nominatim/api/search/db_searches.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 Implementation of the acutal database accesses for forward search.
   9 """
  10 from typing import List, Tuple, AsyncIterator, Dict, Any, Callable
  11 import abc
  12
  13 import sqlalchemy as sa
  14 from sqlalchemy.dialects.postgresql import ARRAY, array_agg
  15
  16 from nominatim.typing import SaFromClause, SaScalarSelect, SaColumn, \
  17                              SaExpression, SaSelect, SaLambdaSelect, SaRow, SaBind
  18 from nominatim.api.connection import SearchConnection
  19 from nominatim.api.types import SearchDetails, DataLayer, GeometryFormat, Bbox
  20 import nominatim.api.results as nres
  21 from nominatim.api.search.db_search_fields import SearchData, WeightedCategories
  22 from nominatim.db.sqlalchemy_types import Geometry
  23
  24 #pylint: disable=singleton-comparison,not-callable
  25 #pylint: disable=too-many-branches,too-many-arguments,too-many-locals,too-many-statements
  26
  27 def _details_to_bind_params(details: SearchDetails) -> Dict[str, Any]:
  28     """ Create a dictionary from search parameters that can be used
  29         as bind parameter for SQL execute.
  30     """
  31     return {'limit': details.max_results,
  32             'min_rank': details.min_rank,
  33             'max_rank': details.max_rank,
  34             'viewbox': details.viewbox,
  35             'viewbox2': details.viewbox_x2,
  36             'near': details.near,
  37             'near_radius': details.near_radius,
  38             'excluded': details.excluded,
  39             'countries': details.countries}
  40
  41
  42 LIMIT_PARAM: SaBind = sa.bindparam('limit')
  43 MIN_RANK_PARAM: SaBind = sa.bindparam('min_rank')
  44 MAX_RANK_PARAM: SaBind = sa.bindparam('max_rank')
  45 VIEWBOX_PARAM: SaBind = sa.bindparam('viewbox', type_=Geometry)
  46 VIEWBOX2_PARAM: SaBind = sa.bindparam('viewbox2', type_=Geometry)
  47 NEAR_PARAM: SaBind = sa.bindparam('near', type_=Geometry)
  48 NEAR_RADIUS_PARAM: SaBind = sa.bindparam('near_radius')
  49 COUNTRIES_PARAM: SaBind = sa.bindparam('countries')
  50
  51 def _within_near(t: SaFromClause) -> Callable[[], SaExpression]:
  52     return lambda: t.c.geometry.ST_DWithin(NEAR_PARAM, NEAR_RADIUS_PARAM)
  53
  54 def _exclude_places(t: SaFromClause) -> Callable[[], SaExpression]:
  55     return lambda: t.c.place_id.not_in(sa.bindparam('excluded'))
  56
  57 def _select_placex(t: SaFromClause) -> SaSelect:
  58     return sa.select(t.c.place_id, t.c.osm_type, t.c.osm_id, t.c.name,
  59                      t.c.class_, t.c.type,
  60                      t.c.address, t.c.extratags,
  61                      t.c.housenumber, t.c.postcode, t.c.country_code,
  62                      t.c.importance, t.c.wikipedia,
  63                      t.c.parent_place_id, t.c.rank_address, t.c.rank_search,
  64                      t.c.centroid,
  65                      t.c.geometry.ST_Expand(0).label('bbox'))
  66
  67
  68 def _add_geometry_columns(sql: SaLambdaSelect, col: SaColumn, details: SearchDetails) -> SaSelect:
  69     out = []
  70
  71     if details.geometry_simplification > 0.0:
  72         col = sa.func.ST_SimplifyPreserveTopology(col, details.geometry_simplification)
  73
  74     if details.geometry_output & GeometryFormat.GEOJSON:
  75         out.append(sa.func.ST_AsGeoJSON(col).label('geometry_geojson'))
  76     if details.geometry_output & GeometryFormat.TEXT:
  77         out.append(sa.func.ST_AsText(col).label('geometry_text'))
  78     if details.geometry_output & GeometryFormat.KML:
  79         out.append(sa.func.ST_AsKML(col).label('geometry_kml'))
  80     if details.geometry_output & GeometryFormat.SVG:
  81         out.append(sa.func.ST_AsSVG(col).label('geometry_svg'))
  82
  83     return sql.add_columns(*out)
  84
  85
  86 def _make_interpolation_subquery(table: SaFromClause, inner: SaFromClause,
  87                                  numerals: List[int], details: SearchDetails) -> SaScalarSelect:
  88     all_ids = array_agg(table.c.place_id) # type: ignore[no-untyped-call]
  89     sql = sa.select(all_ids).where(table.c.parent_place_id == inner.c.place_id)
  90
  91     if len(numerals) == 1:
