psycopg2/tests/test_cursor.py

639 lines
24 KiB
Python
Raw Normal View History

#!/usr/bin/env python
# test_cursor.py - unit test for cursor attributes
#
# Copyright (C) 2010-2011 Daniele Varrazzo <daniele.varrazzo@gmail.com>
#
# psycopg2 is free software: you can redistribute it and/or modify it
# under the terms of the GNU Lesser General Public License as published
# by the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# In addition, as a special exception, the copyright holders give
# permission to link this program with the OpenSSL library (or with
# modified versions of OpenSSL that use the same license as OpenSSL),
# and distribute linked combinations including the two.
#
# You must obey the GNU Lesser General Public License in all respects for
# all of the code used other than OpenSSL.
#
# psycopg2 is distributed in the hope that it will be useful, but WITHOUT
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
# FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
# License for more details.
import time
import pickle
import psycopg2
import psycopg2.extensions
from testutils import (unittest, ConnectingTestCase, skip_before_postgres,
skip_if_no_namedtuple, skip_if_no_getrefcount, slow, skip_if_no_superuser,
skip_if_windows)
Always detect when a connection is closed behind psycopg2's back. There's a race condition that only seems to happen over Unix-domain sockets. Sometimes, the closed socket is reported by the kernel to libpq like this (captured with strace): sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = 29 recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, 0x12d0330, 16384, 0, 0, 0) = -1 ECONNRESET (Connection reset by peer) That is, psycopg2/libpq sees no error when sending the first query after the connection is closed, but gets an error reading the result. In that case, everything worked fine. But sometimes, the error manifests like this: sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = -1 EPIPE (Broken pipe) recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 i.e. libpq received an error when sending the query. This manifests as a slightly different exception from a slightly different place. More importantly, in this case connection.closed is left at 0 rather than being set to 2, and that is the bug I'm fixing here. Note that we see almost identical behaviour for sync and async connections, and the fixes are the same. So I added extremely similar test cases. Finally, there is still a bug here: for async connections, we sometimes raise DatabaseError (incorrect) and sometimes raise OperationalError (correct). Will fix that next.
2016-06-29 00:46:21 +03:00
import psycopg2.extras
2016-10-11 02:10:53 +03:00
class CursorTests(ConnectingTestCase):
def test_close_idempotent(self):
cur = self.conn.cursor()
cur.close()
cur.close()
self.assert_(cur.closed)
def test_empty_query(self):
cur = self.conn.cursor()
self.assertRaises(psycopg2.ProgrammingError, cur.execute, "")
self.assertRaises(psycopg2.ProgrammingError, cur.execute, " ")
self.assertRaises(psycopg2.ProgrammingError, cur.execute, ";")
def test_executemany_propagate_exceptions(self):
conn = self.conn
cur = conn.cursor()
cur.execute("create temp table test_exc (data int);")
2016-10-11 02:10:53 +03:00
def buggygen():
2016-10-11 02:10:53 +03:00
yield 1 // 0
self.assertRaises(ZeroDivisionError,
cur.executemany, "insert into test_exc values (%s)", buggygen())
cur.close()
def test_mogrify_unicode(self):
conn = self.conn
cur = conn.cursor()
