move filling of postcode table to python

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

diff --git a/nominatim/tools/refresh.py b/nominatim/tools/refresh.py
index b59c37bf9d5b46f1e74053ea3fd2c14621f5714b..805bd6348543de46b4ca8bf3f7dc0643e3389397 100644 (file)
--- a/nominatim/tools/refresh.py
+++ b/nominatim/tools/refresh.py
@@ -3,21 +3,16 @@ 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 textwrap import dedent
 
 from psycopg2.extras import execute_values
 

-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(dsn, 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(dsn, sql_dir / 'update-postcodes.sql')
-

 
 def recompute_word_counts(dsn, sql_dir):
     """ Compute the frequency of full-word search terms.
 
 def recompute_word_counts(dsn, sql_dir):
     """ Compute the frequency of full-word search terms.


@@ -75,100 +70,16 @@ 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,
-                     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 = (


@@ -187,14 +98,11 @@ PHP_CONST_DEFS = (
     ('Default_Language', 'DEFAULT_LANGUAGE', str),
     ('Log_DB', 'LOG_DB', bool),
     ('Log_File', 'LOG_FILE', str),
     ('Default_Language', 'DEFAULT_LANGUAGE', str),
     ('Log_DB', 'LOG_DB', bool),
     ('Log_File', 'LOG_FILE', str),

-    ('Max_Word_Frequency', 'MAX_WORD_FREQUENCY', int),

     ('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),
 )


@@ -247,7 +155,7 @@ def recompute_importance(conn):
     conn.commit()
 
 
     conn.commit()
 
 

-def setup_website(basedir, phplib_dir, config):
+def setup_website(basedir, config):

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


@@ -258,9 +166,12 @@ 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:
         if var_type == bool:
 
     for php_name, conf_name, var_type in PHP_CONST_DEFS:
         if var_type == bool:


@@ -274,7 +185,7 @@ def setup_website(basedir, phplib_dir, config):
 
         template += "@define('CONST_{}', {});\n".format(php_name, varout)
 
 
         template += "@define('CONST_{}', {});\n".format(php_name, varout)
 

-    template += "\nrequire_once('{}/website/{{}}');\n".format(phplib_dir)
+    template += "\nrequire_once('{}/website/{{}}');\n".format(config.lib_dir.php)

 
     for script in WEBSITE_SCRIPTS:
         (basedir / script).write_text(template.format(script), 'utf-8')
 
     for script in WEBSITE_SCRIPTS:
         (basedir / script).write_text(template.format(script), 'utf-8')