convert admin --analyse-indexing to new indexing method

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

diff --git a/nominatim/tools/migration.py b/nominatim/tools/migration.py
index bcf8f1425050a3fd82249af27c9c49109d870fef..28a14455066f689fad6b93a05a78ddd8c0014fe0 100644 (file)
--- a/nominatim/tools/migration.py
+++ b/nominatim/tools/migration.py
@@ -1,11 +1,19 @@
+# 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 database migration to newer software versions.
 """
 import logging
 
 """
 Functions for database migration to newer software versions.
 """
 import logging
 

+from psycopg2 import sql as pysql
+

 from nominatim.db import properties
 from nominatim.db.connection import connect
 from nominatim.db import properties
 from nominatim.db.connection import connect

-from nominatim.version import NOMINATIM_VERSION
+from nominatim.version import NOMINATIM_VERSION, version_str

 from nominatim.tools import refresh
 from nominatim.tokenizer import factory as tokenizer_factory
 from nominatim.errors import UsageError
 from nominatim.tools import refresh
 from nominatim.tokenizer import factory as tokenizer_factory
 from nominatim.errors import UsageError


@@ -41,7 +49,7 @@ def migrate(config, paths):
         for version, func in _MIGRATION_FUNCTIONS:
             if db_version <= version:
                 LOG.warning("Runnning: %s (%s)", func.__doc__.split('\n', 1)[0],
         for version, func in _MIGRATION_FUNCTIONS:
             if db_version <= version:
                 LOG.warning("Runnning: %s (%s)", func.__doc__.split('\n', 1)[0],

-                            '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(version))
+                            version_str(version))

                 kwargs = dict(conn=conn, config=config, paths=paths)
                 func(**kwargs)
                 conn.commit()
                 kwargs = dict(conn=conn, config=config, paths=paths)
                 func(**kwargs)
                 conn.commit()


@@ -53,8 +61,7 @@ def migrate(config, paths):
             tokenizer = tokenizer_factory.get_tokenizer_for_db(config)
             tokenizer.update_sql_functions(config)
 
             tokenizer = tokenizer_factory.get_tokenizer_for_db(config)
             tokenizer.update_sql_functions(config)
 

-        properties.set_property(conn, 'database_version',
-                                '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(NOMINATIM_VERSION))
+        properties.set_property(conn, 'database_version', version_str())

 
         conn.commit()
 
 
         conn.commit()
 


@@ -119,11 +126,11 @@ def add_nominatim_property_table(conn, config, **_):
     """
     if not conn.table_exists('nominatim_properties'):
         with conn.cursor() as cur:
     """
     if not conn.table_exists('nominatim_properties'):
         with conn.cursor() as cur:

-            cur.execute("""CREATE TABLE nominatim_properties (
-                               property TEXT,
-                               value TEXT);
-                           GRANT SELECT ON TABLE nominatim_properties TO "{}";
-                        """.format(config.DATABASE_WEBUSER))
+            cur.execute(pysql.SQL("""CREATE TABLE nominatim_properties (
+                                        property TEXT,
+                                        value TEXT);
+                                     GRANT SELECT ON TABLE nominatim_properties TO {};
+                                  """).format(pysql.Identifier(config.DATABASE_WEBUSER)))

 
 @_migration(3, 6, 0, 0)
 def change_housenumber_transliteration(conn, **_):
 
 @_migration(3, 6, 0, 0)
 def change_housenumber_transliteration(conn, **_):


@@ -188,7 +195,8 @@ def install_legacy_tokenizer(conn, config, **_):
                                            and column_name = 'token_info'""",
                                         (table, ))
                 if has_column == 0:
                                            and column_name = 'token_info'""",
                                         (table, ))
                 if has_column == 0:

-                    cur.execute('ALTER TABLE {} ADD COLUMN token_info JSONB'.format(table))
+                    cur.execute(pysql.SQL('ALTER TABLE {} ADD COLUMN token_info JSONB')
+                                .format(pysql.Identifier(table)))

         tokenizer = tokenizer_factory.create_tokenizer(config, init_db=False,
                                                        module_name='legacy')
 
         tokenizer = tokenizer_factory.create_tokenizer(config, init_db=False,
                                                        module_name='legacy')
 


