]> git.openstreetmap.org Git - nominatim.git/blobdiff - test/bdd/steps/db_ops.py
Do not have postcode node appear in addresses directly
[nominatim.git] / test / bdd / steps / db_ops.py
index c4384f07beea639af473087a7cf23d54695d6952..b50f1d59baebb1c1556b0a60a8edae3f046bc31e 100644 (file)
@@ -8,7 +8,7 @@ import psycopg2.extras
 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
@@ -21,7 +21,9 @@ class PlaceColumn:
         elif key.startswith('extra+'):
             self.add_hstore('extratags', key[6:], value)
         elif key.startswith('addr+'):
-            self.add_hstore('address', key[6:], value)
+            self.add_hstore('address', key[5:], value)
+        elif key in ('name', 'address', 'extratags'):
+            self.columns[key] = eval('{' + value + '}')
         else:
             assert_in(key, ('class', 'type'))
             self.columns[key] = None if value == '' else value
@@ -40,16 +42,24 @@ class PlaceColumn:
         self.columns['admin_level'] = int(value)
 
     def set_key_housenr(self, value):
-        self.add_hstore('address', '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):
-        self.add_hstore('address', 'street', None if value == '' else value)
+        if value:
+            self.add_hstore('address', 'street', value)
 
     def set_key_addr_place(self, value):
-        self.add_hstore('address', 'place', None if value == '' else value)
+        if value:
+            self.add_hstore('address', 'place', value)
 
     def set_key_country(self, value):
-        self.add_hstore('address', 'country', 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)
@@ -81,6 +91,59 @@ class PlaceColumn:
                      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 "<null>"
+
+        cur = self.conn.cursor()
+        cur.execute("""SELECT osm_type, osm_id, class
+                       FROM placex WHERE place_id = %s""",
+                    (self.pid, ))
+        eq_(1, cur.rowcount, "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':
+        eq_(0, result,
+            LazyFmt("Bad place id in column %s. Expected: 0, got: %s.",
+                    column, PlaceObjName(result, context.db)))
+    elif expected == '-':
+        assert_is_none(result,
+                LazyFmt("bad place id in column %s: %s.",
+                        column, PlaceObjName(result, context.db)))
+    else:
+        eq_(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""")
+    eq_(0, cur.fetchone()[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
@@ -229,16 +292,17 @@ 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,
-           address, 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(
             """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('index', 'index-noanalyse')
+    context.nominatim.run_setup_script('calculate-postcodes', 'index', 'index-noanalyse')
+    check_database_integrity(context)
 
 @when("updating places")
 def update_place_table(context):
@@ -263,6 +327,8 @@ def update_place_table(context):
         if cur.rowcount == 0:
             break
 
+    check_database_integrity(context)
+
 @when("marking for delete (?P<oids>.*)")
 def delete_places(context, oids):
     context.nominatim.run_setup_script(
@@ -299,20 +365,28 @@ def check_placex_contents(context, exact):
             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_is_none(res[h])
+                    else:
+                        vdict = eval('{' + row[h] + '}')
+                        assert_equals(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])
                 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])
+                elif h.startswith('addr+'):
+                    if row[h] == '-':
+                        if res['address'] is not None:
+                            assert_not_in(h[5:], res['address'])
                     else:
-                        eq_(NominatimID(row[h]).get_place_id(context.db.cursor()),
-                            res[h])
+                        assert_in(h[5:], res['address'], "column " + h)
+                        assert_equals(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)
 
@@ -358,13 +432,7 @@ def check_placex_contents(context, exact):
                     else:
                         assert_equals(res['address'][h[5:]], row[h], msg)
                 elif h in ('linked_place_id', 'parent_place_id'):
-                    if row[h] == '0':
-                        assert_equals(0, res[h], msg)
-                    elif row[h] == '-':
-                        assert_is_none(res[h], msg)
-                    else:
-                        assert_equals(NominatimID(row[h]).get_place_id(context.db.cursor()),
-                                      res[h], msg)
+                    compare_place_id(row[h], res[h], h, context)
                 else:
                     assert_db_column(res, h, row[h], context)
 
@@ -374,8 +442,8 @@ def check_placex_contents(context, exact):
 
     context.db.commit()
 
-@then("search_name contains")
-def check_search_name_contents(context):
+@then("search_name contains(?P<exclude> not)?")
+def check_search_name_contents(context, exclude):
     cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor)
 
     for row in context.table:
@@ -393,10 +461,16 @@ def check_search_name_contents(context):
                                       FROM word, (SELECT unnest(%s) as term) t
                                       WHERE word_token = make_standard_name(t.term)""",
                                    (terms,))
-                    ok_(subcur.rowcount >= len(terms))
+                    if not exclude:
+                        ok_(subcur.rowcount >= len(terms),
+                            "No word entry found for " + row[h])
                     for wid in subcur:
-                        assert_in(wid[0], res[h],
-                                  "Missing term for %s/%s: %s" % (pid, h, wid[1]))
+                        if exclude:
+                            assert_not_in(wid[0], res[h],
+                                          "Found term for %s/%s: %s" % (pid, h, wid[1]))
+                        else:
+                            assert_in(wid[0], res[h],
+                                      "Missing term for %s/%s: %s" % (pid, h, wid[1]))
                 else:
                     assert_db_column(res, h, row[h], context)
 
@@ -424,6 +498,21 @@ def check_place_addressline(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))
+        eq_(0, cur.rowcount,
+            "Row found for place %s and address %s" % (row['object'], row['address']))
+
+    context.db.commit()
+
 @then("(?P<oid>\w+) expands to(?P<neg> no)? interpolation")
 def check_location_property_osmline(context, oid, neg):
     cur = context.db.cursor(cursor_factory=psycopg2.extras.DictCursor)
@@ -455,13 +544,7 @@ def check_location_property_osmline(context, oid, neg):
             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)