nominatim/tools/exec_utils.py

   1 """
   2 Helper functions for executing external programs.
   3 """
   4 import logging
   5 import os
   6 import subprocess
   7 import urllib.request as urlrequest
   8 from urllib.parse import urlencode
   9
  10 from psycopg2.extensions import parse_dsn
  11
  12 from ..version import NOMINATIM_VERSION
  13 from ..db.connection import get_pg_env
  14
  15 LOG = logging.getLogger()
  16
  17 def run_legacy_script(script, *args, nominatim_env=None, throw_on_fail=False):
  18     """ Run a Nominatim PHP script with the given arguments.
  19
  20         Returns the exit code of the script. If `throw_on_fail` is True
  21         then throw a `CalledProcessError` on a non-zero exit.
  22     """
  23     cmd = ['/usr/bin/env', 'php', '-Cq',
  24            nominatim_env.phplib_dir / 'admin' / script]
  25     cmd.extend([str(a) for a in args])
  26
  27     env = nominatim_env.config.get_os_env()
  28     env['NOMINATIM_DATADIR'] = str(nominatim_env.data_dir)
  29     env['NOMINATIM_SQLDIR'] = str(nominatim_env.sqllib_dir)
  30     env['NOMINATIM_CONFIGDIR'] = str(nominatim_env.config_dir)
  31     env['NOMINATIM_DATABASE_MODULE_SRC_PATH'] = nominatim_env.module_dir
  32     if not env['NOMINATIM_OSM2PGSQL_BINARY']:
  33         env['NOMINATIM_OSM2PGSQL_BINARY'] = nominatim_env.osm2pgsql_path
  34
  35     proc = subprocess.run(cmd, cwd=str(nominatim_env.project_dir), env=env,
  36                           check=throw_on_fail)
  37
  38     return proc.returncode
  39
  40 def run_api_script(endpoint, project_dir, extra_env=None, phpcgi_bin=None,
  41                    params=None):
  42     """ Execute a Nominiatim API function.
  43
  44         The function needs a project directory that contains the website
  45         directory with the scripts to be executed. The scripts will be run
  46         using php_cgi. Query parameters can be added as named arguments.
  47
  48         Returns the exit code of the script.
  49     """
  50     log = logging.getLogger()
  51     webdir = str(project_dir / 'website')
  52     query_string = urlencode(params or {})
  53
  54     env = dict(QUERY_STRING=query_string,
  55                SCRIPT_NAME='/{}.php'.format(endpoint),
  56                REQUEST_URI='/{}.php?{}'.format(endpoint, query_string),
  57                CONTEXT_DOCUMENT_ROOT=webdir,
  58                SCRIPT_FILENAME='{}/{}.php'.format(webdir, endpoint),
  59                HTTP_HOST='localhost',
  60                HTTP_USER_AGENT='nominatim-tool',
  61                REMOTE_ADDR='0.0.0.0',
  62                DOCUMENT_ROOT=webdir,
  63                REQUEST_METHOD='GET',
  64                SERVER_PROTOCOL='HTTP/1.1',
  65                GATEWAY_INTERFACE='CGI/1.1',
  66                REDIRECT_STATUS='CGI')
  67
  68     if extra_env:
  69         env.update(extra_env)
  70
  71     if phpcgi_bin is None:
  72         cmd = ['/usr/bin/env', 'php-cgi']
  73     else:
  74         cmd = [str(phpcgi_bin)]
  75
  76     proc = subprocess.run(cmd, cwd=str(project_dir), env=env, capture_output=True,
  77                           check=False)
  78
  79     if proc.returncode != 0 or proc.stderr:
  80         if proc.stderr:
  81             log.error(proc.stderr.decode('utf-8').replace('\\n', '\n'))
  82         else:
  83             log.error(proc.stdout.decode('utf-8').replace('\\n', '\n'))
  84         return proc.returncode or 1
  85
  86     result = proc.stdout.decode('utf-8')
  87     content_start = result.find('\r\n\r\n')
  88
  89     print(result[content_start + 4:].replace('\\n', '\n'))
  90
  91     return 0
  92
  93
  94 def run_php_server(server_address, base_dir):
  95     """ Run the built-in server from the given directory.
  96     """
  97     subprocess.run(['/usr/bin/env', 'php', '-S', server_address],
  98                    cwd=str(base_dir), check=True)
  99
 100
 101 def run_osm2pgsql(options):
 102     """ Run osm2pgsql with the given options.
 103     """
 104     env = get_pg_env(options['dsn'])
 105     cmd = [options['osm2pgsql'],
 106            '--hstore', '--latlon', '--slim',
 107            '--with-forward-dependencies', 'false',
 108            '--log-progress', 'true',
 109            '--number-processes', str(options['threads']),
 110            '--cache', str(options['osm2pgsql_cache']),
 111            '--output', 'gazetteer',
 112            '--style', str(options['osm2pgsql_style'])
 113           ]
 114     if options['append']:
 115         cmd.append('--append')
 116
 117     if options['flatnode_file']:
 118         cmd.extend(('--flat-nodes', options['flatnode_file']))
 119
 120     if options.get('disable_jit', False):
 121         env['PGOPTIONS'] = '-c jit=off -c max_parallel_workers_per_gather=0'
 122
 123     cmd.append(str(options['import_file']))
 124
 125     subprocess.run(cmd, cwd=options.get('cwd', '.'), env=env, check=True)
 126
 127
 128 def get_url(url):
 129     """ Get the contents from the given URL and return it as a UTF-8 string.
 130     """
 131     headers = {"User-Agent" : "Nominatim/" + NOMINATIM_VERSION}
 132
 133     try:
 134         with urlrequest.urlopen(urlrequest.Request(url, headers=headers)) as response:
 135             return response.read().decode('utf-8')
 136     except:
 137         LOG.fatal('Failed to load URL: %s', url)
 138         raise