From: Sarah Hoffmann Date: Thu, 13 May 2021 20:11:41 +0000 (+0200) Subject: use tokenizer during Tiger data import X-Git-Tag: v4.0.0~82^2~2 X-Git-Url: https://git.openstreetmap.org./nominatim.git/commitdiff_plain/35efe3b41ceb525565c230b42d953b794d03fe5e?ds=inline;hp=-c use tokenizer during Tiger data import This also changes the required import format to CSV. --- 35efe3b41ceb525565c230b42d953b794d03fe5e diff --git a/.pylintrc b/.pylintrc index 6cf97be9..022243ad 100644 --- a/.pylintrc +++ b/.pylintrc @@ -12,4 +12,4 @@ ignored-modules=icu ignored-classes=NominatimArgs,closing disable=too-few-public-methods,duplicate-code -good-names=i,x,y +good-names=i,x,y,fd diff --git a/lib-sql/tiger_import_finish.sql b/lib-sql/tiger_import_finish.sql index 85bb5e3e..a084a2e2 100644 --- a/lib-sql/tiger_import_finish.sql +++ b/lib-sql/tiger_import_finish.sql @@ -12,4 +12,6 @@ ALTER TABLE location_property_tiger_import RENAME TO location_property_tiger; ALTER INDEX IF EXISTS idx_location_property_tiger_parent_place_id_imp RENAME TO idx_location_property_tiger_housenumber_parent_place_id; ALTER INDEX IF EXISTS idx_location_property_tiger_place_id_imp RENAME TO idx_location_property_tiger_place_id; -DROP FUNCTION tiger_line_import (linegeo geometry, in_startnumber integer, in_endnumber integer, interpolationtype text, in_street text, in_isin text, in_postcode text); +DROP FUNCTION tiger_line_import (linegeo GEOMETRY, in_startnumber INTEGER, + in_endnumber INTEGER, interpolationtype TEXT, + token_info JSONB, in_postcode TEXT); diff --git a/lib-sql/tiger_import_start.sql b/lib-sql/tiger_import_start.sql index 4b9c33fc..faa4efbb 100644 --- a/lib-sql/tiger_import_start.sql +++ b/lib-sql/tiger_import_start.sql @@ -1,9 +1,9 @@ DROP TABLE IF EXISTS location_property_tiger_import; CREATE TABLE location_property_tiger_import (linegeo GEOMETRY, place_id BIGINT, partition INTEGER, parent_place_id BIGINT, startnumber INTEGER, endnumber INTEGER, interpolationtype TEXT, postcode TEXT); -CREATE OR REPLACE FUNCTION tiger_line_import(linegeo GEOMETRY, in_startnumber INTEGER, - in_endnumber INTEGER, interpolationtype TEXT, - in_street TEXT, in_isin TEXT, in_postcode TEXT) RETURNS INTEGER +CREATE OR REPLACE FUNCTION tiger_line_import(linegeo GEOMETRY, in_startnumber INTEGER, + in_endnumber INTEGER, interpolationtype TEXT, + token_info JSONB, in_postcode TEXT) RETURNS INTEGER AS $$ DECLARE startnumber INTEGER; @@ -27,13 +27,13 @@ BEGIN END IF; IF startnumber < 0 THEN - RAISE WARNING 'Negative house number range (% to %) on %, %', startnumber, endnumber, in_street, in_isin; + RAISE WARNING 'Negative house number range (% to %)', startnumber, endnumber; RETURN 0; END IF; numberrange := endnumber - startnumber; - IF (interpolationtype = 'odd' AND startnumber%2 = 0) OR (interpolationtype = 'even' AND startnumber%2 = 1) THEN + IF (interpolationtype = 'odd' AND startnumber % 2 = 0) OR (interpolationtype = 'even' AND startnumber % 2 = 1) THEN startnumber := startnumber + 1; stepsize := 2; ELSE @@ -45,10 +45,10 @@ BEGIN END IF; -- Filter out really broken tiger data - IF numberrange > 0 AND (numberrange::float/stepsize::float > 500) + IF numberrange > 0 AND (numberrange::float/stepsize::float > 500) AND ST_length(linegeo)/(numberrange::float/stepsize::float) < 0.000001 THEN - RAISE WARNING 'Road too short for number range % to % on %, % (%)',startnumber,endnumber,in_street,in_isin, - ST_length(linegeo)/(numberrange::float/stepsize::float); + RAISE WARNING 'Road too short for number range % to % (%)',startnumber,endnumber, + ST_length(linegeo)/(numberrange::float/stepsize::float); RETURN 0; END IF; @@ -56,7 +56,7 @@ BEGIN out_partition := get_partition('us'); out_parent_place_id := null; - address_street_word_ids := word_ids_from_name(in_street); + address_street_word_ids := token_addr_street_match_tokens(token_info); IF address_street_word_ids IS NOT NULL THEN out_parent_place_id := getNearestNamedRoadPlaceId(out_partition, place_centroid, address_street_word_ids); diff --git a/nominatim/cli.py b/nominatim/cli.py index 20a9c5f1..c66e78d3 100644 --- a/nominatim/cli.py +++ b/nominatim/cli.py @@ -13,7 +13,6 @@ from nominatim.tools.exec_utils import run_legacy_script, run_php_server from nominatim.errors import UsageError from nominatim import clicmd from nominatim.clicmd.args import NominatimArgs -from nominatim.tools import tiger_data LOG = logging.getLogger() @@ -147,9 +146,14 @@ class UpdateAddData: @staticmethod def run(args): + from nominatim.tokenizer import factory as tokenizer_factory + from nominatim.tools import tiger_data + if args.tiger_data: + tokenizer = tokenizer_factory.get_tokenizer_for_db(args.config) return tiger_data.add_tiger_data(args.tiger_data, - args.config, args.threads or 1) + args.config, args.threads or 1, + tokenizer) params = ['update.php'] if args.file: diff --git a/nominatim/tools/tiger_data.py b/nominatim/tools/tiger_data.py index fbcdb077..ffb55886 100644 --- a/nominatim/tools/tiger_data.py +++ b/nominatim/tools/tiger_data.py @@ -1,10 +1,14 @@ """ Functions for importing tiger data and handling tarbar and directory files """ +import csv +import io import logging import os import tarfile +import psycopg2.extras + from nominatim.db.connection import connect from nominatim.db.async_connection import WorkerPool from nominatim.db.sql_preprocessor import SQLPreprocessor @@ -20,30 +24,40 @@ def handle_tarfile_or_directory(data_dir): tar = None if data_dir.endswith('.tar.gz'): tar = tarfile.open(data_dir) - sql_files = [i for i in tar.getmembers() if i.name.endswith('.sql')] - LOG.warning("Found %d SQL files in tarfile with path %s", len(sql_files), data_dir) - if not sql_files: + csv_files = [i for i in tar.getmembers() if i.name.endswith('.csv')] + LOG.warning("Found %d CSV files in tarfile with path %s", len(csv_files), data_dir) + if not csv_files: LOG.warning("Tiger data import selected but no files in tarfile's path %s", data_dir) return None, None else: files = os.listdir(data_dir) - sql_files = [os.path.join(data_dir, i) for i in files if i.endswith('.sql')] - LOG.warning("Found %d SQL files in path %s", len(sql_files), data_dir) - if not sql_files: + csv_files = [os.path.join(data_dir, i) for i in files if i.endswith('.csv')] + LOG.warning("Found %d CSV files in path %s", len(csv_files), data_dir) + if not csv_files: LOG.warning("Tiger data import selected but no files found in path %s", data_dir) return None, None - return sql_files, tar + return csv_files, tar -def handle_threaded_sql_statements(pool, file): +def handle_threaded_sql_statements(pool, fd, analyzer): """ Handles sql statement with multiplexing """ - lines = 0 # Using pool of database connections to execute sql statements - for sql_query in file: - pool.next_free_worker().perform(sql_query) + + sql = "SELECT tiger_line_import(%s, %s, %s, %s, %s, %s)" + + for row in csv.DictReader(fd, delimiter=';'): + try: + address = dict(street=row['street'], postcode=row['postcode']) + args = ('SRID=4326;' + row['geometry'], + int(row['from']), int(row['to']), row['interpolation'], + psycopg2.extras.Json(analyzer.process_place(dict(address=address))), + analyzer.normalize_postcode(row['postcode'])) + except ValueError: + continue + pool.next_free_worker().perform(sql, args=args) lines += 1 if lines == 1000: @@ -51,31 +65,34 @@ def handle_threaded_sql_statements(pool, file): lines = 0 -def add_tiger_data(data_dir, config, threads): +def add_tiger_data(data_dir, config, threads, tokenizer): """ Import tiger data from directory or tar file `data dir`. """ dsn = config.get_libpq_dsn() - sql_files, tar = handle_tarfile_or_directory(data_dir) + files, tar = handle_tarfile_or_directory(data_dir) - if not sql_files: + if not files: return with connect(dsn) as conn: sql = SQLPreprocessor(conn, config) sql.run_sql_file(conn, 'tiger_import_start.sql') - # Reading sql_files and then for each file line handling + # Reading files and then for each file line handling # sql_query in chunks. place_threads = max(1, threads - 1) with WorkerPool(dsn, place_threads, ignore_sql_errors=True) as pool: - for sql_file in sql_files: - if not tar: - file = open(sql_file) - else: - file = tar.extractfile(sql_file) + with tokenizer.name_analyzer() as analyzer: + for fname in files: + if not tar: + fd = open(fname) + else: + fd = io.TextIOWrapper(tar.extractfile(fname)) + + handle_threaded_sql_statements(pool, fd, analyzer) - handle_threaded_sql_statements(pool, file) + fd.close() if tar: tar.close()