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.linked_place_id, t.c.admin_level,
  65                      t.c.centroid,
  66                      t.c.geometry.ST_Expand(0).label('bbox'))
  67
  68
  69 def _add_geometry_columns(sql: SaLambdaSelect, col: SaColumn, details: SearchDetails) -> SaSelect:
  70     out = []
  71
  72     if details.geometry_simplification > 0.0:
  73         col = sa.func.ST_SimplifyPreserveTopology(col, details.geometry_simplification)
  74
  75     if details.geometry_output & GeometryFormat.GEOJSON:
  76         out.append(sa.func.ST_AsGeoJSON(col).label('geometry_geojson'))
  77     if details.geometry_output & GeometryFormat.TEXT:
  78         out.append(sa.func.ST_AsText(col).label('geometry_text'))
  79     if details.geometry_output & GeometryFormat.KML:
  80         out.append(sa.func.ST_AsKML(col).label('geometry_kml'))
  81     if details.geometry_output & GeometryFormat.SVG:
  82         out.append(sa.func.ST_AsSVG(col).label('geometry_svg'))
  83
  84     return sql.add_columns(*out)
  85
  86
  87 def _make_interpolation_subquery(table: SaFromClause, inner: SaFromClause,
  88                                  numerals: List[int], details: SearchDetails) -> SaScalarSelect:
  89     all_ids = array_agg(table.c.place_id) # type: ignore[no-untyped-call]
  90     sql = sa.select(all_ids).where(table.c.parent_place_id == inner.c.place_id)
  91
  92     if len(numerals) == 1:
  93         sql = sql.where(sa.between(numerals[0], table.c.startnumber, table.c.endnumber))\
  94                  .where((numerals[0] - table.c.startnumber) % table.c.step == 0)
  95     else:
  96         sql = sql.where(sa.or_(
  97                 *(sa.and_(sa.between(n, table.c.startnumber, table.c.endnumber),
  98                           (n - table.c.startnumber) % table.c.step == 0)
  99                   for n in numerals)))
 100
 101     if details.excluded:
 102         sql = sql.where(_exclude_places(table))
 103
 104     return sql.scalar_subquery()
 105
 106
 107 def _filter_by_layer(table: SaFromClause, layers: DataLayer) -> SaColumn:
 108     orexpr: List[SaExpression] = []
 109     if layers & DataLayer.ADDRESS and layers & DataLayer.POI:
 110         orexpr.append(table.c.rank_address.between(1, 30))
 111     elif layers & DataLayer.ADDRESS:
 112         orexpr.append(table.c.rank_address.between(1, 29))
 113         orexpr.append(sa.and_(table.c.rank_address == 30,
 114                               sa.or_(table.c.housenumber != None,
 115                                      table.c.address.has_key('addr:housename'))))
 116     elif layers & DataLayer.POI:
 117         orexpr.append(sa.and_(table.c.rank_address == 30,
 118                               table.c.class_.not_in(('place', 'building'))))
 119
 120     if layers & DataLayer.MANMADE:
 121         exclude = []
 122         if not layers & DataLayer.RAILWAY:
 123             exclude.append('railway')
 124         if not layers & DataLayer.NATURAL:
 125             exclude.extend(('natural', 'water', 'waterway'))
 126         orexpr.append(sa.and_(table.c.class_.not_in(tuple(exclude)),
 127                               table.c.rank_address == 0))
 128     else:
 129         include = []
 130         if layers & DataLayer.RAILWAY:
 131             include.append('railway')
 132         if layers & DataLayer.NATURAL:
 133             include.extend(('natural', 'water', 'waterway'))
 134         orexpr.append(sa.and_(table.c.class_.in_(tuple(include)),
 135                               table.c.rank_address == 0))
 136
 137     if len(orexpr) == 1:
 138         return orexpr[0]
 139
 140     return sa.or_(*orexpr)
 141
 142
 143 def _interpolated_position(table: SaFromClause, nr: SaColumn) -> SaColumn:
 144     pos = sa.cast(nr - table.c.startnumber, sa.Float) / (table.c.endnumber - table.c.startnumber)
 145     return sa.case(
 146             (table.c.endnumber == table.c.startnumber, table.c.linegeo.ST_Centroid()),
 147             else_=table.c.linegeo.ST_LineInterpolatePoint(pos)).label('centroid')
