Merge pull request #2244 from AntoJvlt/import-special-phrases-tests-cleaning

[nominatim.git] / nominatim / tools / refresh.py

diff --git a/nominatim/tools/refresh.py b/nominatim/tools/refresh.py
index 33efe8f875121dad7acd3868432756e79c1a8452..581c69e8a110875406dfc76cf4a0f245ef0e3d9a 100644 (file)
--- a/nominatim/tools/refresh.py
+++ b/nominatim/tools/refresh.py
@@ -3,12 +3,13 @@ Functions for bringing auxiliary data in the database up-to-date.
 """
 import json
 import logging
 """
 import json
 import logging

-import re

 from textwrap import dedent
 
 from psycopg2.extras import execute_values
 
 from ..db.utils import execute_file
 from textwrap import dedent
 
 from psycopg2.extras import execute_values
 
 from ..db.utils import execute_file

+from ..db.sql_preprocessor import SQLPreprocessor
+from ..version import NOMINATIM_VERSION

 
 LOG = logging.getLogger()
 
 
 LOG = logging.getLogger()
 


@@ -75,100 +76,17 @@ def load_address_levels_from_file(conn, config_file):
     with config_file.open('r') as fdesc:
         load_address_levels(conn, 'address_levels', json.load(fdesc))
 
     with config_file.open('r') as fdesc:
         load_address_levels(conn, 'address_levels', json.load(fdesc))
 

-PLPGSQL_BASE_MODULES = (
-    'utils.sql',
-    'normalization.sql',
-    'ranking.sql',
-    'importance.sql',
-    'address_lookup.sql',
-    'interpolation.sql'
-)
-
-PLPGSQL_TABLE_MODULES = (
-    ('place', 'place_triggers.sql'),
-    ('placex', 'placex_triggers.sql'),
-    ('location_postcode', 'postcode_triggers.sql')
-)
-
-def _get_standard_function_sql(conn, config, sql_dir, enable_diff_updates, enable_debug):
-    """ Read all applicable SQLs containing PL/pgSQL functions, replace
-        placefolders and execute them.
-    """
-    sql_func_dir = sql_dir / 'functions'
-    sql = ''
-
-    # Get the basic set of functions that is always imported.
-    for sql_file in PLPGSQL_BASE_MODULES:
-        with (sql_func_dir / sql_file).open('r') as fdesc:
-            sql += fdesc.read()
-
-    # Some files require the presence of a certain table
-    for table, fname in PLPGSQL_TABLE_MODULES:
-        if conn.table_exists(table):
-            with (sql_func_dir / fname).open('r') as fdesc:
-                sql += fdesc.read()
-
-    # Replace placeholders.
-    sql = sql.replace('{modulepath}',
-                      config.DATABASE_MODULE_PATH or str((config.project_dir / 'module').resolve()))
-
-    if enable_diff_updates:
-        sql = sql.replace('RETURN NEW; -- %DIFFUPDATES%', '--')
-
-    if enable_debug:
-        sql = sql.replace('--DEBUG:', '')
-
-    if config.get_bool('LIMIT_REINDEXING'):
-        sql = sql.replace('--LIMIT INDEXING:', '')
-
-    if not config.get_bool('USE_US_TIGER_DATA'):
-        sql = sql.replace('-- %NOTIGERDATA% ', '')
-
-    if not config.get_bool('USE_AUX_LOCATION_DATA'):
-        sql = sql.replace('-- %NOAUXDATA% ', '')
-
-    reverse_only = 'false' if conn.table_exists('search_name') else 'true'
-
-    return sql.replace('%REVERSE-ONLY%', reverse_only)
-
-
-def replace_partition_string(sql, partitions):
-    """ Replace a partition template with the actual partition code.
-    """
-    for match in re.findall('^-- start(.*?)^-- end', sql, re.M | re.S):
-        repl = ''
-        for part in partitions:
-            repl += match.replace('-partition-', str(part))
-        sql = sql.replace(match, repl)
-
-    return sql

 
 

