X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/c42273a4db2d7b4fe05a0be9210901d35e038887..b54ff7d7664eadf3fdf018622540d6fce835502e:/test/python/api/search/test_db_search_builder.py diff --git a/test/python/api/search/test_db_search_builder.py b/test/python/api/search/test_db_search_builder.py index 9631850e..5d984014 100644 --- a/test/python/api/search/test_db_search_builder.py +++ b/test/python/api/search/test_db_search_builder.py @@ -9,11 +9,11 @@ Tests for creating abstract searches from token assignments. """ import pytest -from nominatim.api.search.query import Token, TokenRange, BreakType, PhraseType, TokenType, QueryStruct, Phrase -from nominatim.api.search.db_search_builder import SearchBuilder -from nominatim.api.search.token_assignment import TokenAssignment -from nominatim.api.types import SearchDetails -import nominatim.api.search.db_searches as dbs +from nominatim_api.search.query import Token, TokenRange, BreakType, PhraseType, TokenType, QueryStruct, Phrase +from nominatim_api.search.db_search_builder import SearchBuilder +from nominatim_api.search.token_assignment import TokenAssignment +from nominatim_api.types import SearchDetails +import nominatim_api.search.db_searches as dbs class MyToken(Token): def get_category(self): @@ -21,21 +21,20 @@ class MyToken(Token): def make_query(*args): - q = None + q = QueryStruct([Phrase(PhraseType.NONE, '')]) - for tlist in args: - if q is None: - q = QueryStruct([Phrase(PhraseType.NONE, '')]) - else: - q.add_node(BreakType.WORD, PhraseType.NONE) + for _ in range(max(inner[0] for tlist in args for inner in tlist)): + q.add_node(BreakType.WORD, PhraseType.NONE) + q.add_node(BreakType.END, PhraseType.NONE) - start = len(q.nodes) - 1 + for start, tlist in enumerate(args): for end, ttype, tinfo in tlist: for tid, word in tinfo: q.add_token(TokenRange(start, end), ttype, - MyToken(0.5 if ttype == TokenType.PARTIAL else 0.0, tid, 1, word, True)) + MyToken(penalty=0.5 if ttype == TokenType.PARTIAL else 0.0, + token=tid, count=1, addr_count=1, + lookup_word=word, is_indexed=True)) - q.add_node(BreakType.END, PhraseType.NONE) return q @@ -68,7 +67,7 @@ def test_country_search_with_country_restriction(): assert set(search.countries.values) == {'en'} -def test_country_search_with_confllicting_country_restriction(): +def test_country_search_with_conflicting_country_restriction(): q = make_query([(1, TokenType.COUNTRY, [(2, 'de'), (3, 'en')])]) builder = SearchBuilder(q, SearchDetails.from_kwargs({'countries': 'fr'})) @@ -150,27 +149,27 @@ def test_postcode_with_address_with_full_word(): @pytest.mark.parametrize('kwargs', [{'viewbox': '0,0,1,1', 'bounded_viewbox': True}, {'near': '10,10'}]) -def test_category_only(kwargs): - q = make_query([(1, TokenType.CATEGORY, [(2, 'foo')])]) +def test_near_item_only(kwargs): + q = make_query([(1, TokenType.NEAR_ITEM, [(2, 'foo')])]) builder = SearchBuilder(q, SearchDetails.from_kwargs(kwargs)) - searches = list(builder.build(TokenAssignment(category=TokenRange(0, 1)))) + searches = list(builder.build(TokenAssignment(near_item=TokenRange(0, 1)))) assert len(searches) == 1 search = searches[0] assert isinstance(search, dbs.PoiSearch) - assert search.categories.values == [('this', 'that')] + assert search.qualifiers.values == [('this', 'that')] @pytest.mark.parametrize('kwargs', [{'viewbox': '0,0,1,1'}, {}]) -def test_category_skipped(kwargs): - q = make_query([(1, TokenType.CATEGORY, [(2, 'foo')])]) +def test_near_item_skipped(kwargs): + q = make_query([(1, TokenType.NEAR_ITEM, [(2, 'foo')])]) builder = SearchBuilder(q, SearchDetails.from_kwargs(kwargs)) - searches = list(builder.build(TokenAssignment(category=TokenRange(0, 1)))) + searches = list(builder.build(TokenAssignment(near_item=TokenRange(0, 1)))) assert len(searches) == 0 @@ -287,13 +286,13 @@ def test_name_and_complex_address(): def test_name_only_near_search(): - q = make_query([(1, TokenType.CATEGORY, [(88, 'g')])], + q = make_query([(1, TokenType.NEAR_ITEM, [(88, 'g')])], [(2, TokenType.PARTIAL, [(1, 'a')]), (2, TokenType.WORD, [(100, 'a')])]) builder = SearchBuilder(q, SearchDetails()) searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2), - category=TokenRange(0, 1)))) + near_item=TokenRange(0, 1)))) assert len(searches) == 1 search = searches[0] @@ -312,10 +311,68 @@ def test_name_only_search_with_category(): assert len(searches) == 1 search = searches[0] + assert isinstance(search, dbs.PlaceSearch) + assert search.qualifiers.values == [('foo', 'bar')] + + +def test_name_with_near_item_search_with_category_mismatch(): + q = make_query([(1, TokenType.NEAR_ITEM, [(88, 'g')])], + [(2, TokenType.PARTIAL, [(1, 'a')]), + (2, TokenType.WORD, [(100, 'a')])]) + builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar')]})) + + searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2), + near_item=TokenRange(0, 1)))) + + assert len(searches) == 0 + + +def test_name_with_near_item_search_with_category_match(): + q = make_query([(1, TokenType.NEAR_ITEM, [(88, 'g')])], + [(2, TokenType.PARTIAL, [(1, 'a')]), + (2, TokenType.WORD, [(100, 