X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/7f4e7a257940088e17046487f618e59f6aeb8ce9..4aba70caeef6f574e59ad05972ac4981d07ea26d:/test/bdd/steps/db_ops.py diff --git a/test/bdd/steps/db_ops.py b/test/bdd/steps/db_ops.py index da65984d..078e29fe 100644 --- a/test/bdd/steps/db_ops.py +++ b/test/bdd/steps/db_ops.py @@ -2,13 +2,14 @@ import base64 import random import string import re -from nose.tools import * # for assert functions import psycopg2.extras +from check_functions import Almost + class PlaceColumn: def __init__(self, context, force_name): - self.columns = { 'admin_level' : 100} + self.columns = { 'admin_level' : 15} self.force_name = force_name self.context = context self.geometry = None @@ -20,17 +21,20 @@ class PlaceColumn: self.add_hstore('name', key[5:], value) elif key.startswith('extra+'): self.add_hstore('extratags', key[6:], value) + elif key.startswith('addr+'): + self.add_hstore('address', key[5:], value) + elif key in ('name', 'address', 'extratags'): + self.columns[key] = eval('{' + value + '}') else: - assert_in(key, ('class', 'type', 'street', 'addr_place', - 'isin', 'postcode')) + assert key in ('class', 'type') self.columns[key] = None if value == '' else value def set_key_name(self, value): self.add_hstore('name', 'name', value) def set_key_osm(self, value): - assert_in(value[0], 'NRW') - ok_(value[1:].isdigit()) + assert value[0] in 'NRW' + assert value[1:].isdigit() self.columns['osm_type'] = value[0] self.columns['osm_id'] = int(value[1:]) @@ -39,14 +43,28 @@ class PlaceColumn: self.columns['admin_level'] = int(value) def set_key_housenr(self, value): - self.columns['housenumber'] = None if value == '' else value + if value: + self.add_hstore('address', 'housenumber', value) + + def set_key_postcode(self, value): + if value: + self.add_hstore('address', 'postcode', value) + + def set_key_street(self, value): + if value: + self.add_hstore('address', 'street', value) + + def set_key_addr_place(self, value): + if value: + self.add_hstore('address', 'place', value) def set_key_country(self, value): - self.columns['country_code'] = None if value == '' else value + if value: + self.add_hstore('address', 'country', value) def set_key_geometry(self, value): self.geometry = self.context.osm.parse_geometry(value, self.context.scene) - assert_is_not_none(self.geometry) + assert self.geometry is not None def add_hstore(self, column, key, value): if column in self.columns: @@ -55,22 +73,81 @@ class PlaceColumn: self.columns[column] = { key : value } def db_insert(self, cursor): - assert_in('osm_type', self.columns) + assert 'osm_type' in self.columns if self.force_name and 'name' not in self.columns: self.add_hstore('name', 'name', ''.join(random.choice(string.printable) for _ in range(int(random.random()*30)))) if self.columns['osm_type'] == 'N' and self.geometry is None: - self.geometry = "ST_SetSRID(ST_Point(%f, %f), 4326)" % ( - random.random()*360 - 180, random.random()*180 - 90) + pt = self.context.osm.grid_node(self.columns['osm_id']) + if pt is None: + pt = (random.random()*360 - 180, random.random()*180 - 90) + + self.geometry = "ST_SetSRID(ST_Point(%f, %f), 4326)" % pt else: - assert_is_not_none(self.geometry, "Geometry missing") + assert self.geometry is not None, "Geometry missing" query = 'INSERT INTO place (%s, geometry) values(%s, %s)' % ( ','.join(self.columns.keys()), ','.join(['%s' for x in range(len(self.columns))]), self.geometry) cursor.execute(query, list(self.columns.values())) +class LazyFmt(object): + + def __init__(self, fmtstr, *args): + self.fmt = fmtstr + self.args = args + + def __str__(self): + return self.fmt % self.args + +class PlaceObjName(object): + + def __init__(self, placeid, conn): + self.pid = placeid + self.conn = conn + + def __str__(self): + if self.pid is None: + return "" + + if self.pid == 0: + return "place ID 0" + + cur = self.conn.cursor() + cur.execute("""SELECT osm_type, osm_id, class + FROM placex WHERE place_id = %s""", + (self.pid, )) + assert cur.rowcount == 1, "No entry found for place id %s" % self.pid + + return "%s%s:%s" % cur.fetchone() + +def compare_place_id(expected, result, column, context): + if expected == '0': + assert result == 0, \ + LazyFmt("Bad place id in column %s. Expected: 0, got: %s.", + column, PlaceObjName(result, context.db)) + elif expected == '-': + assert result is None, \ + LazyFmt("bad place id in column %s: %s.", + column, PlaceObjName(result, context.db)) + else: + assert NominatimID(expected).get_place_id(context.db.cursor()) == result, \ + LazyFmt("Bad place id in column %s. Expected: %s, got: %s.", + column, expected, PlaceObjName(result, context.db)) + +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" + + class NominatimID: """ Splits a unique identifier for places into its components. As place_ids cannot be used for testing, we use a unique @@ -84,12 +161,18 @@ class NominatimID: if oid is not None: m = self.id_regex.fullmatch(oid) - assert_is_not_none(m, "ID '%s' not of form [:]" % oid) + assert m is not None, "ID '%s' not of form [:]" % oid self.typ = m.group('tp') self.oid = m.group('id') self.cls = m.group('cls') + def __str__(self): + if self.cls is None: + return self.typ + self.oid + + return '%s%d:%s' % (self.typ, self.oid, self.cls) + def table_select(self): """ Return where clause and parameter list to select the object from a Nominatim table. @@ -98,7 +181,7 @@ class NominatimID: params = [self.typ, self. oid] if self.cls is not None: - where += ' class = %s' + where += ' and class = %s' params.append(self.cls) return where, params @@ -106,7 +189,8 @@ class NominatimID: def get_place_id(self, cur): where, params = self.table_select() cur.execute("SELECT place_id FROM placex WHERE %s" % where, params) - eq_(1, cur.rowcount, "Expected exactly 1 entry in placex found %s" % cur.rowcount) + assert cur.rowcount == 1, \ + "Expected exactly 1 entry in placex for %s found %s" % (str(self), cur.rowcount) return cur.fetchone()[0] @@ -116,20 +200,30 @@ def assert_db_column(row, column, value, context): return if column.startswith('centroid'): - fac = float(column[9:]) if h.startswith('centroid*') else 1.0 - x, y = value.split(' ') - assert_almost_equal(float(x) * fac, row['cx']) - assert_almost_equal(float(y) * fac, row['cy']) + if value == 'in geometry': + query = """SELECT ST_Within(ST_SetSRID(ST_Point({}, {}), 4326), + ST_SetSRID('{}'::geometry, 4326))""".format( + row['cx'], row['cy'], row['geomtxt']) + cur = context.db.cursor() + cur.execute(query) + assert cur.fetchone()[0], "(Row %s failed: %s)" % (column, query) + else: + fac = float(column[9:]) if column.startswith('centroid*') else 1.0 + x, y = value.split(' ') + assert Almost(float(x) * fac) == row['cx'], "Bad x coordinate" + assert Almost(float(y) * fac) == row['cy'], "Bad y coordinate" elif column == 'geometry': geom = context.osm.parse_geometry(value, context.scene) cur = context.db.cursor() - cur.execute("SELECT ST_MaxDistance(%s, ST_SetSRID(%%s::geometry, 4326))" % geom, - (row['geomtxt'],)) - assert_less(cur.fetchone()[0], 0.005) + query = "SELECT ST_Equals(ST_SnapToGrid(%s, 0.00001, 0.00001), ST_SnapToGrid(ST_SetSRID('%s'::geometry, 4326), 0.00001, 0.00001))" % ( + geom, row['geomtxt'],) + cur.execute(query) + assert cur.fetchone()[0], "(Row %s failed: %s)" % (column, query) + elif value == '-': + assert row[column] is None, "Row %s" % column else: - eq_(value, str(row[column]), - "Row '%s': expected: %s, got: %s" - % (column, value, str(row[column]))) + assert value == str(row[column]), \ + "Row '%s': expected: %s, got: %s" % (column, value, str(row[column])) ################################ STEPS ################################## @@ -208,18 +302,64 @@ def import_and_index_data_from_place_table(context): context.nominatim.run_setup_script('create-functions', 'create-partition-functions') cur = context.db.cursor() cur.execute( - """insert into placex (osm_type, osm_id, class, type, name, admin_level, - housenumber, street, addr_place, isin, postcode, country_code, extratags, - geometry) - select * from place where not (class='place' and type='houses' and osm_type='W')""") + """insert into placex (osm_type, osm_id, class, type, name, admin_level, address, extratags, geometry) + select osm_type, osm_id, class, type, name, admin_level, address, extratags, geometry + from place where not (class='place' and type='houses' and osm_type='W')""") cur.execute( - """select insert_osmline (osm_id, housenumber, street, addr_place, - postcode, country_code, geometry) - from place where class='place' and type='houses' and osm_type='W'""") + """insert into location_property_osmline (osm_id, address, linegeo) + SELECT osm_id, address, geometry from place + WHERE class='place' and type='houses' and osm_type='W' + and ST_GeometryType(geometry) = 'ST_LineString'""") + context.db.commit() + context.nominatim.run_setup_script('calculate-postcodes', 'index', 'index-noanalyse') + check_database_integrity(context) + +@when("updating places") +def update_place_table(context): + context.nominatim.run_setup_script( + 'create-functions', 'create-partition-functions', 'enable-diff-updates') + cur = context.db.cursor() + for r in context.table: + col = PlaceColumn(context, False) + + for h in r.headings: + col.add(h, r[h]) + + col.db_insert(cur) + + context.db.commit() + + while True: + context.nominatim.run_update_script('index') + + cur = context.db.cursor() + cur.execute("SELECT 'a' FROM placex WHERE indexed_status != 0 LIMIT 1") + if cur.rowcount == 0: + break + + check_database_integrity(context) + +@when("updating postcodes") +def update_postcodes(context): + context.nominatim.run_update_script('calculate-postcodes') + +@when("marking for delete (?P.*)") +def delete_places(context, oids): + context.nominatim.run_setup_script( + 'create-functions', 'create-partition-functions', 'enable-diff-updates') + cur = context.db.cursor() + for oid in oids.split(','): + where, params = NominatimID(oid).table_select() + cur.execute("DELETE FROM place WHERE " + where, params) context.db.commit() - context.nominatim.run_setup_script('index', 'index-noanalyse') + while True: + context.nominatim.run_update_script('index') + cur = context.db.cursor() + cur.execute("SELECT 'a' FROM placex WHERE indexed_status != 0 LIMIT 1") + if cur.rowcount == 0: + break @then("placex contains(?P exactly)?") def check_placex_contents(context, exact): @@ -233,74 +373,217 @@ def check_placex_contents(context, exact): ST_X(centroid) as cx, ST_Y(centroid) as cy FROM placex where %s""" % where, params) + assert cur.rowcount > 0, "No rows found for " + row['object'] for res in cur: if exact: expected_content.add((res['osm_type'], res['osm_id'], res['class'])) for h in row.headings: - if h.startswith('name'): + if h in ('extratags', 'address'): + if row[h] == '-': + assert res[h] is None + else: + vdict = eval('{' + row[h] + '}') + assert vdict == res[h] + elif h.startswith('name'): name = h[5:] if h.startswith('name+') else 'name' - assert_in(name, res['name']) - eq_(res['name'][name], row[h]) + assert name in res['name'] + assert res['name'][name] == row[h] elif h.startswith('extratags+'): - eq_(res['extratags'][h[10:]], row[h]) - elif h in ('linked_place_id', 'parent_place_id'): - if row[h] == '0': - eq_(0, res[h]) - elif row[h] == '-': - assert_is_none(res[h]) + assert res['extratags'][h[10:]] == row[h] + elif h.startswith('addr+'): + if row[h] == '-': + if res['address'] is not None: + assert h[5:] not in res['address'] else: - eq_(NominatimID(row[h]).get_place_id(context.db.cursor()), - res[h]) + assert h[5:] in res['address'], "column " + h + assert res['address'][h[5:]] == row[h], "column %s" % h + elif h in ('linked_place_id', 'parent_place_id'): + compare_place_id(row[h], res[h], h, context) else: assert_db_column(res, h, row[h], context) if exact: cur.execute('SELECT osm_type, osm_id, class from placex') - eq_(expected_content, set([(r[0], r[1], r[2]) for r in cur])) + assert expected_content == set([(r[0], r[1], r[2]) for r in cur]) + + context.db.commit() + +@then("place contains(?P exactly)?") +def check_placex_contents(context, exact): + cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) + + expected_content = set() + for row in context.table: + nid = NominatimID(row['object']) + where, params = nid.table_select() + cur.execute("""SELECT *, ST_AsText(geometry) as geomtxt, + ST_GeometryType(geometry) as geometrytype + FROM place where %s""" % where, + params) + assert cur.rowcount > 0, "No rows found for " + row['object'] + + for res in cur: + if exact: + expected_content.add((res['osm_type'], res['osm_id'], res['class'])) + for h in row.headings: + msg = "%s: %s" % (row['object'], h) + if h in ('name', 'extratags', 'address'): + if row[h] == '-': + assert res[h] is None, msg + else: + vdict = eval('{' + row[h] + '}') + assert vdict == res[h], msg + elif h.startswith('name+'): + assert res['name'][h[5:]] == row[h], msg + elif h.startswith('extratags+'): + assert res['extratags'][h[10:]] == row[h], msg + elif h.startswith('addr+'): + if row[h] == '-': + if res['address'] is not None: + assert h[5:] not in res['address'] + else: + assert res['address'][h[5:]] == row[h], msg + elif h in ('linked_place_id', 'parent_place_id'): + compare_place_id(row[h], res[h], h, context) + else: + assert_db_column(res, h, row[h], context) + + if exact: + cur.execute('SELECT osm_type, osm_id, class from place') + assert expected_content, set([(r[0], r[1], r[2]) for r in cur]) context.db.commit() -@then("search_name contains") -def check_search_name_contents(context): +@then("search_name contains(?P not)?") +def check_search_name_contents(context, exclude): cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) for row in context.table: pid = NominatimID(row['object']).get_place_id(cur) cur.execute("""SELECT *, ST_X(centroid) as cx, ST_Y(centroid) as cy FROM search_name WHERE place_id = %s""", (pid, )) + assert cur.rowcount > 0, "No rows found for " + row['object'] for res in cur: for h in row.headings: if h in ('name_vector', 'nameaddress_vector'): - terms = [x.strip().replace('#', ' ') for x in row[h].split(',')] + terms = [x.strip() for x in row[h].split(',') if not x.strip().startswith('#')] + words = [x.strip()[1:] for x in row[h].split(',') if x.strip().startswith('#')] subcur = context.db.cursor() - subcur.execute("""SELECT word_id, word_token - FROM word, (SELECT unnest(%s) as term) t - WHERE word_token = make_standard_name(t.term)""", - (terms,)) - ok_(subcur.rowcount >= len(terms)) + 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 + """, + (terms, words)) + if not exclude: + assert subcur.rowcount >= len(terms) + len(words), \ + "No word entry found for " + row[h] + ". Entries found: " + str(subcur.rowcount) for wid in subcur: - assert_in(wid[0], res[h], - "Missing term for %s/%s: %s" % (pid, h, wid[1])) + if exclude: + assert wid[0] not in res[h], "Found term for %s/%s: %s" % (pid, h, wid[1]) + else: + assert wid[0] in res[h], "Missing term for %s/%s: %s" % (pid, h, wid[1]) else: assert_db_column(res, h, row[h], context) context.db.commit() -@then("(?P\w+) expands to interpolation") -def check_location_property_osmline(context, oid): +@then("location_postcode contains exactly") +def