1 # SPDX-License-Identifier: GPL-3.0-or-later
3 # This file is part of Nominatim. (https://nominatim.org)
5 # Copyright (C) 2023 by the Nominatim developer community.
6 # For a full list of authors see the git log.
8 Tests for creating abstract searches from token assignments.
12 from nominatim_api.search.query import Token, TokenRange, QueryStruct, Phrase
13 import nominatim_api.search.query as qmod
14 from nominatim_api.search.db_search_builder import SearchBuilder
15 from nominatim_api.search.token_assignment import TokenAssignment
16 from nominatim_api.types import SearchDetails
17 import nominatim_api.search.db_searches as dbs
21 def get_category(self):
25 def make_query(*args):
26 q = QueryStruct([Phrase(qmod.PHRASE_ANY, '')])
28 for _ in range(max(inner[0] for tlist in args for inner in tlist)):
29 q.add_node(qmod.BREAK_WORD, qmod.PHRASE_ANY)
30 q.add_node(qmod.BREAK_END, qmod.PHRASE_ANY)
32 for start, tlist in enumerate(args):
33 for end, ttype, tinfo in tlist:
34 for tid, word in tinfo:
35 q.add_token(TokenRange(start, end), ttype,
36 MyToken(penalty=0.5 if ttype == qmod.TOKEN_PARTIAL else 0.0,
37 token=tid, count=1, addr_count=1,
43 def test_country_search():
44 q = make_query([(1, qmod.TOKEN_COUNTRY, [(2, 'de'), (3, 'en')])])
45 builder = SearchBuilder(q, SearchDetails())
47 searches = list(builder.build(TokenAssignment(country=TokenRange(0, 1))))
49 assert len(searches) == 1
53 assert isinstance(search, dbs.CountrySearch)
54 assert set(search.countries.values) == {'de', 'en'}
57 def test_country_search_with_country_restriction():
58 q = make_query([(1, qmod.TOKEN_COUNTRY, [(2, 'de'), (3, 'en')])])
59 builder = SearchBuilder(q, SearchDetails.from_kwargs({'countries': 'en,fr'}))
61 searches = list(builder.build(TokenAssignment(country=TokenRange(0, 1))))
63 assert len(searches) == 1
67 assert isinstance(search, dbs.CountrySearch)
68 assert set(search.countries.values) == {'en'}
71 def test_country_search_with_conflicting_country_restriction():
72 q = make_query([(1, qmod.TOKEN_COUNTRY, [(2, 'de'), (3, 'en')])])
73 builder = SearchBuilder(q, SearchDetails.from_kwargs({'countries': 'fr'}))
75 searches = list(builder.build(TokenAssignment(country=TokenRange(0, 1))))
77 assert len(searches) == 0
80 def test_postcode_search_simple():
81 q = make_query([(1, qmod.TOKEN_POSTCODE, [(34, '2367')])])
82 builder = SearchBuilder(q, SearchDetails())
84 searches = list(builder.build(TokenAssignment(postcode=TokenRange(0, 1))))
86 assert len(searches) == 1
89 assert isinstance(search, dbs.PostcodeSearch)
90 assert search.postcodes.values == ['2367']
91 assert not search.countries.values
92 assert not search.lookups
93 assert not search.rankings
96 def test_postcode_with_country():
97 q = make_query([(1, qmod.TOKEN_POSTCODE, [(34, '2367')])],
98 [(2, qmod.TOKEN_COUNTRY, [(1, 'xx')])])
99 builder = SearchBuilder(q, SearchDetails())
101 searches = list(builder.build(TokenAssignment(postcode=TokenRange(0, 1),
102 country=TokenRange(1, 2))))
104 assert len(searches) == 1
107 assert isinstance(search, dbs.PostcodeSearch)
108 assert search.postcodes.values == ['2367']
109 assert search.countries.values == ['xx']
110 assert not search.lookups
111 assert not search.rankings
114 def test_postcode_with_address():
115 q = make_query([(1, qmod.TOKEN_POSTCODE, [(34, '2367')])],
116 [(2, qmod.TOKEN_PARTIAL, [(100, 'word')])])
117 builder = SearchBuilder(q, SearchDetails())
119 searches = list(builder.build(TokenAssignment(postcode=TokenRange(0, 1),
