nominatim/tools/refresh.py

   1 """
   2 Functions for bringing auxiliary data in the database up-to-date.
   3 """
   4 import json
   5 import logging
   6 from textwrap import dedent
   7
   8 from psycopg2.extras import execute_values
   9
  10 from nominatim.db.utils import execute_file
  11 from nominatim.db.sql_preprocessor import SQLPreprocessor
  12 from nominatim.version import NOMINATIM_VERSION
  13
  14 LOG = logging.getLogger()
  15
  16
  17 def recompute_word_counts(dsn, sql_dir):
  18     """ Compute the frequency of full-word search terms.
  19     """
  20     execute_file(dsn, sql_dir / 'words_from_search_name.sql')
  21
  22
  23 def _add_address_level_rows_from_entry(rows, entry):
  24     """ Converts a single entry from the JSON format for address rank
  25         descriptions into a flat format suitable for inserting into a
  26         PostgreSQL table and adds these lines to `rows`.
  27     """
  28     countries = entry.get('countries') or (None, )
  29     for key, values in entry['tags'].items():
  30         for value, ranks in values.items():
  31             if isinstance(ranks, list):
  32                 rank_search, rank_address = ranks
  33             else:
  34                 rank_search = rank_address = ranks
  35             if not value:
  36                 value = None
  37             for country in countries:
  38                 rows.append((country, key, value, rank_search, rank_address))
  39
  40 def load_address_levels(conn, table, levels):
  41     """ Replace the `address_levels` table with the contents of `levels'.
  42
  43         A new table is created any previously existing table is dropped.
  44         The table has the following columns:
  45             country, class, type, rank_search, rank_address
  46     """
  47     rows = []
  48     for entry in levels:
  49         _add_address_level_rows_from_entry(rows, entry)
  50
  51     with conn.cursor() as cur:
  52         cur.drop_table(table)
  53
  54         cur.execute("""CREATE TABLE {} (country_code varchar(2),
  55                                         class TEXT,
  56                                         type TEXT,
  57                                         rank_search SMALLINT,
  58                                         rank_address SMALLINT)""".format(table))
  59
  60         execute_values(cur, "INSERT INTO {} VALUES %s".format(table), rows)
  61
  62         cur.execute('CREATE UNIQUE INDEX ON {} (country_code, class, type)'.format(table))
  63
  64     conn.commit()
  65
  66 def load_address_levels_from_file(conn, config_file):
  67     """ Replace the `address_levels` table with the contents of the config
  68         file.
  69     """
  70     with config_file.open('r') as fdesc:
  71         load_address_levels(conn, 'address_levels', json.load(fdesc))
  72
  73
  74 def create_functions(conn, config, enable_diff_updates=True, enable_debug=False):
  75     """ (Re)create the PL/pgSQL functions.
  76     """
  77     sql = SQLPreprocessor(conn, config)
  78
  79     sql.run_sql_file(conn, 'functions.sql',
  80                      disable_diff_updates=not enable_diff_updates,
  81                      debug=enable_debug)
  82
  83
  84
  85 WEBSITE_SCRIPTS = (
  86     'deletable.php',
  87     'details.php',
  88     'lookup.php',
  89     'polygons.php',
  90     'reverse.php',
  91     'search.php',
  92     'status.php'
  93 )
  94
  95 # constants needed by PHP scripts: PHP name, config name, type
  96 PHP_CONST_DEFS = (
  97     ('Database_DSN', 'DATABASE_DSN', str),
  98     ('Default_Language', 'DEFAULT_LANGUAGE', str),
  99     ('Log_DB', 'LOG_DB', bool),
 100     ('Log_File', 'LOG_FILE', str),
 101     ('NoAccessControl', 'CORS_NOACCESSCONTROL', bool),
 102     ('Places_Max_ID_count', 'LOOKUP_MAX_COUNT', int),
 103     ('PolygonOutput_MaximumTypes', 'POLYGON_OUTPUT_MAX_TYPES', int),
 104     ('Search_BatchMode', 'SEARCH_BATCH_MODE', bool),
 105     ('Search_NameOnlySearchFrequencyThreshold', 'SEARCH_NAME_ONLY_THRESHOLD', str),
 106     ('Use_US_Tiger_Data', 'USE_US_TIGER_DATA', bool),
 107     ('MapIcon_URL', 'MAPICON_URL', str),
 108 )
 109
 110
 111 def import_wikipedia_articles(dsn, data_path, ignore_errors=False):
 112     """ Replaces the wikipedia importance tables with new data.
