER)]), + (qmod.BREAK_WORD, qmod.PHRASE_ANY, [(4, qmod.TOKEN_PARTIAL)]), + (qmod.BREAK_PHRASE, qmod.PHRASE_ANY, [(5, qmod.TOKEN_PARTIAL)])) check_assignments(yield_token_assignments(q))