port address level computation to Python

Also adds simple tests for correct table creation.

nominatim/cli.py

@@ -372,33 +372,38 @@ class UpdateRefresh:
     def run(args):
         import nominatim.tools.refresh
 
-        with psycopg2.connect(args.config.get_libpq_dsn()) as conn:
-            if args.postcodes:
-                LOG.warning("Update postcodes centroid")
-                nominatim.tools.refresh.update_postcodes(conn, args.data_dir)
-            if args.word_counts:
-                LOG.warning('Recompute frequency of full-word search terms')
-                nominatim.tools.refresh.recompute_word_counts(conn, args.data_dir)
-            if args.address_levels:
-                run_legacy_script('update.php', '--update-address-levels',
-                                  nominatim_env=args, throw_on_fail=True)
-            if args.functions:
-                params = ['setup.php', '--create-functions', '--create-partition-functions']
-                if args.diffs:
-                    params.append('--enable-diff-updates')
-                if args.enable_debug_statements:
-                    params.append('--enable-debug-statements')
-                run_legacy_script(*params, nominatim_env=args, throw_on_fail=True)
-            if args.wiki_data:
-                run_legacy_script('setup.php', '--import-wikipedia-articles',
-                                  nominatim_env=args, throw_on_fail=True)
-            # Attention: importance MUST come after wiki data import.
-            if args.importance:
-                run_legacy_script('update.php', '--recompute-importance',
-                                  nominatim_env=args, throw_on_fail=True)
-            if args.website:
-                run_legacy_script('setup.php', '--setup-website',
-                                  nominatim_env=args, throw_on_fail=True)
+        conn = psycopg2.connect(args.config.get_libpq_dsn())
+
+        if args.postcodes:
+            LOG.warning("Update postcodes centroid")
+            nominatim.tools.refresh.update_postcodes(conn, args.data_dir)
+        if args.word_counts:
+            LOG.warning('Recompute frequency of full-word search terms')
+            nominatim.tools.refresh.recompute_word_counts(conn, args.data_dir)
+        if args.address_levels:
+            cfg = Path(args.config.ADDRESS_LEVEL_CONFIG)
+            LOG.warning('Updating address levels from %s', cfg)
+            nominatim.tools.refresh.load_address_levels_from_file(conn, cfg)
+        if args.functions:
+            params = ['setup.php', '--create-functions', '--create-partition-functions']
+            if args.diffs:
+                params.append('--enable-diff-updates')
+            if args.enable_debug_statements:
+                params.append('--enable-debug-statements')
+            run_legacy_script(*params, nominatim_env=args, throw_on_fail=True)
+        if args.wiki_data:
+            run_legacy_script('setup.php', '--import-wikipedia-articles',
+                              nominatim_env=args, throw_on_fail=True)
+        # Attention: importance MUST come after wiki data import.
+        if args.importance:
+            run_legacy_script('update.php', '--recompute-importance',
+                              nominatim_env=args, throw_on_fail=True)
+        if args.website:
+            run_legacy_script('setup.php', '--setup-website',
+                              nominatim_env=args, throw_on_fail=True)
+
+        conn.close()
+
         return 0
 
 

nominatim/config.py

@@ -24,6 +24,13 @@ class Configuration:
         if project_dir is not None:
             self._config.update(dotenv_values(str((project_dir / '.env').resolve())))
 
+        # Add defaults for variables that are left empty to set the default.
+        # They may still be overwritten by environment variables.
+        if not self._config['NOMINATIM_ADDRESS_LEVEL_CONFIG']:
+            self._config['NOMINATIM_ADDRESS_LEVEL_CONFIG'] = \
+                str(config_dir / 'address-levels.json')
+
+
     def __getattr__(self, name):
         name = 'NOMINATIM_' + name
 

nominatim/db/utils.py

@@ -9,3 +9,4 @@ def execute_file(conn, fname):
         sql = fdesc.read()
     with conn.cursor() as cur:
         cur.execute(sql)
+    conn.commit()

nominatim/tools/refresh.py

@@ -1,6 +1,10 @@
 """
 Functions for bringing auxiliary data in the database up-to-date.
 """
+import json
+
+from psycopg2.extras import execute_values
+
 from ..db.utils import execute_file
 
 def update_postcodes(conn, datadir):
@@ -14,3 +18,54 @@ def recompute_word_counts(conn, datadir):
     """ Compute the frequency of full-word search terms.
     """
     execute_file(conn, datadir / 'sql' / 'words_from_search_name.sql')
+
+
+def _add_address_level_rows_from_entry(rows, entry):
+    """ Converts a single entry from the JSON format for address rank
+        descriptions into a flat format suitable for inserting into a
+        PostgreSQL table and adds these lines to `rows`.
+    """
+    countries = entry.get('countries') or (None, )
+    for key, values in entry['tags'].items():
+        for value, ranks in values.items():
+            if isinstance(ranks, list):
+                rank_search, rank_address = ranks
+            else:
+                rank_search = rank_address = ranks
+            if not value:
+                value = None
+            for country in countries:
+                rows.append((country, key, value, rank_search, rank_address))
+
+def load_address_levels(conn, table, levels):
+    """ Replace the `address_levels` table with the contents of `levels'.
+
+        A new table is created any previously existing table is dropped.
+        The table has the following columns:
+            country, class, type, rank_search, rank_address
+    """
+    rows = []
+    for entry in levels:
+        _add_address_level_rows_from_entry(rows, entry)
+
+    with conn.cursor() as cur:
+        cur.execute('DROP TABLE IF EXISTS {}'.format(table))
+
+        cur.execute("""CREATE TABLE {} (country_code varchar(2),
+                                        class TEXT,
+                                        type TEXT,
+                                        rank_search SMALLINT,
+                                        rank_address SMALLINT)""".format(table))
+
+        execute_values(cur, "INSERT INTO {} VALUES %s".format(table), rows)
+
+        cur.execute('CREATE UNIQUE INDEX ON {} (country_code, class, type)'.format(table))
+
+    conn.commit()
+
+def load_address_levels_from_file(conn, config_file):
+    """ Replace the `address_levels` table with the contents of the config
+        file.
+    """
+    with config_file.open('r') as fdesc:
+        load_address_levels(conn, 'address_levels', json.load(fdesc))