class MongoTest(unittest.TestCase):
'''
Test the mongodb.py module. Uses a library that fakes
a MongoDB server. See https://pypi.python.org/pypi/mongomock/1.0.1
'''
def setUp(self):
self.objs = [{
"fname": "Franco",
"lname": "Corelli"
}, {
"fname": "Leonardo",
"lname": "DaVinci",
"age": 300
}, {
"fname": "Franco",
"lname": "Gandolpho"
}]
self.mongodb = MongoDB(dbName='unittest', collection='unittest')
self.mongodb.clearCollection(collection="unittest")
self.mongodb.clearCollection(collection="new_coll")
self.mongodb.setCollection("unittest")
def tearDown(self):
self.mongodb.dropCollection(collection='unittest')
self.mongodb.dropCollection(collection='new_coll')
self.mongodb.close()
@unittest.skipIf(not TEST_ALL, "Skipping")
def test_update_and_find_one(self):
self.mongodb.insert(self.objs[0])
# Get a generator for the results:
resGen = self.mongodb.query({"fname": "Franco"},
limit=1,
collection="unittest")
res = resGen.next()
self.assertEqual(
'Corelli', res['lname'],
"Failed retrieval of single obj; expected '%s' but got '%s'" %
('Corelli', res['lname']))
@unittest.skipIf(not TEST_ALL, "Skipping")
def test_set_coll_use_different_coll(self):
# Insert into unittest:
self.mongodb.insert(self.objs[0])
# Switch to new_coll:
self.mongodb.setCollection('new_coll')
self.mongodb.insert({"recommendation": "Hawaii"})
# We're in new_coll; the following should be empty result:
self.mongodb.query({"fname": "Franco"}, limit=1)
resCount = self.mongodb.resultCount({"fname": "Franco"})
self.assertIsNone(
resCount, "Got non-null result that should be null: %s" % resCount)
# But this search is within new_coll, and should succeed:
resGen = self.mongodb.query({"recommendation": {
'$regex': '.*'
}},
limit=1)
res = resGen.next()
self.assertEqual(
'Hawaii', res['recommendation'],
"Failed retrieval of single obj; expected '%s' but got '%s'" %
('Hawaii', res['recommendation']))
# Try inline collection switch:
resGen = self.mongodb.query({"fname": "Franco"},
limit=1,
collection="unittest")
res = resGen.next()
self.assertEqual(
'Corelli', res['lname'],
"Failed retrieval of single obj; expected '%s' but got '%s'" %
('Corelli', res['lname']))
# But the default collection should still be new_coll,
# so a search with unspecified coll should be in new_coll:
resGen = self.mongodb.query({"recommendation": {
'$regex': '.*'
}},
limit=1)
res = resGen.next()
self.assertEqual(
'Hawaii', res['recommendation'],
"Failed retrieval of single obj; expected '%s' but got '%s'" %
('Hawaii', res['recommendation']))
@unittest.skipIf(not TEST_ALL, "Skipping")
def test_multi_result(self):
# Insert two docs with fname == Franco:
self.mongodb.insert(self.objs[0])
self.mongodb.insert(self.objs[2])
resGen = self.mongodb.query({"fname": "Franco"})
# To get result count, must retrieve at least one result first:
resGen.next()
resCount = self.mongodb.resultCount({"fname": "Franco"})
if resCount != 2:
self.fail("Added two Franco objects, but only %s are found." %
str(resCount))
@unittest.skipIf(not TEST_ALL, "Skipping")
def test_clear_collection(self):
self.mongodb.insert({"foo": 10})
resGen = self.mongodb.query({"foo": 10}, limit=1)
res = resGen.next()
self.assertIsNotNone(res,
"Did not find document that was just inserted.")
