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 import sql as pysql
   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         cur.execute_values(pysql.SQL("INSERT INTO {} VALUES %s")
  61                            .format(pysql.Identifier(table)), rows)
  62
  63         cur.execute('CREATE UNIQUE INDEX ON {} (country_code, class, type)'.format(table))
  64
  65     conn.commit()
  66
  67 def load_address_levels_from_file(conn, config_file):
  68     """ Replace the `address_levels` table with the contents of the config
  69         file.
  70     """
  71     with config_file.open('r') as fdesc:
  72         load_address_levels(conn, 'address_levels', json.load(fdesc))
  73
  74
  75 def create_functions(conn, config, enable_diff_updates=True, enable_debug=False):
  76     """ (Re)create the PL/pgSQL functions.
  77     """
  78     sql = SQLPreprocessor(conn, config)
  79
  80     sql.run_sql_file(conn, 'functions.sql',
  81                      disable_diff_updates=not enable_diff_updates,
  82                      debug=enable_debug)
  83
  84
  85
  86 WEBSITE_SCRIPTS = (
  87     'deletable.php',
  88     'details.php',
  89     'lookup.php',
  90     'polygons.php',
  91     'reverse.php',
  92     'search.php',
  93     'status.php'
  94 )
  95
  96 # constants needed by PHP scripts: PHP name, config name, type
  97 PHP_CONST_DEFS = (
  98     ('Database_DSN', 'DATABASE_DSN', str),
  99     ('Default_Language', 'DEFAULT_LANGUAGE', str),
 100     ('Log_DB', 'LOG_DB', bool),
 101     ('Log_File', 'LOG_FILE', str),
 102     ('NoAccessControl', 'CORS_NOACCESSCONTROL', bool),
 103     ('Places_Max_ID_count', 'LOOKUP_MAX_COUNT', int),
 104     ('PolygonOutput_MaximumTypes', 'POLYGON_OUTPUT_MAX_TYPES', int),
 105     ('Search_BatchMode', 'SEARCH_BATCH_MODE', bool),
 106     ('Search_NameOnlySearchFrequencyThreshold', 'SEARCH_NAME_ONLY_THRESHOLD', str),
 107     ('Use_US_Tiger_Data', 'USE_US_TIGER_DATA', bool),
 108     ('MapIcon_URL', 'MAPICON_URL', str),
 109 )
 110
 111
 112 def import_wikipedia_articles(dsn, data_path, ignore_errors=False):
 113     """ Replaces the wikipedia importance tables with new data.
 114         The import is run in a single transaction so that the new data
 115         is replace seemlessly.
 116
 117         Returns 0 if all was well and 1 if the importance file could not
 118         be found. Throws an exception if there was an error reading the file.
 119     """
 120     datafile = data_path / 'wikimedia-importance.sql.gz'
 121
 122     if not datafile.exists():
 123         return 1
 124
 125     pre_code = """BEGIN;
 126                   DROP TABLE IF EXISTS "wikipedia_article";
 127                   DROP TABLE IF EXISTS "wikipedia_redirect"
 128                """
 129     post_code = "COMMIT"
 130     execute_file(dsn, datafile, ignore_errors=ignore_errors,
 131                  pre_code=pre_code, post_code=post_code)
 132
 133     return 0
 134
 135
 136 def recompute_importance(conn):
 137     """ Recompute wikipedia links and importance for all entries in placex.
 138         This is a long-running operations that must not be executed in
 139         parallel with updates.
 140     """
 141     with conn.cursor() as cur:
 142         cur.execute('ALTER TABLE placex DISABLE TRIGGER ALL')
 143         cur.execute("""
 144             UPDATE placex SET (wikipedia, importance) =
 145                (SELECT wikipedia, importance
 146                 FROM compute_importance(extratags, country_code, osm_type, osm_id))
 147             """)
 148         cur.execute("""
 149             UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance
 150              FROM placex d
 151              WHERE s.place_id = d.linked_place_id and d.wikipedia is not null
 152                    and (s.wikipedia is null or s.importance < d.importance);
 153             """)
 154
 155         cur.execute('ALTER TABLE placex ENABLE TRIGGER ALL')
 156     conn.commit()
 157
 158
 159 def _quote_php_variable(var_type, config, conf_name):
 160     if var_type == bool:
 161         return 'true' if config.get_bool(conf_name) else 'false'
 162
 163     if var_type == int:
 164         return getattr(config, conf_name)
 165
 166     if not getattr(config, conf_name):
 167         return 'false'
 168
 169     quoted = getattr(config, conf_name).replace("'", "\\'")
 170     return f"'{quoted}'"
 171
 172
 173 def setup_website(basedir, config, conn):
 174     """ Create the website script stubs.
 175     """
 176     if not basedir.exists():
 177         LOG.info('Creating website directory.')
 178         basedir.mkdir()
 179
 180     template = dedent("""\
 181                       <?php
 182
 183                       @define('CONST_Debug', $_GET['debug'] ?? false);
 184                       @define('CONST_LibDir', '{0}');
 185                       @define('CONST_TokenizerDir', '{2}');
 186                       @define('CONST_NominatimVersion', '{1[0]}.{1[1]}.{1[2]}-{1[3]}');
 187
 188                       """.format(config.lib_dir.php, NOMINATIM_VERSION,
 189                                  config.project_dir / 'tokenizer'))
 190
 191     for php_name, conf_name, var_type in PHP_CONST_DEFS:
 192         varout = _quote_php_variable(var_type, config, conf_name)
 193
 194         template += f"@define('CONST_{php_name}', {varout});\n"
 195
 196     template += f"\nrequire_once('{config.lib_dir.php}/website/{{}}');\n"
 197
 198     search_name_table_exists = bool(conn and conn.table_exists('search_name'))
 199
 200     for script in WEBSITE_SCRIPTS:
 201         if not search_name_table_exists and script == 'search.php':
 202             (basedir / script).write_text(template.format('reverse-only-search.php'), 'utf-8')
 203         else:
 204             (basedir / script).write_text(template.format(script), 'utf-8')