Merge pull request #2614 from lonvia/reorganise-country-names

[nominatim.git] / test / bdd / steps / steps_db_ops.py

diff --git a/test/bdd/steps/steps_db_ops.py b/test/bdd/steps/steps_db_ops.py
index 72a610eb123733db313ee74d510b633afdac5fb3..8df5d6170e741e4807b8ddfac541c72d6437b89e 100644 (file)
--- a/test/bdd/steps/steps_db_ops.py
+++ b/test/bdd/steps/steps_db_ops.py
@@ -1,3 +1,9 @@
+# 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.

 import logging
 from itertools import chain
 
 import logging
 from itertools import chain
 


@@ -7,6 +13,7 @@ from place_inserter import PlaceColumn
 from table_compare import NominatimID, DBRow
 
 from nominatim.indexer import indexer
 from table_compare import NominatimID, DBRow
 
 from nominatim.indexer import indexer

+from nominatim.tokenizer import factory as tokenizer_factory

 
 def check_database_integrity(context):
     """ Check some generic constraints on the tables.
 
 def check_database_integrity(context):
     """ Check some generic constraints on the tables.


@@ -86,26 +93,8 @@ def add_data_to_planet_ways(context):
 def import_and_index_data_from_place_table(context):
     """ Import data previously set up in the place table.
     """
 def import_and_index_data_from_place_table(context):
     """ Import data previously set up in the place table.
     """

-    context.nominatim.copy_from_place(context.db)
-
-    # XXX use tool function as soon as it is ported
-    with context.db.cursor() as cur:
-        with (context.nominatim.src_dir / 'lib-sql' / 'postcode_tables.sql').open('r') as fd:
-            cur.execute(fd.read())
-        cur.execute("""
-            INSERT INTO location_postcode
-             (place_id, indexed_status, country_code, postcode, geometry)
-            SELECT nextval('seq_place'), 1, country_code,
-                   upper(trim (both ' ' from address->'postcode')) as pc,
-                   ST_Centroid(ST_Collect(ST_Centroid(geometry)))
-              FROM placex
-             WHERE address ? 'postcode' AND address->'postcode' NOT SIMILAR TO '%(,|;)%'
-                   AND geometry IS NOT null
-             GROUP BY country_code, pc""")
-
-    # Call directly as the refresh function does not include postcodes.
-    indexer.LOG.setLevel(logging.ERROR)
-    indexer.Indexer(context.nominatim.get_libpq_dsn(), 1).index_full(analyse=False)
+    context.nominatim.run_nominatim('import', '--continue', 'load-data',
+                                              '--index-noanalyse', '-q')

 
     check_database_integrity(context)
 
 
     check_database_integrity(context)
 


@@ -195,44 +184,35 @@ def check_search_name_contents(context, exclude):
         have an identifier of the form '<NRW><osm id>[:<class>]'. All
         expected rows are expected to be present with at least one database row.
     """
         have an identifier of the form '<NRW><osm id>[:<class>]'. All
         expected rows are expected to be present with at least one database row.
     """

-    with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
-        for row in context.table:
-            nid = NominatimID(row['object'])
-            nid.row_by_place_id(cur, 'search_name',
-                                ['ST_X(centroid) as cx', 'ST_Y(centroid) as cy'])
-            assert cur.rowcount > 0, "No rows found for " + row['object']
+    tokenizer = tokenizer_factory.get_tokenizer_for_db(context.nominatim.get_test_config())
+
+    with tokenizer.name_analyzer() as analyzer:
+        with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
+            for row in context.table:
+                nid = NominatimID(row['object'])
+                nid.row_by_place_id(cur, 'search_name',
+                                    ['ST_X(centroid) as cx', 'ST_Y(centroid) as cy'])
+                assert cur.rowcount > 0, "No rows found for " + row['object']
+
+                for res in cur:
+                    db_row = DBRow(nid, res, context)
+                    for name, value in zip(row.headings, row.cells):
+                        if name in ('name_vector', 'nameaddress_vector'):
+                            items = [x.strip() for x in value.split(',')]
+                            tokens = analyzer.get_word_token_info(items)

 
 

-            for res in cur:
-                db_row = DBRow(nid, res, context)
-                for name, value in zip(row.headings, row.cells):
-                    if name in ('name_vector', 'nameaddress_vector'):
-                        items = [x.strip() for x in value.split(',')]
-                        with context.db.cursor() as subcur:
-                            subcur.execute(""" SELECT word_id, word_token
-                                               FROM word, (SELECT unnest(%s::TEXT[]) as term) t
-                                               WHERE word_token = make_standard_name(t.term)
-                                                     and class is null and country_code is null
-                                                     and operator is null
-                                              UNION
-                                               SELECT word_id, word_token
-                                               FROM word, (SELECT unnest(%s::TEXT[]) as term) t
-                                               WHERE word_token = ' ' || make_standard_name(t.term)
-                                                     and class is null and country_code is null
-                                                     and operator is null
-                                           """,
-                                           (list(filter(lambda x: not x.startswith('#'), items)),
-                                            list(filter(lambda x: x.startswith('#'), items))))

                             if not exclude:
                             if not exclude:

-                                assert subcur.rowcount >= len(items), \
-                                    "No word entry found for {}. Entries found: {!s}".format(value, subcur.rowcount)
-                            for wid in subcur:
-                                present = wid[0] in res[name]
+                                assert len(tokens) >= len(items), \
+                                       "No word entry found for {}. Entries found: {!s}".format(value, len(tokens))
+                            for word, token, wid in tokens:

                                 if exclude:
                                 if exclude:

-                                    assert not present, "Found term for {}/{}: {}".format(row['object'], name, wid[1])
+                                    assert wid not in res[name], \
+                                           "Found term for {}/{}: {}".format(nid, name, wid)

                                 else:
                                 else:

-                                    assert present, "Missing term for {}/{}: {}".fromat(row['object'], name, wid[1])
-                    elif name != 'object':
-                        assert db_row.contains(name, value), db_row.assert_msg(name, value)
+                                    assert wid in res[name], \
+                                           "Missing term for {}/{}: {}".format(nid, name, wid)
+                        elif name != 'object':
+                            assert db_row.contains(name, value), db_row.assert_msg(name, value)

 
 @then("search_name has no entry for (?P<oid>.*)")
 def check_search_name_has_entry(context, oid):
 
 @then("search_name has no entry for (?P<oid>.*)")
 def check_search_name_has_entry(context, oid):


@@ -256,7 +236,7 @@ def check_location_postcode(context):
     with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
         cur.execute("SELECT *, ST_AsText(geometry) as geomtxt FROM location_postcode")
         assert cur.rowcount == len(list(context.table)), \
     with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
         cur.execute("SELECT *, ST_AsText(geometry) as geomtxt FROM location_postcode")
         assert cur.rowcount == len(list(context.table)), \

-            "Postcode table has {} rows, expected {}.".foramt(cur.rowcount, len(list(context.table)))
+            "Postcode table has {} rows, expected {}.".format(cur.rowcount, len(list(context.table)))

 
         results = {}
         for row in cur:
 
         results = {}
         for row in cur:


@@ -267,24 +247,40 @@ def check_location_postcode(context):
         for row in context.table:
             db_row = results.get((row['country'],row['postcode']))
             assert db_row is not None, \
         for row in context.table:
             db_row = results.get((row['country'],row['postcode']))
             assert db_row is not None, \

-                "Missing row for country '{r['country']}' postcode '{r['postcode']}'.".format(r=row)
+                f"Missing row for country '{row['country']}' postcode '{row['postcode']}'."

 
             db_row.assert_row(row, ('country', 'postcode'))
 
 
             db_row.assert_row(row, ('country', 'postcode'))
 

-@then("word contains(?P<exclude> not)?")
-def check_word_table(context, exclude):
-    """ Check the contents of the word table. Each row represents a table row
-        and all data must match. Data not present in the expected table, may
-        be arbitry. The rows are identified via all given columns.
+@then("there are(?P<exclude> no)? word tokens for postcodes (?P<postcodes>.*)")
+def check_word_table_for_postcodes(context, exclude, postcodes):
+    """ Check that the tokenizer produces postcode tokens for the given
+        postcodes. The postcodes are a comma-separated list of postcodes.
+        Whitespace matters.

     """
     """

+    nctx = context.nominatim
+    tokenizer = tokenizer_factory.get_tokenizer_for_db(nctx.get_test_config())
+    with tokenizer.name_analyzer() as ana:
+        plist = [ana.normalize_postcode(p) for p in postcodes.split(',')]
+
+    plist.sort()
+

     with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
     with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:

-        for row in context.table:
-            wheres = ' AND '.join(["{} = %s".format(h) for h in row.headings])
-            cur.execute("SELECT * from word WHERE " + wheres, list(row.cells))
-            if exclude:
-                assert cur.rowcount == 0, "Row still in word table: %s" % '/'.join(values)
-            else:
-                assert cur.rowcount > 0, "Row not in word table: %s" % '/'.join(values)
+        if nctx.tokenizer == 'icu':
+            cur.execute("SELECT word FROM word WHERE type = 'P' and word = any(%s)",
+                        (plist,))
+        else:
+            cur.execute("""SELECT word FROM word WHERE word = any(%s)
+                             and class = 'place' and type = 'postcode'""",
+                        (plist,))
+
+        found = [row[0] for row in cur]
+        assert len(found) == len(set(found)), f"Duplicate rows for postcodes: {found}"
+
+    if exclude:
+        assert len(found) == 0, f"Unexpected postcodes: {found}"
+    else:
+        assert set(found) == set(plist), \
+        f"Missing postcodes {set(plist) - set(found)}. Found: {found}"

 
 @then("place_addressline contains")
 def check_place_addressline(context):
 
 @then("place_addressline contains")
 def check_place_addressline(context):


@@ -316,12 +312,13 @@ def check_place_addressline_exclude(context):
     with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
         for row in context.table:
             pid = NominatimID(row['object']).get_place_id(cur)
     with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
         for row in context.table:
             pid = NominatimID(row['object']).get_place_id(cur)

-            apid = NominatimID(row['address']).get_place_id(cur)
-            cur.execute(""" SELECT * FROM place_addressline
-                            WHERE place_id = %s AND address_place_id = %s""",
-                        (pid, apid))
-            assert cur.rowcount == 0, \
-                "Row found for place %s and address %s" % (row['object'], row['address'])
+            apid = NominatimID(row['address']).get_place_id(cur, allow_empty=True)
+            if apid is not None:
+                cur.execute(""" SELECT * FROM place_addressline
+                                WHERE place_id = %s AND address_place_id = %s""",
+                            (pid, apid))
+                assert cur.rowcount == 0, \
+                    "Row found for place %s and address %s" % (row['object'], row['address'])

 
 @then("W(?P<oid>\d+) expands to(?P<neg> no)? interpolation")
 def check_location_property_osmline(context, oid, neg):
 
 @then("W(?P<oid>\d+) expands to(?P<neg> no)? interpolation")
 def check_location_property_osmline(context, oid, neg):