# test consistency between execute and mogrify.
# unicode query containing only ascii data
cur.execute(u"SELECT 'foo';")
self.assertEqual('foo', cur.fetchone()[0])
self.assertEqual(b"SELECT 'foo';", cur.mogrify(u"SELECT 'foo';"))
conn.set_client_encoding('UTF8')
snowman = u"\u2603"
2016-10-11 02:22:23 +03:00
def b(s):
if isinstance(s, unicode):
return s.encode('utf8')
else:
return s
# unicode query with non-ascii data
cur.execute(u"SELECT '%s';" % snowman)
2010-12-29 05:47:29 +03:00
self.assertEqual(snowman.encode('utf8'), b(cur.fetchone()[0]))
self.assertQuotedEqual(("SELECT '%s';" % snowman).encode('utf8'),
cur.mogrify(u"SELECT '%s';" % snowman))
# unicode args
cur.execute("SELECT %s;", (snowman,))
2010-12-29 05:47:29 +03:00
self.assertEqual(snowman.encode("utf-8"), b(cur.fetchone()[0]))
self.assertQuotedEqual(("SELECT '%s';" % snowman).encode('utf8'),
cur.mogrify("SELECT %s;", (snowman,)))
# unicode query and args
cur.execute(u"SELECT %s;", (snowman,))
2010-12-29 05:47:29 +03:00
self.assertEqual(snowman.encode("utf-8"), b(cur.fetchone()[0]))
self.assertQuotedEqual(("SELECT '%s';" % snowman).encode('utf8'),
cur.mogrify(u"SELECT %s;", (snowman,)))
def test_mogrify_decimal_explodes(self):
# issue #7: explodes on windows with python 2.5 and psycopg 2.2.2
try:
from decimal import Decimal
except:
return
conn = self.conn
cur = conn.cursor()
self.assertEqual(b'SELECT 10.3;',
cur.mogrify("SELECT %s;", (Decimal("10.3"),)))
@skip_if_no_getrefcount
def test_mogrify_leak_on_multiple_reference(self):
# issue #81: reference leak when a parameter value is referenced
# more than once from a dict.
cur = self.conn.cursor()
2016-10-11 02:10:53 +03:00
foo = (lambda x: x)('foo') * 10
import sys
nref1 = sys.getrefcount(foo)
cur.mogrify("select %(foo)s, %(foo)s, %(foo)s", {'foo': foo})
nref2 = sys.getrefcount(foo)
self.assertEqual(nref1, nref2)
def test_modify_closed(self):
cur = self.conn.cursor()
cur.close()
sql = cur.mogrify("select %s", (10,))
self.assertEqual(sql, b"select 10")
def test_bad_placeholder(self):
cur = self.conn.cursor()
self.assertRaises(psycopg2.ProgrammingError,
cur.mogrify, "select %(foo", {})
self.assertRaises(psycopg2.ProgrammingError,
cur.mogrify, "select %(foo", {'foo': 1})
self.assertRaises(psycopg2.ProgrammingError,
cur.mogrify, "select %(foo, %(bar)", {'foo': 1})
self.assertRaises(psycopg2.ProgrammingError,
cur.mogrify, "select %(foo, %(bar)", {'foo': 1, 'bar': 2})
def test_cast(self):
curs = self.conn.cursor()
self.assertEqual(42, curs.cast(20, '42'))
self.assertAlmostEqual(3.14, curs.cast(700, '3.14'))
try:
from decimal import Decimal
except ImportError:
self.assertAlmostEqual(123.45, curs.cast(1700, '123.45'))
else:
self.assertEqual(Decimal('123.45'), curs.cast(1700, '123.45'))
from datetime import date
2016-10-11 02:10:53 +03:00
self.assertEqual(date(2011, 1, 2), curs.cast(1082, '2011-01-02'))
self.assertEqual("who am i?", curs.cast(705, 'who am i?')) # unknown
def test_cast_specificity(self):
curs = self.conn.cursor()
self.assertEqual("foo", curs.cast(705, 'foo'))
D = psycopg2.extensions.new_type((705,), "DOUBLING", lambda v, c: v * 2)
psycopg2.extensions.register_type(D, self.conn)
self.assertEqual("foofoo", curs.cast(705, 'foo'))
T = psycopg2.extensions.new_type((705,), "TREBLING", lambda v, c: v * 3)
psycopg2.extensions.register_type(T, curs)
self.assertEqual("foofoofoo", curs.cast(705, 'foo'))
curs2 = self.conn.cursor()
