X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/296a66558ffb2ee2c43e653b00ff24097ee3525c..cf98cff2a166eb35ed2c946e03f1610069bdd1d4:/nominatim/tools/database_import.py

diff --git a/nominatim/tools/database_import.py b/nominatim/tools/database_import.py
index 324a82cf..75483971 100644
--- a/nominatim/tools/database_import.py
+++ b/nominatim/tools/database_import.py
@@ -8,6 +8,7 @@ import subprocess
 from pathlib import Path
 
 import psutil
+import psycopg2.extras
 
 from nominatim.db.connection import connect, get_pg_env
 from nominatim.db import utils as db_utils
@@ -129,9 +130,8 @@ def import_osm_data(osm_file, options, drop=False, ignore_errors=False):
         if drop:
             conn.drop_table('planet_osm_nodes')
 
-    if drop:
-        if options['flatnode_file']:
-            Path(options['flatnode_file']).unlink()
+    if drop and options['flatnode_file']:
+        Path(options['flatnode_file']).unlink()
 
 
 def create_tables(conn, config, reverse_only=False):
@@ -160,11 +160,10 @@ def create_partition_tables(conn, config):
     sql.run_sql_file(conn, 'partition-tables.src.sql')
 
 
-def truncate_data_tables(conn, max_word_frequency=None):
+def truncate_data_tables(conn):
     """ Truncate all data tables to prepare for a fresh load.
     """
     with conn.cursor() as cur:
-        cur.execute('TRUNCATE word')
         cur.execute('TRUNCATE placex')
         cur.execute('TRUNCATE place_addressline')
         cur.execute('TRUNCATE location_area')
@@ -183,23 +182,15 @@ def truncate_data_tables(conn, max_word_frequency=None):
         for table in [r[0] for r in list(cur)]:
             cur.execute('TRUNCATE ' + table)
 
-        if max_word_frequency is not None:
-            # Used by getorcreate_word_id to ignore frequent partial words.
-            cur.execute("""CREATE OR REPLACE FUNCTION get_maxwordfreq()
-                           RETURNS integer AS $$
-                             SELECT {} as maxwordfreq;
-                           $$ LANGUAGE SQL IMMUTABLE
-                        """.format(max_word_frequency))
-        conn.commit()
+    conn.commit()
+
 
 _COPY_COLUMNS = 'osm_type, osm_id, class, type, name, admin_level, address, extratags, geometry'
 
-def load_data(dsn, data_dir, threads):
+
+def load_data(dsn, threads):
     """ Copy data into the word and placex table.
     """
-    # Pre-calculate the most important terms in the word list.
-    db_utils.execute_file(dsn, data_dir / 'words.sql')
-
     sel = selectors.DefaultSelector()
     # Then copy data from place to placex in <threads - 1> chunks.
     place_threads = max(1, threads - 1)
@@ -209,7 +200,7 @@ def load_data(dsn, data_dir, threads):
         conn.perform("""INSERT INTO placex ({0})
                          SELECT {0} FROM place
                          WHERE osm_id % {1} = {2}
-                           AND NOT (class='place' and type='houses')
+                           AND NOT (class='place' and (type='houses' or type='postcode'))
                            AND ST_IsValid(geometry)
                      """.format(_COPY_COLUMNS, place_threads, imod))
         sel.register(conn, selectors.EVENT_READ, conn)
@@ -261,34 +252,37 @@ def create_search_indices(conn, config, drop=False):
 
     sql.run_sql_file(conn, 'indices.sql', drop=drop)
 
-def create_country_names(conn, config):
-    """ Create search index for default country names.
+
+def create_country_names(conn, tokenizer, languages=None):
+    """ Add default country names to search index. `languages` is a comma-
+        separated list of language codes as used in OSM. If `languages` is not
+        empty then only name translations for the given languages are added
+        to the index.
     """
+    if languages:
+        languages = languages.split(',')
+
+    def _include_key(key):
+        return key == 'name' or \
+               (key.startswith('name:') and (not languages or key[5:] in languages))
 
     with conn.cursor() as cur:
-        cur.execute("""SELECT getorcreate_country(make_standard_name('uk'), 'gb')""")
-        cur.execute("""SELECT getorcreate_country(make_standard_name('united states'), 'us')""")
-        cur.execute("""SELECT COUNT(*) FROM
-                       (SELECT getorcreate_country(make_standard_name(country_code),
-                       country_code) FROM country_name WHERE country_code is not null) AS x""")
-        cur.execute("""SELECT COUNT(*) FROM
-                       (SELECT getorcreate_country(make_standard_name(name->'name'), country_code) 
-                       FROM country_name WHERE name ? 'name') AS x""")
-        sql_statement = """SELECT COUNT(*) FROM (SELECT getorcreate_country(make_standard_name(v),
-                           country_code) FROM (SELECT country_code, skeys(name)
-                           AS k, svals(name) AS v FROM country_name) x WHERE k"""
-
-        languages = config.LANGUAGES
-
-        if languages:
-            sql_statement = "{} IN (".format(sql_statement)
-            delim = ''
-            for language in languages.split(','):
-                sql_statement = "{}{}'name:{}'".format(sql_statement, delim, language)
-                delim = ', '
-            sql_statement = '{})'.format(sql_statement)
-        else:
-            sql_statement = "{} LIKE 'name:%'".format(sql_statement)
-        sql_statement = "{}) v".format(sql_statement)
-        cur.execute(sql_statement)
+        psycopg2.extras.register_hstore(cur)
+        cur.execute("""SELECT country_code, name FROM country_name
+                       WHERE country_code is not null""")
+
+        with tokenizer.name_analyzer() as analyzer:
+            for code, name in cur:
+                names = {'countrycode': code}
+                if code == 'gb':
+                    names['short_name'] = 'UK'
+                if code == 'us':
+                    names['short_name'] = 'United States'
+
+                # country names (only in languages as provided)
+                if name:
+                    names.update(((k, v) for k, v in name.items() if _include_key(k)))
+
+                analyzer.add_country_names(code, names)
+
     conn.commit()