test/python/api/search/test_search_places.py

   1 # SPDX-License-Identifier: GPL-3.0-or-later
   2 #
   3 # This file is part of Nominatim. (https://nominatim.org)
   4 #
   5 # Copyright (C) 2023 by the Nominatim developer community.
   6 # For a full list of authors see the git log.
   7 """
   8 Tests for running the generic place searcher.
   9 """
  10 import json
  11
  12 import pytest
  13
  14 import nominatim.api as napi
  15 from nominatim.api.types import SearchDetails
  16 from nominatim.api.search.db_searches import PlaceSearch
  17 from nominatim.api.search.db_search_fields import WeightedStrings, WeightedCategories,\
  18                                                   FieldLookup, FieldRanking, RankedTokens
  19 from nominatim.api.search.db_search_lookups import LookupAll, LookupAny, Restrict
  20
  21 def run_search(apiobj, global_penalty, lookup, ranking, count=2,
  22                hnrs=[], pcs=[], ccodes=[], quals=[],
  23                details=SearchDetails()):
  24     class MySearchData:
  25         penalty = global_penalty
  26         postcodes = WeightedStrings(pcs, [0.0] * len(pcs))
  27         countries = WeightedStrings(ccodes, [0.0] * len(ccodes))
  28         housenumbers = WeightedStrings(hnrs, [0.0] * len(hnrs))
  29         qualifiers = WeightedCategories(quals, [0.0] * len(quals))
  30         lookups = lookup
  31         rankings = ranking
  32
  33     search = PlaceSearch(0.0, MySearchData(), count)
  34
  35     async def run():
  36         async with apiobj.api._async_api.begin() as conn:
  37             return await search.lookup(conn, details)
  38
  39     results = apiobj.async_to_sync(run())
  40     results.sort(key=lambda r: r.accuracy)
  41
  42     return results
  43
  44
  45 class TestNameOnlySearches:
  46
  47     @pytest.fixture(autouse=True)
  48     def fill_database(self, apiobj):
  49         apiobj.add_placex(place_id=100, country_code='us',
  50                           centroid=(5.6, 4.3))
  51         apiobj.add_search_name(100, names=[1,2,10,11], country_code='us',
  52                                centroid=(5.6, 4.3))
  53         apiobj.add_placex(place_id=101, country_code='mx',
  54                           centroid=(-10.3, 56.9))
  55         apiobj.add_search_name(101, names=[1,2,20,21], country_code='mx',
  56                                centroid=(-10.3, 56.9))
  57
  58
  59     @pytest.mark.parametrize('lookup_type', [LookupAll, Restrict])
  60     @pytest.mark.parametrize('rank,res', [([10], [100, 101]),
  61                                           ([20], [101, 100])])
  62     def test_lookup_all_match(self, apiobj, lookup_type, rank, res):
  63         lookup = FieldLookup('name_vector', [1,2], lookup_type)
  64         ranking = FieldRanking('name_vector', 0.9, [RankedTokens(0.0, rank)])
  65
  66         results = run_search(apiobj, 0.1, [lookup], [ranking])
  67
  68         assert [r.place_id for r in results] == res
  69
  70
  71     @pytest.mark.parametrize('lookup_type', [LookupAll, Restrict])
  72     def test_lookup_all_partial_match(self, apiobj, lookup_type):
  73         lookup = FieldLookup('name_vector', [1,20], lookup_type)
  74         ranking = FieldRanking('name_vector', 0.9, [RankedTokens(0.0, [21])])
  75
  76         results = run_search(apiobj, 0.1, [lookup], [ranking])
  77
  78         assert len(results) == 1
  79         assert results[0].place_id == 101
  80
  81     @pytest.mark.parametrize('rank,res', [([10], [100, 101]),
  82                                           ([20], [101, 100])])
  83     def test_lookup_any_match(self, apiobj, rank, res):
  84         lookup = FieldLookup('name_vector', [11,21], LookupAny)
  85         ranking = FieldRanking('name_vector', 0.9, [RankedTokens(0.0, rank)])
  86
  87         results = run_search(apiobj, 0.1, [lookup], [ranking])
  88
  89         assert [r.place_id for r in results] == res
  90
  91
  92     def test_lookup_any_partial_match(self, apiobj):
  93         lookup = FieldLookup('name_vector', [20], LookupAll)
  94         ranking = FieldRanking('name_vector', 0.9, [RankedTokens(0.0, [21])])
  95
  96         results = run_search(apiobj, 0.1, [lookup], [ranking])
  97
