django1/tests/regressiontests/backends/tests.py

228 lines
10 KiB
Python
Raw Normal View History

# -*- coding: utf-8 -*-
# Unit and doctests for specific database backends.
import datetime
import unittest
from django.conf import settings
from django.core import management
from django.core.management.color import no_style
from django.db import backend, connection, connections, DEFAULT_DB_ALIAS
from django.db.backends.signals import connection_created
from django.db.backends.postgresql import version as pg_version
from django.test import TestCase
from regressiontests.backends import models
class Callproc(unittest.TestCase):
def test_dbms_session(self):
# If the backend is Oracle, test that we can call a standard
# stored procedure through our cursor wrapper.
if settings.DATABASES[DEFAULT_DB_ALIAS]['ENGINE'] == 'django.db.backends.oracle':
convert_unicode = backend.convert_unicode
cursor = connection.cursor()
cursor.callproc(convert_unicode('DBMS_SESSION.SET_IDENTIFIER'),
[convert_unicode('_django_testing!'),])
return True
else:
return True
def test_cursor_var(self):
# If the backend is Oracle, test that we can pass cursor variables
# as query parameters.
if settings.DATABASES[DEFAULT_DB_ALIAS]['ENGINE'] == 'django.db.backends.oracle':
cursor = connection.cursor()
var = cursor.var(backend.Database.STRING)
cursor.execute("BEGIN %s := 'X'; END; ", [var])
self.assertEqual(var.getvalue(), 'X')
class LongString(unittest.TestCase):
def test_long_string(self):
# If the backend is Oracle, test that we can save a text longer
# than 4000 chars and read it properly
if settings.DATABASES[DEFAULT_DB_ALIAS]['ENGINE'] == 'django.db.backends.oracle':
c = connection.cursor()
c.execute('CREATE TABLE ltext ("TEXT" NCLOB)')
long_str = ''.join([unicode(x) for x in xrange(4000)])
c.execute('INSERT INTO ltext VALUES (%s)',[long_str])
c.execute('SELECT text FROM ltext')
row = c.fetchone()
self.assertEquals(long_str, row[0].read())
c.execute('DROP TABLE ltext')
class DateQuotingTest(TestCase):
def test_django_date_trunc(self):
"""
Test the custom ``django_date_trunc method``, in particular against
fields which clash with strings passed to it (e.g. 'year') - see
#12818__.
__: http://code.djangoproject.com/ticket/12818
"""
updated = datetime.datetime(2010, 2, 20)
models.SchoolClass.objects.create(year=2009, last_updated=updated)
years = models.SchoolClass.objects.dates('last_updated', 'year')
self.assertEqual(list(years), [datetime.datetime(2010, 1, 1, 0, 0)])
def test_django_extract(self):
"""
Test the custom ``django_extract method``, in particular against fields
which clash with strings passed to it (e.g. 'day') - see #12818__.
__: http://code.djangoproject.com/ticket/12818
"""
updated = datetime.datetime(2010, 2, 20)
models.SchoolClass.objects.create(year=2009, last_updated=updated)
classes = models.SchoolClass.objects.filter(last_updated__day=20)
self.assertEqual(len(classes), 1)
class ParameterHandlingTest(TestCase):
def test_bad_parameter_count(self):
"An executemany call with too many/not enough parameters will raise an exception (Refs #12612)"
cursor = connection.cursor()
query = ('INSERT INTO %s (%s, %s) VALUES (%%s, %%s)' % (
connection.introspection.table_name_converter('backends_square'),
connection.ops.quote_name('root'),
connection.ops.quote_name('square')
))
self.assertRaises(Exception, cursor.executemany, query, [(1,2,3),])
self.assertRaises(Exception, cursor.executemany, query, [(1,),])
# Unfortunately, the following tests would be a good test to run on all
# backends, but it breaks MySQL hard. Until #13711 is fixed, it can't be run
# everywhere (although it would be an effective test of #13711).
if settings.DATABASES[DEFAULT_DB_ALIAS]['ENGINE'] != 'django.db.backends.mysql':
class LongNameTest(TestCase):
"""Long primary keys and model names can result in a sequence name
that exceeds the database limits, which will result in truncation
on certain databases (e.g., Postgres). The backend needs to use
the correct sequence name in last_insert_id and other places, so
check it is. Refs #8901.
