change updates to handle delete/insert workflow

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

diff --git a/nominatim/tools/migration.py b/nominatim/tools/migration.py
index 284bd3166f8648cf18b0554c9368dc0368579ccf..7854154c92c3e7ffb0746775fbe9f852a8bde0de 100644 (file)
--- a/nominatim/tools/migration.py
+++ b/nominatim/tools/migration.py
@@ -7,20 +7,26 @@
 """
 Functions for database migration to newer software versions.
 """
 """
 Functions for database migration to newer software versions.
 """

+from typing import List, Tuple, Callable, Any

 import logging
 
 import logging
 

+from psycopg2 import sql as pysql
+
+from nominatim.config import Configuration

 from nominatim.db import properties
 from nominatim.db import properties

-from nominatim.db.connection import connect
-from nominatim.version import NOMINATIM_VERSION
+from nominatim.db.connection import connect, Connection
+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
 
 LOG = logging.getLogger()
 
 from nominatim.tools import refresh
 from nominatim.tokenizer import factory as tokenizer_factory
 from nominatim.errors import UsageError
 
 LOG = logging.getLogger()
 

-_MIGRATION_FUNCTIONS = []
+VersionTuple = Tuple[int, int, int, int]
+
+_MIGRATION_FUNCTIONS : List[Tuple[VersionTuple, Callable[..., None]]] = []

 
 

-def migrate(config, paths):
+def migrate(config: Configuration, paths: Any) -> int:

     """ Check for the current database version and execute migrations,
         if necesssary.
     """
     """ Check for the current database version and execute migrations,
         if necesssary.
     """


@@ -46,8 +52,9 @@ def migrate(config, paths):
         has_run_migration = False
         for version, func in _MIGRATION_FUNCTIONS:
             if db_version <= version:
         has_run_migration = False
         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))
+                title = func.__doc__ or ''
+                LOG.warning("Running: %s (%s)", title.split('\n', 1)[0],
+                            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()


@@ -59,15 +66,14 @@ 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()
 
     return 0
 
 
 
         conn.commit()
 
     return 0
 
 

-def _guess_version(conn):
+def _guess_version(conn: Connection) -> VersionTuple:

     """ Guess a database version when there is no property table yet.
         Only migrations for 3.6 and later are supported, so bail out
         when the version seems older.
     """ Guess a database version when there is no property table yet.
         Only migrations for 3.6 and later are supported, so bail out
         when the version seems older.


@@ -87,7 +93,8 @@ def _guess_version(conn):
 
 
 
 
 
 

-def _migration(major, minor, patch=0, dbpatch=0):
+def _migration(major: int, minor: int, patch: int = 0,
+               dbpatch: int = 0) -> Callable[[Callable[..., None]], Callable[..., None]]:

     """ Decorator for a single migration step. The parameters describe the
         version after which the migration is applicable, i.e before changing
         from the given version to the next, the migration is required.
     """ Decorator for a single migration step. The parameters describe the
         version after which the migration is applicable, i.e before changing
         from the given version to the next, the migration is required.


@@ -100,7 +107,7 @@ def _migration(major, minor, patch=0, dbpatch=0):
         process, so the migration functions may leave a temporary state behind
         there.
     """
         process, so the migration functions may leave a temporary state behind
         there.
     """

-    def decorator(func):
+    def decorator(func: Callable[..., None]) -> Callable[..., None]:

         _MIGRATION_FUNCTIONS.append(((major, minor, patch, dbpatch), func))
         return func
 
         _MIGRATION_FUNCTIONS.append(((major, minor, patch, dbpatch), func))
         return func
 


@@ -108,7 +115,7 @@ def _migration(major, minor, patch=0, dbpatch=0):
 
 
 @_migration(3, 5, 0, 99)
 
 
 @_migration(3, 5, 0, 99)

-def import_status_timestamp_change(conn, **_):
+def import_status_timestamp_change(conn: Connection, **_: Any) -> None:

     """ Add timezone to timestamp in status table.
 