  98         assert len(results) == 1
  99         assert results[0].place_id == 101
 100
 101
 102     @pytest.mark.parametrize('cc,res', [('us', 100), ('mx', 101)])
 103     def test_lookup_restrict_country(self, apiobj, cc, res):
 104         lookup = FieldLookup('name_vector', [1,2], LookupAll)
 105         ranking = FieldRanking('name_vector', 0.9, [RankedTokens(0.0, [10])])
 106
 107         results = run_search(apiobj, 0.1, [lookup], [ranking], ccodes=[cc])
 108
 109         assert [r.place_id for r in results] == [res]
 110
 111
 112     def test_lookup_restrict_placeid(self, apiobj):
 113         lookup = FieldLookup('name_vector', [1,2], LookupAll)
 114         ranking = FieldRanking('name_vector', 0.9, [RankedTokens(0.0, [10])])
 115
 116         results = run_search(apiobj, 0.1, [lookup], [ranking],
 117                              details=SearchDetails(excluded=[101]))
 118
 119         assert [r.place_id for r in results] == [100]
 120
 121
 122     @pytest.mark.parametrize('geom', [napi.GeometryFormat.GEOJSON,
 123                                       napi.GeometryFormat.KML,
 124                                       napi.GeometryFormat.SVG,
 125                                       napi.GeometryFormat.TEXT])
 126     def test_return_geometries(self, apiobj, geom):
 127         lookup = FieldLookup('name_vector', [20], LookupAll)
 128         ranking = FieldRanking('name_vector', 0.9, [RankedTokens(0.0, [21])])
 129
 130         results = run_search(apiobj, 0.1, [lookup], [ranking],
 131                              details=SearchDetails(geometry_output=geom))
 132
 133         assert geom.name.lower() in results[0].geometry
 134
 135
 136     @pytest.mark.parametrize('factor,npoints', [(0.0, 3), (1.0, 2)])
 137     def test_return_simplified_geometry(self, apiobj, factor, npoints):
 138         apiobj.add_placex(place_id=333, country_code='us',
 139                           centroid=(9.0, 9.0),
 140                           geometry='LINESTRING(8.9 9.0, 9.0 9.0, 9.1 9.0)')
 141         apiobj.add_search_name(333, names=[55], country_code='us',
 142                                centroid=(5.6, 4.3))
 143
 144         lookup = FieldLookup('name_vector', [55], LookupAll)
 145         ranking = FieldRanking('name_vector', 0.9, [RankedTokens(0.0, [21])])
 146
 147         results = run_search(apiobj, 0.1, [lookup], [ranking],
 148                              details=SearchDetails(geometry_output=napi.GeometryFormat.GEOJSON,
 149                                                    geometry_simplification=factor))
 150
 151         assert len(results) == 1
 152         result = results[0]
 153         geom = json.loads(result.geometry['geojson'])
 154
 155         assert result.place_id == 333
 156         assert len(geom['coordinates']) == npoints
 157
 158
 159     @pytest.mark.parametrize('viewbox', ['5.0,4.0,6.0,5.0', '5.7,4.0,6.0,5.0'])
 160     @pytest.mark.parametrize('wcount,rids', [(2, [100, 101]), (20000, [100])])
 161     def test_prefer_viewbox(self, apiobj, viewbox, wcount, rids):
 162         lookup = FieldLookup('name_vector', [1, 2], LookupAll)
 163         ranking = FieldRanking('name_vector', 0.2, [RankedTokens(0.0, [21])])
 164
 165         results = run_search(apiobj, 0.1, [lookup], [ranking])
 166         assert [r.place_id for r in results] == [101, 100]
 167
 168         results = run_search(apiobj, 0.1, [lookup], [ranking], count=wcount,
 169                              details=SearchDetails.from_kwargs({'viewbox': viewbox}))
 170         assert [r.place_id for r in results] == rids
 171
 172
 173     @pytest.mark.parametrize('viewbox', ['5.0,4.0,6.0,5.0', '5.55,4.27,5.62,4.31'])
 174     def test_force_viewbox(self, apiobj, viewbox):
 175         lookup = FieldLookup('name_vector', [1, 2], LookupAll)
 176
 177         details=SearchDetails.from_kwargs({'viewbox': viewbox,
 178                                            'bounded_viewbox': True})
 179
 180         results = run_search(apiobj, 0.1, [lookup], [], details=details)
 181         assert [r.place_id for r in results] == [100]
 182
 183
 184     def test_prefer_near(self, apiobj):
 185         lookup = FieldLookup('name_vector', [1, 2], LookupAll)
