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