120 address=[TokenRange(1, 2)])))
122 assert len(searches) == 1
125 assert isinstance(search, dbs.PostcodeSearch)
126 assert search.postcodes.values == ['2367']
127 assert not search.countries
128 assert search.lookups
129 assert not search.rankings
132 def test_postcode_with_address_with_full_word():
133 q = make_query([(1, qmod.TOKEN_POSTCODE, [(34, '2367')])],
134 [(2, qmod.TOKEN_PARTIAL, [(100, 'word')]),
135 (2, qmod.TOKEN_WORD, [(1, 'full')])])
136 builder = SearchBuilder(q, SearchDetails())
138 searches = list(builder.build(TokenAssignment(postcode=TokenRange(0, 1),
139 address=[TokenRange(1, 2)])))
141 assert len(searches) == 1
144 assert isinstance(search, dbs.PostcodeSearch)
145 assert search.postcodes.values == ['2367']
146 assert not search.countries
147 assert search.lookups
148 assert len(search.rankings) == 1
151 @pytest.mark.parametrize('kwargs', [{'viewbox': '0,0,1,1', 'bounded_viewbox': True},
153 def test_near_item_only(kwargs):
154 q = make_query([(1, qmod.TOKEN_NEAR_ITEM, [(2, 'foo')])])
155 builder = SearchBuilder(q, SearchDetails.from_kwargs(kwargs))
157 searches = list(builder.build(TokenAssignment(near_item=TokenRange(0, 1))))
159 assert len(searches) == 1
163 assert isinstance(search, dbs.PoiSearch)
164 assert search.qualifiers.values == [('this', 'that')]
167 @pytest.mark.parametrize('kwargs', [{'viewbox': '0,0,1,1'},
169 def test_near_item_skipped(kwargs):
170 q = make_query([(1, qmod.TOKEN_NEAR_ITEM, [(2, 'foo')])])
171 builder = SearchBuilder(q, SearchDetails.from_kwargs(kwargs))
173 searches = list(builder.build(TokenAssignment(near_item=TokenRange(0, 1))))
175 assert len(searches) == 0
178 def test_name_only_search():
179 q = make_query([(1, qmod.TOKEN_PARTIAL, [(1, 'a')]),
180 (1, qmod.TOKEN_WORD, [(100, 'a')])])
181 builder = SearchBuilder(q, SearchDetails())
183 searches = list(builder.build(TokenAssignment(name=TokenRange(0, 1))))
185 assert len(searches) == 1
188 assert isinstance(search, dbs.PlaceSearch)
189 assert not search.postcodes.values
190 assert not search.countries.values
191 assert not search.housenumbers.values
192 assert not search.qualifiers.values
193 assert len(search.lookups) == 1
194 assert len(search.rankings) == 1
197 def test_name_with_qualifier():
198 q = make_query([(1, qmod.TOKEN_PARTIAL, [(1, 'a')]),
199 (1, qmod.TOKEN_WORD, [(100, 'a')])],
200 [(2, qmod.TOKEN_QUALIFIER, [(55, 'hotel')])])
201 builder = SearchBuilder(q, SearchDetails())
203 searches = list(builder.build(TokenAssignment(name=TokenRange(0, 1),
204 qualifier=TokenRange(1, 2))))
206 assert len(searches) == 1
209 assert isinstance(search, dbs.PlaceSearch)
210 assert not search.postcodes.values
211 assert not search.countries.values
212 assert not search.housenumbers.values
213 assert search.qualifiers.values == [('this', 'that')]
214 assert len(search.lookups) == 1
215 assert len(search.rankings) == 1
218 def test_name_with_housenumber_search():
219 q = make_query([(1, qmod.TOKEN_PARTIAL, [(1, 'a')]),
220 (1, qmod.TOKEN_WORD, [(100, 'a')])],
221 [(2, qmod.TOKEN_HOUSENUMBER, [(66, '66')])])
222 builder = SearchBuilder(q, SearchDetails())
224 searches = list(builder.build(TokenAssignment(name=TokenRange(0, 1),
225 housenumber=TokenRange(1, 2))))
227 assert len(searches) == 1
230 assert isinstance(search, dbs.PlaceSearch)
231 assert not search.postcodes.values
232 assert not search.countries.values
233 assert search.housenumbers.values == ['66']
234 assert len(search.lookups) == 1
235 assert len(search.rankings) == 1
