+
+
+def create_tables(conn, config, reverse_only=False):
+ """ Create the set of basic tables.
+ When `reverse_only` is True, then the main table for searching will
+ be skipped and only reverse search is possible.
+ """
+ sql = SQLPreprocessor(conn, config)
+ sql.env.globals['db']['reverse_only'] = reverse_only
+
+ sql.run_sql_file(conn, 'tables.sql')
+
+
+def create_table_triggers(conn, config):
+ """ Create the triggers for the tables. The trigger functions must already
+ have been imported with refresh.create_functions().
+ """
+ sql = SQLPreprocessor(conn, config)
+ sql.run_sql_file(conn, 'table-triggers.sql')
+
+
+def create_partition_tables(conn, config):
+ """ Create tables that have explicit partitioning.
+ """
+ sql = SQLPreprocessor(conn, config)
+ sql.run_sql_file(conn, 'partition-tables.src.sql')
+
+
+def truncate_data_tables(conn):
+ """ Truncate all data tables to prepare for a fresh load.
+ """
+ with conn.cursor() as cur:
+ cur.execute('TRUNCATE placex')
+ cur.execute('TRUNCATE place_addressline')
+ cur.execute('TRUNCATE location_area')
+ cur.execute('TRUNCATE location_area_country')
+ cur.execute('TRUNCATE location_property_tiger')
+ cur.execute('TRUNCATE location_property_osmline')
+ cur.execute('TRUNCATE location_postcode')
+ if conn.table_exists('search_name'):
+ cur.execute('TRUNCATE search_name')
+ cur.execute('DROP SEQUENCE IF EXISTS seq_place')
+ cur.execute('CREATE SEQUENCE seq_place start 100000')
+
+ cur.execute("""SELECT tablename FROM pg_tables
+ WHERE tablename LIKE 'location_road_%'""")
+
+ for table in [r[0] for r in list(cur)]:
+ cur.execute('TRUNCATE ' + table)
+
+ conn.commit()
+
+_COPY_COLUMNS = 'osm_type, osm_id, class, type, name, admin_level, address, extratags, geometry'
+
+def load_data(dsn, threads):
+ """ Copy data into the word and placex table.
+ """
+ sel = selectors.DefaultSelector()
+ # Then copy data from place to placex in <threads - 1> chunks.
+ place_threads = max(1, threads - 1)
+ for imod in range(place_threads):
+ conn = DBConnection(dsn)
+ conn.connect()
+ conn.perform("""INSERT INTO placex ({0})
+ SELECT {0} FROM place
+ WHERE osm_id % {1} = {2}
+ AND NOT (class='place' and type='houses')
+ AND ST_IsValid(geometry)
+ """.format(_COPY_COLUMNS, place_threads, imod))
+ sel.register(conn, selectors.EVENT_READ, conn)
+
+ # Address interpolations go into another table.
+ conn = DBConnection(dsn)
+ conn.connect()
+ conn.perform("""INSERT INTO location_property_osmline (osm_id, address, linegeo)
+ SELECT osm_id, address, geometry FROM place
+ WHERE class='place' and type='houses' and osm_type='W'
+ and ST_GeometryType(geometry) = 'ST_LineString'
+ """)
+ sel.register(conn, selectors.EVENT_READ, conn)
+
+ # Now wait for all of them to finish.
+ todo = place_threads + 1
+ while todo > 0:
+ for key, _ in sel.select(1):
+ conn = key.data
+ sel.unregister(conn)
+ conn.wait()
+ conn.close()
+ todo -= 1
+ print('.', end='', flush=True)
+ print('\n')
+
+ with connect(dsn) as conn:
+ with conn.cursor() as cur:
+ cur.execute('ANALYSE')
+
+
+def create_search_indices(conn, config, drop=False):
+ """ Create tables that have explicit partitioning.
+ """
+
+ # If index creation failed and left an index invalid, they need to be
+ # cleaned out first, so that the script recreates them.
+ with conn.cursor() as cur:
+ cur.execute("""SELECT relname FROM pg_class, pg_index
+ WHERE pg_index.indisvalid = false
+ AND pg_index.indexrelid = pg_class.oid""")
+ bad_indices = [row[0] for row in list(cur)]
+ for idx in bad_indices:
+ LOG.info("Drop invalid index %s.", idx)
+ cur.execute('DROP INDEX "{}"'.format(idx))
+ conn.commit()
+
+ sql = SQLPreprocessor(conn, config)
+
+ sql.run_sql_file(conn, 'indices.sql', drop=drop)
+
+def create_country_names(conn, tokenizer, languages=None):
+ """ Add default country names to search index. `languages` is a comma-
+ separated list of language codes as used in OSM. If `languages` is not
+ empty then only name translations for the given languages are added
+ to the index.
+ """
+ if languages:
+ languages = languages.split(',')
+
+ def _include_key(key):
+ return key == 'name' or \
+ (key.startswith('name:') \
+ and (not languages or key[5:] in languages))
+
+ with conn.cursor() as cur:
+ psycopg2.extras.register_hstore(cur)
+ cur.execute("""SELECT country_code, name FROM country_name
+ WHERE country_code is not null""")
+
+ with tokenizer.name_analyzer() as analyzer:
+ for code, name in cur:
+ names = [code]
+ if code == 'gb':
+ names.append('UK')
+ if code == 'us':
+ names.append('United States')
+
+ # country names (only in languages as provided)
+ if name:
+ names.extend((v for k, v in name.items() if _include_key(k)))
+
+ analyzer.add_country_names(code, names)
+
+ conn.commit()