Merge pull request #3345 from lonvia/simplify-large-geometries

[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 c549f3eb5476e144f3921b3e1d92ea445755f528..e6122bfe4050dd9093f56b76f7486e3196311239 100644 (file)
--- a/test/bdd/steps/steps_db_ops.py
+++ b/test/bdd/steps/steps_db_ops.py
@@ -1,3 +1,10 @@
+# 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 psycopg2.extras
 from itertools import chain
 
 import psycopg2.extras


@@ -5,17 +12,25 @@ import psycopg2.extras
 from place_inserter import PlaceColumn
 from table_compare import NominatimID, DBRow
 
 from place_inserter import PlaceColumn
 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.
     """

-    # place_addressline should not have duplicate (place_id, address_place_id)
-    cur = context.db.cursor()
-    cur.execute("""SELECT count(*) FROM
-                    (SELECT place_id, address_place_id, count(*) as c
-                     FROM place_addressline GROUP BY place_id, address_place_id) x
-                   WHERE c > 1""")
-    assert cur.fetchone()[0] == 0, "Duplicates found in place_addressline"
+    with context.db.cursor() as cur:
+        # place_addressline should not have duplicate (place_id, address_place_id)
+        cur.execute("""SELECT count(*) FROM
+                        (SELECT place_id, address_place_id, count(*) as c
+                         FROM place_addressline GROUP BY place_id, address_place_id) x
+                       WHERE c > 1""")
+        assert cur.fetchone()[0] == 0, "Duplicates found in place_addressline"
+
+        # word table must not have empty word_tokens
+        if context.nominatim.tokenizer != 'legacy':
+            cur.execute("SELECT count(*) FROM word WHERE word_token = ''")
+            assert cur.fetchone()[0] == 0, "Empty word tokens found in word table"
+

 
 
 ################################ GIVEN ##################################
 
 
 ################################ GIVEN ##################################


@@ -37,33 +52,52 @@ def add_data_to_planet_relations(context):
         for tests on data that looks up members.
     """
     with context.db.cursor() as cur:
         for tests on data that looks up members.
     """
     with context.db.cursor() as cur:

-        for r in context.table:
-            last_node = 0
-            last_way = 0
-            parts = []
-            if r['members']:
-                members = []
-                for m in r['members'].split(','):
-                    mid = NominatimID(m)
-                    if mid.typ == 'N':
-                        parts.insert(last_node, int(mid.oid))
-                        last_node += 1
-                        last_way += 1
-                    elif mid.typ == 'W':
-                        parts.insert(last_way, int(mid.oid))
-                        last_way += 1
-                    else:
-                        parts.append(int(mid.oid))
-
-                    members.extend((mid.typ.lower() + mid.oid, mid.cls or ''))
-            else:
-                members = None
-
-            tags = chain.from_iterable([(h[5:], r[h]) for h in r.headings if h.startswith("tags+")])
-
-            cur.execute("""INSERT INTO planet_osm_rels (id, way_off, rel_off, parts, members, tags)
-                           VALUES (%s, %s, %s, %s, %s, %s)""",
-                        (r['id'], last_node, last_way, parts, members, list(tags)))
+        cur.execute("SELECT value FROM osm2pgsql_properties WHERE property = 'db_format'")
+        row = cur.fetchone()
+        if row is None or row[0] == '1':
+            for r in context.table:
+                last_node = 0
+                last_way = 0
+                parts = []
+                if r['members']:
+                    members = []
+                    for m in r['members'].split(','):
+                        mid = NominatimID(m)
+                        if mid.typ == 'N':
+                            parts.insert(last_node, int(mid.oid))
+                            last_node += 1
+                            last_way += 1
+                        elif mid.typ == 'W':
+                            parts.insert(last_way, int(mid.oid))
+                            last_way += 1
+                        else:
+                            parts.append(int(mid.oid))
+
+                        members.extend((mid.typ.lower() + mid.oid, mid.cls or ''))
+                else:
+                    members = None
+
+                tags = chain.from_iterable([(h[5:], r[h]) for h in r.headings if h.startswith("tags+")])
+
+                cur.execute("""INSERT INTO planet_osm_rels (id, way_off, rel_off, parts, members, tags)
+                               VALUES (%s, %s, %s, %s, %s, %s)""",
+                            (r['id'], last_node, last_way, parts, members, list(tags)))
+        else:
+            for r in context.table:
+                if r['members']:
+                    members = []
+                    for m in r['members'].split(','):
+                        mid = NominatimID(m)
+                        members.append({'ref': mid.oid, 'role': mid.cls or '', 'type': mid.typ})
+                else:
+                    members = []
+
+                tags = {h[5:]: r[h] for h in r.headings if h.startswith("tags+")}
+
+                cur.execute("""INSERT INTO planet_osm_rels (id, tags, members)
+                               VALUES (%s, %s, %s)""",
+                            (r['id'], psycopg2.extras.Json(tags),
+                             psycopg2.extras.Json(members)))

 
 @given("the ways")
 def add_data_to_planet_ways(context):
 
 @given("the ways")
 def add_data_to_planet_ways(context):


@@ -71,12 +105,19 @@ def add_data_to_planet_ways(context):
         tests on that that looks up node ids in this table.
     """
     with context.db.cursor() as cur:
         tests on that that looks up node ids in this table.
     """
     with context.db.cursor() as cur:

+        cur.execute("SELECT value FROM osm2pgsql_properties WHERE property = 'db_format'")
+        row = cur.fetchone()
+        json_tags = row is not None and row[0] != '1'

         for r in context.table:
         for r in context.table:

-            tags = chain.from_iterable([(h[5:], r[h]) for h in r.headings if h.startswith("tags+")])
+            if json_tags:
+                tags = psycopg2.extras.Json({h[5:]: r[h] for h in r.headings if h.startswith("tags+")})
+            else:
+                tags = list(chain.from_iterable([(h[5:], r[h])
+                                                 for h in r.headings if h.startswith("tags+")]))

             nodes = [ int(x.strip()) for x in r['nodes'].split(',') ]
 
             cur.execute("INSERT INTO planet_osm_ways (id, nodes, tags) VALUES (%s, %s, %s)",
             nodes = [ int(x.strip()) for x in r['nodes'].split(',') ]
 
             cur.execute("INSERT INTO planet_osm_ways (id, nodes, tags) VALUES (%s, %s, %s)",

-                        (r['id'], nodes, list(tags)))
+                        (r['id'], nodes, tags))

 
 ################################ WHEN ##################################
 
 
 ################################ WHEN ##################################
 


@@ -84,29 +125,44 @@ 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)
-    context.nominatim.run_setup_script('calculate-postcodes', 'index', 'index-noanalyse')
+    context.nominatim.run_nominatim('import', '--continue', 'load-data',
+                                              '--index-noanalyse', '-q',
+                                              '--offline')
+

     check_database_integrity(context)
 
     check_database_integrity(context)
 

+    # Remove the output of the input, when all was right. Otherwise it will be
+    # output when there are errors that had nothing to do with the import
+    # itself.
+    context.log_capture.buffer.clear()
+

 @when("updating places")
 def update_place_table(context):
     """ Update the place table with the given data. Also runs all triggers
         related to updates and reindexes the new data.
     """
 @when("updating places")
 def update_place_table(context):
     """ Update the place table with the given data. Also runs all triggers
         related to updates and reindexes the new data.
     """

-    context.nominatim.run_setup_script(
-        'create-functions', 'create-partition-functions', 'enable-diff-updates')
+    context.nominatim.run_nominatim('refresh', '--functions')

     with context.db.cursor() as cur:
         for row in context.table:
     with context.db.cursor() as cur:
         for row in context.table:

-            PlaceColumn(context).add_row(row, False).db_insert(cur)
+            col = PlaceColumn(context).add_row(row, False)
+            col.db_delete(cur)
+            col.db_insert(cur)
+        cur.execute('SELECT flush_deleted_places()')

 
     context.nominatim.reindex_placex(context.db)
     check_database_integrity(context)
 
 
     context.nominatim.reindex_placex(context.db)
     check_database_integrity(context)
 

+    # Remove the output of the input, when all was right. Otherwise it will be
+    # output when there are errors that had nothing to do with the import
+    # itself.
+    context.log_capture.buffer.clear()
+
+

 @when("updating postcodes")
 def update_postcodes(context):
     """ Rerun the calculation of postcodes.
     """
 @when("updating postcodes")
 def update_postcodes(context):
     """ Rerun the calculation of postcodes.
     """

-    context.nominatim.run_update_script('calculate-postcodes')
+    context.nominatim.run_nominatim('refresh', '--postcodes')

 
 @when("marking for delete (?P<oids>.*)")
 def delete_places(context, oids):
 
 @when("marking for delete (?P<oids>.*)")
 def delete_places(context, oids):