 148
 149
 150 async def _get_placex_housenumbers(conn: SearchConnection,
 151                                    place_ids: List[int],
 152                                    details: SearchDetails) -> AsyncIterator[nres.SearchResult]:
 153     t = conn.t.placex
 154     sql = _select_placex(t).where(t.c.place_id.in_(place_ids))
 155
 156     if details.geometry_output:
 157         sql = _add_geometry_columns(sql, t.c.geometry, details)
 158
 159     for row in await conn.execute(sql):
 160         result = nres.create_from_placex_row(row, nres.SearchResult)
 161         assert result
 162         result.bbox = Bbox.from_wkb(row.bbox)
 163         yield result
 164
 165
 166 async def _get_osmline(conn: SearchConnection, place_ids: List[int],
 167                        numerals: List[int],
 168                        details: SearchDetails) -> AsyncIterator[nres.SearchResult]:
 169     t = conn.t.osmline
 170     values = sa.values(sa.Column('nr', sa.Integer()), name='housenumber')\
 171                .data([(n,) for n in numerals])
 172     sql = sa.select(t.c.place_id, t.c.osm_id,
 173                     t.c.parent_place_id, t.c.address,
 174                     values.c.nr.label('housenumber'),
 175                     _interpolated_position(t, values.c.nr),
 176                     t.c.postcode, t.c.country_code)\
 177             .where(t.c.place_id.in_(place_ids))\
 178             .join(values, values.c.nr.between(t.c.startnumber, t.c.endnumber))
 179
 180     if details.geometry_output:
 181         sub = sql.subquery()
 182         sql = _add_geometry_columns(sa.select(sub), sub.c.centroid, details)
 183
 184     for row in await conn.execute(sql):
 185         result = nres.create_from_osmline_row(row, nres.SearchResult)
 186         assert result
 187         yield result
 188
 189
 190 async def _get_tiger(conn: SearchConnection, place_ids: List[int],
 191                      numerals: List[int], osm_id: int,
 192                      details: SearchDetails) -> AsyncIterator[nres.SearchResult]:
 193     t = conn.t.tiger
 194     values = sa.values(sa.Column('nr', sa.Integer()), name='housenumber')\
 195                .data([(n,) for n in numerals])
 196     sql = sa.select(t.c.place_id, t.c.parent_place_id,
 197                     sa.literal('W').label('osm_type'),
 198                     sa.literal(osm_id).label('osm_id'),
 199                     values.c.nr.label('housenumber'),
 200                     _interpolated_position(t, values.c.nr),
 201                     t.c.postcode)\
 202             .where(t.c.place_id.in_(place_ids))\
 203             .join(values, values.c.nr.between(t.c.startnumber, t.c.endnumber))
 204
 205     if details.geometry_output:
 206         sub = sql.subquery()
 207         sql = _add_geometry_columns(sa.select(sub), sub.c.centroid, details)
 208
 209     for row in await conn.execute(sql):
 210         result = nres.create_from_tiger_row(row, nres.SearchResult)
 211         assert result
 212         yield result
 213
 214
 215 class AbstractSearch(abc.ABC):
 216     """ Encapuslation of a single lookup in the database.
 217     """
 218
 219     def __init__(self, penalty: float) -> None:
 220         self.penalty = penalty
 221
 222     @abc.abstractmethod
 223     async def lookup(self, conn: SearchConnection,
 224                      details: SearchDetails) -> nres.SearchResults:
 225         """ Find results for the search in the database.
 226         """
 227
 228
 229 class NearSearch(AbstractSearch):
 230     """ Category search of a place type near the result of another search.
 231     """
 232     def __init__(self, penalty: float, categories: WeightedCategories,
 233                  search: AbstractSearch) -> None:
 234         super().__init__(penalty)
 235         self.search = search
 236         self.categories = categories
 237
 238
 239     async def lookup(self, conn: SearchConnection,
 240                      details: SearchDetails) -> nres.SearchResults:
 241         """ Find results for the search in the database.