 186         ranking = FieldRanking('name_vector', 0.9, [RankedTokens(0.0, [21])])
 187
 188         results = run_search(apiobj, 0.1, [lookup], [ranking])
 189         assert [r.place_id for r in results] == [101, 100]
 190
 191         results = run_search(apiobj, 0.1, [lookup], [ranking],
 192                              details=SearchDetails.from_kwargs({'near': '5.6,4.3'}))
 193         results.sort(key=lambda r: -r.importance)
 194         assert [r.place_id for r in results] == [100, 101]
 195
 196
 197     @pytest.mark.parametrize('radius', [0.09, 0.11])
 198     def test_force_near(self, apiobj, radius):
 199         lookup = FieldLookup('name_vector', [1, 2], LookupAll)
 200
 201         details=SearchDetails.from_kwargs({'near': '5.6,4.3',
 202                                            'near_radius': radius})
 203
 204         results = run_search(apiobj, 0.1, [lookup], [], details=details)
 205
 206         assert [r.place_id for r in results] == [100]
 207
 208
 209 class TestStreetWithHousenumber:
 210
 211     @pytest.fixture(autouse=True)
 212     def fill_database(self, apiobj):
 213         apiobj.add_placex(place_id=1, class_='place', type='house',
 214                           parent_place_id=1000,
 215                           housenumber='20 a', country_code='es')
 216         apiobj.add_placex(place_id=2, class_='place', type='house',
 217                           parent_place_id=1000,
 218                           housenumber='21;22', country_code='es')
 219         apiobj.add_placex(place_id=1000, class_='highway', type='residential',
 220                           rank_search=26, rank_address=26,
 221                           country_code='es')
 222         apiobj.add_search_name(1000, names=[1,2,10,11],
 223                                search_rank=26, address_rank=26,
 224                                country_code='es')
 225         apiobj.add_placex(place_id=91, class_='place', type='house',
 226                           parent_place_id=2000,
 227                           housenumber='20', country_code='pt')
 228         apiobj.add_placex(place_id=92, class_='place', type='house',
 229                           parent_place_id=2000,
 230                           housenumber='22', country_code='pt')
 231         apiobj.add_placex(place_id=93, class_='place', type='house',
 232                           parent_place_id=2000,
 233                           housenumber='24', country_code='pt')
 234         apiobj.add_placex(place_id=2000, class_='highway', type='residential',
 235                           rank_search=26, rank_address=26,
 236                           country_code='pt')
 237         apiobj.add_search_name(2000, names=[1,2,20,21],
 238                                search_rank=26, address_rank=26,
 239                                country_code='pt')
 240
 241
 242     @pytest.mark.parametrize('hnr,res', [('20', [91, 1]), ('20 a', [1]),
 243                                          ('21', [2]), ('22', [2, 92]),
 244                                          ('24', [93]), ('25', [])])
 245     def test_lookup_by_single_housenumber(self, apiobj, hnr, res):
 246         lookup = FieldLookup('name_vector', [1,2], LookupAll)
 247         ranking = FieldRanking('name_vector', 0.3, [RankedTokens(0.0, [10])])
 248
 249         results = run_search(apiobj, 0.1, [lookup], [ranking], hnrs=[hnr])
 250
 251         assert [r.place_id for r in results] == res + [1000, 2000]
 252
 253
 254     @pytest.mark.parametrize('cc,res', [('es', [2, 1000]), ('pt', [92, 2000])])
 255     def test_lookup_with_country_restriction(self, apiobj, cc, res):
 256         lookup = FieldLookup('name_vector', [1,2], LookupAll)
 257         ranking = FieldRanking('name_vector', 0.3, [RankedTokens(0.0, [10])])
 258
 259         results = run_search(apiobj, 0.1, [lookup], [ranking], hnrs=['22'],
 260                              ccodes=[cc])
 261
 262         assert [r.place_id for r in results] == res
 263
 264
 265     def test_lookup_exclude_housenumber_placeid(self, apiobj):
 266         lookup = FieldLookup('name_vector', [1,2], LookupAll)
 267         ranking = FieldRanking('name_vector', 0.3, [RankedTokens(0.0, [10])])
 268
