1 # SPDX-License-Identifier: GPL-2.0-only
3 # This file is part of Nominatim. (https://nominatim.org)
5 # Copyright (C) 2022 by the Nominatim developer community.
6 # For a full list of authors see the git log.
8 from itertools import chain
10 import psycopg2.extras
12 from place_inserter import PlaceColumn
13 from table_compare import NominatimID, DBRow
15 from nominatim.indexer import indexer
16 from nominatim.tokenizer import factory as tokenizer_factory
18 def check_database_integrity(context):
19 """ Check some generic constraints on the tables.
21 with context.db.cursor() as cur:
22 # place_addressline should not have duplicate (place_id, address_place_id)
23 cur.execute("""SELECT count(*) FROM
24 (SELECT place_id, address_place_id, count(*) as c
25 FROM place_addressline GROUP BY place_id, address_place_id) x
27 assert cur.fetchone()[0] == 0, "Duplicates found in place_addressline"
29 # word table must not have empty word_tokens
30 if context.nominatim.tokenizer != 'legacy':
31 cur.execute("SELECT count(*) FROM word WHERE word_token = ''")
32 assert cur.fetchone()[0] == 0, "Empty word tokens found in word table"
36 ################################ GIVEN ##################################
38 @given("the (?P<named>named )?places")
39 def add_data_to_place_table(context, named):
40 """ Add entries into the place table. 'named places' makes sure that
41 the entries get a random name when none is explicitly given.
43 with context.db.cursor() as cur:
44 cur.execute('ALTER TABLE place DISABLE TRIGGER place_before_insert')
45 for row in context.table:
46 PlaceColumn(context).add_row(row, named is not None).db_insert(cur)
47 cur.execute('ALTER TABLE place ENABLE TRIGGER place_before_insert')
49 @given("the relations")
50 def add_data_to_planet_relations(context):
51 """ Add entries into the osm2pgsql relation middle table. This is needed
52 for tests on data that looks up members.
54 with context.db.cursor() as cur:
55 for r in context.table:
61 for m in r['members'].split(','):
64 parts.insert(last_node, int(mid.oid))
68 parts.insert(last_way, int(mid.oid))
71 parts.append(int(mid.oid))
73 members.extend((mid.typ.lower() + mid.oid, mid.cls or ''))
77 tags = chain.from_iterable([(h[5:], r[h]) for h in r.headings if h.startswith("tags+")])
79 cur.execute("""INSERT INTO planet_osm_rels (id, way_off, rel_off, parts, members, tags)
80 VALUES (%s, %s, %s, %s, %s, %s)""",
81 (r['id'], last_node, last_way, parts, members, list(tags)))
84 def add_data_to_planet_ways(context):
85 """ Add entries into the osm2pgsql way middle table. This is necessary for
86 tests on that that looks up node ids in this table.
88 with context.db.cursor() as cur:
89 for r in context.table:
90 tags = chain.from_iterable([(h[5:], r[h]) for h in r.headings if h.startswith("tags+")])
91 nodes = [ int(x.strip()) for x in r['nodes'].split(',') ]
93 cur.execute("INSERT INTO planet_osm_ways (id, nodes, tags) VALUES (%s, %s, %s)",
94 (r['id'], nodes, list(tags)))
96 ################################ WHEN ##################################
99 def import_and_index_data_from_place_table(context):
100 """ Import data previously set up in the place table.
102 context.nominatim.run_nominatim('import', '--continue', 'load-data',
103 '--index-noanalyse', '-q',
106 check_database_integrity(context)
108 # Remove the output of the input, when all was right. Otherwise it will be
109 # output when there are errors that had nothing to do with the import
111 context.log_capture.buffer.clear()
113 @when("updating places")
114 def update_place_table(context):
115 """ Update the place table with the given data. Also runs all triggers
116 related to updates and reindexes the new data.
118 context.nominatim.run_nominatim('refresh', '--functions')
119 with context.db.cursor() as cur:
120 for row in context.table:
121 PlaceColumn(context).add_row(row, False).db_insert(cur)
123 context.nominatim.reindex_placex(context.db)
124 check_database_integrity(context)
126 # Remove the output of the input, when all was right. Otherwise it will be
127 # output when there are errors that had nothing to do with the import
129 context.log_capture.buffer.clear()
132 @when("updating postcodes")
133 def update_postcodes(context):
134 """ Rerun the calculation of postcodes.
136 context.nominatim.run_nominatim('refresh', '--postcodes')
138 @when("marking for delete (?P<oids>.*)")
139 def delete_places(context, oids):
140 """ Remove entries from the place table. Multiple ids may be given
141 separated by commas. Also runs all triggers
142 related to updates and reindexes the new data.
