safety-screen/Nominatim
1
0
Fork 0
forked from hans/Nominatim
Code Pull Requests Activity

move default country name creation to tokenizer

Browse Source 
The new function is also used, when a country us updated. All SQL
function related to country names have been removed.
This commit is contained in:
Sarah Hoffmann
2021-04-27 11:37:18 +02:00
parent dc700c25b6
commit bef300305e
8 changed files with 105 additions and 134 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
nominatim/clicmd/setup.py
View File

@@ -133,7 +133,8 @@ class SetupAll:
database_import.create_search_indices(conn, args.config,
drop=args.no_updates)
LOG.warning('Create search index for default country names.')
database_import.create_country_names(conn, args.config)
database_import.create_country_names(conn, tokenizer,
args.config.LANGUAGES)
webdir = args.project_dir / 'website'
LOG.warning('Setup website at %s', webdir)

37
nominatim/tokenizer/legacy_tokenizer.py
View File

@@ -223,6 +223,21 @@ class LegacyNameAnalyzer:
FROM (SELECT distinct(postcode) as pc
FROM location_postcode) x""")
def add_country_names(self, country_code, names):
""" Add names for the given country to the search index.
"""
with self.conn.cursor() as cur:
cur.execute(
"""INSERT INTO word (word_id, word_token, country_code)
(SELECT nextval('seq_word'), lookup_token, %s
FROM (SELECT ' ' || make_standard_name(n) as lookup_token
FROM unnest(%s)n) y
WHERE NOT EXISTS(SELECT * FROM word
WHERE word_token = lookup_token and country_code = %s))
""", (country_code, names, country_code))
def process_place(self, place):
""" Determine tokenizer information about the given place.
@@ -231,7 +246,14 @@ class LegacyNameAnalyzer:
"""
token_info = _TokenInfo(self._cache)
token_info.add_names(self.conn, place.get('name'), place.get('country_feature'))
names = place.get('name')
if names:
token_info.add_names(self.conn, names)
country_feature = place.get('country_feature')
if country_feature and re.fullmatch(r'[A-Za-z][A-Za-z]', country_feature):
self.add_country_names(country_feature.lower(), list(names.values()))
address = place.get('address')
@@ -279,22 +301,14 @@ class _TokenInfo:
self.data = {}
def add_names(self, conn, names, country_feature):
def add_names(self, conn, names):
""" Add token information for the names of the place.
"""
if not names:
return
with conn.cursor() as cur:
# Create the token IDs for all names.
self.data['names'] = cur.scalar("SELECT make_keywords(%s)::text",
(names, ))
# Add country tokens to word table if necessary.
if country_feature and re.fullmatch(r'[A-Za-z][A-Za-z]', country_feature):
cur.execute("SELECT create_country(%s, %s)",
(names, country_feature.lower()))
def add_housenumbers(self, conn, hnrs):
""" Extract housenumber information from the address.
@@ -334,7 +348,8 @@ class _TokenInfo:
"""
def _get_place(name):
with conn.cursor() as cur:
cur.execute("""SELECT (addr_ids_from_name(%s) || getorcreate_name_id(make_standard_name(%s), ''))::text,
cur.execute("""SELECT (addr_ids_from_name(%s)
|| getorcreate_name_id(make_standard_name(%s), ''))::text,
word_ids_from_name(%s)::text""",
(name, name, name))
return cur.fetchone()

54
nominatim/tools/database_import.py
View File

@@ -8,6 +8,7 @@ import subprocess
from pathlib import Path
import psutil
import psycopg2.extras
from nominatim.db.connection import connect, get_pg_env
from nominatim.db import utils as db_utils
@@ -250,34 +251,37 @@ def create_search_indices(conn, config, drop=False):
sql.run_sql_file(conn, 'indices.sql', drop=drop)
def create_country_names(conn, config):
""" Create search index for default country names.
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:
cur.execute("""SELECT getorcreate_country(make_standard_name('uk'), 'gb')""")
cur.execute("""SELECT getorcreate_country(make_standard_name('united states'), 'us')""")
cur.execute("""SELECT COUNT(*) FROM
(SELECT getorcreate_country(make_standard_name(country_code),
country_code) FROM country_name WHERE country_code is not null) AS x""")
cur.execute("""SELECT COUNT(*) FROM
(SELECT getorcreate_country(make_standard_name(name->'name'), country_code)
FROM country_name WHERE name ? 'name') AS x""")
sql_statement = """SELECT COUNT(*) FROM (SELECT getorcreate_country(make_standard_name(v),
country_code) FROM (SELECT country_code, skeys(name)
AS k, svals(name) AS v FROM country_name) x WHERE k"""
psycopg2.extras.register_hstore(cur)
cur.execute("""SELECT country_code, name FROM country_name
WHERE country_code is not null""")
languages = config.LANGUAGES
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)
if languages:
sql_statement = "{} IN (".format(sql_statement)
delim = ''
for language in languages.split(','):
sql_statement = "{}{}'name:{}'".format(sql_statement, delim, language)
delim = ', '
sql_statement = '{})'.format(sql_statement)
else:
sql_statement = "{} LIKE 'name:%'".format(sql_statement)
sql_statement = "{}) v".format(sql_statement)
cur.execute(sql_statement)
conn.commit()