 269         results = run_search(apiobj, 0.1, [lookup], [ranking], hnrs=['22'],
 270                              details=SearchDetails(excluded=[92]))
 271
 272         assert [r.place_id for r in results] == [2, 1000, 2000]
 273
 274
 275     def test_lookup_exclude_street_placeid(self, apiobj):
 276         lookup = FieldLookup('name_vector', [1,2], LookupAll)
 277         ranking = FieldRanking('name_vector', 0.3, [RankedTokens(0.0, [10])])
 278
 279         results = run_search(apiobj, 0.1, [lookup], [ranking], hnrs=['22'],
 280                              details=SearchDetails(excluded=[1000]))
 281
 282         assert [r.place_id for r in results] == [2, 92, 2000]
 283
 284
 285     def test_lookup_only_house_qualifier(self, apiobj):
 286         lookup = FieldLookup('name_vector', [1,2], LookupAll)
 287         ranking = FieldRanking('name_vector', 0.3, [RankedTokens(0.0, [10])])
 288
 289         results = run_search(apiobj, 0.1, [lookup], [ranking], hnrs=['22'],
 290                              quals=[('place', 'house')])
 291
 292         assert [r.place_id for r in results] == [2, 92]
 293
 294
 295     def test_lookup_only_street_qualifier(self, apiobj):
 296         lookup = FieldLookup('name_vector', [1,2], LookupAll)
 297         ranking = FieldRanking('name_vector', 0.3, [RankedTokens(0.0, [10])])
 298
 299         results = run_search(apiobj, 0.1, [lookup], [ranking], hnrs=['22'],
 300                              quals=[('highway', 'residential')])
 301
 302         assert [r.place_id for r in results] == [1000, 2000]
 303
 304
 305     @pytest.mark.parametrize('rank,found', [(26, True), (27, False), (30, False)])
 306     def test_lookup_min_rank(self, apiobj, rank, found):
 307         lookup = FieldLookup('name_vector', [1,2], LookupAll)
 308         ranking = FieldRanking('name_vector', 0.3, [RankedTokens(0.0, [10])])
 309
 310         results = run_search(apiobj, 0.1, [lookup], [ranking], hnrs=['22'],
 311                              details=SearchDetails(min_rank=rank))
 312
 313         assert [r.place_id for r in results] == ([2, 92, 1000, 2000] if found else [2, 92])
 314
 315
 316     @pytest.mark.parametrize('geom', [napi.GeometryFormat.GEOJSON,
 317                                       napi.GeometryFormat.KML,
 318                                       napi.GeometryFormat.SVG,
 319                                       napi.GeometryFormat.TEXT])
 320     def test_return_geometries(self, apiobj, geom):
 321         lookup = FieldLookup('name_vector', [1, 2], LookupAll)
 322
 323         results = run_search(apiobj, 0.1, [lookup], [], hnrs=['20', '21', '22'],
 324                              details=SearchDetails(geometry_output=geom))
 325
 326         assert results
 327         assert all(geom.name.lower() in r.geometry for r in results)
 328
 329
 330 def test_very_large_housenumber(apiobj):
 331     apiobj.add_placex(place_id=93, class_='place', type='house',
 332                       parent_place_id=2000,
 333                       housenumber='2467463524544', country_code='pt')
 334     apiobj.add_placex(place_id=2000, class_='highway', type='residential',
 335                       rank_search=26, rank_address=26,
 336                       country_code='pt')
 337     apiobj.add_search_name(2000, names=[1,2],
 338                            search_rank=26, address_rank=26,
 339                            country_code='pt')
 340
 341     lookup = FieldLookup('name_vector', [1, 2], LookupAll)
 342
 343     results = run_search(apiobj, 0.1, [lookup], [], hnrs=['2467463524544'],
 344                          details=SearchDetails())
 345
 346     assert results
 347     assert [r.place_id for r in results] == [93, 2000]
 348
 349
 350 @pytest.mark.parametrize('wcount,rids', [(2, [990, 991]), (30000, [990])])
 351 def test_name_and_postcode(apiobj, wcount, rids):
 352     apiobj.add_placex(place_id=990, class_='highway', type='service',
 353                       rank_search=27, rank_address=27,
 354                       postcode='11225',
 355                       centroid=(10.0, 10.0),
 356                       geometry='LINESTRING(9.995 10, 10.005 10)')
 357     apiobj.add_search_name(990, names=[111], centroid=(10.0, 10.0),