check_location_postcode(context): + cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) + + cur.execute("SELECT *, ST_AsText(geometry) as geomtxt FROM location_postcode") + assert cur.rowcount == len(list(context.table)), \ + "Postcode table has %d rows, expected %d rows." % (cur.rowcount, len(list(context.table))) + + table = list(cur) + for row in context.table: + for i in range(len(table)): + if table[i]['country_code'] != row['country'] \ + or table[i]['postcode'] != row['postcode']: + continue + for h in row.headings: + if h not in ('country', 'postcode'): + assert_db_column(table[i], h, row[h], context) + +@then("word contains(?P not)?") +def check_word_table(context, exclude): + cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) + + for row in context.table: + wheres = [] + values = [] + for h in row.headings: + wheres.append("%s = %%s" % h) + values.append(row[h]) + cur.execute("SELECT * from word WHERE %s" % ' AND '.join(wheres), values) + 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) + +@then("place_addressline contains") +def check_place_addressline(context): + cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) + + 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, \ + "No rows found for place %s and address %s" % (row['object'], row['address']) + + for res in cur: + for h in row.headings: + if h not in ('address', 'object'): + assert_db_column(res, h, row[h], context) + + context.db.commit() + +@then("place_addressline doesn't contain") +def check_place_addressline_exclude(context): + cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) + + 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']) + + context.db.commit() + +@then("(?P\w+) expands to(?P no)? interpolation") +def check_location_property_osmline(context, oid, neg): cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) nid = NominatimID(oid) - eq_('W', nid.typ, "interpolation must be a way") + assert 'W' == nid.typ, "interpolation must be a way" cur.execute("""SELECT *, ST_AsText(linegeo) as geomtxt - FROM location_property_osmline WHERE osm_id = %s""", + FROM location_property_osmline + WHERE osm_id = %s AND startnumber IS NOT NULL""", (nid.oid, )) - todo = list(range(len([context.table]))) + if neg: + assert cur.rowcount == 0 + return + + todo = list(range(len(list(context.table)))) for res in cur: for i in todo: row = context.table[i] @@ -309,32 +592,26 @@ def check_location_property_osmline(context, oid): todo.remove(i) break else: - assert(False, "Unexpected row %s" % (str(res))) + assert False, "Unexpected row %s" % (str(res)) for h in row.headings: if h in ('start', 'end'): continue elif h == 'parent_place_id': - if row[h] == '0': - eq_(0, res[h]) - elif row[h] == '-': - assert_is_none(res[h]) - else: - eq_(NominatimID(row[h]).get_place_id(context.db.cursor()), - res[h]) + compare_place_id(row[h], res[h], h, context) else: assert_db_column(res, h, row[h], context) - eq_(todo, []) + assert not todo -@then("placex has no entry for (?P.*)") -def check_placex_has_entry(context, oid): +@then("(?Pplacex|place) has no entry for (?P.*)") +def check_placex_has_entry(context, table, oid): cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) nid = NominatimID(oid) where, params = nid.table_select() - cur.execute("SELECT * FROM placex where %s" % where, params) - eq_(0, cur.rowcount) + cur.execute("SELECT * FROM %s where %s" % (table, where), params) + assert cur.rowcount == 0 context.db.commit() @then("search_name has no entry for (?P.*)") @@ -342,5 +619,5 @@ def check_search_name_has_entry(context, oid): cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) pid = NominatimID(oid).get_place_id(cur) cur.execute("SELECT * FROM search_name WHERE place_id = %s", (pid, )) - eq_(0, cur.rowcount) + assert cur.rowcount == 0 context.db.commit()