src/nominatim_api/lookup.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) 2024 by the Nominatim developer community.
   6 # For a full list of authors see the git log.
   7 """
   8 Implementation of place lookup by ID.
   9 """
  10 from typing import Optional, Callable, Tuple, Type
  11 import datetime as dt
  12
  13 import sqlalchemy as sa
  14
  15 from .typing import SaColumn, SaRow, SaSelect
  16 from .connection import SearchConnection
  17 from .logging import log
  18 from . import types as ntyp
  19 from . import results as nres
  20
  21 RowFunc = Callable[[Optional[SaRow], Type[nres.BaseResultT]], Optional[nres.BaseResultT]]
  22
  23 GeomFunc = Callable[[SaSelect, SaColumn], SaSelect]
  24
  25
  26 async def find_in_placex(conn: SearchConnection, place: ntyp.PlaceRef,
  27                          add_geometries: GeomFunc) -> Optional[SaRow]:
  28     """ Search for the given place in the placex table and return the
  29         base information.
  30     """
  31     log().section("Find in placex table")
  32     t = conn.t.placex
  33     sql = sa.select(t.c.place_id, t.c.osm_type, t.c.osm_id, t.c.name,
  34                     t.c.class_, t.c.type, t.c.admin_level,
  35                     t.c.address, t.c.extratags,
  36                     t.c.housenumber, t.c.postcode, t.c.country_code,
  37                     t.c.importance, t.c.wikipedia, t.c.indexed_date,
  38                     t.c.parent_place_id, t.c.rank_address, t.c.rank_search,
  39                     t.c.linked_place_id,
  40                     t.c.geometry.ST_Expand(0).label('bbox'),
  41                     t.c.centroid)
  42
  43     if isinstance(place, ntyp.PlaceID):
  44         sql = sql.where(t.c.place_id == place.place_id)
  45     elif isinstance(place, ntyp.OsmID):
  46         sql = sql.where(t.c.osm_type == place.osm_type)\
  47                  .where(t.c.osm_id == place.osm_id)
  48         if place.osm_class:
  49             sql = sql.where(t.c.class_ == place.osm_class)
  50         else:
  51             sql = sql.order_by(t.c.class_)
  52         sql = sql.limit(1)
  53     else:
  54         return None
  55
  56     return (await conn.execute(add_geometries(sql, t.c.geometry))).one_or_none()
  57
  58
  59 async def find_in_osmline(conn: SearchConnection, place: ntyp.PlaceRef,
  60                           add_geometries: GeomFunc) -> Optional[SaRow]:
  61     """ Search for the given place in the osmline table and return the
  62         base information.
  63     """
  64     log().section("Find in interpolation table")
  65     t = conn.t.osmline
  66     sql = sa.select(t.c.place_id, t.c.osm_id, t.c.parent_place_id,
  67                     t.c.indexed_date, t.c.startnumber, t.c.endnumber,
  68                     t.c.step, t.c.address, t.c.postcode, t.c.country_code,
  69                     t.c.linegeo.ST_Centroid().label('centroid'))
  70
  71     if isinstance(place, ntyp.PlaceID):
  72         sql = sql.where(t.c.place_id == place.place_id)
  73     elif isinstance(place, ntyp.OsmID) and place.osm_type == 'W':
  74         # There may be multiple interpolations for a single way.
  75         # If 'class' contains a number, return the one that belongs to that number.
  76         sql = sql.where(t.c.osm_id == place.osm_id).limit(1)
  77         if place.osm_class and place.osm_class.isdigit():
  78             sql = sql.order_by(sa.func.greatest(0,
  79                                                 int(place.osm_class) - t.c.endnumber,
  80                                                 t.c.startnumber - int(place.osm_class)))
  81     else:
  82         return None
  83
  84     return (await conn.execute(add_geometries(sql, t.c.linegeo))).one_or_none()
  85
  86
  87 async def find_in_tiger(conn: SearchConnection, place: ntyp.PlaceRef,
  88                         add_geometries: GeomFunc) -> Optional[SaRow]:
  89     """ Search for the given place in the table of Tiger addresses and return
  90         the base information. Only lookup by place ID is supported.
  91     """
  92     if not isinstance(place, ntyp.PlaceID):
  93         return None
  94
  95     log().section("Find in TIGER table")
  96     t = conn.t.tiger
  97     parent = conn.t.placex
  98     sql = sa.select(t.c.place_id, t.c.parent_place_id,
  99                     parent.c.osm_type, parent.c.osm_id,
 100                     t.c.startnumber, t.c.endnumber, t.c.step,
 101                     t.c.postcode,
 102                     t.c.linegeo.ST_Centroid().label('centroid'))\
 103             .where(t.c.place_id == place.place_id)\
 104             .join(parent, t.c.parent_place_id == parent.c.place_id, isouter=True)
 105
 106     return (await conn.execute(add_geometries(sql, t.c.linegeo))).one_or_none()
 107
 108
 109 async def find_in_postcode(conn: SearchConnection, place: ntyp.PlaceRef,
 110                            add_geometries: GeomFunc) -> Optional[SaRow]:
 111     """ Search for the given place in the postcode table and return the
 112         base information. Only lookup by place ID is supported.
 113     """
 114     if not isinstance(place, ntyp.PlaceID):
 115         return None
 116
 117     log().section("Find in postcode table")
 118     t = conn.t.postcode
 119     sql = sa.select(t.c.place_id, t.c.parent_place_id,
 120                     t.c.rank_search, t.c.rank_address,
 121                     t.c.indexed_date, t.c.postcode, t.c.country_code,
 122                     t.c.geometry.label('centroid')) \
 123             .where(t.c.place_id == place.place_id)
 124
 125     return (await conn.execute(add_geometries(sql, t.c.geometry))).one_or_none()
 126
 127
 128 async def find_in_all_tables(conn: SearchConnection, place: ntyp.PlaceRef,
 129                              add_geometries: GeomFunc
 130                              ) -> Tuple[Optional[SaRow], RowFunc[nres.BaseResultT]]:
 131     """ Search for the given place in all data tables
 132         and return the base information.