'a')])]) + builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar'), + ('this', 'that')]})) + + searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2), + near_item=TokenRange(0, 1)))) + + assert len(searches) == 1 + search = searches[0] + assert isinstance(search, dbs.NearSearch) assert isinstance(search.search, dbs.PlaceSearch) +def test_name_with_qualifier_search_with_category_mismatch(): + q = make_query([(1, TokenType.QUALIFIER, [(88, 'g')])], + [(2, TokenType.PARTIAL, [(1, 'a')]), + (2, TokenType.WORD, [(100, 'a')])]) + builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar')]})) + + searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2), + qualifier=TokenRange(0, 1)))) + + assert len(searches) == 0 + + +def test_name_with_qualifier_search_with_category_match(): + q = make_query([(1, TokenType.QUALIFIER, [(88, 'g')])], + [(2, TokenType.PARTIAL, [(1, 'a')]), + (2, TokenType.WORD, [(100, 'a')])]) + builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar'), + ('this', 'that')]})) + + searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2), + qualifier=TokenRange(0, 1)))) + + assert len(searches) == 1 + search = searches[0] + + assert isinstance(search, dbs.PlaceSearch) + assert search.qualifiers.values == [('this', 'that')] + + def test_name_only_search_with_countries(): q = make_query([(1, TokenType.PARTIAL, [(1, 'a')]), (1, TokenType.WORD, [(100, 'a')])]) @@ -332,25 +389,27 @@ def test_name_only_search_with_countries(): assert not search.housenumbers.values -def make_counted_searches(name_part, name_full, address_part, address_full): +def make_counted_searches(name_part, name_full, address_part, address_full, + num_address_parts=1): q = QueryStruct([Phrase(PhraseType.NONE, '')]) - for i in range(2): + for i in range(1 + num_address_parts): q.add_node(BreakType.WORD, PhraseType.NONE) q.add_node(BreakType.END, PhraseType.NONE) q.add_token(TokenRange(0, 1), TokenType.PARTIAL, - MyToken(0.5, 1, name_part, 'name_part', True)) + MyToken(0.5, 1, name_part, 1, 'name_part', True)) q.add_token(TokenRange(0, 1), TokenType.WORD, - MyToken(0, 101, name_full, 'name_full', True)) - q.add_token(TokenRange(1, 2), TokenType.PARTIAL, - MyToken(0.5, 2, address_part, 'address_part', True)) - q.add_token(TokenRange(1, 2), TokenType.WORD, - MyToken(0, 102, address_full, 'address_full', True)) + MyToken(0, 101, name_full, 1, 'name_full', True)) + for i in range(num_address_parts): + q.add_token(TokenRange(i + 1, i + 2), TokenType.PARTIAL, + MyToken(0.5, 2, address_part, 1, 'address_part', True)) + q.add_token(TokenRange(i + 1, i + 2), TokenType.WORD, + MyToken(0, 102, address_full, 1, 'address_full', True)) builder = SearchBuilder(q, SearchDetails()) return list(builder.build(TokenAssignment(name=TokenRange(0, 1), - address=[TokenRange(1, 2)]))) + address=[TokenRange(1, 1 + num_address_parts)]))) def test_infrequent_partials_in_name(): @@ -363,33 +422,31 @@ def test_infrequent_partials_in_name(): assert len(search.lookups) == 2 assert len(search.rankings) == 2 - assert set((l.column, l.lookup_type) for l in search.lookups) == \ - {('name_vector', 'lookup_all'), ('nameaddress_vector', 'restrict')} + assert set((l.column, l.lookup_type.__name__) for l in search.lookups) == \ + {('name_vector', 'LookupAll'), ('nameaddress_vector', 'Restrict')} -def test_frequent_partials_in_name_but_not_in_address(): - searches = make_counted_searches(10000, 1, 1, 1) +def test_frequent_partials_in_name_and_address(): + searches = make_counted_searches(9999, 1, 9999, 1) - assert len(searches) == 1 - search = searches[0] + assert len(searches) == 2 - assert isinstance(search, dbs.PlaceSearch) - assert len(search.lookups) == 2 - assert len(search.rankings) == 2 + assert all(isinstance(s, dbs.PlaceSearch) for s in searches) + searches.sort(key=lambda s: s.penalty) - assert set((l.column, l.lookup_type) for l in search.lookups) == \ - {('nameaddress_vector', 'lookup_all'), ('name_vector', 'restrict')} + assert set((l.column, l.lookup_type.__name__) for l in searches[0].lookups) == \ + {('name_vector', 'LookupAny'), ('nameaddress_vector', 'Restrict')} + assert set((l.column, l.lookup_type.__name__) for l in searches[1].lookups) == \ + {('nameaddress_vector', 'LookupAll'), ('name_vector', 'LookupAll')} -def test_frequent_partials_in_name_and_address(): - searches = make_counted_searches(10000, 1, 10000, 1) +def test_too_frequent_partials_in_name_and_address(): + searches = make_counted_searches(20000, 1, 10000, 1) - assert len(searches) == 2 + assert len(searches) == 1 assert all(isinstance(s, dbs.PlaceSearch) for s in searches) searches.sort(key=lambda s: s.penalty) - assert set((l.column, l.lookup_type) for l in searches[0].lookups) == \ - {('name_vector', 'lookup_any'), ('nameaddress_vector', 'restrict')} - assert set((l.column, l.lookup_type) for l in searches[1].lookups) == \ - {('nameaddress_vector', 'lookup_all'), ('name_vector', 'lookup_all')} + assert set((l.column, l.lookup_type.__name__) for l in searches[0].lookups) == \ + {('name_vector', 'LookupAny'), ('nameaddress_vector', 'Restrict')}