Merge remote-tracking branch 'upstream/master'

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

diff --git a/nominatim/tools/refresh.py b/nominatim/tools/refresh.py
index f09c0cede2cc62e8603fc905501a4b71469f2259..95be4c0f6a16c9be3c8f437aa3f65e989c0880cb 100644 (file)
--- a/nominatim/tools/refresh.py
+++ b/nominatim/tools/refresh.py
@@ -1,29 +1,24 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# This file is part of Nominatim. (https://nominatim.org)
+#
+# Copyright (C) 2022 by the Nominatim developer community.
+# For a full list of authors see the git log.

 """
 Functions for bringing auxiliary data in the database up-to-date.
 """
 """
 Functions for bringing auxiliary data in the database up-to-date.
 """

-import json

 import logging
 import logging

-import re

 from textwrap import dedent
 from textwrap import dedent

+from pathlib import Path

 
 

-from psycopg2.extras import execute_values
+from psycopg2 import sql as pysql

 
 

-from ..db.utils import execute_file
+from nominatim.db.utils import execute_file
+from nominatim.db.sql_preprocessor import SQLPreprocessor
+from nominatim.version import NOMINATIM_VERSION

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

-def update_postcodes(conn, sql_dir):
-    """ Recalculate postcode centroids and add, remove and update entries in the
-        location_postcode table. `conn` is an opne connection to the database.
-    """
-    execute_file(conn, sql_dir / 'update-postcodes.sql')
-
-
-def recompute_word_counts(conn, sql_dir):
-    """ Compute the frequency of full-word search terms.
-    """
-    execute_file(conn, sql_dir / '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
 
 def _add_address_level_rows_from_entry(rows, entry):
     """ Converts a single entry from the JSON format for address rank


@@ -54,7 +49,7 @@ def load_address_levels(conn, table, levels):
         _add_address_level_rows_from_entry(rows, entry)
 
     with conn.cursor() as cur:
         _add_address_level_rows_from_entry(rows, entry)
 
     with conn.cursor() as cur:

-        cur.execute('DROP TABLE IF EXISTS {}'.format(table))
+        cur.drop_table(table)

 
         cur.execute("""CREATE TABLE {} (country_code varchar(2),
                                         class TEXT,
 
         cur.execute("""CREATE TABLE {} (country_code varchar(2),
                                         class TEXT,


@@ -62,113 +57,33 @@ def load_address_levels(conn, table, levels):
                                         rank_search SMALLINT,
                                         rank_address SMALLINT)""".format(table))
 
                                         rank_search SMALLINT,
                                         rank_address SMALLINT)""".format(table))
 

-        execute_values(cur, "INSERT INTO {} VALUES %s".format(table), rows)
+        cur.execute_values(pysql.SQL("INSERT INTO {} VALUES %s")
+                           .format(pysql.Identifier(table)), rows)

 
         cur.execute('CREATE UNIQUE INDEX ON {} (country_code, class, type)'.format(table))
 
     conn.commit()
 
 
         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))
-
-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.
+def load_address_levels_from_config(conn, config):
+    """ Replace the `address_levels` table with the content as
+        defined in the given configuration. Uses the parameter
+        NOMINATIM_ADDRESS_LEVEL_CONFIG to determine the location of the
+        configuration file.

     """
     """

-    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()
+    cfg = config.load_sub_configuration('', config='ADDRESS_LEVEL_CONFIG')
+    load_address_levels(conn, 'address_levels', cfg)

 
 

-    return replace_partition_string(sql, sorted(partitions))

 
 

-def create_functions(conn, config, sql_dir,
-                     enable_diff_updates=True, enable_debug=False):
+def create_functions(conn, config, enable_diff_updates=True, enable_debug=False):

     """ (Re)create the PL/pgSQL functions.
     """
     """ (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)

 
 

-    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 = (


@@ -186,21 +101,84 @@ PHP_CONST_DEFS = (
     ('Database_DSN', 'DATABASE_DSN', str),
     ('Default_Language', 'DEFAULT_LANGUAGE', str),
     ('Log_DB', 'LOG_DB', bool),
     ('Database_DSN', 'DATABASE_DSN', str),
     ('Default_Language', 'DEFAULT_LANGUAGE', str),
     ('Log_DB', 'LOG_DB', bool),