self.assertEqual("foofoo", curs2.cast(705, 'foo'))
2011-01-03 14:47:01 +03:00
def test_weakref(self):
from weakref import ref
curs = self.conn.cursor()
w = ref(curs)
del curs
2016-10-11 02:10:53 +03:00
import gc
gc.collect()
2011-01-03 14:47:01 +03:00
self.assert_(w() is None)
def test_null_name(self):
curs = self.conn.cursor(None)
self.assertEqual(curs.name, None)
def test_invalid_name(self):
curs = self.conn.cursor()
curs.execute("create temp table invname (data int);")
2016-10-11 02:10:53 +03:00
for i in (10, 20, 30):
curs.execute("insert into invname values (%s)", (i,))
curs.close()
curs = self.conn.cursor(r'1-2-3 \ "test"')
curs.execute("select data from invname order by data")
self.assertEqual(curs.fetchall(), [(10,), (20,), (30,)])
def _create_withhold_table(self):
curs = self.conn.cursor()
try:
curs.execute("drop table withhold")
except psycopg2.ProgrammingError:
self.conn.rollback()
curs.execute("create table withhold (data int)")
for i in (10, 20, 30):
curs.execute("insert into withhold values (%s)", (i,))
curs.close()
def test_withhold(self):
self.assertRaises(psycopg2.ProgrammingError, self.conn.cursor,
withhold=True)
self._create_withhold_table()
curs = self.conn.cursor("W")
2016-10-11 02:10:53 +03:00
self.assertEqual(curs.withhold, False)
curs.withhold = True
2016-10-11 02:10:53 +03:00
self.assertEqual(curs.withhold, True)
curs.execute("select data from withhold order by data")
self.conn.commit()
self.assertEqual(curs.fetchall(), [(10,), (20,), (30,)])
curs.close()
curs = self.conn.cursor("W", withhold=True)
2016-10-11 02:10:53 +03:00
self.assertEqual(curs.withhold, True)
curs.execute("select data from withhold order by data")
self.conn.commit()
self.assertEqual(curs.fetchall(), [(10,), (20,), (30,)])
curs = self.conn.cursor()
curs.execute("drop table withhold")
self.conn.commit()
def test_withhold_no_begin(self):
self._create_withhold_table()
curs = self.conn.cursor("w", withhold=True)
curs.execute("select data from withhold order by data")
self.assertEqual(curs.fetchone(), (10,))
self.assertEqual(self.conn.status, psycopg2.extensions.STATUS_BEGIN)
self.assertEqual(self.conn.get_transaction_status(),
psycopg2.extensions.TRANSACTION_STATUS_INTRANS)
self.conn.commit()
self.assertEqual(self.conn.status, psycopg2.extensions.STATUS_READY)
self.assertEqual(self.conn.get_transaction_status(),
psycopg2.extensions.TRANSACTION_STATUS_IDLE)
self.assertEqual(curs.fetchone(), (20,))
self.assertEqual(self.conn.status, psycopg2.extensions.STATUS_READY)
self.assertEqual(self.conn.get_transaction_status(),
psycopg2.extensions.TRANSACTION_STATUS_IDLE)
curs.close()
self.assertEqual(self.conn.status, psycopg2.extensions.STATUS_READY)
self.assertEqual(self.conn.get_transaction_status(),
psycopg2.extensions.TRANSACTION_STATUS_IDLE)
def test_withhold_autocommit(self):
self._create_withhold_table()
self.conn.commit()
self.conn.autocommit = True
curs = self.conn.cursor("w", withhold=True)
curs.execute("select data from withhold order by data")
self.assertEqual(curs.fetchone(), (10,))
self.assertEqual(self.conn.status, psycopg2.extensions.STATUS_READY)
self.assertEqual(self.conn.get_transaction_status(),
psycopg2.extensions.TRANSACTION_STATUS_IDLE)
self.conn.commit()
self.assertEqual(self.conn.status, psycopg2.extensions.STATUS_READY)
self.assertEqual(self.conn.get_transaction_status(),
psycopg2.extensions.TRANSACTION_STATUS_IDLE)