         The import_status table has been changed to include timezone information
     """ Add timezone to timestamp in status table.
 
         The import_status table has been changed to include timezone information


@@ -120,19 +127,19 @@ def import_status_timestamp_change(conn, **_):
 
 
 @_migration(3, 5, 0, 99)
 
 
 @_migration(3, 5, 0, 99)

-def add_nominatim_property_table(conn, config, **_):
+def add_nominatim_property_table(conn: Connection, config: Configuration, **_: Any) -> None:

     """ Add nominatim_property table.
     """
     if not conn.table_exists('nominatim_properties'):
         with conn.cursor() as cur:
     """ Add nominatim_property table.
     """
     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)
 
 @_migration(3, 6, 0, 0)

-def change_housenumber_transliteration(conn, **_):
+def change_housenumber_transliteration(conn: Connection, **_: Any) -> None:

     """ Transliterate housenumbers.
 
         The database schema switched from saving raw housenumbers in
     """ Transliterate housenumbers.
 
         The database schema switched from saving raw housenumbers in


@@ -163,7 +170,7 @@ def change_housenumber_transliteration(conn, **_):
 
 
 @_migration(3, 7, 0, 0)
 
 
 @_migration(3, 7, 0, 0)

-def switch_placenode_geometry_index(conn, **_):
+def switch_placenode_geometry_index(conn: Connection, **_: Any) -> None:

     """ Replace idx_placex_geometry_reverse_placeNode index.
 
         Make the index slightly more permissive, so that it can also be used
     """ Replace idx_placex_geometry_reverse_placeNode index.
 
         Make the index slightly more permissive, so that it can also be used


@@ -180,7 +187,7 @@ def switch_placenode_geometry_index(conn, **_):
 
 
 @_migration(3, 7, 0, 1)
 
 
 @_migration(3, 7, 0, 1)

-def install_legacy_tokenizer(conn, config, **_):
+def install_legacy_tokenizer(conn: Connection, config: Configuration, **_: Any) -> None:

     """ Setup legacy tokenizer.
 
         If no other tokenizer has been configured yet, then create the
     """ Setup legacy tokenizer.
 
         If no other tokenizer has been configured yet, then create the


@@ -194,15 +201,16 @@ 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')
 

-        tokenizer.migrate_database(config)
+        tokenizer.migrate_database(config) # type: ignore[attr-defined]

 
 
 @_migration(4, 0, 99, 0)
 
 
 @_migration(4, 0, 99, 0)

-def create_tiger_housenumber_index(conn, **_):
+def create_tiger_housenumber_index(conn: Connection, **_: Any) -> None:

     """ Create idx_location_property_tiger_parent_place_id with included
         house number.
 
     """ Create idx_location_property_tiger_parent_place_id with included
         house number.
 


@@ -219,7 +227,7 @@ def create_tiger_housenumber_index(conn, **_):
 
 
 @_migration(4, 0, 99, 1)
 
 
 @_migration(4, 0, 99, 1)

-def create_interpolation_index_on_place(conn, **_):
+def create_interpolation_index_on_place(conn: Connection, **_: Any) -> None:

     """ Create idx_place_interpolations for lookup of interpolation lines
         on updates.
     """
     """ Create idx_place_interpolations for lookup of interpolation lines
         on updates.
     """


@@ -227,3 +235,83 @@ def create_interpolation_index_on_place(conn, **_):
         cur.execute("""CREATE INDEX IF NOT EXISTS idx_place_interpolations
                        ON place USING gist(geometry)
                        WHERE osm_type = 'W' and address ? 'interpolation'""")
         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: Connection, **_: Any) -> None:
+    """ Add a new column 'step' to the interpolations table.
+
+        Also converts 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: Connection, **_: Any) -> None:
+    """ 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: Connection, **_: Any) -> None:
+    """ 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: Connection, config: Configuration, **_: Any) -> None:
+    """ 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)