-    ('Log_File', 'LOG_FILE', str),
-    ('Max_Word_Frequency', 'MAX_WORD_FREQUENCY', int),
+    ('Log_File', 'LOG_FILE', Path),

     ('NoAccessControl', 'CORS_NOACCESSCONTROL', bool),
     ('Places_Max_ID_count', 'LOOKUP_MAX_COUNT', int),
     ('PolygonOutput_MaximumTypes', 'POLYGON_OUTPUT_MAX_TYPES', int),
     ('Search_BatchMode', 'SEARCH_BATCH_MODE', bool),
     ('Search_NameOnlySearchFrequencyThreshold', 'SEARCH_NAME_ONLY_THRESHOLD', str),
     ('NoAccessControl', 'CORS_NOACCESSCONTROL', bool),
     ('Places_Max_ID_count', 'LOOKUP_MAX_COUNT', int),
     ('PolygonOutput_MaximumTypes', 'POLYGON_OUTPUT_MAX_TYPES', int),
     ('Search_BatchMode', 'SEARCH_BATCH_MODE', bool),
     ('Search_NameOnlySearchFrequencyThreshold', 'SEARCH_NAME_ONLY_THRESHOLD', str),

-    ('Term_Normalization_Rules', 'TERM_NORMALIZATION', str),
-    ('Use_Aux_Location_data', 'USE_AUX_LOCATION_DATA', bool),

     ('Use_US_Tiger_Data', 'USE_US_TIGER_DATA', bool),
     ('MapIcon_URL', 'MAPICON_URL', str),
 )
 
 
     ('Use_US_Tiger_Data', 'USE_US_TIGER_DATA', bool),
     ('MapIcon_URL', 'MAPICON_URL', str),
 )
 
 

-def setup_website(basedir, phplib_dir, config):
+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 _quote_php_variable(var_type, config, conf_name):
+    if var_type == bool:
+        return 'true' if config.get_bool(conf_name) else 'false'
+
+    if var_type == int:
+        return getattr(config, conf_name)
+
+    if not getattr(config, conf_name):
+        return 'false'
+
+    if var_type == Path:
+        value = str(config.get_path(conf_name))
+    else:
+        value = getattr(config, conf_name)
+
+    quoted = value.replace("'", "\\'")
+    return f"'{quoted}'"
+
+
+def setup_website(basedir, config, conn):

     """ Create the website script stubs.
     """
     if not basedir.exists():
     """ Create the website script stubs.
     """
     if not basedir.exists():


@@ -211,23 +189,24 @@ 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_TokenizerDir', '{2}');
+                      @define('CONST_NominatimVersion', '{1[0]}.{1[1]}.{1[2]}-{1[3]}');

 
 

-                      """.format(phplib_dir))
+                      """.format(config.lib_dir.php, NOMINATIM_VERSION,
+                                 config.project_dir / 'tokenizer'))

 
     for php_name, conf_name, var_type in PHP_CONST_DEFS:
 
     for php_name, conf_name, var_type in PHP_CONST_DEFS:

-        if var_type == bool:
-            varout = 'true' if config.get_bool(conf_name) else 'false'
-        elif var_type == int:
-            varout = getattr(config, conf_name)
-        elif not getattr(config, conf_name):
-            varout = 'false'
-        else:
-            varout = "'{}'".format(getattr(config, conf_name).replace("'", "\\'"))
+        varout = _quote_php_variable(var_type, config, conf_name)
+
+        template += f"@define('CONST_{php_name}', {varout});\n"

 
 

-        template += "@define('CONST_{}', {});\n".format(php_name, varout)
+    template += f"\nrequire_once('{config.lib_dir.php}/website/{{}}');\n"

 
 

-    template += "\nrequire_once('{}/website/{{}}');\n".format(phplib_dir)
+    search_name_table_exists = bool(conn and conn.table_exists('search_name'))

 
     for script in WEBSITE_SCRIPTS:
 
     for script in WEBSITE_SCRIPTS:

-        (basedir / script).write_text(template.format(script), 'utf-8')
+        if not search_name_table_exists and script == 'search.php':
+            (basedir / script).write_text(template.format('reverse-only-search.php'), 'utf-8')
+        else:
+            (basedir / script).write_text(template.format(script), 'utf-8')