curs.close()
self.assertEqual(self.conn.status, psycopg2.extensions.STATUS_READY)
self.assertEqual(self.conn.get_transaction_status(),
psycopg2.extensions.TRANSACTION_STATUS_IDLE)
2012-08-15 04:11:26 +04:00
def test_scrollable(self):
self.assertRaises(psycopg2.ProgrammingError, self.conn.cursor,
scrollable=True)
curs = self.conn.cursor()
curs.execute("create table scrollable (data int)")
curs.executemany("insert into scrollable values (%s)",
2016-10-11 02:10:53 +03:00
[(i,) for i in range(100)])
2012-08-15 04:11:26 +04:00
curs.close()
for t in range(2):
if not t:
curs = self.conn.cursor("S")
2016-10-11 02:10:53 +03:00
self.assertEqual(curs.scrollable, None)
2012-08-15 04:11:26 +04:00
curs.scrollable = True
else:
curs = self.conn.cursor("S", scrollable=True)
2016-10-11 02:10:53 +03:00
self.assertEqual(curs.scrollable, True)
2012-08-15 04:11:26 +04:00
curs.itersize = 10
# complex enough to make postgres cursors declare without
# scroll/no scroll to fail
curs.execute("""
select x.data
from scrollable x
join scrollable y on x.data = y.data
order by y.data""")
for i, (n,) in enumerate(curs):
self.assertEqual(i, n)
curs.scroll(-1)
for i in range(99, -1, -1):
curs.scroll(-1)
self.assertEqual(i, curs.fetchone()[0])
curs.scroll(-1)
curs.close()
def test_not_scrollable(self):
self.assertRaises(psycopg2.ProgrammingError, self.conn.cursor,
scrollable=False)
curs = self.conn.cursor()
curs.execute("create table scrollable (data int)")
curs.executemany("insert into scrollable values (%s)",
2016-10-11 02:10:53 +03:00
[(i,) for i in range(100)])
curs.close()
curs = self.conn.cursor("S") # default scrollability
curs.execute("select * from scrollable")
self.assertEqual(curs.scrollable, None)
curs.scroll(2)
try:
curs.scroll(-1)
except psycopg2.OperationalError:
return self.skipTest("can't evaluate non-scrollable cursor")
curs.close()
curs = self.conn.cursor("S", scrollable=False)
self.assertEqual(curs.scrollable, False)
curs.execute("select * from scrollable")
curs.scroll(2)
self.assertRaises(psycopg2.OperationalError, curs.scroll, -1)
@slow
@skip_before_postgres(8, 2)
def test_iter_named_cursor_efficient(self):
curs = self.conn.cursor('tmp')
# if these records are fetched in the same roundtrip their
# timestamp will not be influenced by the pause in Python world.
curs.execute("""select clock_timestamp() from generate_series(1,2)""")
i = iter(curs)
t1 = (i.next())[0] # the brackets work around a 2to3 bug
time.sleep(0.2)
t2 = (i.next())[0]
self.assert_((t2 - t1).microseconds * 1e-6 < 0.1,
"named cursor records fetched in 2 roundtrips (delta: %s)"
% (t2 - t1))
@skip_before_postgres(8, 0)
def test_iter_named_cursor_default_itersize(self):
curs = self.conn.cursor('tmp')
curs.execute('select generate_series(1,50)')
2016-10-11 02:10:53 +03:00
rv = [(r[0], curs.rownumber) for r in curs]
# everything swallowed in one gulp
2016-10-11 02:10:53 +03:00
self.assertEqual(rv, [(i, i) for i in range(1, 51)])
@skip_before_postgres(8, 0)
def test_iter_named_cursor_itersize(self):
curs = self.conn.cursor('tmp')
curs.itersize = 30
curs.execute('select generate_series(1,50)')
2016-10-11 02:10:53 +03:00
rv = [(r[0], curs.rownumber) for r in curs]
# everything swallowed in two gulps
2016-10-11 02:10:53 +03:00
self.assertEqual(rv, [(i, ((i - 1) % 30) + 1) for i in range(1, 51)])
@skip_before_postgres(8, 0)
def test_iter_named_cursor_rownumber(self):
curs = self.conn.cursor('tmp')