@@ -210,3 +218,94 @@ def create_tiger_housenumber_index(conn, **_):
                             ON location_property_tiger
                             USING btree(parent_place_id)
                             INCLUDE (startnumber, endnumber) """)
                             ON location_property_tiger
                             USING btree(parent_place_id)
                             INCLUDE (startnumber, endnumber) """)

+
+
+@_migration(4, 0, 99, 1)
+def create_interpolation_index_on_place(conn, **_):
+    """ Create idx_place_interpolations for lookup of interpolation lines
+        on updates.
+    """
+    with conn.cursor() as cur:
+        cur.execute("""CREATE INDEX IF NOT EXISTS idx_place_interpolations
+                       ON place USING gist(geometry)
+                       WHERE osm_type = 'W' and address ? 'interpolation'""")
+
+
+@_migration(4, 0, 99, 2)
+def add_step_column_for_interpolation(conn, **_):
+    """ Add a new column 'step' to the interpolations table.
+
+        Also convers the data into the stricter format which requires that
+        startnumbers comply with the odd/even requirements.
+    """
+    if conn.table_has_column('location_property_osmline', 'step'):
+        return
+
+    with conn.cursor() as cur:
+        # Mark invalid all interpolations with no intermediate numbers.
+        cur.execute("""UPDATE location_property_osmline SET startnumber = null
+                       WHERE endnumber - startnumber <= 1 """)
+        # Align the start numbers where odd/even does not match.
+        cur.execute("""UPDATE location_property_osmline
+                       SET startnumber = startnumber + 1,
+                           linegeo = ST_LineSubString(linegeo,
+                                                      1.0 / (endnumber - startnumber)::float,
+                                                      1)
+                       WHERE (interpolationtype = 'odd' and startnumber % 2 = 0)
+                              or (interpolationtype = 'even' and startnumber % 2 = 1)
+                    """)
+        # Mark invalid odd/even interpolations with no intermediate numbers.
+        cur.execute("""UPDATE location_property_osmline SET startnumber = null
+                       WHERE interpolationtype in ('odd', 'even')
+                             and endnumber - startnumber = 2""")
+        # Finally add the new column and populate it.
+        cur.execute("ALTER TABLE location_property_osmline ADD COLUMN step SMALLINT")
+        cur.execute("""UPDATE location_property_osmline
+                         SET step = CASE WHEN interpolationtype = 'all'
+                                         THEN 1 ELSE 2 END
+                    """)
+
+
+@_migration(4, 0, 99, 3)
+def add_step_column_for_tiger(conn, **_):
+    """ Add a new column 'step' to the tiger data table.
+    """
+    if conn.table_has_column('location_property_tiger', 'step'):
+        return
+
+    with conn.cursor() as cur:
+        cur.execute("ALTER TABLE location_property_tiger ADD COLUMN step SMALLINT")
+        cur.execute("""UPDATE location_property_tiger
+                         SET step = CASE WHEN interpolationtype = 'all'
+                                         THEN 1 ELSE 2 END
+                    """)
+
+
+@_migration(4, 0, 99, 4)
+def add_derived_name_column_for_country_names(conn, **_):
+    """ Add a new column 'derived_name' which in the future takes the
+        country names as imported from OSM data.
+    """
+    if not conn.table_has_column('country_name', 'derived_name'):
+        with conn.cursor() as cur:
+            cur.execute("ALTER TABLE country_name ADD COLUMN derived_name public.HSTORE")
+
+
+@_migration(4, 0, 99, 5)
+def mark_internal_country_names(conn, config, **_):
+    """ Names from the country table should be marked as internal to prevent
+        them from being deleted. Only necessary for ICU tokenizer.
+    """
+    import psycopg2.extras # pylint: disable=import-outside-toplevel
+
+    tokenizer = tokenizer_factory.get_tokenizer_for_db(config)
+    with tokenizer.name_analyzer() as analyzer:
+        with conn.cursor() as cur:
+            psycopg2.extras.register_hstore(cur)
+            cur.execute("SELECT country_code, name FROM country_name")
+
+            for country_code, names in cur:
+                if not names:
+                    names = {}
+                names['countrycode'] = country_code
+                analyzer.add_country_names(country_code, names)