 242         """
 243         results = nres.SearchResults()
 244         base = await self.search.lookup(conn, details)
 245
 246         if not base:
 247             return results
 248
 249         base.sort(key=lambda r: (r.accuracy, r.rank_search))
 250         max_accuracy = base[0].accuracy + 0.5
 251         base = nres.SearchResults(r for r in base if r.source_table == nres.SourceTable.PLACEX
 252                                                      and r.accuracy <= max_accuracy
 253                                                      and r.bbox and r.bbox.area < 20)
 254
 255         if base:
 256             baseids = [b.place_id for b in base[:5] if b.place_id]
 257
 258             for category, penalty in self.categories:
 259                 await self.lookup_category(results, conn, baseids, category, penalty, details)
 260                 if len(results) >= details.max_results:
 261                     break
 262
 263         return results
 264
 265
 266     async def lookup_category(self, results: nres.SearchResults,
 267                               conn: SearchConnection, ids: List[int],
 268                               category: Tuple[str, str], penalty: float,
 269                               details: SearchDetails) -> None:
 270         """ Find places of the given category near the list of
 271             place ids and add the results to 'results'.
 272         """
 273         table = await conn.get_class_table(*category)
 274
 275         t = conn.t.placex
 276         tgeom = conn.t.placex.alias('pgeom')
 277
 278         sql = _select_placex(t).where(tgeom.c.place_id.in_(ids))\
 279                                .where(t.c.class_ == category[0])\
 280                                .where(t.c.type == category[1])
 281
 282         if table is None:
 283             # No classtype table available, do a simplified lookup in placex.
 284             sql = sql.join(tgeom, t.c.geometry.ST_DWithin(tgeom.c.centroid, 0.01))\
 285                      .order_by(tgeom.c.centroid.ST_Distance(t.c.centroid))
 286         else:
 287             # Use classtype table. We can afford to use a larger
 288             # radius for the lookup.
 289             sql = sql.join(table, t.c.place_id == table.c.place_id)\
 290                      .join(tgeom,
 291                            table.c.centroid.ST_CoveredBy(
 292                                sa.case((sa.and_(tgeom.c.rank_address < 9,
 293                                                 tgeom.c.geometry.is_area()),
 294                                         tgeom.c.geometry),
 295                                        else_ = tgeom.c.centroid.ST_Expand(0.05))))\
 296                      .order_by(tgeom.c.centroid.ST_Distance(table.c.centroid))
 297
 298         sql = sql.where(t.c.rank_address.between(MIN_RANK_PARAM, MAX_RANK_PARAM))
 299         if details.countries:
 300             sql = sql.where(t.c.country_code.in_(COUNTRIES_PARAM))
 301         if details.excluded:
 302             sql = sql.where(_exclude_places(t))
 303         if details.layers is not None:
 304             sql = sql.where(_filter_by_layer(t, details.layers))
 305
 306         sql = sql.limit(LIMIT_PARAM)
 307         for row in await conn.execute(sql, _details_to_bind_params(details)):
 308             result = nres.create_from_placex_row(row, nres.SearchResult)
 309             assert result
 310             result.accuracy = self.penalty + penalty
 311             result.bbox = Bbox.from_wkb(row.bbox)
 312             results.append(result)
 313
 314
 315
 316 class PoiSearch(AbstractSearch):
 317     """ Category search in a geographic area.
 318     """
 319     def __init__(self, sdata: SearchData) -> None:
 320         super().__init__(sdata.penalty)
 321         self.qualifiers = sdata.qualifiers
 322         self.countries = sdata.countries
 323
 324
 325     async def lookup(self, conn: SearchConnection,
 326                      details: SearchDetails) -> nres.SearchResults:
 327         """ Find results for the search in the database.