 358                            search_rank=27, address_rank=27)
 359     apiobj.add_placex(place_id=991, class_='highway', type='service',
 360                       rank_search=27, rank_address=27,
 361                       postcode='11221',
 362                       centroid=(10.1, 10.1),
 363                       geometry='LINESTRING(9.995 10.1, 10.005 10.1)')
 364     apiobj.add_search_name(991, names=[111], centroid=(10.1, 10.1),
 365                            search_rank=27, address_rank=27)
 366     apiobj.add_postcode(place_id=100, country_code='ch', postcode='11225',
 367                         geometry='POINT(10 10)')
 368
 369     lookup = FieldLookup('name_vector', [111], LookupAll)
 370
 371     results = run_search(apiobj, 0.1, [lookup], [], pcs=['11225'], count=wcount,
 372                          details=SearchDetails())
 373
 374     assert results
 375     assert [r.place_id for r in results] == rids
 376
 377
 378 class TestInterpolations:
 379
 380     @pytest.fixture(autouse=True)
 381     def fill_database(self, apiobj):
 382         apiobj.add_placex(place_id=990, class_='highway', type='service',
 383                           rank_search=27, rank_address=27,
 384                           centroid=(10.0, 10.0),
 385                           geometry='LINESTRING(9.995 10, 10.005 10)')
 386         apiobj.add_search_name(990, names=[111],
 387                                search_rank=27, address_rank=27)
 388         apiobj.add_placex(place_id=991, class_='place', type='house',
 389                           parent_place_id=990,
 390                           rank_search=30, rank_address=30,
 391                           housenumber='23',
 392                           centroid=(10.0, 10.00002))
 393         apiobj.add_osmline(place_id=992,
 394                            parent_place_id=990,
 395                            startnumber=21, endnumber=29, step=2,
 396                            centroid=(10.0, 10.00001),
 397                            geometry='LINESTRING(9.995 10.00001, 10.005 10.00001)')
 398
 399
 400     @pytest.mark.parametrize('hnr,res', [('21', [992]), ('22', []), ('23', [991])])
 401     def test_lookup_housenumber(self, apiobj, hnr, res):
 402         lookup = FieldLookup('name_vector', [111], LookupAll)
 403
 404         results = run_search(apiobj, 0.1, [lookup], [], hnrs=[hnr])
 405
 406         assert [r.place_id for r in results] == res + [990]
 407
 408
 409     @pytest.mark.parametrize('geom', [napi.GeometryFormat.GEOJSON,
 410                                       napi.GeometryFormat.KML,
 411                                       napi.GeometryFormat.SVG,
 412                                       napi.GeometryFormat.TEXT])
 413     def test_osmline_with_geometries(self, apiobj, geom):
 414         lookup = FieldLookup('name_vector', [111], LookupAll)
 415
 416         results = run_search(apiobj, 0.1, [lookup], [], hnrs=['21'],
 417                              details=SearchDetails(geometry_output=geom))
 418
 419         assert results[0].place_id == 992
 420         assert geom.name.lower() in results[0].geometry
 421
 422
 423
 424 class TestTiger:
 425
 426     @pytest.fixture(autouse=True)
 427     def fill_database(self, apiobj):
 428         apiobj.add_placex(place_id=990, class_='highway', type='service',
 429                           rank_search=27, rank_address=27,
 430                           country_code='us',
 431                           centroid=(10.0, 10.0),
 432                           geometry='LINESTRING(9.995 10, 10.005 10)')
 433         apiobj.add_search_name(990, names=[111], country_code='us',
 434                                search_rank=27, address_rank=27)
 435         apiobj.add_placex(place_id=991, class_='place', type='house',
 436                           parent_place_id=990,
 437                           rank_search=30, rank_address=30,
 438                           housenumber='23',
 439                           country_code='us',
 440                           centroid=(10.0, 10.00002))
 441         apiobj.add_tiger(place_id=992,
 442                          parent_place_id=990,