  92         sql = sql.where(sa.between(numerals[0], table.c.startnumber, table.c.endnumber))\
  93                  .where((numerals[0] - table.c.startnumber) % table.c.step == 0)
  94     else:
  95         sql = sql.where(sa.or_(
  96                 *(sa.and_(sa.between(n, table.c.startnumber, table.c.endnumber),
  97                           (n - table.c.startnumber) % table.c.step == 0)
  98                   for n in numerals)))
  99
 100     if details.excluded:
 101         sql = sql.where(_exclude_places(table))
 102
 103     return sql.scalar_subquery()
 104
 105
 106 def _filter_by_layer(table: SaFromClause, layers: DataLayer) -> SaColumn:
 107     orexpr: List[SaExpression] = []
 108     if layers & DataLayer.ADDRESS and layers & DataLayer.POI:
 109         orexpr.append(table.c.rank_address.between(1, 30))
 110     elif layers & DataLayer.ADDRESS:
 111         orexpr.append(table.c.rank_address.between(1, 29))
 112         orexpr.append(sa.and_(table.c.rank_address == 30,
 113                               sa.or_(table.c.housenumber != None,
 114                                      table.c.address.has_key('housename'))))
 115     elif layers & DataLayer.POI:
 116         orexpr.append(sa.and_(table.c.rank_address == 30,
 117                               table.c.class_.not_in(('place', 'building'))))
 118
 119     if layers & DataLayer.MANMADE:
 120         exclude = []
 121         if not layers & DataLayer.RAILWAY:
 122             exclude.append('railway')
 123         if not layers & DataLayer.NATURAL:
 124             exclude.extend(('natural', 'water', 'waterway'))
 125         orexpr.append(sa.and_(table.c.class_.not_in(tuple(exclude)),
 126                               table.c.rank_address == 0))
 127     else:
 128         include = []
 129         if layers & DataLayer.RAILWAY:
 130             include.append('railway')
 131         if layers & DataLayer.NATURAL:
 132             include.extend(('natural', 'water', 'waterway'))
 133         orexpr.append(sa.and_(table.c.class_.in_(tuple(include)),
 134                               table.c.rank_address == 0))
 135
 136     if len(orexpr) == 1:
 137         return orexpr[0]
 138
 139     return sa.or_(*orexpr)
 140
 141
 142 def _interpolated_position(table: SaFromClause, nr: SaColumn) -> SaColumn:
 143     pos = sa.cast(nr - table.c.startnumber, sa.Float) / (table.c.endnumber - table.c.startnumber)
 144     return sa.case(
 145             (table.c.endnumber == table.c.startnumber, table.c.linegeo.ST_Centroid()),
 146             else_=table.c.linegeo.ST_LineInterpolatePoint(pos)).label('centroid')
 147
 148
 149 async def _get_placex_housenumbers(conn: SearchConnection,
 150                                    place_ids: List[int],
 151                                    details: SearchDetails) -> AsyncIterator[nres.SearchResult]:
 152     t = conn.t.placex
 153     sql = _select_placex(t).where(t.c.place_id.in_(place_ids))
 154
 155     if details.geometry_output:
 156         sql = _add_geometry_columns(sql, t.c.geometry, details)
 157
 158     for row in await conn.execute(sql):
 159         result = nres.create_from_placex_row(row, nres.SearchResult)
 160         assert result
 161         result.bbox = Bbox.from_wkb(row.bbox)
 162         yield result
 163
 164
 165 async def _get_osmline(conn: SearchConnection, place_ids: List[int],
 166                        numerals: List[int],
 167                        details: SearchDetails) -> AsyncIterator[nres.SearchResult]:
 168     t = conn.t.osmline
 169     values = sa.values(sa.Column('nr', sa.Integer()), name='housenumber')\
 170                .data([(n,) for n in numerals])
 171     sql = sa.select(t.c.place_id, t.c.osm_id,
 172                     t.c.parent_place_id, t.c.address,
 173                     values.c.nr.label('housenumber'),
 174                     _interpolated_position(t, values.c.nr),
 175                     t.c.postcode, t.c.country_code)\
 176             .where(t.c.place_id.in_(place_ids))\
 177             .join(values, values.c.nr.between(t.c.startnumber, t.c.endnumber))
 178
 179     if details.geometry_output:
 180         sub = sql.subquery()
 181         sql = _add_geometry_columns(sa.select(sub), sub.c.centroid, details)
 182
 183     for row in await conn.execute(sql):