-def _get_partition_function_sql(conn, sql_dir):
-    """ Create functions that work on partition tables.
-    """
-    with conn.cursor() as cur:
-        cur.execute('SELECT distinct partition FROM country_name')
-        partitions = set([0])
-        for row in cur:
-            partitions.add(row[0])
-
-    with (sql_dir / 'partition-functions.src.sql').open('r') as fdesc:
-        sql = fdesc.read()
-
-    return replace_partition_string(sql, sorted(partitions))
-
-def create_functions(conn, config, sql_dir,
+def create_functions(conn, config, sqllib_dir,

                      enable_diff_updates=True, enable_debug=False):
     """ (Re)create the PL/pgSQL functions.
     """
                      enable_diff_updates=True, enable_debug=False):
     """ (Re)create the PL/pgSQL functions.
     """

-    sql = _get_standard_function_sql(conn, config, sql_dir,
-                                     enable_diff_updates, enable_debug)
-    sql += _get_partition_function_sql(conn, sql_dir)
+    sql = SQLPreprocessor(conn, config, sqllib_dir)

 
 

-    with conn.cursor() as cur:
-        cur.execute(sql)
+    sql.run_sql_file(conn, 'functions.sql',
+                     disable_diff_updates=not enable_diff_updates,
+                     debug=enable_debug)

 
 

-    conn.commit()

 
 
 WEBSITE_SCRIPTS = (
 
 
 WEBSITE_SCRIPTS = (


@@ -200,6 +118,53 @@ PHP_CONST_DEFS = (
 )
 
 
 )
 
 

+def import_wikipedia_articles(dsn, data_path, ignore_errors=False):
+    """ Replaces the wikipedia importance tables with new data.
+        The import is run in a single transaction so that the new data
+        is replace seemlessly.
+
+        Returns 0 if all was well and 1 if the importance file could not
+        be found. Throws an exception if there was an error reading the file.
+    """
+    datafile = data_path / 'wikimedia-importance.sql.gz'
+
+    if not datafile.exists():
+        return 1
+
+    pre_code = """BEGIN;
+                  DROP TABLE IF EXISTS "wikipedia_article";
+                  DROP TABLE IF EXISTS "wikipedia_redirect"
+               """
+    post_code = "COMMIT"
+    execute_file(dsn, datafile, ignore_errors=ignore_errors,
+                 pre_code=pre_code, post_code=post_code)
+
+    return 0
+
+
+def recompute_importance(conn):
+    """ Recompute wikipedia links and importance for all entries in placex.
+        This is a long-running operations that must not be executed in
+        parallel with updates.
+    """
+    with conn.cursor() as cur:
+        cur.execute('ALTER TABLE placex DISABLE TRIGGER ALL')
+        cur.execute("""
+            UPDATE placex SET (wikipedia, importance) =
+               (SELECT wikipedia, importance
+                FROM compute_importance(extratags, country_code, osm_type, osm_id))
+            """)
+        cur.execute("""
+            UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance
+             FROM placex d
+             WHERE s.place_id = d.linked_place_id and d.wikipedia is not null
+                   and (s.wikipedia is null or s.importance < d.importance);
+            """)
+
+        cur.execute('ALTER TABLE placex ENABLE TRIGGER ALL')
+    conn.commit()
+
+

 def setup_website(basedir, phplib_dir, config):
     """ Create the website script stubs.
     """
 def setup_website(basedir, phplib_dir, config):
     """ Create the website script stubs.
     """


@@ -211,9 +176,10 @@ def setup_website(basedir, phplib_dir, config):
                       <?php
 
                       @define('CONST_Debug', $_GET['debug'] ?? false);
                       <?php
 
                       @define('CONST_Debug', $_GET['debug'] ?? false);

-                      @define('CONST_LibDir', '{}');
+                      @define('CONST_LibDir', '{0}');
+                      @define('CONST_NominatimVersion', '{1[0]}.{1[1]}.{1[2]}-{1[3]}');

 
 

-                      """.format(phplib_dir))
+                      """.format(phplib_dir, NOMINATIM_VERSION))

 
     for php_name, conf_name, var_type in PHP_CONST_DEFS:
         if var_type == bool:
 
     for php_name, conf_name, var_type in PHP_CONST_DEFS:
         if var_type == bool: