Merge remote-tracking branch 'upstream/master'

[nominatim.git] / src / nominatim_db / tools / check_database.py

diff --git a/src/nominatim_db/tools/check_database.py b/src/nominatim_db/tools/check_database.py
index ef28a0e5a61653b6a2f2c250137050130caff084..79770142494821fc23bee2c63b72940911e43300 100644 (file)
--- a/src/nominatim_db/tools/check_database.py
+++ b/src/nominatim_db/tools/check_database.py
@@ -12,7 +12,8 @@ from enum import Enum
 from textwrap import dedent
 
 from ..config import Configuration
 from textwrap import dedent
 
 from ..config import Configuration

-from ..db.connection import connect, Connection
+from ..db.connection import connect, Connection, server_version_tuple, \
+                            index_exists, table_exists, execute_scalar

 from ..db import properties
 from ..errors import UsageError
 from ..tokenizer import factory as tokenizer_factory
 from ..db import properties
 from ..errors import UsageError
 from ..tokenizer import factory as tokenizer_factory


@@ -21,6 +22,7 @@ from ..version import NOMINATIM_VERSION, parse_version
 
 CHECKLIST = []
 
 
 CHECKLIST = []
 

+

 class CheckState(Enum):
     """ Possible states of a check. FATAL stops check execution entirely.
     """
 class CheckState(Enum):
     """ Possible states of a check. FATAL stops check execution entirely.
     """


@@ -30,9 +32,11 @@ class CheckState(Enum):
     NOT_APPLICABLE = 3
     WARN = 4
 
     NOT_APPLICABLE = 3
     WARN = 4
 

+

 CheckResult = Union[CheckState, Tuple[CheckState, Mapping[str, Any]]]
 CheckFunc = Callable[[Connection, Configuration], CheckResult]
 
 CheckResult = Union[CheckState, Tuple[CheckState, Mapping[str, Any]]]
 CheckFunc = Callable[[Connection, Configuration], CheckResult]
 

+

 def _check(hint: Optional[str] = None) -> Callable[[CheckFunc], CheckFunc]:
     """ Decorator for checks. It adds the function to the list of
         checks to execute and adds the code for printing progress messages.
 def _check(hint: Optional[str] = None) -> Callable[[CheckFunc], CheckFunc]:
     """ Decorator for checks. It adds the function to the list of
         checks to execute and adds the code for printing progress messages.


@@ -67,6 +71,7 @@ def _check(hint: Optional[str] = None) -> Callable[[CheckFunc], CheckFunc]:
 
     return decorator
 
 
     return decorator
 

+

 class _BadConnection:
 
     def __init__(self, msg: str) -> None:
 class _BadConnection:
 
     def __init__(self, msg: str) -> None:


@@ -76,13 +81,14 @@ class _BadConnection:
         """ Dummy function to provide the implementation.
         """
 
         """ Dummy function to provide the implementation.
         """
 

+

 def check_database(config: Configuration) -> int:
     """ Run a number of checks on the database and return the status.
     """
     try:
 def check_database(config: Configuration) -> int:
     """ Run a number of checks on the database and return the status.
     """
     try:

-        conn = connect(config.get_libpq_dsn()).connection
+        conn = connect(config.get_libpq_dsn())

     except UsageError as err:
     except UsageError as err:

-        conn = _BadConnection(str(err)) # type: ignore[assignment]
+        conn = _BadConnection(str(err))  # type: ignore[assignment]

 
     overall_result = 0
     for check in CHECKLIST:
 
     overall_result = 0
     for check in CHECKLIST:


@@ -102,21 +108,26 @@ def _get_indexes(conn: Connection) -> List[str]:
                'idx_placex_rank_search',
                'idx_placex_rank_address',
                'idx_placex_parent_place_id',
                'idx_placex_rank_search',
                'idx_placex_rank_address',
                'idx_placex_parent_place_id',

+               'idx_placex_geometry_reverse_lookupplacenode',

                'idx_placex_geometry_reverse_lookuppolygon',
                'idx_placex_geometry_placenode',
                'idx_osmline_parent_place_id',
                'idx_osmline_parent_osm_id',
                'idx_postcode_id',
                'idx_postcode_postcode'
                'idx_placex_geometry_reverse_lookuppolygon',
                'idx_placex_geometry_placenode',
                'idx_osmline_parent_place_id',
                'idx_osmline_parent_osm_id',
                'idx_postcode_id',
                'idx_postcode_postcode'

-              ]
-    if conn.table_exists('search_name'):
+               ]
+
+    # These won't exist if --reverse-only import was used
+    if table_exists(conn, 'search_name'):

         indexes.extend(('idx_search_name_nameaddress_vector',
                         'idx_search_name_name_vector',
                         'idx_search_name_centroid'))
         indexes.extend(('idx_search_name_nameaddress_vector',
                         'idx_search_name_name_vector',
                         'idx_search_name_centroid'))

-        if conn.server_version_tuple() >= (11, 0, 0):
+        if server_version_tuple(conn) >= (11, 0, 0):

             indexes.extend(('idx_placex_housenumber',
                             'idx_osmline_parent_osm_id_with_hnr'))
             indexes.extend(('idx_placex_housenumber',
                             'idx_osmline_parent_osm_id_with_hnr'))

-    if conn.table_exists('place'):
+
+    # These won't exist if --no-updates import was used
+    if table_exists(conn, 'place'):

         indexes.extend(('idx_location_area_country_place_id',
                         'idx_place_osm_unique',
                         'idx_placex_rank_address_sector',
         indexes.extend(('idx_location_area_country_place_id',
                         'idx_place_osm_unique',
                         'idx_placex_rank_address_sector',


@@ -148,12 +159,13 @@ def check_connection(conn: Any, config: Configuration) -> CheckResult:
 
     return CheckState.OK
 
 
     return CheckState.OK
 

+

 @_check(hint="""\
              Database version ({db_version}) doesn't match Nominatim version ({nom_version})
 
              Hints:
              * Are you connecting to the correct database?
 @_check(hint="""\
              Database version ({db_version}) doesn't match Nominatim version ({nom_version})
 
              Hints:
              * Are you connecting to the correct database?

-             
+

              {instruction}
 
              Check the Migration chapter of the Administration Guide.
              {instruction}
 
              Check the Migration chapter of the Administration Guide.


@@ -165,7 +177,7 @@ def check_database_version(conn: Connection, config: Configuration) -> CheckResu
     """ Checking database_version matches Nominatim software version
     """
 
     """ Checking database_version matches Nominatim software version
     """
 

-    if conn.table_exists('nominatim_properties'):
+    if table_exists(conn, 'nominatim_properties'):

         db_version_str = properties.get_property(conn, 'database_version')
     else:
         db_version_str = None
         db_version_str = properties.get_property(conn, 'database_version')
     else:
         db_version_str = None


@@ -189,6 +201,7 @@ def check_database_version(conn: Connection, config: Configuration) -> CheckResu
                                   instruction=instruction,
                                   config=config)
 
                                   instruction=instruction,
                                   config=config)
 

+

 @_check(hint="""\
              placex table not found
 
 @_check(hint="""\
              placex table not found
 


@@ -202,7 +215,7 @@ def check_database_version(conn: Connection, config: Configuration) -> CheckResu
 def check_placex_table(conn: Connection, config: Configuration) -> CheckResult:
     """ Checking for placex table
     """
 def check_placex_table(conn: Connection, config: Configuration) -> CheckResult:
     """ Checking for placex table
     """

-    if conn.table_exists('placex'):
+    if table_exists(conn, 'placex'):

         return CheckState.OK
 
     return CheckState.FATAL, dict(config=config)
         return CheckState.OK
 
     return CheckState.FATAL, dict(config=config)


@@ -212,8 +225,7 @@ def check_placex_table(conn: Connection, config: Configuration) -> CheckResult:
 def check_placex_size(conn: Connection, _: Configuration) -> CheckResult:
     """ Checking for placex content
     """
 def check_placex_size(conn: Connection, _: Configuration) -> CheckResult:
     """ Checking for placex content
     """

-    with conn.cursor() as cur:
-        cnt = cur.scalar('SELECT count(*) FROM (SELECT * FROM placex LIMIT 100) x')
+    cnt = execute_scalar(conn, 'SELECT count(*) FROM (SELECT * FROM placex LIMIT 100) x')

 
     return CheckState.OK if cnt > 0 else CheckState.FATAL
 
 
     return CheckState.OK if cnt > 0 else CheckState.FATAL
 


@@ -244,16 +256,15 @@ def check_tokenizer(_: Connection, config: Configuration) -> CheckResult:
 def check_existance_wikipedia(conn: Connection, _: Configuration) -> CheckResult:
     """ Checking for wikipedia/wikidata data
     """
 def check_existance_wikipedia(conn: Connection, _: Configuration) -> CheckResult:
     """ Checking for wikipedia/wikidata data
     """