 184         result = nres.create_from_osmline_row(row, nres.SearchResult)
 185         assert result
 186         yield result
 187
 188
 189 async def _get_tiger(conn: SearchConnection, place_ids: List[int],
 190                      numerals: List[int], osm_id: int,
 191                      details: SearchDetails) -> AsyncIterator[nres.SearchResult]:
 192     t = conn.t.tiger
 193     values = sa.values(sa.Column('nr', sa.Integer()), name='housenumber')\
 194                .data([(n,) for n in numerals])
 195     sql = sa.select(t.c.place_id, t.c.parent_place_id,
 196                     sa.literal('W').label('osm_type'),
 197                     sa.literal(osm_id).label('osm_id'),
 198                     values.c.nr.label('housenumber'),
 199                     _interpolated_position(t, values.c.nr),
 200                     t.c.postcode)\
 201             .where(t.c.place_id.in_(place_ids))\
 202             .join(values, values.c.nr.between(t.c.startnumber, t.c.endnumber))
 203
 204     if details.geometry_output:
 205         sub = sql.subquery()
 206         sql = _add_geometry_columns(sa.select(sub), sub.c.centroid, details)
 207
 208     for row in await conn.execute(sql):
 209         result = nres.create_from_tiger_row(row, nres.SearchResult)
 210         assert result
 211         yield result
 212
 213
 214 class AbstractSearch(abc.ABC):
 215     """ Encapuslation of a single lookup in the database.
 216     """
 217
 218     def __init__(self, penalty: float) -> None:
 219         self.penalty = penalty
 220
 221     @abc.abstractmethod
 222     async def lookup(self, conn: SearchConnection,
 223                      details: SearchDetails) -> nres.SearchResults:
 224         """ Find results for the search in the database.
 225         """
 226
 227
 228 class NearSearch(AbstractSearch):
 229     """ Category search of a place type near the result of another search.
 230     """
 231     def __init__(self, penalty: float, categories: WeightedCategories,
 232                  search: AbstractSearch) -> None:
 233         super().__init__(penalty)
 234         self.search = search
 235         self.categories = categories
 236
 237
 238     async def lookup(self, conn: SearchConnection,
 239                      details: SearchDetails) -> nres.SearchResults:
 240         """ Find results for the search in the database.
 241         """
 242         results = nres.SearchResults()
 243         base = await self.search.lookup(conn, details)
 244
 245         if not base:
 246             return results
 247
 248         base.sort(key=lambda r: (r.accuracy, r.rank_search))
 249         max_accuracy = base[0].accuracy + 0.5
 250         base = nres.SearchResults(r for r in base if r.source_table == nres.SourceTable.PLACEX
 251                                                      and r.accuracy <= max_accuracy
 252                                                      and r.bbox and r.bbox.area < 20)
 253
 254         if base:
 255             baseids = [b.place_id for b in base[:5] if b.place_id]
 256
 257             for category, penalty in self.categories:
 258                 await self.lookup_category(results, conn, baseids, category, penalty, details)
 259                 if len(results) >= details.max_results:
 260                     break
 261
 262         return results
 263
 264
 265     async def lookup_category(self, results: nres.SearchResults,
 266                               conn: SearchConnection, ids: List[int],
 267                               category: Tuple[str, str], penalty: float,
 268                               details: SearchDetails) -> None:
 269         """ Find places of the given category near the list of
 270             place ids and add the results to 'results'.
 271         """
 272         table = await conn.get_class_table(*category)
 273
 274         t = conn.t.placex.alias('p')
 275         tgeom = conn.t.placex.alias('pgeom')
 276
 277         sql = _select_placex(t).where(tgeom.c.place_id.in_(ids))\
 278                                .where(t.c.class_ == category[0])\
 279                                .where(t.c.type == category[1])
 280
 281         if table is None:
 282             # No classtype table available, do a simplified lookup in placex.