# note: this fails if itersize < dataset: internally we check
# rownumber == rowcount to detect when to read anoter page, so we
# would need an extra attribute to have a monotonic rownumber.
curs.itersize = 20
curs.execute('select generate_series(1,10)')
for i, rec in enumerate(curs):
self.assertEqual(i + 1, curs.rownumber)
@skip_if_no_namedtuple
def test_namedtuple_description(self):
curs = self.conn.cursor()
curs.execute("""select
3.14::decimal(10,2) as pi,
'hello'::text as hi,
'2010-02-18'::date as now;
""")
self.assertEqual(len(curs.description), 3)
for c in curs.description:
self.assertEqual(len(c), 7) # DBAPI happy
for a in ('name', 'type_code', 'display_size', 'internal_size',
'precision', 'scale', 'null_ok'):
self.assert_(hasattr(c, a), a)
c = curs.description[0]
self.assertEqual(c.name, 'pi')
self.assert_(c.type_code in psycopg2.extensions.DECIMAL.values)
self.assert_(c.internal_size > 0)
self.assertEqual(c.precision, 10)
self.assertEqual(c.scale, 2)
c = curs.description[1]
self.assertEqual(c.name, 'hi')
self.assert_(c.type_code in psycopg2.STRING.values)
self.assert_(c.internal_size < 0)
self.assertEqual(c.precision, None)
self.assertEqual(c.scale, None)
c = curs.description[2]
self.assertEqual(c.name, 'now')
self.assert_(c.type_code in psycopg2.extensions.DATE.values)
self.assert_(c.internal_size > 0)
self.assertEqual(c.precision, None)
self.assertEqual(c.scale, None)
def test_pickle_description(self):
curs = self.conn.cursor()
curs.execute('SELECT 1 AS foo')
description = curs.description
pickled = pickle.dumps(description, pickle.HIGHEST_PROTOCOL)
unpickled = pickle.loads(pickled)
self.assertEqual(description, unpickled)
@skip_before_postgres(8, 0)
def test_named_cursor_stealing(self):
# you can use a named cursor to iterate on a refcursor created
# somewhere else
cur1 = self.conn.cursor()
cur1.execute("DECLARE test CURSOR WITHOUT HOLD "
" FOR SELECT generate_series(1,7)")
cur2 = self.conn.cursor('test')
# can call fetch without execute
self.assertEqual((1,), cur2.fetchone())
self.assertEqual([(2,), (3,), (4,)], cur2.fetchmany(3))
self.assertEqual([(5,), (6,), (7,)], cur2.fetchall())
@skip_before_postgres(8, 2)
def test_named_noop_close(self):
cur = self.conn.cursor('test')
cur.close()
@skip_before_postgres(8, 2)
def test_stolen_named_cursor_close(self):
cur1 = self.conn.cursor()
cur1.execute("DECLARE test CURSOR WITHOUT HOLD "
" FOR SELECT generate_series(1,7)")
cur2 = self.conn.cursor('test')
cur2.close()
cur1.execute("DECLARE test CURSOR WITHOUT HOLD "
" FOR SELECT generate_series(1,7)")
cur2 = self.conn.cursor('test')
cur2.close()
2011-12-26 23:06:23 +04:00
@skip_before_postgres(8, 0)
def test_scroll(self):
cur = self.conn.cursor()
cur.execute("select generate_series(0,9)")
cur.scroll(2)
self.assertEqual(cur.fetchone(), (2,))
cur.scroll(2)
self.assertEqual(cur.fetchone(), (5,))
cur.scroll(2, mode='relative')
self.assertEqual(cur.fetchone(), (8,))
cur.scroll(-1)
self.assertEqual(cur.fetchone(), (8,))
cur.scroll(-2)
self.assertEqual(cur.fetchone(), (7,))
cur.scroll(2, mode='absolute')
self.assertEqual(cur.fetchone(), (2,))
# on the boundary
cur.scroll(0, mode='absolute')
self.assertEqual(cur.fetchone(), (0,))
self.assertRaises((IndexError, psycopg2.ProgrammingError),
cur.scroll, -1, mode='absolute')
cur.scroll(0, mode='absolute')
self.assertRaises((IndexError, psycopg2.ProgrammingError),
cur.scroll, -1)
cur.scroll(9, mode='absolute')
self.assertEqual(cur.fetchone(), (9,))
self.assertRaises((IndexError, psycopg2.ProgrammingError),
cur.scroll, 10, mode='absolute')
cur.scroll(9, mode='absolute')
self.assertRaises((IndexError, psycopg2.ProgrammingError),
cur.scroll, 1)