144 context.nominatim.run_nominatim('refresh', '--functions')
145 with context.db.cursor() as cur:
146 for oid in oids.split(','):
147 NominatimID(oid).query_osm_id(cur, 'DELETE FROM place WHERE {}')
149 context.nominatim.reindex_placex(context.db)
151 # Remove the output of the input, when all was right. Otherwise it will be
152 # output when there are errors that had nothing to do with the import
154 context.log_capture.buffer.clear()
156 ################################ THEN ##################################
158 @then("(?P<table>placex|place) contains(?P<exact> exactly)?")
159 def check_place_contents(context, table, exact):
160 """ Check contents of place/placex tables. Each row represents a table row
161 and all data must match. Data not present in the expected table, may
162 be arbitry. The rows are identified via the 'object' column which must
163 have an identifier of the form '<NRW><osm id>[:<class>]'. When multiple
164 rows match (for example because 'class' was left out and there are
165 multiple entries for the given OSM object) then all must match. All
166 expected rows are expected to be present with at least one database row.
167 When 'exactly' is given, there must not be additional rows in the database.
169 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
170 expected_content = set()
171 for row in context.table:
172 nid = NominatimID(row['object'])
173 query = 'SELECT *, ST_AsText(geometry) as geomtxt, ST_GeometryType(geometry) as geometrytype'
174 if table == 'placex':
175 query += ' ,ST_X(centroid) as cx, ST_Y(centroid) as cy'
176 query += " FROM %s WHERE {}" % (table, )
177 nid.query_osm_id(cur, query)
178 assert cur.rowcount > 0, "No rows found for " + row['object']
182 expected_content.add((res['osm_type'], res['osm_id'], res['class']))
184 DBRow(nid, res, context).assert_row(row, ['object'])
187 cur.execute('SELECT osm_type, osm_id, class from {}'.format(table))
188 assert expected_content == set([(r[0], r[1], r[2]) for r in cur])
191 @then("(?P<table>placex|place) has no entry for (?P<oid>.*)")
192 def check_place_has_entry(context, table, oid):
193 """ Ensure that no database row for the given object exists. The ID
194 must be of the form '<NRW><osm id>[:<class>]'.
196 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
197 NominatimID(oid).query_osm_id(cur, "SELECT * FROM %s where {}" % table)
198 assert cur.rowcount == 0, \
199 "Found {} entries for ID {}".format(cur.rowcount, oid)
202 @then("search_name contains(?P<exclude> not)?")
203 def check_search_name_contents(context, exclude):
204 """ Check contents of place/placex tables. Each row represents a table row
205 and all data must match. Data not present in the expected table, may
206 be arbitry. The rows are identified via the 'object' column which must
207 have an identifier of the form '<NRW><osm id>[:<class>]'. All
208 expected rows are expected to be present with at least one database row.
210 tokenizer = tokenizer_factory.get_tokenizer_for_db(context.nominatim.get_test_config())
212 with tokenizer.name_analyzer() as analyzer:
213 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
214 for row in context.table:
215 nid = NominatimID(row['object'])
216 nid.row_by_place_id(cur, 'search_name',
217 ['ST_X(centroid) as cx', 'ST_Y(centroid) as cy'])
218 assert cur.rowcount > 0, "No rows found for " + row['object']
221 db_row = DBRow(nid, res, context)
222 for name, value in zip(row.headings, row.cells):
223 if name in ('name_vector', 'nameaddress_vector'):
224 items = [x.strip() for x in value.split(',')]
225 tokens = analyzer.get_word_token_info(items)
228 assert len(tokens) >= len(items), \
229 "No word entry found for {}. Entries found: {!s}".format(value, len(tokens))
230 for word, token, wid in tokens:
232 assert wid not in res[name], \
233 "Found term for {}/{}: {}".format(nid, name, wid)
235 assert wid in res[name], \
236 "Missing term for {}/{}: {}".format(nid, name, wid)
237 elif name != 'object':
238 assert db_row.contains(name, value), db_row.assert_msg(name, value)
240 @then("search_name has no entry for (?P<oid>.*)")
241 def check_search_name_has_entry(context, oid):
242 """ Check that there is noentry in the search_name table for the given
243 objects. IDs are in format '<NRW><osm id>[:<class>]'.