 283             sql = sql.join(tgeom, t.c.geometry.ST_DWithin(tgeom.c.centroid, 0.01))\
 284                      .order_by(tgeom.c.centroid.ST_Distance(t.c.centroid))
 285         else:
 286             # Use classtype table. We can afford to use a larger
 287             # radius for the lookup.
 288             sql = sql.join(table, t.c.place_id == table.c.place_id)\
 289                      .join(tgeom,
 290                            sa.case((sa.and_(tgeom.c.rank_address < 9,
 291                                             tgeom.c.geometry.is_area()),
 292                                     tgeom.c.geometry.ST_Contains(table.c.centroid)),
 293                                    else_ = tgeom.c.centroid.ST_DWithin(table.c.centroid, 0.05)))\
 294                      .order_by(tgeom.c.centroid.ST_Distance(table.c.centroid))
 295
 296         sql = sql.where(t.c.rank_address.between(MIN_RANK_PARAM, MAX_RANK_PARAM))
 297         if details.countries:
 298             sql = sql.where(t.c.country_code.in_(COUNTRIES_PARAM))
 299         if details.excluded:
 300             sql = sql.where(_exclude_places(t))
 301         if details.layers is not None:
 302             sql = sql.where(_filter_by_layer(t, details.layers))
 303
 304         sql = sql.limit(LIMIT_PARAM)
 305         for row in await conn.execute(sql, _details_to_bind_params(details)):
 306             result = nres.create_from_placex_row(row, nres.SearchResult)
 307             assert result
 308             result.accuracy = self.penalty + penalty
 309             result.bbox = Bbox.from_wkb(row.bbox)
 310             results.append(result)
 311
 312
 313
 314 class PoiSearch(AbstractSearch):
 315     """ Category search in a geographic area.
 316     """
 317     def __init__(self, sdata: SearchData) -> None:
 318         super().__init__(sdata.penalty)
 319         self.categories = sdata.qualifiers
 320         self.countries = sdata.countries
 321
 322
 323     async def lookup(self, conn: SearchConnection,
 324                      details: SearchDetails) -> nres.SearchResults:
 325         """ Find results for the search in the database.
 326         """
 327         bind_params = _details_to_bind_params(details)
 328         t = conn.t.placex
 329
 330         rows: List[SaRow] = []
 331
 332         if details.near and details.near_radius is not None and details.near_radius < 0.2:
 333             # simply search in placex table
 334             def _base_query() -> SaSelect:
 335                 return _select_placex(t) \
 336                            .where(t.c.linked_place_id == None) \
 337                            .where(t.c.geometry.ST_DWithin(NEAR_PARAM, NEAR_RADIUS_PARAM)) \
 338                            .order_by(t.c.centroid.ST_Distance(NEAR_PARAM)) \
 339                            .limit(LIMIT_PARAM)
 340
 341             classtype = self.categories.values
 342             if len(classtype) == 1:
 343                 cclass, ctype = classtype[0]
 344                 sql: SaLambdaSelect = sa.lambda_stmt(lambda: _base_query()
 345                                                  .where(t.c.class_ == cclass)
 346                                                  .where(t.c.type == ctype))
 347             else:
 348                 sql = _base_query().where(sa.or_(*(sa.and_(t.c.class_ == cls, t.c.type == typ)
 349                                                    for cls, typ in classtype)))
 350
 351             if self.countries:
 352                 sql = sql.where(t.c.country_code.in_(self.countries.values))
 353
 354             if details.viewbox is not None and details.bounded_viewbox:
 355                 sql = sql.where(t.c.geometry.intersects(VIEWBOX_PARAM))
 356
 357             rows.extend(await conn.execute(sql, bind_params))
 358         else:
 359             # use the class type tables
 360             for category in self.categories.values:
 361                 table = await conn.get_class_table(*category)
 362                 if table is not None:
 363                     sql = _select_placex(t)\
 364                                .join(table, t.c.place_id == table.c.place_id)\
 365                                .where(t.c.class_ == category[0])\
 366                                .where(t.c.type == category[1])
 367
 368                     if details.viewbox is not None and details.bounded_viewbox:
 369                         sql = sql.where(table.c.centroid.intersects(VIEWBOX_PARAM))
 370
 371                     if details.near and details.near_radius is not None:
 372                         sql = sql.order_by(table.c.centroid.ST_Distance(NEAR_PARAM))\
 373                                  .where(table.c.centroid.ST_DWithin(NEAR_PARAM,
 374                                                                     NEAR_RADIUS_PARAM))
 375
 376                     if self.countries:
 377                         sql = sql.where(t.c.country_code.in_(self.countries.values))
 378
 379                     sql = sql.limit(LIMIT_PARAM)
 380                     rows.extend(await conn.execute(sql, bind_params))
 381
 382         results = nres.SearchResults()
 383         for row in rows:
 384             result = nres.create_from_placex_row(row, nres.SearchResult)
 385             assert result
 386             result.accuracy = self.penalty + self.categories.get_penalty((row.class_, row.type))
 387             result.bbox = Bbox.from_wkb(row.bbox)
 388             results.append(result)
 389
 390         return results
 391
 392
 393 class CountrySearch(AbstractSearch):
 394     """ Search for a country name or country code.
