2 from itertools import chain
6 from place_inserter import PlaceColumn
7 from table_compare import NominatimID, DBRow
9 from nominatim.indexer import indexer
10 from nominatim.tokenizer import factory as tokenizer_factory
12 def check_database_integrity(context):
13 """ Check some generic constraints on the tables.
15 # place_addressline should not have duplicate (place_id, address_place_id)
16 cur = context.db.cursor()
17 cur.execute("""SELECT count(*) FROM
18 (SELECT place_id, address_place_id, count(*) as c
19 FROM place_addressline GROUP BY place_id, address_place_id) x
21 assert cur.fetchone()[0] == 0, "Duplicates found in place_addressline"
24 ################################ GIVEN ##################################
26 @given("the (?P<named>named )?places")
27 def add_data_to_place_table(context, named):
28 """ Add entries into the place table. 'named places' makes sure that
29 the entries get a random name when none is explicitly given.
31 with context.db.cursor() as cur:
32 cur.execute('ALTER TABLE place DISABLE TRIGGER place_before_insert')
33 for row in context.table:
34 PlaceColumn(context).add_row(row, named is not None).db_insert(cur)
35 cur.execute('ALTER TABLE place ENABLE TRIGGER place_before_insert')
37 @given("the relations")
38 def add_data_to_planet_relations(context):
39 """ Add entries into the osm2pgsql relation middle table. This is needed
40 for tests on data that looks up members.
42 with context.db.cursor() as cur:
43 for r in context.table:
49 for m in r['members'].split(','):
52 parts.insert(last_node, int(mid.oid))
56 parts.insert(last_way, int(mid.oid))
59 parts.append(int(mid.oid))
61 members.extend((mid.typ.lower() + mid.oid, mid.cls or ''))
65 tags = chain.from_iterable([(h[5:], r[h]) for h in r.headings if h.startswith("tags+")])
67 cur.execute("""INSERT INTO planet_osm_rels (id, way_off, rel_off, parts, members, tags)
68 VALUES (%s, %s, %s, %s, %s, %s)""",
69 (r['id'], last_node, last_way, parts, members, list(tags)))
72 def add_data_to_planet_ways(context):
73 """ Add entries into the osm2pgsql way middle table. This is necessary for
74 tests on that that looks up node ids in this table.
76 with context.db.cursor() as cur:
77 for r in context.table:
78 tags = chain.from_iterable([(h[5:], r[h]) for h in r.headings if h.startswith("tags+")])
79 nodes = [ int(x.strip()) for x in r['nodes'].split(',') ]
81 cur.execute("INSERT INTO planet_osm_ways (id, nodes, tags) VALUES (%s, %s, %s)",
82 (r['id'], nodes, list(tags)))
84 ################################ WHEN ##################################
87 def import_and_index_data_from_place_table(context):
88 """ Import data previously set up in the place table.
90 nctx = context.nominatim
92 tokenizer = tokenizer_factory.create_tokenizer(nctx.get_test_config())
93 context.nominatim.copy_from_place(context.db)
95 # XXX use tool function as soon as it is ported
96 with context.db.cursor() as cur:
97 with (context.nominatim.src_dir / 'lib-sql' / 'postcode_tables.sql').open('r') as fd:
98 cur.execute(fd.read())
100 INSERT INTO location_postcode
101 (place_id, indexed_status, country_code, postcode, geometry)
102 SELECT nextval('seq_place'), 1, country_code,
103 upper(trim (both ' ' from address->'postcode')) as pc,
104 ST_Centroid(ST_Collect(ST_Centroid(geometry)))
106 WHERE address ? 'postcode' AND address->'postcode' NOT SIMILAR TO '%(,|;)%'
107 AND geometry IS NOT null
108 GROUP BY country_code, pc""")
110 # Call directly as the refresh function does not include postcodes.
111 indexer.LOG.setLevel(logging.ERROR)
112 indexer.Indexer(context.nominatim.get_libpq_dsn(), tokenizer, 1).index_full(analyse=False)
114 check_database_integrity(context)
116 @when("updating places")
117 def update_place_table(context):
118 """ Update the place table with the given data. Also runs all triggers
119 related to updates and reindexes the new data.
121 context.nominatim.run_nominatim('refresh', '--functions')
122 with context.db.cursor() as cur:
123 for row in context.table:
124 PlaceColumn(context).add_row(row, False).db_insert(cur)
126 context.nominatim.reindex_placex(context.db)
127 check_database_integrity(context)
129 @when("updating postcodes")
130 def update_postcodes(context):
131 """ Rerun the calculation of postcodes.