 328         """
 329         bind_params = _details_to_bind_params(details)
 330         t = conn.t.placex
 331
 332         rows: List[SaRow] = []
 333
 334         if details.near and details.near_radius is not None and details.near_radius < 0.2:
 335             # simply search in placex table
 336             def _base_query() -> SaSelect:
 337                 return _select_placex(t) \
 338                            .where(t.c.linked_place_id == None) \
 339                            .where(t.c.geometry.ST_DWithin(NEAR_PARAM, NEAR_RADIUS_PARAM)) \
 340                            .order_by(t.c.centroid.ST_Distance(NEAR_PARAM)) \
 341                            .limit(LIMIT_PARAM)
 342
 343             classtype = self.qualifiers.values
 344             if len(classtype) == 1:
 345                 cclass, ctype = classtype[0]
 346                 sql: SaLambdaSelect = sa.lambda_stmt(lambda: _base_query()
 347                                                  .where(t.c.class_ == cclass)
 348                                                  .where(t.c.type == ctype))
 349             else:
 350                 sql = _base_query().where(sa.or_(*(sa.and_(t.c.class_ == cls, t.c.type == typ)
 351                                                    for cls, typ in classtype)))
 352
 353             if self.countries:
 354                 sql = sql.where(t.c.country_code.in_(self.countries.values))
 355
 356             if details.viewbox is not None and details.bounded_viewbox:
 357                 sql = sql.where(t.c.geometry.intersects(VIEWBOX_PARAM))
 358
 359             rows.extend(await conn.execute(sql, bind_params))
 360         else:
 361             # use the class type tables
 362             for category in self.qualifiers.values:
 363                 table = await conn.get_class_table(*category)
 364                 if table is not None:
 365                     sql = _select_placex(t)\
 366                                .join(table, t.c.place_id == table.c.place_id)\
 367                                .where(t.c.class_ == category[0])\
 368                                .where(t.c.type == category[1])
 369
 370                     if details.viewbox is not None and details.bounded_viewbox:
 371                         sql = sql.where(table.c.centroid.intersects(VIEWBOX_PARAM))
 372
 373                     if details.near and details.near_radius is not None:
 374                         sql = sql.order_by(table.c.centroid.ST_Distance(NEAR_PARAM))\
 375                                  .where(table.c.centroid.ST_DWithin(NEAR_PARAM,
 376                                                                     NEAR_RADIUS_PARAM))
 377
 378                     if self.countries:
 379                         sql = sql.where(t.c.country_code.in_(self.countries.values))
 380
 381                     sql = sql.limit(LIMIT_PARAM)
 382                     rows.extend(await conn.execute(sql, bind_params))
 383
 384         results = nres.SearchResults()
 385         for row in rows:
 386             result = nres.create_from_placex_row(row, nres.SearchResult)
 387             assert result
 388             result.accuracy = self.penalty + self.qualifiers.get_penalty((row.class_, row.type))
 389             result.bbox = Bbox.from_wkb(row.bbox)
 390             results.append(result)
 391
 392         return results
 393
 394
 395 class CountrySearch(AbstractSearch):
 396     """ Search for a country name or country code.
 397     """
 398     def __init__(self, sdata: SearchData) -> None:
 399         super().__init__(sdata.penalty)
 400         self.countries = sdata.countries
 401
 402
 403     async def lookup(self, conn: SearchConnection,
 404                      details: SearchDetails) -> nres.SearchResults:
 405         """ Find results for the search in the database.
 406         """
 407         t = conn.t.placex
 408
 409         ccodes = self.countries.values
 410         sql = _select_placex(t)\
 411                 .where(t.c.country_code.in_(ccodes))\
 412                 .where(t.c.rank_address == 4)
 413
 414         if details.geometry_output:
 415             sql = _add_geometry_columns(sql, t.c.geometry, details)
 416
 417         if details.excluded:
 418             sql = sql.where(_exclude_places(t))
 419
 420         if details.viewbox is not None and details.bounded_viewbox:
 421             sql = sql.where(lambda: t.c.geometry.intersects(VIEWBOX_PARAM))
 422
 423         if details.near is not None and details.near_radius is not None:
 424             sql = sql.where(_within_near(t))
 425
 426         results = nres.SearchResults()
 427         for row in await conn.execute(sql, _details_to_bind_params(details)):
 428             result = nres.create_from_placex_row(row, nres.SearchResult)
 429             assert result
 430             result.accuracy = self.penalty + self.countries.get_penalty(row.country_code, 5.0)
 431             result.bbox = Bbox.from_wkb(row.bbox)
 432             results.append(result)
 433
 434         return results or await self.lookup_in_country_table(conn, details)
 435
 436
 437     async def lookup_in_country_table(self, conn: SearchConnection,
 438                                       details: SearchDetails) -> nres.SearchResults:
 439         """ Look up the country in the fallback country tables.