 443                          startnumber=21, endnumber=29, step=2,
 444                          centroid=(10.0, 10.00001),
 445                          geometry='LINESTRING(9.995 10.00001, 10.005 10.00001)')
 446
 447
 448     @pytest.mark.parametrize('hnr,res', [('21', [992]), ('22', []), ('23', [991])])
 449     def test_lookup_housenumber(self, apiobj, hnr, res):
 450         lookup = FieldLookup('name_vector', [111], LookupAll)
 451
 452         results = run_search(apiobj, 0.1, [lookup], [], hnrs=[hnr])
 453
 454         assert [r.place_id for r in results] == res + [990]
 455
 456
 457     @pytest.mark.parametrize('geom', [napi.GeometryFormat.GEOJSON,
 458                                       napi.GeometryFormat.KML,
 459                                       napi.GeometryFormat.SVG,
 460                                       napi.GeometryFormat.TEXT])
 461     def test_tiger_with_geometries(self, apiobj, geom):
 462         lookup = FieldLookup('name_vector', [111], LookupAll)
 463
 464         results = run_search(apiobj, 0.1, [lookup], [], hnrs=['21'],
 465                              details=SearchDetails(geometry_output=geom))
 466
 467         assert results[0].place_id == 992
 468         assert geom.name.lower() in results[0].geometry
 469
 470
 471 class TestLayersRank30:
 472
 473     @pytest.fixture(autouse=True)
 474     def fill_database(self, apiobj):
 475         apiobj.add_placex(place_id=223, class_='place', type='house',
 476                           housenumber='1',
 477                           rank_address=30,
 478                           rank_search=30)
 479         apiobj.add_search_name(223, names=[34],
 480                                importance=0.0009,
 481                                address_rank=30, search_rank=30)
 482         apiobj.add_placex(place_id=224, class_='amenity', type='toilet',
 483                           rank_address=30,
 484                           rank_search=30)
 485         apiobj.add_search_name(224, names=[34],
 486                                importance=0.0008,
 487                                address_rank=30, search_rank=30)
 488         apiobj.add_placex(place_id=225, class_='man_made', type='tower',
 489                           rank_address=0,
 490                           rank_search=30)
 491         apiobj.add_search_name(225, names=[34],
 492                                importance=0.0007,
 493                                address_rank=0, search_rank=30)
 494         apiobj.add_placex(place_id=226, class_='railway', type='station',
 495                           rank_address=0,
 496                           rank_search=30)
 497         apiobj.add_search_name(226, names=[34],
 498                                importance=0.0006,
 499                                address_rank=0, search_rank=30)
 500         apiobj.add_placex(place_id=227, class_='natural', type='cave',
 501                           rank_address=0,
 502                           rank_search=30)
 503         apiobj.add_search_name(227, names=[34],
 504                                importance=0.0005,
 505                                address_rank=0, search_rank=30)
 506
 507
 508     @pytest.mark.parametrize('layer,res', [(napi.DataLayer.ADDRESS, [223]),
 509                                            (napi.DataLayer.POI, [224]),
 510                                            (napi.DataLayer.ADDRESS | napi.DataLayer.POI, [223, 224]),
 511                                            (napi.DataLayer.MANMADE, [225]),
 512                                            (napi.DataLayer.RAILWAY, [226]),
 513                                            (napi.DataLayer.NATURAL, [227]),
 514                                            (napi.DataLayer.MANMADE | napi.DataLayer.NATURAL, [225, 227]),
 515                                            (napi.DataLayer.MANMADE | napi.DataLayer.RAILWAY, [225, 226])])
 516     def test_layers_rank30(self, apiobj, layer, res):
 517         lookup = FieldLookup('name_vector', [34], LookupAny)
 518
 519         results = run_search(apiobj, 0.1, [lookup], [],
 520                              details=SearchDetails(layers=layer))
 521
 522         assert [r.place_id for r in results] == res