133 context.nominatim.run_nominatim('refresh', '--postcodes')
135 @when("marking for delete (?P<oids>.*)")
136 def delete_places(context, oids):
137 """ Remove entries from the place table. Multiple ids may be given
138 separated by commas. Also runs all triggers
139 related to updates and reindexes the new data.
141 context.nominatim.run_nominatim('refresh', '--functions')
142 with context.db.cursor() as cur:
143 for oid in oids.split(','):
144 NominatimID(oid).query_osm_id(cur, 'DELETE FROM place WHERE {}')
146 context.nominatim.reindex_placex(context.db)
148 ################################ THEN ##################################
150 @then("(?P<table>placex|place) contains(?P<exact> exactly)?")
151 def check_place_contents(context, table, exact):
152 """ Check contents of place/placex tables. Each row represents a table row
153 and all data must match. Data not present in the expected table, may
154 be arbitry. The rows are identified via the 'object' column which must
155 have an identifier of the form '<NRW><osm id>[:<class>]'. When multiple
156 rows match (for example because 'class' was left out and there are
157 multiple entries for the given OSM object) then all must match. All
158 expected rows are expected to be present with at least one database row.
159 When 'exactly' is given, there must not be additional rows in the database.
161 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
162 expected_content = set()
163 for row in context.table:
164 nid = NominatimID(row['object'])
165 query = 'SELECT *, ST_AsText(geometry) as geomtxt, ST_GeometryType(geometry) as geometrytype'
166 if table == 'placex':
167 query += ' ,ST_X(centroid) as cx, ST_Y(centroid) as cy'
168 query += " FROM %s WHERE {}" % (table, )
169 nid.query_osm_id(cur, query)
170 assert cur.rowcount > 0, "No rows found for " + row['object']
174 expected_content.add((res['osm_type'], res['osm_id'], res['class']))
176 DBRow(nid, res, context).assert_row(row, ['object'])
179 cur.execute('SELECT osm_type, osm_id, class from {}'.format(table))
180 assert expected_content == set([(r[0], r[1], r[2]) for r in cur])
183 @then("(?P<table>placex|place) has no entry for (?P<oid>.*)")
184 def check_place_has_entry(context, table, oid):
185 """ Ensure that no database row for the given object exists. The ID
186 must be of the form '<NRW><osm id>[:<class>]'.
188 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
189 NominatimID(oid).query_osm_id(cur, "SELECT * FROM %s where {}" % table)
190 assert cur.rowcount == 0, \
191 "Found {} entries for ID {}".format(cur.rowcount, oid)
194 @then("search_name contains(?P<exclude> not)?")
195 def check_search_name_contents(context, exclude):
196 """ Check contents of place/placex tables. Each row represents a table row
197 and all data must match. Data not present in the expected table, may
198 be arbitry. The rows are identified via the 'object' column which must
199 have an identifier of the form '<NRW><osm id>[:<class>]'. All
200 expected rows are expected to be present with at least one database row.
202 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
203 for row in context.table:
204 nid = NominatimID(row['object'])
205 nid.row_by_place_id(cur, 'search_name',
206 ['ST_X(centroid) as cx', 'ST_Y(centroid) as cy'])
207 assert cur.rowcount > 0, "No rows found for " + row['object']
210 db_row = DBRow(nid, res, context)
211 for name, value in zip(row.headings, row.cells):
212 if name in ('name_vector', 'nameaddress_vector'):
213 items = [x.strip() for x in value.split(',')]
214 with context.db.cursor() as subcur:
215 subcur.execute(""" SELECT word_id, word_token
216 FROM word, (SELECT unnest(%s::TEXT[]) as term) t
217 WHERE word_token = make_standard_name(t.term)
218 and class is null and country_code is null
221 SELECT word_id, word_token
222 FROM word, (SELECT unnest(%s::TEXT[]) as term) t
223 WHERE word_token = ' ' || make_standard_name(t.term)
224 and class is null and country_code is null
227 (list(filter(lambda x: not x.startswith('#'), items)),
228 list(filter(lambda x: x.startswith('#'), items))))
230 assert subcur.rowcount >= len(items), \
231 "No word entry found for {}. Entries found: {!s}".format(value, subcur.rowcount)
233 present = wid[0] in res[name]
235 assert not present, "Found term for {}/{}: {}".format(row['object'], name, wid[1])
237 assert present, "Missing term for {}/{}: {}".format(row['object'], name, wid[1])
238 elif name != 'object':
239 assert db_row.contains(name, value), db_row.assert_msg(name, value)
241 @then("search_name has no entry for (?P<oid>.*)")
242 def check_search_name_has_entry(context, oid):
243 """ Check that there is noentry in the search_name table for the given
244 objects. IDs are in format '<NRW><osm id>[:<class>]'.