@skip_before_postgres(8, 0)
def test_scroll_named(self):
cur = self.conn.cursor('tmp', scrollable=True)
2011-12-26 23:06:23 +04:00
cur.execute("select generate_series(0,9)")
cur.scroll(2)
self.assertEqual(cur.fetchone(), (2,))
cur.scroll(2)
self.assertEqual(cur.fetchone(), (5,))
cur.scroll(2, mode='relative')
self.assertEqual(cur.fetchone(), (8,))
cur.scroll(9, mode='absolute')
self.assertEqual(cur.fetchone(), (9,))
def test_bad_subclass(self):
# check that we get an error message instead of a segfault
# for badly written subclasses.
# see http://stackoverflow.com/questions/22019341/
class StupidCursor(psycopg2.extensions.cursor):
def __init__(self, *args, **kwargs):
# I am stupid so not calling superclass init
pass
cur = StupidCursor()
self.assertRaises(psycopg2.InterfaceError, cur.execute, 'select 1')
self.assertRaises(psycopg2.InterfaceError, cur.executemany,
'select 1', [])
def test_callproc_badparam(self):
cur = self.conn.cursor()
self.assertRaises(TypeError, cur.callproc, 'lower', 42)
# It would be inappropriate to test callproc's named parameters in the
# DBAPI2.0 test section because they are a psycopg2 extension.
@skip_before_postgres(9, 0)
def test_callproc_dict(self):
# This parameter name tests for injection and quote escaping
paramname = '''
Robert'); drop table "students" --
'''.strip()
escaped_paramname = '"%s"' % paramname.replace('"', '""')
procname = 'pg_temp.randall'
cur = self.conn.cursor()
# Set up the temporary function
cur.execute('''
CREATE FUNCTION %s(%s INT)
RETURNS INT AS
'SELECT $1 * $1'
LANGUAGE SQL
2017-03-14 15:06:46 +03:00
''' % (procname, escaped_paramname))
# Make sure callproc works right
2017-03-14 15:06:46 +03:00
cur.callproc(procname, {paramname: 2})
self.assertEquals(cur.fetchone()[0], 4)
# Make sure callproc fails right
failing_cases = [
2017-03-14 15:06:46 +03:00
({paramname: 2, 'foo': 'bar'}, psycopg2.ProgrammingError),
({paramname: '2'}, psycopg2.ProgrammingError),
({paramname: 'two'}, psycopg2.ProgrammingError),
({u'bj\xc3rn': 2}, psycopg2.ProgrammingError),
({3: 2}, TypeError),
({self: 2}, TypeError),
]
for parameter_sequence, exception in failing_cases:
self.assertRaises(exception, cur.callproc, procname, parameter_sequence)
self.conn.rollback()
@skip_if_no_superuser
@skip_if_windows
@skip_before_postgres(8, 4)
Always detect when a connection is closed behind psycopg2's back. There's a race condition that only seems to happen over Unix-domain sockets. Sometimes, the closed socket is reported by the kernel to libpq like this (captured with strace): sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = 29 recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, 0x12d0330, 16384, 0, 0, 0) = -1 ECONNRESET (Connection reset by peer) That is, psycopg2/libpq sees no error when sending the first query after the connection is closed, but gets an error reading the result. In that case, everything worked fine. But sometimes, the error manifests like this: sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = -1 EPIPE (Broken pipe) recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 i.e. libpq received an error when sending the query. This manifests as a slightly different exception from a slightly different place. More importantly, in this case connection.closed is left at 0 rather than being set to 2, and that is the bug I'm fixing here. Note that we see almost identical behaviour for sync and async connections, and the fixes are the same. So I added extremely similar test cases. Finally, there is still a bug here: for async connections, we sometimes raise DatabaseError (incorrect) and sometimes raise OperationalError (correct). Will fix that next.