 113         The import is run in a single transaction so that the new data
 114         is replace seemlessly.
 115
 116         Returns 0 if all was well and 1 if the importance file could not
 117         be found. Throws an exception if there was an error reading the file.
 118     """
 119     datafile = data_path / 'wikimedia-importance.sql.gz'
 120
 121     if not datafile.exists():
 122         return 1
 123
 124     pre_code = """BEGIN;
 125                   DROP TABLE IF EXISTS "wikipedia_article";
 126                   DROP TABLE IF EXISTS "wikipedia_redirect"
 127                """
 128     post_code = "COMMIT"
 129     execute_file(dsn, datafile, ignore_errors=ignore_errors,
 130                  pre_code=pre_code, post_code=post_code)
 131
 132     return 0
 133
 134
 135 def recompute_importance(conn):
 136     """ Recompute wikipedia links and importance for all entries in placex.
 137         This is a long-running operations that must not be executed in
 138         parallel with updates.
 139     """
 140     with conn.cursor() as cur:
 141         cur.execute('ALTER TABLE placex DISABLE TRIGGER ALL')
 142         cur.execute("""
 143             UPDATE placex SET (wikipedia, importance) =
 144                (SELECT wikipedia, importance
 145                 FROM compute_importance(extratags, country_code, osm_type, osm_id))
 146             """)
 147         cur.execute("""
 148             UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance
 149              FROM placex d
 150              WHERE s.place_id = d.linked_place_id and d.wikipedia is not null
 151                    and (s.wikipedia is null or s.importance < d.importance);
 152             """)
 153
 154         cur.execute('ALTER TABLE placex ENABLE TRIGGER ALL')
 155     conn.commit()
 156
 157
 158 def _quote_php_variable(var_type, config, conf_name):
 159     if var_type == bool:
 160         return 'true' if config.get_bool(conf_name) else 'false'
 161
 162     if var_type == int:
 163         return getattr(config, conf_name)
 164
 165     if not getattr(config, conf_name):
 166         return 'false'
 167
 168     quoted = getattr(config, conf_name).replace("'", "\\'")
 169     return f"'{quoted}'"
 170
 171
 172 def setup_website(basedir, config, conn):
 173     """ Create the website script stubs.
 174     """
 175     if not basedir.exists():
 176         LOG.info('Creating website directory.')
 177         basedir.mkdir()
 178
 179     template = dedent("""\
 180                       <?php
 181
 182                       @define('CONST_Debug', $_GET['debug'] ?? false);
 183                       @define('CONST_LibDir', '{0}');
 184                       @define('CONST_TokenizerDir', '{2}');
 185                       @define('CONST_NominatimVersion', '{1[0]}.{1[1]}.{1[2]}-{1[3]}');
 186
 187                       """.format(config.lib_dir.php, NOMINATIM_VERSION,
 188                                  config.project_dir / 'tokenizer'))
 189
 190     for php_name, conf_name, var_type in PHP_CONST_DEFS:
 191         varout = _quote_php_variable(var_type, config, conf_name)
 192
 193         template += f"@define('CONST_{php_name}', {varout});\n"
 194
 195     template += f"\nrequire_once('{config.lib_dir.php}/website/{{}}');\n"
 196
 197     search_name_table_exists = bool(conn and conn.table_exists('search_name'))
 198
 199     for script in WEBSITE_SCRIPTS:
 200         if not search_name_table_exists and script == 'search.php':
 201             (basedir / script).write_text(template.format('reverse-only-search.php'), 'utf-8')
 202         else:
 203             (basedir / script).write_text(template.format(script), 'utf-8')