@@ -114,14 +170,20 @@ def delete_places(context, oids):
         separated by commas. Also runs all triggers
         related to updates and reindexes the new data.
     """
         separated by commas. Also runs all triggers
         related to updates and reindexes the new data.
     """

-    context.nominatim.run_setup_script(
-        'create-functions', 'create-partition-functions', 'enable-diff-updates')
+    context.nominatim.run_nominatim('refresh', '--functions')

     with context.db.cursor() as cur:
     with context.db.cursor() as cur:

+        cur.execute('TRUNCATE place_to_be_deleted')

         for oid in oids.split(','):
             NominatimID(oid).query_osm_id(cur, 'DELETE FROM place WHERE {}')
         for oid in oids.split(','):
             NominatimID(oid).query_osm_id(cur, 'DELETE FROM place WHERE {}')

+        cur.execute('SELECT flush_deleted_places()')

 
     context.nominatim.reindex_placex(context.db)
 
 
     context.nominatim.reindex_placex(context.db)
 

+    # Remove the output of the input, when all was right. Otherwise it will be
+    # output when there are errors that had nothing to do with the import
+    # itself.
+    context.log_capture.buffer.clear()
+

 ################################ THEN ##################################
 
 @then("(?P<table>placex|place) contains(?P<exact> exactly)?")
 ################################ THEN ##################################
 
 @then("(?P<table>placex|place) contains(?P<exact> exactly)?")


@@ -154,7 +216,10 @@ def check_place_contents(context, table, exact):
 
         if exact:
             cur.execute('SELECT osm_type, osm_id, class from {}'.format(table))
 
         if exact:
             cur.execute('SELECT osm_type, osm_id, class from {}'.format(table))

-            assert expected_content == set([(r[0], r[1], r[2]) for r in cur])
+            actual = set([(r[0], r[1], r[2]) for r in cur])
+            assert expected_content == actual, \
+                   f"Missing entries: {expected_content - actual}\n" \
+                   f"Not expected in table: {actual - expected_content}"

 
 
 @then("(?P<table>placex|place) has no entry for (?P<oid>.*)")
 
 
 @then("(?P<table>placex|place) has no entry for (?P<oid>.*)")


@@ -176,44 +241,35 @@ 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):


@@ -237,7 +293,7 @@ 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:


@@ -248,24 +304,40 @@ 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 != 'legacy':
+            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):


@@ -297,12 +369,13 @@ 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):


@@ -331,6 +404,51 @@ def check_location_property_osmline(context, oid, neg):
 
             DBRow(oid, res, context).assert_row(row, ('start', 'end'))
 
 
             DBRow(oid, res, context).assert_row(row, ('start', 'end'))
 

-        assert not todo
+        assert not todo, f"Unmatched lines in table: {list(context.table[i] for i in todo)}"
+
+@then("location_property_osmline contains(?P<exact> exactly)?")
+def check_place_contents(context, exact):
+    """ Check contents of the interpolation 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 the 'object' column which must
+        have an identifier of the form '<osm id>[:<startnumber>]'. When multiple
+        rows match (for example because 'startnumber' was left out and there are
+        multiple entries for the given OSM object) then all must match. All
+        expected rows are expected to be present with at least one database row.
+        When 'exactly' is given, there must not be additional rows in the database.
+    """
+    with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
+        expected_content = set()
+        for row in context.table:
+            if ':' in row['object']:
+                nid, start = row['object'].split(':', 2)
+                start = int(start)
+            else:
+                nid, start = row['object'], None
+
+            query = """SELECT *, ST_AsText(linegeo) as geomtxt,
+                              ST_GeometryType(linegeo) as geometrytype
+                       FROM location_property_osmline WHERE osm_id=%s"""
+
+            if ':' in row['object']:
+                query += ' and startnumber = %s'
+                params = [int(val) for val in row['object'].split(':', 2)]
+            else:
+                params = (int(row['object']), )
+
+            cur.execute(query, params)
+            assert cur.rowcount > 0, "No rows found for " + row['object']
+
+            for res in cur:
+                if exact:
+                    expected_content.add((res['osm_id'], res['startnumber']))
+
+                DBRow(nid, res, context).assert_row(row, ['object'])

 
 

+        if exact:
+            cur.execute('SELECT osm_id, startnumber from location_property_osmline')
+            actual = set([(r[0], r[1]) for r in cur])
+            assert expected_content == actual, \
+                   f"Missing entries: {expected_content - actual}\n" \
+                   f"Not expected in table: {actual - expected_content}"