 395     """
 396     def __init__(self, sdata: SearchData) -> None:
 397         super().__init__(sdata.penalty)
 398         self.countries = sdata.countries
 399
 400
 401     async def lookup(self, conn: SearchConnection,
 402                      details: SearchDetails) -> nres.SearchResults:
 403         """ Find results for the search in the database.
 404         """
 405         t = conn.t.placex
 406
 407         ccodes = self.countries.values
 408         sql: SaLambdaSelect = sa.lambda_stmt(lambda: _select_placex(t)\
 409                 .where(t.c.country_code.in_(ccodes))\
 410                 .where(t.c.rank_address == 4))
 411
 412         if details.geometry_output:
 413             sql = _add_geometry_columns(sql, t.c.geometry, details)
 414
 415         if details.excluded:
 416             sql = sql.where(_exclude_places(t))
 417
 418         if details.viewbox is not None and details.bounded_viewbox:
 419             sql = sql.where(lambda: t.c.geometry.intersects(VIEWBOX_PARAM))
 420
 421         if details.near is not None and details.near_radius is not None:
 422             sql = sql.where(_within_near(t))
 423
 424         results = nres.SearchResults()
 425         for row in await conn.execute(sql, _details_to_bind_params(details)):
 426             result = nres.create_from_placex_row(row, nres.SearchResult)
 427             assert result
 428             result.accuracy = self.penalty + self.countries.get_penalty(row.country_code, 5.0)
 429             results.append(result)
 430
 431         return results or await self.lookup_in_country_table(conn, details)
 432
 433
 434     async def lookup_in_country_table(self, conn: SearchConnection,
 435                                       details: SearchDetails) -> nres.SearchResults:
 436         """ Look up the country in the fallback country tables.
 437         """
 438         # Avoid the fallback search when this is a more search. Country results
 439         # usually are in the first batch of results and it is not possible
 440         # to exclude these fallbacks.
 441         if details.excluded:
 442             return nres.SearchResults()
 443
 444         t = conn.t.country_name
 445         tgrid = conn.t.country_grid
 446
 447         sql = sa.select(tgrid.c.country_code,
 448                         tgrid.c.geometry.ST_Centroid().ST_Collect().ST_Centroid()
 449                               .label('centroid'))\
 450                 .where(tgrid.c.country_code.in_(self.countries.values))\
 451                 .group_by(tgrid.c.country_code)
 452
 453         if details.viewbox is not None and details.bounded_viewbox:
 454             sql = sql.where(tgrid.c.geometry.intersects(VIEWBOX_PARAM))
 455         if details.near is not None and details.near_radius is not None:
 456             sql = sql.where(_within_near(tgrid))
 457
 458         sub = sql.subquery('grid')
 459
 460         sql = sa.select(t.c.country_code,
 461                         (t.c.name
 462                          + sa.func.coalesce(t.c.derived_name,
 463                                             sa.cast('', type_=conn.t.types.Composite))
 464                         ).label('name'),
 465                         sub.c.centroid)\
 466                 .join(sub, t.c.country_code == sub.c.country_code)
 467
 468         results = nres.SearchResults()
 469         for row in await conn.execute(sql, _details_to_bind_params(details)):
 470             result = nres.create_from_country_row(row, nres.SearchResult)
 471             assert result
 472             result.accuracy = self.penalty + self.countries.get_penalty(row.country_code, 5.0)
 473             results.append(result)
 474
 475         return results
 476
 477
 478
 479 class PostcodeSearch(AbstractSearch):
 480     """ Search for a postcode.