 440         """
 441         # Avoid the fallback search when this is a more search. Country results
 442         # usually are in the first batch of results and it is not possible
 443         # to exclude these fallbacks.
 444         if details.excluded:
 445             return nres.SearchResults()
 446
 447         t = conn.t.country_name
 448         tgrid = conn.t.country_grid
 449
 450         sql = sa.select(tgrid.c.country_code,
 451                         tgrid.c.geometry.ST_Centroid().ST_Collect().ST_Centroid()
 452                               .label('centroid'),
 453                         tgrid.c.geometry.ST_Collect().ST_Expand(0).label('bbox'))\
 454                 .where(tgrid.c.country_code.in_(self.countries.values))\
 455                 .group_by(tgrid.c.country_code)
 456
 457         if details.viewbox is not None and details.bounded_viewbox:
 458             sql = sql.where(tgrid.c.geometry.intersects(VIEWBOX_PARAM))
 459         if details.near is not None and details.near_radius is not None:
 460             sql = sql.where(_within_near(tgrid))
 461
 462         sub = sql.subquery('grid')
 463
 464         sql = sa.select(t.c.country_code,
 465                         (t.c.name
 466                          + sa.func.coalesce(t.c.derived_name,
 467                                             sa.cast('', type_=conn.t.types.Composite))
 468                         ).label('name'),
 469                         sub.c.centroid, sub.c.bbox)\
 470                 .join(sub, t.c.country_code == sub.c.country_code)
 471
 472         if details.geometry_output:
 473             sql = _add_geometry_columns(sql, sub.c.centroid, details)
 474
 475         results = nres.SearchResults()
 476         for row in await conn.execute(sql, _details_to_bind_params(details)):
 477             result = nres.create_from_country_row(row, nres.SearchResult)
 478             assert result
 479             result.bbox = Bbox.from_wkb(row.bbox)
 480             result.accuracy = self.penalty + self.countries.get_penalty(row.country_code, 5.0)
 481             results.append(result)
 482
 483         return results
 484
 485
 486
 487 class PostcodeSearch(AbstractSearch):
 488     """ Search for a postcode.
 489     """
 490     def __init__(self, extra_penalty: float, sdata: SearchData) -> None:
 491         super().__init__(sdata.penalty + extra_penalty)
 492         self.countries = sdata.countries
 493         self.postcodes = sdata.postcodes
 494         self.lookups = sdata.lookups
 495         self.rankings = sdata.rankings
 496
 497
 498     async def lookup(self, conn: SearchConnection,
 499                      details: SearchDetails) -> nres.SearchResults:
 500         """ Find results for the search in the database.
