speed up reverse lookup of place nodes

[nominatim.git] / nominatim / tools / exec_utils.py

diff --git a/nominatim/tools/exec_utils.py b/nominatim/tools/exec_utils.py
index 9e16e29382473694d8ba7de193eba4cce36cf44e..566ac06edc57d4374b4979095866c64c1e959d60 100644 (file)
--- a/nominatim/tools/exec_utils.py
+++ b/nominatim/tools/exec_utils.py
@@ -1,36 +1,57 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# This file is part of Nominatim. (https://nominatim.org)
+#
+# Copyright (C) 2022 by the Nominatim developer community.
+# For a full list of authors see the git log.

 """
 Helper functions for executing external programs.
 """
 """
 Helper functions for executing external programs.
 """

+from typing import Any, Union, Optional, Mapping, IO
+from pathlib import Path

 import logging
 import logging

+import os

 import subprocess
 import subprocess

+import urllib.request as urlrequest

 from urllib.parse import urlencode
 
 from urllib.parse import urlencode
 

-def run_legacy_script(script, *args, nominatim_env=None, throw_on_fail=False):
+from nominatim.config import Configuration
+from nominatim.typing import StrPath
+from nominatim.version import NOMINATIM_VERSION
+from nominatim.db.connection import get_pg_env
+
+LOG = logging.getLogger()
+
+def run_legacy_script(script: StrPath, *args: Union[int, str],
+                      config: Configuration,
+                      throw_on_fail: bool = False) -> int:

     """ Run a Nominatim PHP script with the given arguments.
 
         Returns the exit code of the script. If `throw_on_fail` is True
         then throw a `CalledProcessError` on a non-zero exit.
     """
     cmd = ['/usr/bin/env', 'php', '-Cq',
     """ Run a Nominatim PHP script with the given arguments.
 
         Returns the exit code of the script. If `throw_on_fail` is True
         then throw a `CalledProcessError` on a non-zero exit.
     """
     cmd = ['/usr/bin/env', 'php', '-Cq',

-           nominatim_env.phplib_dir / 'admin' / script]
+           str(config.lib_dir.php / 'admin' / script)]

     cmd.extend([str(a) for a in args])
 
     cmd.extend([str(a) for a in args])
 

-    env = nominatim_env.config.get_os_env()
-    env['NOMINATIM_DATADIR'] = str(nominatim_env.data_dir)
-    env['NOMINATIM_BINDIR'] = str(nominatim_env.data_dir / 'utils')
-    if not env['NOMINATIM_DATABASE_MODULE_PATH']:
-        env['NOMINATIM_DATABASE_MODULE_PATH'] = nominatim_env.module_dir
+    env = config.get_os_env()
+    env['NOMINATIM_DATADIR'] = str(config.lib_dir.data)
+    env['NOMINATIM_SQLDIR'] = str(config.lib_dir.sql)
+    env['NOMINATIM_CONFIGDIR'] = str(config.config_dir)
+    env['NOMINATIM_DATABASE_MODULE_SRC_PATH'] = str(config.lib_dir.module)

     if not env['NOMINATIM_OSM2PGSQL_BINARY']:
     if not env['NOMINATIM_OSM2PGSQL_BINARY']:

-        env['NOMINATIM_OSM2PGSQL_BINARY'] = nominatim_env.osm2pgsql_path
+        env['NOMINATIM_OSM2PGSQL_BINARY'] = str(config.lib_dir.osm2pgsql)

 
 

-    proc = subprocess.run(cmd, cwd=str(nominatim_env.project_dir), env=env,
+    proc = subprocess.run(cmd, cwd=str(config.project_dir), env=env,

                           check=throw_on_fail)
 
     return proc.returncode
 
                           check=throw_on_fail)
 
     return proc.returncode
 

-def run_api_script(endpoint, project_dir, extra_env=None, phpcgi_bin=None,
-                   params=None):
-    """ Execute a Nominiatim API function.
+def run_api_script(endpoint: str, project_dir: Path,
+                   extra_env: Optional[Mapping[str, str]] = None,
+                   phpcgi_bin: Optional[Path] = None,
+                   params: Optional[Mapping[str, Any]] = None) -> int:
+    """ Execute a Nominatim API function.