246 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
247 NominatimID(oid).row_by_place_id(cur, 'search_name')
249 assert cur.rowcount == 0, \
250 "Found {} entries for ID {}".format(cur.rowcount, oid)
252 @then("location_postcode contains exactly")
253 def check_location_postcode(context):
254 """ Check full contents for location_postcode table. Each row represents a table row
255 and all data must match. Data not present in the expected table, may
256 be arbitry. The rows are identified via 'country' and 'postcode' columns.
257 All rows must be present as excepted and there must not be additional
260 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
261 cur.execute("SELECT *, ST_AsText(geometry) as geomtxt FROM location_postcode")
262 assert cur.rowcount == len(list(context.table)), \
263 "Postcode table has {} rows, expected {}.".foramt(cur.rowcount, len(list(context.table)))
267 key = (row['country_code'], row['postcode'])
268 assert key not in results, "Postcode table has duplicate entry: {}".format(row)
269 results[key] = DBRow((row['country_code'],row['postcode']), row, context)
271 for row in context.table:
272 db_row = results.get((row['country'],row['postcode']))
273 assert db_row is not None, \
274 "Missing row for country '{r['country']}' postcode '{r['postcode']}'.".format(r=row)
276 db_row.assert_row(row, ('country', 'postcode'))
278 @then("word contains(?P<exclude> not)?")
279 def check_word_table(context, exclude):
280 """ Check the contents of the word table. Each row represents a table row
281 and all data must match. Data not present in the expected table, may
282 be arbitry. The rows are identified via all given columns.
284 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
285 for row in context.table:
286 wheres = ' AND '.join(["{} = %s".format(h) for h in row.headings])
287 cur.execute("SELECT * from word WHERE " + wheres, list(row.cells))
289 assert cur.rowcount == 0, "Row still in word table: %s" % '/'.join(values)
291 assert cur.rowcount > 0, "Row not in word table: %s" % '/'.join(values)
293 @then("place_addressline contains")
294 def check_place_addressline(context):
295 """ Check the contents of the place_addressline table. Each row represents
296 a table row and all data must match. Data not present in the expected
297 table, may be arbitry. The rows are identified via the 'object' column,
298 representing the addressee and the 'address' column, representing the
301 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
302 for row in context.table:
303 nid = NominatimID(row['object'])
304 pid = nid.get_place_id(cur)
305 apid = NominatimID(row['address']).get_place_id(cur)
306 cur.execute(""" SELECT * FROM place_addressline
307 WHERE place_id = %s AND address_place_id = %s""",
309 assert cur.rowcount > 0, \
310 "No rows found for place %s and address %s" % (row['object'], row['address'])
313 DBRow(nid, res, context).assert_row(row, ('address', 'object'))
315 @then("place_addressline doesn't contain")
316 def check_place_addressline_exclude(context):
317 """ Check that the place_addressline doesn't contain any entries for the
318 given addressee/address item pairs.
320 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
321 for row in context.table:
322 pid = NominatimID(row['object']).get_place_id(cur)
323 apid = NominatimID(row['address']).get_place_id(cur)
324 cur.execute(""" SELECT * FROM place_addressline
325 WHERE place_id = %s AND address_place_id = %s""",
327 assert cur.rowcount == 0, \
328 "Row found for place %s and address %s" % (row['object'], row['address'])
330 @then("W(?P<oid>\d+) expands to(?P<neg> no)? interpolation")
331 def check_location_property_osmline(context, oid, neg):
332 """ Check that the given way is present in the interpolation table.
334 with context.db.cursor(cursor_factory=psycopg2.extras.DictCursor) as cur:
335 cur.execute("""SELECT *, ST_AsText(linegeo) as geomtxt
336 FROM location_property_osmline
337 WHERE osm_id = %s AND startnumber IS NOT NULL""",
341 assert cur.rowcount == 0, "Interpolation found for way {}.".format(oid)
344 todo = list(range(len(list(context.table))))
347 row = context.table[i]
348 if (int(row['start']) == res['startnumber']
349 and int(row['end']) == res['endnumber']):
353 assert False, "Unexpected row " + str(res)
355 DBRow(oid, res, context).assert_row(row, ('start', 'end'))