 481     """
 482     def __init__(self, extra_penalty: float, sdata: SearchData) -> None:
 483         super().__init__(sdata.penalty + extra_penalty)
 484         self.countries = sdata.countries
 485         self.postcodes = sdata.postcodes
 486         self.lookups = sdata.lookups
 487         self.rankings = sdata.rankings
 488
 489
 490     async def lookup(self, conn: SearchConnection,
 491                      details: SearchDetails) -> nres.SearchResults:
 492         """ Find results for the search in the database.
 493         """
 494         t = conn.t.postcode
 495         pcs = self.postcodes.values
 496
 497         sql: SaLambdaSelect = sa.lambda_stmt(lambda:
 498                 sa.select(t.c.place_id, t.c.parent_place_id,
 499                         t.c.rank_search, t.c.rank_address,
 500                         t.c.postcode, t.c.country_code,
 501                         t.c.geometry.label('centroid'))
 502                 .where(t.c.postcode.in_(pcs)))
 503
 504         if details.geometry_output:
 505             sql = _add_geometry_columns(sql, t.c.geometry, details)
 506
 507         penalty: SaExpression = sa.literal(self.penalty)
 508
 509         if details.viewbox is not None:
 510             if details.bounded_viewbox:
 511                 sql = sql.where(t.c.geometry.intersects(VIEWBOX_PARAM))
 512             else:
 513                 penalty += sa.case((t.c.geometry.intersects(VIEWBOX_PARAM), 0.0),
 514                                    (t.c.geometry.intersects(VIEWBOX2_PARAM), 1.0),
 515                                    else_=2.0)
 516
 517         if details.near is not None:
 518             if details.near_radius is not None:
 519                 sql = sql.where(_within_near(t))
 520             sql = sql.order_by(t.c.geometry.ST_Distance(NEAR_PARAM))
 521
 522         if self.countries:
 523             sql = sql.where(t.c.country_code.in_(self.countries.values))
 524
 525         if details.excluded:
 526             sql = sql.where(_exclude_places(t))
 527
 528         if self.lookups:
 529             assert len(self.lookups) == 1
 530             assert self.lookups[0].lookup_type == 'restrict'
 531             tsearch = conn.t.search_name
 532             sql = sql.where(tsearch.c.place_id == t.c.parent_place_id)\
 533                      .where(sa.func.array_cat(tsearch.c.name_vector,
 534                                               tsearch.c.nameaddress_vector,
 535                                               type_=ARRAY(sa.Integer))
 536                                     .contains(self.lookups[0].tokens))
 537
 538         for ranking in self.rankings:
 539             penalty += ranking.sql_penalty(conn.t.search_name)
 540         penalty += sa.case(*((t.c.postcode == v, p) for v, p in self.postcodes),
 541                        else_=1.0)
 542
 543
 544         sql = sql.add_columns(penalty.label('accuracy'))
 545         sql = sql.order_by('accuracy').limit(LIMIT_PARAM)
 546
 547         results = nres.SearchResults()
 548         for row in await conn.execute(sql, _details_to_bind_params(details)):
 549             result = nres.create_from_postcode_row(row, nres.SearchResult)
 550             assert result
 551             result.accuracy = row.accuracy
 552             results.append(result)
 553
 554         return results
 555
 556
 557
 558 class PlaceSearch(AbstractSearch):
 559     """ Generic search for an address or named place.
 560     """
 561     def __init__(self, extra_penalty: float, sdata: SearchData, expected_count: int) -> None:
 562         super().__init__(sdata.penalty + extra_penalty)
 563         self.countries = sdata.countries
 564         self.postcodes = sdata.postcodes
 565         self.housenumbers = sdata.housenumbers
 566         self.qualifiers = sdata.qualifiers
 567         self.lookups = sdata.lookups
 568         self.rankings = sdata.rankings
 569         self.expected_count = expected_count
 570
 571
 572     async def lookup(self, conn: SearchConnection,
 573                      details: SearchDetails) -> nres.SearchResults:
 574         """ Find results for the search in the database.