 
         The function needs a project directory that contains the website
         directory with the scripts to be executed. The scripts will be run
 
         The function needs a project directory that contains the website
         directory with the scripts to be executed. The scripts will be run


@@ -43,10 +64,10 @@ def run_api_script(endpoint, project_dir, extra_env=None, phpcgi_bin=None,
     query_string = urlencode(params or {})
 
     env = dict(QUERY_STRING=query_string,
     query_string = urlencode(params or {})
 
     env = dict(QUERY_STRING=query_string,

-               SCRIPT_NAME='/{}.php'.format(endpoint),
-               REQUEST_URI='/{}.php?{}'.format(endpoint, query_string),
+               SCRIPT_NAME=f'/{endpoint}.php',
+               REQUEST_URI=f'/{endpoint}.php?{query_string}',

                CONTEXT_DOCUMENT_ROOT=webdir,
                CONTEXT_DOCUMENT_ROOT=webdir,

-               SCRIPT_FILENAME='{}/{}.php'.format(webdir, endpoint),
+               SCRIPT_FILENAME=f'{webdir}/{endpoint}.php',

                HTTP_HOST='localhost',
                HTTP_USER_AGENT='nominatim-tool',
                REMOTE_ADDR='0.0.0.0',
                HTTP_HOST='localhost',
                HTTP_USER_AGENT='nominatim-tool',
                REMOTE_ADDR='0.0.0.0',


@@ -64,7 +85,9 @@ def run_api_script(endpoint, project_dir, extra_env=None, phpcgi_bin=None,
     else:
         cmd = [str(phpcgi_bin)]
 
     else:
         cmd = [str(phpcgi_bin)]
 

-    proc = subprocess.run(cmd, cwd=str(project_dir), env=env, capture_output=True,
+    proc = subprocess.run(cmd, cwd=str(project_dir), env=env,
+                          stdout=subprocess.PIPE,
+                          stderr=subprocess.PIPE,

                           check=False)
 
     if proc.returncode != 0 or proc.stderr:
                           check=False)
 
     if proc.returncode != 0 or proc.stderr:


@@ -80,3 +103,71 @@ def run_api_script(endpoint, project_dir, extra_env=None, phpcgi_bin=None,
     print(result[content_start + 4:].replace('\\n', '\n'))
 
     return 0
     print(result[content_start + 4:].replace('\\n', '\n'))
 
     return 0

+
+
+def run_php_server(server_address: str, base_dir: StrPath) -> None:
+    """ Run the built-in server from the given directory.
+    """
+    subprocess.run(['/usr/bin/env', 'php', '-S', server_address],
+                   cwd=str(base_dir), check=True)
+
+
+def run_osm2pgsql(options: Mapping[str, Any]) -> None:
+    """ Run osm2pgsql with the given options.
+    """
+    env = get_pg_env(options['dsn'])
+    cmd = [str(options['osm2pgsql']),
+           '--hstore', '--latlon', '--slim',
+           '--log-progress', 'true',
+           '--number-processes', '1' if options['append'] else str(options['threads']),
+           '--cache', str(options['osm2pgsql_cache']),
+           '--style', str(options['osm2pgsql_style'])
+          ]
+
+    if str(options['osm2pgsql_style']).endswith('.lua'):
+        env['LUA_PATH'] = ';'.join((str(options['osm2pgsql_style_path'] / '?.lua'),
+                                    os.environ.get('LUAPATH', ';')))
+        cmd.extend(('--output', 'flex'))
+    else:
+        cmd.extend(('--output', 'gazetteer'))
+
+    cmd.append('--append' if options['append'] else '--create')
+
+    if options['flatnode_file']:
+        cmd.extend(('--flat-nodes', options['flatnode_file']))
+
+    for key, param in (('slim_data', '--tablespace-slim-data'),
+                       ('slim_index', '--tablespace-slim-index'),
+                       ('main_data', '--tablespace-main-data'),
+                       ('main_index', '--tablespace-main-index')):
+        if options['tablespaces'][key]:
+            cmd.extend((param, options['tablespaces'][key]))
+
+    if options.get('disable_jit', False):
+        env['PGOPTIONS'] = '-c jit=off -c max_parallel_workers_per_gather=0'
+
+    if 'import_data' in options:
+        cmd.extend(('-r', 'xml', '-'))
+    elif isinstance(options['import_file'], list):
+        for fname in options['import_file']:
+            cmd.append(str(fname))
+    else:
+        cmd.append(str(options['import_file']))
+
+    subprocess.run(cmd, cwd=options.get('cwd', '.'),
+                   input=options.get('import_data'),
+                   env=env, check=True)
+
+
+def get_url(url: str) -> str:
+    """ Get the contents from the given URL and return it as a UTF-8 string.
+    """
+    headers = {"User-Agent": f"Nominatim/{NOMINATIM_VERSION!s}"}
+
+    try:
+        request = urlrequest.Request(url, headers=headers)
+        with urlrequest.urlopen(request) as response: # type: IO[bytes]
+            return response.read().decode('utf-8')
+    except Exception:
+        LOG.fatal('Failed to load URL: %s', url)
+        raise