1 # SPDX-License-Identifier: GPL-3.0-or-later
3 # This file is part of Nominatim. (https://nominatim.org)
5 # Copyright (C) 2024 by the Nominatim developer community.
6 # For a full list of authors see the git log.
8 Preprocessing of SQL files.
10 from typing import Set, Dict, Any, cast
13 from .connection import Connection, server_version_tuple, postgis_version_tuple
14 from .async_connection import WorkerPool
15 from ..config import Configuration
17 def _get_partitions(conn: Connection) -> Set[int]:
18 """ Get the set of partitions currently in use.
20 with conn.cursor() as cur:
21 cur.execute('SELECT DISTINCT partition FROM country_name')
24 partitions.add(row[0])
29 def _get_tables(conn: Connection) -> Set[str]:
30 """ Return the set of tables currently in use.
32 with conn.cursor() as cur:
33 cur.execute("SELECT tablename FROM pg_tables WHERE schemaname = 'public'")
35 return set((row[0] for row in list(cur)))
37 def _get_middle_db_format(conn: Connection, tables: Set[str]) -> str:
38 """ Returns the version of the slim middle tables.
40 if 'osm2pgsql_properties' not in tables:
43 with conn.cursor() as cur:
44 cur.execute("SELECT value FROM osm2pgsql_properties WHERE property = 'db_format'")
47 return cast(str, row[0]) if row is not None else '1'
50 def _setup_tablespace_sql(config: Configuration) -> Dict[str, str]:
51 """ Returns a dict with tablespace expressions for the different tablespace
52 kinds depending on whether a tablespace is configured or not.
55 for subset in ('ADDRESS', 'SEARCH', 'AUX'):
56 for kind in ('DATA', 'INDEX'):
57 tspace = getattr(config, f'TABLESPACE_{subset}_{kind}')
59 tspace = f'TABLESPACE "{tspace}"'
60 out[f'{subset.lower()}_{kind.lower()}'] = tspace
65 def _setup_postgresql_features(conn: Connection) -> Dict[str, Any]:
66 """ Set up a dictionary with various optional Postgresql/Postgis features that
67 depend on the database version.
69 pg_version = server_version_tuple(conn)
70 postgis_version = postgis_version_tuple(conn)
71 pg11plus = pg_version >= (11, 0, 0)
72 ps3 = postgis_version >= (3, 0)
74 'has_index_non_key_column': pg11plus,
75 'spgist_geom' : 'SPGIST' if pg11plus and ps3 else 'GIST'
78 class SQLPreprocessor:
79 """ A environment for preprocessing SQL files from the
82 The preprocessor provides a number of default filters and variables.
83 The variables may be overwritten when rendering an SQL file.
85 The preprocessing is currently based on the jinja2 templating library
86 and follows its syntax.
89 def __init__(self, conn: Connection, config: Configuration) -> None:
90 self.env = jinja2.Environment(autoescape=False,
91 loader=jinja2.FileSystemLoader(str(config.lib_dir.sql)))
93 db_info: Dict[str, Any] = {}
94 db_info['partitions'] = _get_partitions(conn)
95 db_info['tables'] = _get_tables(conn)
96 db_info['reverse_only'] = 'search_name' not in db_info['tables']
97 db_info['tablespace'] = _setup_tablespace_sql(config)
98 db_info['middle_db_format'] = _get_middle_db_format(conn, db_info['tables'])
100 self.env.globals['config'] = config
101 self.env.globals['db'] = db_info
102 self.env.globals['postgres'] = _setup_postgresql_features(conn)
105 def run_string(self, conn: Connection, template: str, **kwargs: Any) -> None:
106 """ Execute the given SQL template string on the connection.
107 The keyword arguments may supply additional parameters
110 sql = self.env.from_string(template).render(**kwargs)
112 with conn.cursor() as cur:
117 def run_sql_file(self, conn: Connection, name: str, **kwargs: Any) -> None:
118 """ Execute the given SQL file on the connection. The keyword arguments
119 may supply additional parameters for preprocessing.
121 sql = self.env.get_template(name).render(**kwargs)
123 with conn.cursor() as cur:
128 def run_parallel_sql_file(self, dsn: str, name: str, num_threads: int = 1,
129 **kwargs: Any) -> None:
130 """ Execute the given SQL files using parallel asynchronous connections.
131 The keyword arguments may supply additional parameters for
134 After preprocessing the SQL code is cut at lines containing only
135 '---'. Each chunk is sent to one of the `num_threads` workers.
137 sql = self.env.get_template(name).render(**kwargs)
139 parts = sql.split('\n---\n')
141 with WorkerPool(dsn, num_threads) as pool:
143 pool.next_free_worker().perform(part)