238 def test_name_and_address():
239 q = make_query([(1, qmod.TOKEN_PARTIAL, [(1, 'a')]),
240 (1, qmod.TOKEN_WORD, [(100, 'a')])],
241 [(2, qmod.TOKEN_PARTIAL, [(2, 'b')]),
242 (2, qmod.TOKEN_WORD, [(101, 'b')])],
243 [(3, qmod.TOKEN_PARTIAL, [(3, 'c')]),
244 (3, qmod.TOKEN_WORD, [(102, 'c')])])
245 builder = SearchBuilder(q, SearchDetails())
247 searches = list(builder.build(TokenAssignment(name=TokenRange(0, 1),
248 address=[TokenRange(1, 2),
251 assert len(searches) == 1
254 assert isinstance(search, dbs.PlaceSearch)
255 assert not search.postcodes.values
256 assert not search.countries.values
257 assert not search.housenumbers.values
258 assert len(search.lookups) == 2
259 assert len(search.rankings) == 3
262 def test_name_and_complex_address():
263 q = make_query([(1, qmod.TOKEN_PARTIAL, [(1, 'a')]),
264 (1, qmod.TOKEN_WORD, [(100, 'a')])],
265 [(2, qmod.TOKEN_PARTIAL, [(2, 'b')]),
266 (3, qmod.TOKEN_WORD, [(101, 'bc')])],
267 [(3, qmod.TOKEN_PARTIAL, [(3, 'c')])],
268 [(4, qmod.TOKEN_PARTIAL, [(4, 'd')]),
269 (4, qmod.TOKEN_WORD, [(103, 'd')])])
270 builder = SearchBuilder(q, SearchDetails())
272 searches = list(builder.build(TokenAssignment(name=TokenRange(0, 1),
273 address=[TokenRange(1, 2),
276 assert len(searches) == 1
279 assert isinstance(search, dbs.PlaceSearch)
280 assert not search.postcodes.values
281 assert not search.countries.values
282 assert not search.housenumbers.values
283 assert len(search.lookups) == 2
284 assert len(search.rankings) == 2
287 def test_name_only_near_search():
288 q = make_query([(1, qmod.TOKEN_NEAR_ITEM, [(88, 'g')])],
289 [(2, qmod.TOKEN_PARTIAL, [(1, 'a')]),
290 (2, qmod.TOKEN_WORD, [(100, 'a')])])
291 builder = SearchBuilder(q, SearchDetails())
293 searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2),
294 near_item=TokenRange(0, 1))))
296 assert len(searches) == 1
299 assert isinstance(search, dbs.NearSearch)
300 assert isinstance(search.search, dbs.PlaceSearch)
303 def test_name_only_search_with_category():
304 q = make_query([(1, qmod.TOKEN_PARTIAL, [(1, 'a')]),
305 (1, qmod.TOKEN_WORD, [(100, 'a')])])
306 builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar')]}))
308 searches = list(builder.build(TokenAssignment(name=TokenRange(0, 1))))
310 assert len(searches) == 1
313 assert isinstance(search, dbs.PlaceSearch)
314 assert search.qualifiers.values == [('foo', 'bar')]
317 def test_name_with_near_item_search_with_category_mismatch():
318 q = make_query([(1, qmod.TOKEN_NEAR_ITEM, [(88, 'g')])],
319 [(2, qmod.TOKEN_PARTIAL, [(1, 'a')]),
320 (2, qmod.TOKEN_WORD, [(100, 'a')])])
321 builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar')]}))
323 searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2),
324 near_item=TokenRange(0, 1))))
326 assert len(searches) == 0
329 def test_name_with_near_item_search_with_category_match():
330 q = make_query([(1, qmod.TOKEN_NEAR_ITEM, [(88, 'g')])],
331 [(2, qmod.TOKEN_PARTIAL, [(1, 'a')]),
332 (2, qmod.TOKEN_WORD, [(100, 'a')])])
333 builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar'),
336 searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2),
337 near_item=TokenRange(0, 1))))
339 assert len(searches) == 1
342 assert isinstance(search, dbs.NearSearch)
343 assert isinstance(search.search, dbs.PlaceSearch)
346 def test_name_with_qualifier_search_with_category_mismatch():
347 q = make_query([(1, qmod.TOKEN_QUALIFIER, [(88, 'g')])],
348 [(2, qmod.TOKEN_PARTIAL, [(1, 'a')]),
349 (2, qmod.TOKEN_WORD, [(100, 'a')])])