 575         """
 576         t = conn.t.placex
 577         tsearch = conn.t.search_name
 578
 579         sql: SaLambdaSelect = sa.lambda_stmt(lambda:
 580                   sa.select(t.c.place_id, t.c.osm_type, t.c.osm_id, t.c.name,
 581                             t.c.class_, t.c.type,
 582                             t.c.address, t.c.extratags,
 583                             t.c.housenumber, t.c.postcode, t.c.country_code,
 584                             t.c.wikipedia,
 585                             t.c.parent_place_id, t.c.rank_address, t.c.rank_search,
 586                             t.c.centroid,
 587                             t.c.geometry.ST_Expand(0).label('bbox'))
 588                    .where(t.c.place_id == tsearch.c.place_id))
 589
 590
 591         if details.geometry_output:
 592             sql = _add_geometry_columns(sql, t.c.geometry, details)
 593
 594         penalty: SaExpression = sa.literal(self.penalty)
 595         for ranking in self.rankings:
 596             penalty += ranking.sql_penalty(tsearch)
 597
 598         for lookup in self.lookups:
 599             sql = sql.where(lookup.sql_condition(tsearch))
 600
 601         if self.countries:
 602             sql = sql.where(tsearch.c.country_code.in_(self.countries.values))
 603
 604         if self.postcodes:
 605             # if a postcode is given, don't search for state or country level objects
 606             sql = sql.where(tsearch.c.address_rank > 9)
 607             tpc = conn.t.postcode
 608             pcs = self.postcodes.values
 609             if self.expected_count > 1000:
 610                 # Many results expected. Restrict by postcode.
 611                 sql = sql.where(lambda: sa.select(tpc.c.postcode)
 612                                   .where(tpc.c.postcode.in_(pcs))
 613                                   .where(tsearch.c.centroid.ST_DWithin(tpc.c.geometry, 0.12))
 614                                   .exists())
 615
 616             # Less results, only have a preference for close postcodes
 617             pc_near = sa.select(sa.func.min(tpc.c.geometry.ST_Distance(tsearch.c.centroid)))\
 618                       .where(tpc.c.postcode.in_(pcs))\
 619                       .scalar_subquery()
 620             penalty += sa.case((t.c.postcode.in_(pcs), 0.0),
 621                                else_=sa.func.coalesce(pc_near, 2.0))
 622
 623         if details.viewbox is not None:
 624             if details.bounded_viewbox:
 625                 sql = sql.where(tsearch.c.centroid.intersects(VIEWBOX_PARAM))
 626             else:
 627                 penalty += sa.case((t.c.geometry.intersects(VIEWBOX_PARAM), 0.0),
 628                                    (t.c.geometry.intersects(VIEWBOX2_PARAM), 1.0),
 629                                    else_=2.0)
 630
 631         if details.near is not None:
 632             if details.near_radius is not None:
 633                 sql = sql.where(tsearch.c.centroid.ST_DWithin(NEAR_PARAM, NEAR_RADIUS_PARAM))
 634             sql = sql.add_columns(-tsearch.c.centroid.ST_Distance(NEAR_PARAM)
 635                                       .label('importance'))
 636             sql = sql.order_by(sa.desc(sa.text('importance')))
 637         else:
 638             sql = sql.order_by(penalty - sa.case((tsearch.c.importance > 0, tsearch.c.importance),
 639                                   else_=0.75001-(sa.cast(tsearch.c.search_rank, sa.Float())/40)))
 640             sql = sql.add_columns(t.c.importance)
 641
 642
 643         sql = sql.add_columns(penalty.label('accuracy'))\
 644                  .order_by(sa.text('accuracy'))
 645
 646         if self.housenumbers:
 647             hnr_regexp = f"\\m({'|'.join(self.housenumbers.values)})\\M"
 648             sql = sql.where(tsearch.c.address_rank.between(16, 30))\
 649                      .where(sa.or_(tsearch.c.address_rank < 30,
 650                                    t.c.housenumber.op('~*')(hnr_regexp)))
 651
 652             # Cross check for housenumbers, need to do that on a rather large
 653             # set. Worst case there are 40.000 main streets in OSM.