"""
def test_sequence_name_length_limits_create(self):
"""Test creation of model with long name and long pk name doesn't error. Ref #8901"""
models.VeryLongModelNameZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ.objects.create()
def test_sequence_name_length_limits_m2m(self):
"""Test an m2m save of a model with a long name and a long m2m field name doesn't error as on Django >=1.2 this now uses object saves. Ref #8901"""
obj = models.VeryLongModelNameZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ.objects.create()
rel_obj = models.Person.objects.create(first_name='Django', last_name='Reinhardt')
obj.m2m_also_quite_long_zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz.add(rel_obj)
def test_sequence_name_length_limits_flush(self):
"""Test that sequence resetting as part of a flush with model with long name and long pk name doesn't error. Ref #8901"""
# A full flush is expensive to the full test, so we dig into the
# internals to generate the likely offending SQL and run it manually
# Some convenience aliases
VLM = models.VeryLongModelNameZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ
VLM_m2m = VLM.m2m_also_quite_long_zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz.through
tables = [
VLM._meta.db_table,
VLM_m2m._meta.db_table,
]
sequences = [
{
'column': VLM._meta.pk.column,
'table': VLM._meta.db_table
},
]
cursor = connection.cursor()
for statement in connection.ops.sql_flush(no_style(), tables, sequences):
cursor.execute(statement)
class SequenceResetTest(TestCase):
def test_generic_relation(self):
"Sequence names are correct when resetting generic relations (Ref #13941)"
# Create an object with a manually specified PK
models.Post.objects.create(id=10, name='1st post', text='hello world')
# Reset the sequences for the database
cursor = connection.cursor()
commands = connections[DEFAULT_DB_ALIAS].ops.sequence_reset_sql(no_style(), [models.Post])
for sql in commands:
cursor.execute(sql)
# If we create a new object now, it should have a PK greater
# than the PK we specified manually.
obj = models.Post.objects.create(name='New post', text='goodbye world')
self.assertTrue(obj.pk > 10)
class PostgresVersionTest(TestCase):
def assert_parses(self, version_string, version):
self.assertEqual(pg_version._parse_version(version_string), version)
def test_parsing(self):
self.assert_parses("PostgreSQL 8.3 beta4", (8, 3, None))
self.assert_parses("PostgreSQL 8.3", (8, 3, None))
self.assert_parses("EnterpriseDB 8.3", (8, 3, None))
self.assert_parses("PostgreSQL 8.3.6", (8, 3, 6))
self.assert_parses("PostgreSQL 8.4beta1", (8, 4, None))
self.assert_parses("PostgreSQL 8.3.1 on i386-apple-darwin9.2.2, compiled by GCC i686-apple-darwin9-gcc-4.0.1 (GCC) 4.0.1 (Apple Inc. build 5478)", (8, 3, 1))
# Unfortunately with sqlite3 the in-memory test database cannot be
# closed, and so it cannot be re-opened during testing, and so we
# sadly disable this test for now.
if settings.DATABASES[DEFAULT_DB_ALIAS]["ENGINE"] != "django.db.backends.sqlite3":
class ConnectionCreatedSignalTest(TestCase):
def test_signal(self):
data = {}
def receiver(sender, connection, **kwargs):
data["connection"] = connection
connection_created.connect(receiver)
connection.close()
cursor = connection.cursor()
self.assertTrue(data["connection"] is connection)
connection_created.disconnect(receiver)
data.clear()
cursor = connection.cursor()
self.assertTrue(data == {})
class BackendTestCase(TestCase):
def test_cursor_executemany(self):
#4896: Test cursor.executemany
cursor = connection.cursor()
qn = connection.ops.quote_name
opts = models.Square._meta
f1, f2 = opts.get_field('root'), opts.get_field('square')
query = ('INSERT INTO %s (%s, %s) VALUES (%%s, %%s)'
% (connection.introspection.table_name_converter(opts.db_table), qn(f1.column), qn(f2.column)))
cursor.executemany(query, [(i, i**2) for i in range(-5, 6)])
self.assertEqual(models.Square.objects.count(), 11)
for i in range(-5, 6):
square = models.Square.objects.get(root=i)
self.assertEqual(square.square, i**2)
#4765: executemany with params=[] does nothing
cursor.executemany(query, [])
self.assertEqual(models.Square.objects.count(), 11)
def test_unicode_fetches(self):
#6254: fetchone, fetchmany, fetchall return strings as unicode objects
qn = connection.ops.quote_name
models.Person(first_name="John", last_name="Doe").save()
models.Person(first_name="Jane", last_name="Doe").save()
models.Person(first_name="Mary", last_name="Agnelline").save()
models.Person(first_name="Peter", last_name="Parker").save()
models.Person(first_name="Clark", last_name="Kent").save()
opts2 = models.Person._meta
f3, f4 = opts2.get_field('first_name'), opts2.get_field('last_name')
query2 = ('SELECT %s, %s FROM %s ORDER BY %s'
% (qn(f3.column), qn(f4.column), connection.introspection.table_name_converter(opts2.db_table),
qn(f3.column)))
cursor = connection.cursor()
cursor.execute(query2)
self.assertEqual(cursor.fetchone(), (u'Clark', u'Kent'))
self.assertEqual(list(cursor.fetchmany(2)), [(u'Jane', u'Doe'), (u'John', u'Doe')])
self.assertEqual(list(cursor.fetchall()), [(u'Mary', u'Agnelline'), (u'Peter', u'Parker')])