nominatim/tools/exec_utils.py

   1 """
   2 Helper functions for executing external programs.
   3 """
   4 import logging
   5 import subprocess
   6 import urllib.request as urlrequest
   7 from urllib.parse import urlencode
   8
   9 from nominatim.version import NOMINATIM_VERSION
  10 from nominatim.db.connection import get_pg_env
  11
  12 LOG = logging.getLogger()
  13
  14 def run_legacy_script(script, *args, nominatim_env=None, throw_on_fail=False):
  15     """ Run a Nominatim PHP script with the given arguments.
  16
  17         Returns the exit code of the script. If `throw_on_fail` is True
  18         then throw a `CalledProcessError` on a non-zero exit.
  19     """
  20     cmd = ['/usr/bin/env', 'php', '-Cq',
  21            str(nominatim_env.phplib_dir / 'admin' / script)]
  22     cmd.extend([str(a) for a in args])
  23
  24     env = nominatim_env.config.get_os_env()
  25     env['NOMINATIM_DATADIR'] = str(nominatim_env.data_dir)
  26     env['NOMINATIM_SQLDIR'] = str(nominatim_env.sqllib_dir)
  27     env['NOMINATIM_CONFIGDIR'] = str(nominatim_env.config_dir)
  28     env['NOMINATIM_DATABASE_MODULE_SRC_PATH'] = str(nominatim_env.module_dir)
  29     if not env['NOMINATIM_OSM2PGSQL_BINARY']:
  30         env['NOMINATIM_OSM2PGSQL_BINARY'] = str(nominatim_env.osm2pgsql_path)
  31
  32     proc = subprocess.run(cmd, cwd=str(nominatim_env.project_dir), env=env,
  33                           check=throw_on_fail)
  34
  35     return proc.returncode
  36
  37 def run_api_script(endpoint, project_dir, extra_env=None, phpcgi_bin=None,
  38                    params=None):
  39     """ Execute a Nominiatim API function.
  40
  41         The function needs a project directory that contains the website
  42         directory with the scripts to be executed. The scripts will be run
  43         using php_cgi. Query parameters can be added as named arguments.
  44
  45         Returns the exit code of the script.
  46     """
  47     log = logging.getLogger()
  48     webdir = str(project_dir / 'website')
  49     query_string = urlencode(params or {})
  50
  51     env = dict(QUERY_STRING=query_string,
  52                SCRIPT_NAME='/{}.php'.format(endpoint),
  53                REQUEST_URI='/{}.php?{}'.format(endpoint, query_string),
  54                CONTEXT_DOCUMENT_ROOT=webdir,
  55                SCRIPT_FILENAME='{}/{}.php'.format(webdir, endpoint),
  56                HTTP_HOST='localhost',
  57                HTTP_USER_AGENT='nominatim-tool',
  58                REMOTE_ADDR='0.0.0.0',
  59                DOCUMENT_ROOT=webdir,
  60                REQUEST_METHOD='GET',
  61                SERVER_PROTOCOL='HTTP/1.1',
  62                GATEWAY_INTERFACE='CGI/1.1',
  63                REDIRECT_STATUS='CGI')
  64
  65     if extra_env:
  66         env.update(extra_env)
  67
  68     if phpcgi_bin is None:
  69         cmd = ['/usr/bin/env', 'php-cgi']
  70     else:
  71         cmd = [str(phpcgi_bin)]
  72
  73     proc = subprocess.run(cmd, cwd=str(project_dir), env=env,
  74                           stdout=subprocess.PIPE,
  75                           stderr=subprocess.PIPE,
  76                           check=False)
  77
  78     if proc.returncode != 0 or proc.stderr:
  79         if proc.stderr:
  80             log.error(proc.stderr.decode('utf-8').replace('\\n', '\n'))
  81         else:
  82             log.error(proc.stdout.decode('utf-8').replace('\\n', '\n'))
  83         return proc.returncode or 1
  84
  85     result = proc.stdout.decode('utf-8')
  86     content_start = result.find('\r\n\r\n')
  87
  88     print(result[content_start + 4:].replace('\\n', '\n'))
  89
  90     return 0
  91
  92
  93 def run_php_server(server_address, base_dir):
  94     """ Run the built-in server from the given directory.
  95     """
  96     subprocess.run(['/usr/bin/env', 'php', '-S', server_address],
  97                    cwd=str(base_dir), check=True)
  98
  99
 100 def run_osm2pgsql(options):
 101     """ Run osm2pgsql with the given options.
 102     """
 103     env = get_pg_env(options['dsn'])
 104     cmd = [str(options['osm2pgsql']),
 105            '--hstore', '--latlon', '--slim',
 106            '--with-forward-dependencies', 'false',
 107            '--log-progress', 'true',
 108            '--number-processes', str(options['threads']),
 109            '--cache', str(options['osm2pgsql_cache']),
 110            '--output', 'gazetteer',
 111            '--style', str(options['osm2pgsql_style'])
 112           ]
 113     if options['append']:
 114         cmd.append('--append')
 115     else:
 116         cmd.append('--create')
 117
 118     if options['flatnode_file']:
 119         cmd.extend(('--flat-nodes', options['flatnode_file']))
 120
 121     for key, param in (('slim_data', '--tablespace-slim-data'),
 122                        ('slim_index', '--tablespace-slim-index'),
 123                        ('main_data', '--tablespace-main-data'),
 124                        ('main_index', '--tablespace-main-index')):
 125         if options['tablespaces'][key]:
 126             cmd.extend((param, options['tablespaces'][key]))
 127
 128     if options.get('disable_jit', False):
 129         env['PGOPTIONS'] = '-c jit=off -c max_parallel_workers_per_gather=0'
 130
 131     if 'import_data' in options:
 132         cmd.extend(('-r', 'xml', '-'))
 133     else:
 134         cmd.append(str(options['import_file']))
 135
 136     subprocess.run(cmd, cwd=options.get('cwd', '.'),
 137                    input=options.get('import_data'),
 138                    env=env, check=True)
 139
 140
 141 def get_url(url):
 142     """ Get the contents from the given URL and return it as a UTF-8 string.
 143     """
 144     headers = {"User-Agent": "Nominatim/{0[0]}.{0[1]}.{0[2]}-{0[3]}".format(NOMINATIM_VERSION)}
 145
 146     try:
 147         with urlrequest.urlopen(urlrequest.Request(url, headers=headers)) as response:
 148             return response.read().decode('utf-8')
 149     except Exception:
 150         LOG.fatal('Failed to load URL: %s', url)
 151         raise