350 builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar')]}))
352 searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2),
353 qualifier=TokenRange(0, 1))))
355 assert len(searches) == 0
358 def test_name_with_qualifier_search_with_category_match():
359 q = make_query([(1, qmod.TOKEN_QUALIFIER, [(88, 'g')])],
360 [(2, qmod.TOKEN_PARTIAL, [(1, 'a')]),
361 (2, qmod.TOKEN_WORD, [(100, 'a')])])
362 builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar'),
365 searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2),
366 qualifier=TokenRange(0, 1))))
368 assert len(searches) == 1
371 assert isinstance(search, dbs.PlaceSearch)
372 assert search.qualifiers.values == [('this', 'that')]
375 def test_name_only_search_with_countries():
376 q = make_query([(1, qmod.TOKEN_PARTIAL, [(1, 'a')]),
377 (1, qmod.TOKEN_WORD, [(100, 'a')])])
378 builder = SearchBuilder(q, SearchDetails.from_kwargs({'countries': 'de,en'}))
380 searches = list(builder.build(TokenAssignment(name=TokenRange(0, 1))))
382 assert len(searches) == 1
385 assert isinstance(search, dbs.PlaceSearch)
386 assert not search.postcodes.values
387 assert set(search.countries.values) == {'de', 'en'}
388 assert not search.housenumbers.values
391 def make_counted_searches(name_part, name_full, address_part, address_full,
392 num_address_parts=1):
393 q = QueryStruct([Phrase(qmod.PHRASE_ANY, '')])
394 for i in range(1 + num_address_parts):
395 q.add_node(qmod.BREAK_WORD, qmod.PHRASE_ANY)
396 q.add_node(qmod.BREAK_END, qmod.PHRASE_ANY)
398 q.add_token(TokenRange(0, 1), qmod.TOKEN_PARTIAL,
399 MyToken(0.5, 1, name_part, 1, 'name_part'))
400 q.add_token(TokenRange(0, 1), qmod.TOKEN_WORD,
401 MyToken(0, 101, name_full, 1, 'name_full'))
402 for i in range(num_address_parts):
403 q.add_token(TokenRange(i + 1, i + 2), qmod.TOKEN_PARTIAL,
404 MyToken(0.5, 2, address_part, 1, 'address_part'))
405 q.add_token(TokenRange(i + 1, i + 2), qmod.TOKEN_WORD,
406 MyToken(0, 102, address_full, 1, 'address_full'))
408 builder = SearchBuilder(q, SearchDetails())
410 return list(builder.build(TokenAssignment(name=TokenRange(0, 1),
411 address=[TokenRange(1, 1 + num_address_parts)])))
414 def test_infrequent_partials_in_name():
415 searches = make_counted_searches(1, 1, 1, 1)
417 assert len(searches) == 1
420 assert isinstance(search, dbs.PlaceSearch)
421 assert len(search.lookups) == 2
422 assert len(search.rankings) == 2
424 assert set((s.column, s.lookup_type.__name__) for s in search.lookups) == \
425 {('name_vector', 'LookupAll'), ('nameaddress_vector', 'Restrict')}
428 def test_frequent_partials_in_name_and_address():
429 searches = make_counted_searches(9999, 1, 9999, 1)
431 assert len(searches) == 2
433 assert all(isinstance(s, dbs.PlaceSearch) for s in searches)
434 searches.sort(key=lambda s: s.penalty)
436 assert set((s.column, s.lookup_type.__name__) for s in searches[0].lookups) == \
437 {('name_vector', 'LookupAny'), ('nameaddress_vector', 'Restrict')}
438 assert set((s.column, s.lookup_type.__name__) for s in searches[1].lookups) == \
439 {('nameaddress_vector', 'LookupAll'), ('name_vector', 'LookupAll')}
442 def test_too_frequent_partials_in_name_and_address():
443 searches = make_counted_searches(20000, 1, 10000, 1)
445 assert len(searches) == 1
447 assert all(isinstance(s, dbs.PlaceSearch) for s in searches)
448 searches.sort(key=lambda s: s.penalty)
450 assert set((s.column, s.lookup_type.__name__) for s in searches[0].lookups) == \
451 {('name_vector', 'LookupAny'), ('nameaddress_vector', 'Restrict')}