 501         """
 502         t = conn.t.postcode
 503         pcs = self.postcodes.values
 504
 505         sql = sa.select(t.c.place_id, t.c.parent_place_id,
 506                         t.c.rank_search, t.c.rank_address,
 507                         t.c.postcode, t.c.country_code,
 508                         t.c.geometry.label('centroid'))\
 509                 .where(t.c.postcode.in_(pcs))
 510
 511         if details.geometry_output:
 512             sql = _add_geometry_columns(sql, t.c.geometry, details)
 513
 514         penalty: SaExpression = sa.literal(self.penalty)
 515
 516         if details.viewbox is not None:
 517             if details.bounded_viewbox:
 518                 sql = sql.where(t.c.geometry.intersects(VIEWBOX_PARAM))
 519             else:
 520                 penalty += sa.case((t.c.geometry.intersects(VIEWBOX_PARAM), 0.0),
 521                                    (t.c.geometry.intersects(VIEWBOX2_PARAM), 0.5),
 522                                    else_=1.0)
 523
 524         if details.near is not None:
 525             if details.near_radius is not None:
 526                 sql = sql.where(_within_near(t))
 527             sql = sql.order_by(t.c.geometry.ST_Distance(NEAR_PARAM))
 528
 529         if self.countries:
 530             sql = sql.where(t.c.country_code.in_(self.countries.values))
 531
 532         if details.excluded:
 533             sql = sql.where(_exclude_places(t))
 534
 535         if self.lookups:
 536             assert len(self.lookups) == 1
 537             assert self.lookups[0].lookup_type == 'restrict'
 538             tsearch = conn.t.search_name
 539             sql = sql.where(tsearch.c.place_id == t.c.parent_place_id)\
 540                      .where(sa.func.array_cat(tsearch.c.name_vector,
 541                                               tsearch.c.nameaddress_vector,
 542                                               type_=ARRAY(sa.Integer))
 543                                     .contains(self.lookups[0].tokens))
 544
 545         for ranking in self.rankings:
 546             penalty += ranking.sql_penalty(conn.t.search_name)
 547         penalty += sa.case(*((t.c.postcode == v, p) for v, p in self.postcodes),
 548                        else_=1.0)
 549
 550
 551         sql = sql.add_columns(penalty.label('accuracy'))
 552         sql = sql.order_by('accuracy').limit(LIMIT_PARAM)
 553
 554         results = nres.SearchResults()
 555         for row in await conn.execute(sql, _details_to_bind_params(details)):
 556             result = nres.create_from_postcode_row(row, nres.SearchResult)
 557             assert result
 558             result.accuracy = row.accuracy
 559             results.append(result)
 560
 561         return results
 562
 563
 564
 565 class PlaceSearch(AbstractSearch):
 566     """ Generic search for an address or named place.
 567     """
 568     def __init__(self, extra_penalty: float, sdata: SearchData, expected_count: int) -> None:
 569         super().__init__(sdata.penalty + extra_penalty)
 570         self.countries = sdata.countries
 571         self.postcodes = sdata.postcodes
 572         self.housenumbers = sdata.housenumbers
 573         self.qualifiers = sdata.qualifiers
 574         self.lookups = sdata.lookups
 575         self.rankings = sdata.rankings
 576         self.expected_count = expected_count
 577
 578
 579     async def lookup(self, conn: SearchConnection,
 580                      details: SearchDetails) -> nres.SearchResults:
 581         """ Find results for the search in the database.
 582         """
 583         t = conn.t.placex
 584         tsearch = conn.t.search_name
 585
 586         sql: SaLambdaSelect = sa.lambda_stmt(lambda:
 587                   sa.select(t.c.place_id, t.c.osm_type, t.c.osm_id, t.c.name,
 588                             t.c.class_, t.c.type,
 589                             t.c.address, t.c.extratags, t.c.admin_level,
 590                             t.c.housenumber, t.c.postcode, t.c.country_code,
 591                             t.c.wikipedia,
 592                             t.c.parent_place_id, t.c.rank_address, t.c.rank_search,
 593                             t.c.centroid,
 594                             t.c.geometry.ST_Expand(0).label('bbox'))
 595                    .where(t.c.place_id == tsearch.c.place_id))
 596
 597
 598         if details.geometry_output:
 599             sql = _add_geometry_columns(sql, t.c.geometry, details)
 600
 601         penalty: SaExpression = sa.literal(self.penalty)
 602         for ranking in self.rankings:
 603             penalty += ranking.sql_penalty(tsearch)
 604
 605         for lookup in self.lookups:
 606             sql = sql.where(lookup.sql_condition(tsearch))
 607
 608         if self.countries:
 609             sql = sql.where(tsearch.c.country_code.in_(self.countries.values))
 610
 611         if self.postcodes:
 612             # if a postcode is given, don't search for state or country level objects
 613             sql = sql.where(tsearch.c.address_rank > 9)
 614             tpc = conn.t.postcode
 615             pcs = self.postcodes.values
 616             if self.expected_count > 1000:
 617                 # Many results expected. Restrict by postcode.