2016-06-29 00:46:21 +03:00
def test_external_close_sync(self):
# If a "victim" connection is closed by a "control" connection
# behind psycopg2's back, psycopg2 always handles it correctly:
# raise OperationalError, set conn.closed to 2. This reproduces
# issue #443, a race between control_conn closing victim_conn and
# psycopg2 noticing.
control_conn = self.conn
connect_func = self.connect
wait_func = lambda conn: None
self._test_external_close(control_conn, connect_func, wait_func)
@skip_if_no_superuser
@skip_if_windows
@skip_before_postgres(8, 4)
Always detect when a connection is closed behind psycopg2's back. There's a race condition that only seems to happen over Unix-domain sockets. Sometimes, the closed socket is reported by the kernel to libpq like this (captured with strace): sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = 29 recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, 0x12d0330, 16384, 0, 0, 0) = -1 ECONNRESET (Connection reset by peer) That is, psycopg2/libpq sees no error when sending the first query after the connection is closed, but gets an error reading the result. In that case, everything worked fine. But sometimes, the error manifests like this: sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = -1 EPIPE (Broken pipe) recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 i.e. libpq received an error when sending the query. This manifests as a slightly different exception from a slightly different place. More importantly, in this case connection.closed is left at 0 rather than being set to 2, and that is the bug I'm fixing here. Note that we see almost identical behaviour for sync and async connections, and the fixes are the same. So I added extremely similar test cases. Finally, there is still a bug here: for async connections, we sometimes raise DatabaseError (incorrect) and sometimes raise OperationalError (correct). Will fix that next.
2016-06-29 00:46:21 +03:00
def test_external_close_async(self):
# Issue #443 is in the async code too. Since the fix is duplicated,
# so is the test.
control_conn = self.conn
connect_func = lambda: self.connect(async_=True)
Always detect when a connection is closed behind psycopg2's back. There's a race condition that only seems to happen over Unix-domain sockets. Sometimes, the closed socket is reported by the kernel to libpq like this (captured with strace): sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = 29 recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, 0x12d0330, 16384, 0, 0, 0) = -1 ECONNRESET (Connection reset by peer) That is, psycopg2/libpq sees no error when sending the first query after the connection is closed, but gets an error reading the result. In that case, everything worked fine. But sometimes, the error manifests like this: sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = -1 EPIPE (Broken pipe) recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 i.e. libpq received an error when sending the query. This manifests as a slightly different exception from a slightly different place. More importantly, in this case connection.closed is left at 0 rather than being set to 2, and that is the bug I'm fixing here. Note that we see almost identical behaviour for sync and async connections, and the fixes are the same. So I added extremely similar test cases. Finally, there is still a bug here: for async connections, we sometimes raise DatabaseError (incorrect) and sometimes raise OperationalError (correct). Will fix that next.