 133     """
 134     row = await find_in_placex(conn, place, add_geometries)
 135     log().var_dump('Result (placex)', row)
 136     if row is not None:
 137         return row, nres.create_from_placex_row
 138
 139     row = await find_in_osmline(conn, place, add_geometries)
 140     log().var_dump('Result (osmline)', row)
 141     if row is not None:
 142         return row, nres.create_from_osmline_row
 143
 144     row = await find_in_postcode(conn, place, add_geometries)
 145     log().var_dump('Result (postcode)', row)
 146     if row is not None:
 147         return row, nres.create_from_postcode_row
 148
 149     row = await find_in_tiger(conn, place, add_geometries)
 150     log().var_dump('Result (tiger)', row)
 151     return row, nres.create_from_tiger_row
 152
 153
 154 async def get_detailed_place(conn: SearchConnection, place: ntyp.PlaceRef,
 155                              details: ntyp.LookupDetails) -> Optional[nres.DetailedResult]:
 156     """ Retrieve a place with additional details from the database.
 157     """
 158     log().function('get_detailed_place', place=place, details=details)
 159
 160     if details.geometry_output and details.geometry_output != ntyp.GeometryFormat.GEOJSON:
 161         raise ValueError("lookup only supports geojosn polygon output.")
 162
 163     if details.geometry_output & ntyp.GeometryFormat.GEOJSON:
 164         def _add_geometry(sql: SaSelect, column: SaColumn) -> SaSelect:
 165             return sql.add_columns(sa.func.ST_AsGeoJSON(
 166                                     sa.case((sa.func.ST_NPoints(column) > 5000,
 167                                              sa.func.ST_SimplifyPreserveTopology(column, 0.0001)),
 168                                             else_=column), 7).label('geometry_geojson'))
 169     else:
 170         def _add_geometry(sql: SaSelect, column: SaColumn) -> SaSelect:
 171             return sql.add_columns(sa.func.ST_GeometryType(column).label('geometry_type'))
 172
 173     row_func: RowFunc[nres.DetailedResult]
 174     row, row_func = await find_in_all_tables(conn, place, _add_geometry)
 175
 176     if row is None:
 177         return None
 178
 179     result = row_func(row, nres.DetailedResult)
 180     assert result is not None
 181
 182     # add missing details
 183     assert result is not None
 184     if 'type' in result.geometry:
 185         result.geometry['type'] = GEOMETRY_TYPE_MAP.get(result.geometry['type'],
 186                                                         result.geometry['type'])
 187     indexed_date = getattr(row, 'indexed_date', None)
 188     if indexed_date is not None:
 189         result.indexed_date = indexed_date.replace(tzinfo=dt.timezone.utc)
 190
 191     await nres.add_result_details(conn, [result], details)
 192
 193     return result
 194
 195
 196 async def get_simple_place(conn: SearchConnection, place: ntyp.PlaceRef,
 197                            details: ntyp.LookupDetails) -> Optional[nres.SearchResult]:
 198     """ Retrieve a place as a simple search result from the database.
 199     """
 200     log().function('get_simple_place', place=place, details=details)
 201
 202     def _add_geometry(sql: SaSelect, col: SaColumn) -> SaSelect:
 203         if not details.geometry_output:
 204             return sql
 205
 206         out = []
 207
 208         if details.geometry_simplification > 0.0:
 209             col = sa.func.ST_SimplifyPreserveTopology(col, details.geometry_simplification)
 210
 211         if details.geometry_output & ntyp.GeometryFormat.GEOJSON:
 212             out.append(sa.func.ST_AsGeoJSON(col, 7).label('geometry_geojson'))
 213         if details.geometry_output & ntyp.GeometryFormat.TEXT:
 214             out.append(sa.func.ST_AsText(col).label('geometry_text'))
 215         if details.geometry_output & ntyp.GeometryFormat.KML:
 216             out.append(sa.func.ST_AsKML(col, 7).label('geometry_kml'))
 217         if details.geometry_output & ntyp.GeometryFormat.SVG:
 218             out.append(sa.func.ST_AsSVG(col, 0, 7).label('geometry_svg'))
 219
 220         return sql.add_columns(*out)
 221
 222     row_func: RowFunc[nres.SearchResult]
 223     row, row_func = await find_in_all_tables(conn, place, _add_geometry)
 224
 225     if row is None:
 226         return None
 227
 228     result = row_func(row, nres.SearchResult)
 229     assert result is not None
 230
 231     # add missing details
 232     assert result is not None
 233     if hasattr(row, 'bbox'):
 234         result.bbox = ntyp.Bbox.from_wkb(row.bbox)
 235
 236     await nres.add_result_details(conn, [result], details)
 237
 238     return result
 239
 240
 241 GEOMETRY_TYPE_MAP = {
 242     'POINT': 'ST_Point',
 243     'MULTIPOINT': 'ST_MultiPoint',
 244     'LINESTRING': 'ST_LineString',
 245     'MULTILINESTRING': 'ST_MultiLineString',
 246     'POLYGON': 'ST_Polygon',
 247     'MULTIPOLYGON': 'ST_MultiPolygon',
 248     'GEOMETRYCOLLECTION': 'ST_GeometryCollection'
 249 }