 618                 sql = sql.where(sa.select(tpc.c.postcode)
 619                                   .where(tpc.c.postcode.in_(pcs))
 620                                   .where(tsearch.c.centroid.ST_DWithin(tpc.c.geometry, 0.12))
 621                                   .exists())
 622
 623             # Less results, only have a preference for close postcodes
 624             pc_near = sa.select(sa.func.min(tpc.c.geometry.ST_Distance(tsearch.c.centroid)))\
 625                       .where(tpc.c.postcode.in_(pcs))\
 626                       .scalar_subquery()
 627             penalty += sa.case((t.c.postcode.in_(pcs), 0.0),
 628                                else_=sa.func.coalesce(pc_near, 2.0))
 629
 630         if details.viewbox is not None:
 631             if details.bounded_viewbox:
 632                 if details.viewbox.area < 0.2:
 633                     sql = sql.where(tsearch.c.centroid.intersects(VIEWBOX_PARAM))
 634                 else:
 635                     sql = sql.where(tsearch.c.centroid.ST_Intersects_no_index(VIEWBOX_PARAM))
 636             elif self.expected_count >= 10000:
 637                 if details.viewbox.area < 0.5:
 638                     sql = sql.where(tsearch.c.centroid.intersects(VIEWBOX2_PARAM))
 639                 else:
 640                     sql = sql.where(tsearch.c.centroid.ST_Intersects_no_index(VIEWBOX2_PARAM))
 641             else:
 642                 penalty += sa.case((t.c.geometry.intersects(VIEWBOX_PARAM), 0.0),
 643                                    (t.c.geometry.intersects(VIEWBOX2_PARAM), 0.5),
 644                                    else_=1.0)
 645
 646         if details.near is not None:
 647             if details.near_radius is not None:
 648                 if details.near_radius < 0.1:
 649                     sql = sql.where(tsearch.c.centroid.ST_DWithin(NEAR_PARAM, NEAR_RADIUS_PARAM))
 650                 else:
 651                     sql = sql.where(tsearch.c.centroid.ST_DWithin_no_index(NEAR_PARAM,
 652                                                                            NEAR_RADIUS_PARAM))
 653             sql = sql.add_columns((-tsearch.c.centroid.ST_Distance(NEAR_PARAM))
 654                                       .label('importance'))
 655             sql = sql.order_by(sa.desc(sa.text('importance')))
 656         else:
 657             if self.expected_count < 10000\
 658                or (details.viewbox is not None and details.viewbox.area < 0.5):
 659                 sql = sql.order_by(
 660                         penalty - sa.case((tsearch.c.importance > 0, tsearch.c.importance),
 661                                     else_=0.75001-(sa.cast(tsearch.c.search_rank, sa.Float())/40)))
 662             sql = sql.add_columns(t.c.importance)
 663
 664
 665         sql = sql.add_columns(penalty.label('accuracy'))
 666
 667         if self.expected_count < 10000:
 668             sql = sql.order_by(sa.text('accuracy'))
 669
 670         if self.housenumbers:
 671             hnr_regexp = f"\\m({'|'.join(self.housenumbers.values)})\\M"
 672             sql = sql.where(tsearch.c.address_rank.between(16, 30))\
 673                      .where(sa.or_(tsearch.c.address_rank < 30,
 674                                    t.c.housenumber.op('~*')(hnr_regexp)))
 675
 676             # Cross check for housenumbers, need to do that on a rather large
 677             # set. Worst case there are 40.000 main streets in OSM.