 654             inner = sql.limit(10000).subquery()
 655
 656             # Housenumbers from placex
 657             thnr = conn.t.placex.alias('hnr')
 658             pid_list = array_agg(thnr.c.place_id) # type: ignore[no-untyped-call]
 659             place_sql = sa.select(pid_list)\
 660                           .where(thnr.c.parent_place_id == inner.c.place_id)\
 661                           .where(thnr.c.housenumber.op('~*')(hnr_regexp))\
 662                           .where(thnr.c.linked_place_id == None)\
 663                           .where(thnr.c.indexed_status == 0)
 664
 665             if details.excluded:
 666                 place_sql = place_sql.where(_exclude_places(thnr))
 667             if self.qualifiers:
 668                 place_sql = place_sql.where(self.qualifiers.sql_restrict(thnr))
 669
 670             numerals = [int(n) for n in self.housenumbers.values if n.isdigit()]
 671             interpol_sql: SaColumn
 672             tiger_sql: SaColumn
 673             if numerals and \
 674                (not self.qualifiers or ('place', 'house') in self.qualifiers.values):
 675                 # Housenumbers from interpolations
 676                 interpol_sql = _make_interpolation_subquery(conn.t.osmline, inner,
 677                                                             numerals, details)
 678                 # Housenumbers from Tiger
 679                 tiger_sql = sa.case((inner.c.country_code == 'us',
 680                                      _make_interpolation_subquery(conn.t.tiger, inner,
 681                                                                   numerals, details)
 682                                     ), else_=None)
 683             else:
 684                 interpol_sql = sa.null()
 685                 tiger_sql = sa.null()
 686
 687             unsort = sa.select(inner, place_sql.scalar_subquery().label('placex_hnr'),
 688                                interpol_sql.label('interpol_hnr'),
 689                                tiger_sql.label('tiger_hnr')).subquery('unsort')
 690             sql = sa.select(unsort)\
 691                     .order_by(sa.case((unsort.c.placex_hnr != None, 1),
 692                                       (unsort.c.interpol_hnr != None, 2),
 693                                       (unsort.c.tiger_hnr != None, 3),
 694                                       else_=4),
 695                               unsort.c.accuracy)
 696         else:
 697             sql = sql.where(t.c.linked_place_id == None)\
 698                      .where(t.c.indexed_status == 0)
 699             if self.qualifiers:
 700                 sql = sql.where(self.qualifiers.sql_restrict(t))
 701             if details.excluded:
 702                 sql = sql.where(_exclude_places(tsearch))
 703             if details.min_rank > 0:
 704                 sql = sql.where(sa.or_(tsearch.c.address_rank >= MIN_RANK_PARAM,
 705                                        tsearch.c.search_rank >= MIN_RANK_PARAM))
 706             if details.max_rank < 30:
 707                 sql = sql.where(sa.or_(tsearch.c.address_rank <= MAX_RANK_PARAM,
 708                                        tsearch.c.search_rank <= MAX_RANK_PARAM))
 709             if details.layers is not None:
 710                 sql = sql.where(_filter_by_layer(t, details.layers))
 711
 712         sql = sql.limit(LIMIT_PARAM)
 713
 714         results = nres.SearchResults()
 715         for row in await conn.execute(sql, _details_to_bind_params(details)):
 716             result = nres.create_from_placex_row(row, nres.SearchResult)
 717             assert result
 718             result.bbox = Bbox.from_wkb(row.bbox)
 719             result.accuracy = row.accuracy
 720             if not details.excluded or not result.place_id in details.excluded:
 721                 results.append(result)
 722
 723             if self.housenumbers and row.rank_address < 30:
 724                 if row.placex_hnr:
 725                     subs = _get_placex_housenumbers(conn, row.placex_hnr, details)
 726                 elif row.interpol_hnr:
 727                     subs = _get_osmline(conn, row.interpol_hnr, numerals, details)
 728                 elif row.tiger_hnr:
 729                     subs = _get_tiger(conn, row.tiger_hnr, numerals, row.osm_id, details)
 730                 else:
 731                     subs = None
 732
 733                 if subs is not None:
 734                     async for sub in subs:
 735                         assert sub.housenumber
 736                         sub.accuracy = result.accuracy
 737                         if not any(nr in self.housenumbers.values
 738                                    for nr in sub.housenumber.split(';')):
 739                             sub.accuracy += 0.6
 740                         results.append(sub)
 741
 742                 result.accuracy += 1.0 # penalty for missing housenumber
 743
 744         return results