245 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
246 NominatimID(oid).row_by_place_id(cur, 'search_name')
248 assert cur.rowcount == 0, \
249 "Found {} entries for ID {}".format(cur.rowcount, oid)
251 @then("location_postcode contains exactly")
252 def check_location_postcode(context):
253 """ Check full contents for location_postcode table. Each row represents a table row
254 and all data must match. Data not present in the expected table, may
255 be arbitry. The rows are identified via 'country' and 'postcode' columns.
256 All rows must be present as excepted and there must not be additional
259 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
260 cur.execute("SELECT *, ST_AsText(geometry) as geomtxt FROM location_postcode")
261 assert cur.rowcount == len(list(context.table)), \
262 "Postcode table has {} rows, expected {}.".format(cur.rowcount, len(list(context.table)))
266 key = (row['country_code'], row['postcode'])
267 assert key not in results, "Postcode table has duplicate entry: {}".format(row)
268 results[key] = DBRow((row['country_code'],row['postcode']), row, context)
270 for row in context.table:
271 db_row = results.get((row['country'],row['postcode']))
272 assert db_row is not None, \
273 f"Missing row for country '{row['country']}' postcode '{row['postcode']}'."
275 db_row.assert_row(row, ('country', 'postcode'))
277 @then("there are(?P<exclude> no)? word tokens for postcodes (?P<postcodes>.*)")
278 def check_word_table_for_postcodes(context, exclude, postcodes):
279 """ Check that the tokenizer produces postcode tokens for the given
280 postcodes. The postcodes are a comma-separated list of postcodes.
283 nctx = context.nominatim
284 tokenizer = tokenizer_factory.get_tokenizer_for_db(nctx.get_test_config())
285 with tokenizer.name_analyzer() as ana:
286 plist = [ana.normalize_postcode(p) for p in postcodes.split(',')]
290 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
291 if nctx.tokenizer != 'legacy':
292 cur.execute("SELECT word FROM word WHERE type = 'P' and word = any(%s)",
295 cur.execute("""SELECT word FROM word WHERE word = any(%s)
296 and class = 'place' and type = 'postcode'""",
299 found = [row[0] for row in cur]
300 assert len(found) == len(set(found)), f"Duplicate rows for postcodes: {found}"
303 assert len(found) == 0, f"Unexpected postcodes: {found}"
305 assert set(found) == set(plist), \
306 f"Missing postcodes {set(plist) - set(found)}. Found: {found}"
308 @then("place_addressline contains")
309 def check_place_addressline(context):
310 """ Check the contents of the place_addressline table. Each row represents
311 a table row and all data must match. Data not present in the expected
312 table, may be arbitry. The rows are identified via the 'object' column,
313 representing the addressee and the 'address' column, representing the
316 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
317 for row in context.table:
318 nid = NominatimID(row['object'])
319 pid = nid.get_place_id(cur)
320 apid = NominatimID(row['address']).get_place_id(cur)
321 cur.execute(""" SELECT * FROM place_addressline
322 WHERE place_id = %s AND address_place_id = %s""",
324 assert cur.rowcount > 0, \
325 "No rows found for place %s and address %s" % (row['object'], row['address'])
328 DBRow(nid, res, context).assert_row(row, ('address', 'object'))
330 @then("place_addressline doesn't contain")
331 def check_place_addressline_exclude(context):
332 """ Check that the place_addressline doesn't contain any entries for the
333 given addressee/address item pairs.
335 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
336 for row in context.table:
337 pid = NominatimID(row['object']).get_place_id(cur)
338 apid = NominatimID(row['address']).get_place_id(cur, allow_empty=True)
340 cur.execute(""" SELECT * FROM place_addressline
341 WHERE place_id = %s AND address_place_id = %s""",
343 assert cur.rowcount == 0, \
344 "Row found for place %s and address %s" % (row['object'], row['address'])
346 @then("W(?P<oid>\d+) expands to(?P<neg> no)? interpolation")
347 def check_location_property_osmline(context, oid, neg):
348 """ Check that the given way is present in the interpolation table.
350 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
351 cur.execute("""SELECT *, ST_AsText(linegeo) as geomtxt
352 FROM location_property_osmline
353 WHERE osm_id = %s AND startnumber IS NOT NULL""",
357 assert cur.rowcount == 0, "Interpolation found for way {}.".format(oid)
360 todo = list(range(len(list(context.table))))
363 row = context.table[i]
364 if (int(row['start']) == res['startnumber']
365 and int(row['end']) == res['endnumber']):
369 assert False, "Unexpected row " + str(res)
371 DBRow(oid, res, context).assert_row(row, ('start', 'end'))
373 assert not todo, f"Unmatched lines in table: {list(context.table[i] for i in todo)}"