 678             inner = sql.limit(10000).subquery()
 679
 680             # Housenumbers from placex
 681             thnr = conn.t.placex.alias('hnr')
 682             pid_list = array_agg(thnr.c.place_id) # type: ignore[no-untyped-call]
 683             place_sql = sa.select(pid_list)\
 684                           .where(thnr.c.parent_place_id == inner.c.place_id)\
 685                           .where(thnr.c.housenumber.op('~*')(hnr_regexp))\
 686                           .where(thnr.c.linked_place_id == None)\
 687                           .where(thnr.c.indexed_status == 0)
 688
 689             if details.excluded:
 690                 place_sql = place_sql.where(thnr.c.place_id.not_in(sa.bindparam('excluded')))
 691             if self.qualifiers:
 692                 place_sql = place_sql.where(self.qualifiers.sql_restrict(thnr))
 693
 694             numerals = [int(n) for n in self.housenumbers.values
 695                         if n.isdigit() and len(n) < 8]
 696             interpol_sql: SaColumn
 697             tiger_sql: SaColumn
 698             if numerals and \
 699                (not self.qualifiers or ('place', 'house') in self.qualifiers.values):
 700                 # Housenumbers from interpolations
 701                 interpol_sql = _make_interpolation_subquery(conn.t.osmline, inner,
 702                                                             numerals, details)
 703                 # Housenumbers from Tiger
 704                 tiger_sql = sa.case((inner.c.country_code == 'us',
 705                                      _make_interpolation_subquery(conn.t.tiger, inner,
 706                                                                   numerals, details)
 707                                     ), else_=None)
 708             else:
 709                 interpol_sql = sa.null()
 710                 tiger_sql = sa.null()
 711
 712             unsort = sa.select(inner, place_sql.scalar_subquery().label('placex_hnr'),
 713                                interpol_sql.label('interpol_hnr'),
 714                                tiger_sql.label('tiger_hnr')).subquery('unsort')
 715             sql = sa.select(unsort)\
 716                     .order_by(sa.case((unsort.c.placex_hnr != None, 1),
 717                                       (unsort.c.interpol_hnr != None, 2),
 718                                       (unsort.c.tiger_hnr != None, 3),
 719                                       else_=4),
 720                               unsort.c.accuracy)
 721         else:
 722             sql = sql.where(t.c.linked_place_id == None)\
 723                      .where(t.c.indexed_status == 0)
 724             if self.qualifiers:
 725                 sql = sql.where(self.qualifiers.sql_restrict(t))
 726             if details.excluded:
 727                 sql = sql.where(_exclude_places(tsearch))
 728             if details.min_rank > 0:
 729                 sql = sql.where(sa.or_(tsearch.c.address_rank >= MIN_RANK_PARAM,
 730                                        tsearch.c.search_rank >= MIN_RANK_PARAM))
 731             if details.max_rank < 30:
 732                 sql = sql.where(sa.or_(tsearch.c.address_rank <= MAX_RANK_PARAM,
 733                                        tsearch.c.search_rank <= MAX_RANK_PARAM))
 734             if details.layers is not None:
 735                 sql = sql.where(_filter_by_layer(t, details.layers))
 736
 737         sql = sql.limit(LIMIT_PARAM)
 738
 739         results = nres.SearchResults()
 740         for row in await conn.execute(sql, _details_to_bind_params(details)):
 741             result = nres.create_from_placex_row(row, nres.SearchResult)
 742             assert result
 743             result.bbox = Bbox.from_wkb(row.bbox)
 744             result.accuracy = row.accuracy
 745             if not details.excluded or not result.place_id in details.excluded:
 746                 results.append(result)
 747
 748             if self.housenumbers and row.rank_address < 30:
 749                 if row.placex_hnr:
 750                     subs = _get_placex_housenumbers(conn, row.placex_hnr, details)
 751                 elif row.interpol_hnr:
 752                     subs = _get_osmline(conn, row.interpol_hnr, numerals, details)
 753                 elif row.tiger_hnr:
 754                     subs = _get_tiger(conn, row.tiger_hnr, numerals, row.osm_id, details)
 755                 else:
 756                     subs = None
 757
 758                 if subs is not None:
 759                     async for sub in subs:
 760                         assert sub.housenumber
 761                         sub.accuracy = result.accuracy
 762                         if not any(nr in self.housenumbers.values
 763                                    for nr in sub.housenumber.split(';')):
 764                             sub.accuracy += 0.6
 765                         results.append(sub)
 766
 767                 result.accuracy += 1.0 # penalty for missing housenumber
 768
 769         return results