-    if not conn.table_exists('search_name') or not conn.table_exists('place'):
+    if not table_exists(conn, 'search_name') or not table_exists(conn, 'place'):

         return CheckState.NOT_APPLICABLE
 
         return CheckState.NOT_APPLICABLE
 

-    with conn.cursor() as cur:
-        if conn.table_exists('wikimedia_importance'):
-            cnt = cur.scalar('SELECT count(*) FROM wikimedia_importance')
-        else:
-            cnt = cur.scalar('SELECT count(*) FROM wikipedia_article')
+    if table_exists(conn, 'wikimedia_importance'):
+        cnt = execute_scalar(conn, 'SELECT count(*) FROM wikimedia_importance')
+    else:
+        cnt = execute_scalar(conn, 'SELECT count(*) FROM wikipedia_article')

 
 

-        return CheckState.WARN if cnt == 0 else CheckState.OK
+    return CheckState.WARN if cnt == 0 else CheckState.OK

 
 
 @_check(hint="""\
 
 
 @_check(hint="""\


@@ -264,19 +275,18 @@ def check_existance_wikipedia(conn: Connection, _: Configuration) -> CheckResult
 def check_indexing(conn: Connection, _: Configuration) -> CheckResult:
     """ Checking indexing status
     """
 def check_indexing(conn: Connection, _: Configuration) -> CheckResult:
     """ Checking indexing status
     """

-    with conn.cursor() as cur:
-        cnt = cur.scalar('SELECT count(*) FROM placex WHERE indexed_status > 0')
+    cnt = execute_scalar(conn, 'SELECT count(*) FROM placex WHERE indexed_status > 0')

 
     if cnt == 0:
         return CheckState.OK
 
     if freeze.is_frozen(conn):
 
     if cnt == 0:
         return CheckState.OK
 
     if freeze.is_frozen(conn):

-        index_cmd="""\
+        index_cmd = """\

             Database is marked frozen, it cannot be updated.
             Low counts of unindexed places are fine."""
         return CheckState.WARN, dict(count=cnt, index_cmd=index_cmd)
 
             Database is marked frozen, it cannot be updated.
             Low counts of unindexed places are fine."""
         return CheckState.WARN, dict(count=cnt, index_cmd=index_cmd)
 

-    if conn.index_exists('idx_placex_rank_search'):
+    if index_exists(conn, 'idx_placex_rank_search'):

         # Likely just an interrupted update.
         index_cmd = 'nominatim index'
     else:
         # Likely just an interrupted update.
         index_cmd = 'nominatim index'
     else:


@@ -297,7 +307,7 @@ def check_database_indexes(conn: Connection, _: Configuration) -> CheckResult:
     """
     missing = []
     for index in _get_indexes(conn):
     """
     missing = []
     for index in _get_indexes(conn):

-        if not conn.index_exists(index):
+        if not index_exists(conn, index):

             missing.append(index)
 
     if missing:
             missing.append(index)
 
     if missing:


@@ -340,11 +350,10 @@ def check_tiger_table(conn: Connection, config: Configuration) -> CheckResult:
     if not config.get_bool('USE_US_TIGER_DATA'):
         return CheckState.NOT_APPLICABLE
 
     if not config.get_bool('USE_US_TIGER_DATA'):
         return CheckState.NOT_APPLICABLE
 

-    if not conn.table_exists('location_property_tiger'):
+    if not table_exists(conn, 'location_property_tiger'):

         return CheckState.FAIL, dict(error='TIGER data table not found.')
 
         return CheckState.FAIL, dict(error='TIGER data table not found.')
 

-    with conn.cursor() as cur:
-        if cur.scalar('SELECT count(*) FROM location_property_tiger') == 0:
-            return CheckState.FAIL, dict(error='TIGER data table is empty.')
+    if execute_scalar(conn, 'SELECT count(*) FROM location_property_tiger') == 0:
+        return CheckState.FAIL, dict(error='TIGER data table is empty.')

 
     return CheckState.OK
 
     return CheckState.OK