2016-06-29 00:46:21 +03:00
wait_func = psycopg2.extras.wait_select
self._test_external_close(control_conn, connect_func, wait_func)
def _test_external_close(self, control_conn, connect_func, wait_func):
# The short sleep before using victim_conn the second time makes it
# much more likely to lose the race and see the bug. Repeating the
# test several times makes it even more likely.
for i in range(10):
victim_conn = connect_func()
wait_func(victim_conn)
with victim_conn.cursor() as cur:
cur.execute('select pg_backend_pid()')
wait_func(victim_conn)
pid1 = cur.fetchall()[0][0]
with control_conn.cursor() as cur:
cur.execute('select pg_terminate_backend(%s)', (pid1,))
time.sleep(0.001)
def f():
Always detect when a connection is closed behind psycopg2's back. There's a race condition that only seems to happen over Unix-domain sockets. Sometimes, the closed socket is reported by the kernel to libpq like this (captured with strace): sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = 29 recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, 0x12d0330, 16384, 0, 0, 0) = -1 ECONNRESET (Connection reset by peer) That is, psycopg2/libpq sees no error when sending the first query after the connection is closed, but gets an error reading the result. In that case, everything worked fine. But sometimes, the error manifests like this: sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = -1 EPIPE (Broken pipe) recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 i.e. libpq received an error when sending the query. This manifests as a slightly different exception from a slightly different place. More importantly, in this case connection.closed is left at 0 rather than being set to 2, and that is the bug I'm fixing here. Note that we see almost identical behaviour for sync and async connections, and the fixes are the same. So I added extremely similar test cases. Finally, there is still a bug here: for async connections, we sometimes raise DatabaseError (incorrect) and sometimes raise OperationalError (correct). Will fix that next.
2016-06-29 00:46:21 +03:00
with victim_conn.cursor() as cur:
cur.execute('select 1')
wait_func(victim_conn)
self.assertRaises(psycopg2.OperationalError, f)
Always detect when a connection is closed behind psycopg2's back. There's a race condition that only seems to happen over Unix-domain sockets. Sometimes, the closed socket is reported by the kernel to libpq like this (captured with strace): sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = 29 recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, 0x12d0330, 16384, 0, 0, 0) = -1 ECONNRESET (Connection reset by peer) That is, psycopg2/libpq sees no error when sending the first query after the connection is closed, but gets an error reading the result. In that case, everything worked fine. But sometimes, the error manifests like this: sendto(3, "Q\0\0\0\34select pg_backend_pid()\0", 29, MSG_NOSIGNAL, NULL, 0) = -1 EPIPE (Broken pipe) recvfrom(3, "E\0\0\0mSFATAL\0C57P01\0Mterminating "..., 16384, 0, NULL, NULL) = 110 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 recvfrom(3, "", 16274, 0, NULL, NULL) = 0 i.e. libpq received an error when sending the query. This manifests as a slightly different exception from a slightly different place. More importantly, in this case connection.closed is left at 0 rather than being set to 2, and that is the bug I'm fixing here. Note that we see almost identical behaviour for sync and async connections, and the fixes are the same. So I added extremely similar test cases. Finally, there is still a bug here: for async connections, we sometimes raise DatabaseError (incorrect) and sometimes raise OperationalError (correct). Will fix that next.
2016-06-29 00:46:21 +03:00
self.assertEqual(victim_conn.closed, 2)
@skip_before_postgres(8, 2)
def test_rowcount_on_executemany_returning(self):
cur = self.conn.cursor()
cur.execute("create table execmany(id serial primary key, data int)")
cur.executemany(
"insert into execmany (data) values (%s)",
[(i,) for i in range(4)])
self.assertEqual(cur.rowcount, 4)
cur.executemany(
"insert into execmany (data) values (%s) returning data",
[(i,) for i in range(5)])
self.assertEqual(cur.rowcount, 5)
def test_suite():
return unittest.TestLoader().loadTestsFromName(__name__)
if __name__ == "__main__":
unittest.main()