self.mongodb.clearCollection()
resGen = self.mongodb.query({"foo": 10}, limit=1)
self.assertRaises(StopIteration, resGen.next)
@unittest.skipIf(not TEST_ALL, "Skipping")
def test_only_some_return_columns(self):
# Also tests the suppression of _id col when desired:
self.mongodb.insert(self.objs[0])
self.mongodb.insert(self.objs[1])
resGen = self.mongodb.query({}, ("lname"))
names = []
for lnameDict in resGen:
resCount = self.mongodb.resultCount({})
self.assertEqual(2, resCount)
names.append(lnameDict['lname'])
self.assertItemsEqual(['Corelli', 'DaVinci'], names,
"Did not receive expected lnames: %s" %
str(names))
class TestForumEtl(unittest.TestCase):
# Forum rows have the following columns:
# type, anonymous, anonymous_to_peers, at_position_list, user_int_id, body, course_display_name, created_at, votes, count, down_count, up_count, up, down, comment_thread_id, parent_id, parent_ids, sk
# Correct result for relationization of tinyForum.json
# (in <projDir>/src/forum_etl/data). This result is anonymized and not relatable,
# i.e. poster name UIDs use integers, while other tables use hashes:
tinyForumGoldAnonymized = \
[
# poster Otto van Homberg: body is clean to start with:
('anon_screen_name_redacted','CommentThread', 'False', 'False', '[]', 5L, 'Harmless body', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 20), "{u'count': 10, u'point': -6, u'down_count': 8, u'up': [u'2', u'10'], u'down': [u'1', u'3', u'4', u'5', u'6', u'7', u'8', u'9'], u'up_count': 2}", 10L, 8L, 2L, "['2', '10']", "['1', '3', '4', '5', '6', '7', '8', '9']", None, None, None, None),
# poster Andreas Fritz: body has someone's email:
('anon_screen_name_redacted','Comment', 'False', 'False', '[]', 7L, ' Body with <emailRedac> email.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 10, u'point': -4, u'down_count': 7, u'up': [u'6', u'8', u'10'], u'down': [u'1', u'2', u'3', u'4', u'5', u'7', u'9'], u'up_count': 3}", 10L, 7L, 3L, "['6', '8', '10']", "['1', '2', '3', '4', '5', '7', '9']", '519461545924670200000001', None, '[]', '519461555924670200000006'),
# poster Otto van Homberg: body has 'Otto':
('anon_screen_name_redacted','Comment', 'False', 'False', '[]', 5L, 'Body with poster name <nameRedac_anon_screen_name_redacted> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Andreas Fritz: body has a phone number:
('anon_screen_name_redacted','Comment', 'False', 'False', '[]', 10L, 'Body with <phoneRedac> a phone number.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461545924670200000005', "[u'519461545924670200000005']", '519461545924670200000005-519461555924670200000008'),
# poster Otto van Homberg: body has his screen name (otto_king):
('anon_screen_name_redacted','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name <nameRedac_anon_screen_name_redacted> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Otto van Homberg: body has his full name (Otto van Homberg):
('anon_screen_name_redacted','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name <nameRedac_anon_screen_name_redacted> <nameRedac_anon_screen_name_redacted> <nameRedac_anon_screen_name_redacted> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007')
]
# Gold result for anonymization that allows relating to other tables (i.e. hashes are constant)
tinyForumGoldRelatable = \
[
# poster Otto van Homberg: body is clean to start with:
('abc','CommentThread', 'False', 'False', '[]', 5L, 'Harmless body', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 20), "{u'count': 10, u'point': -6, u'down_count': 8, u'up': [u'2', u'10'], u'down': [u'1', u'3', u'4', u'5', u'6', u'7', u'8', u'9'], u'up_count': 2}", 10L, 8L, 2L, "['2', '10']", "['1', '3', '4', '5', '6', '7', '8', '9']", None, None, None, None),
# poster Andreas Fritz: body has someone's email:
('def','Comment', 'False', 'False', '[]', 7L, ' Body with <emailRedac> email.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 10, u'point': -4, u'down_count': 7, u'up': [u'6', u'8', u'10'], u'down': [u'1', u'2', u'3', u'4', u'5', u'7', u'9'], u'up_count': 3}", 10L, 7L, 3L, "['6', '8', '10']", "['1', '2', '3', '4', '5', '7', '9']", '519461545924670200000001', None, '[]', '519461555924670200000006'),
# poster Otto van Homberg: body has 'Otto':
('abc','Comment', 'False', 'False', '[]', 5L, 'Body with poster name <nameRedac_abc> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Andreas Fritz: body has a phone number:
('ghi','Comment', 'False', 'False', '[]', 10L, 'Body with <phoneRedac> a phone number.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461545924670200000005', "[u'519461545924670200000005']", '519461545924670200000005-519461555924670200000008'),
# poster Otto van Homberg: body has his screen name (otto_king):
('abc','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name <nameRedac_abc> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Otto van Homberg: body has his full name (Otto van Homberg):
('abc','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name <nameRedac_abc> <nameRedac_abc> <nameRedac_abc> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007')
]
# Gold result for non-anonymized forum:
tinyForumGoldClear = \
[
# poster Otto van Homberg: body is clean to start with:
('otto_king','CommentThread', 'False', 'False', '[]', 5L, 'Harmless body', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 20), "{u'count': 10, u'point': -6, u'down_count': 8, u'up': [u'2', u'10'], u'down': [u'1', u'3', u'4', u'5', u'6', u'7', u'8', u'9'], u'up_count': 2}", 10L, 8L, 2L, "['2', '10']", "['1', '3', '4', '5', '6', '7', '8', '9']", None, None, None, None),
# poster Andreas Fritz: body has someone's email:
('fritzL','Comment', 'False', 'False', '[]', 7L, ' Body with [email protected] email.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 10, u'point': -4, u'down_count': 7, u'up': [u'6', u'8', u'10'], u'down': [u'1', u'2', u'3', u'4', u'5', u'7', u'9'], u'up_count': 3}", 10L, 7L, 3L, "['6', '8', '10']", "['1', '2', '3', '4', '5', '7', '9']", '519461545924670200000001', None, '[]', '519461555924670200000006'),
# poster Otto van Homberg: body has 'Otto':
('otto_king','Comment', 'False', 'False', '[]', 5L, 'Body with poster name Otto embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Andreas Fritz: body has a phone number:
('bebeW','Comment', 'False', 'False', '[]', 10L, 'Body with 650-333-4567 a phone number.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461545924670200000005', "[u'519461545924670200000005']", '519461545924670200000005-519461555924670200000008'),
# poster Otto van Homberg: body has his screen name (otto_king):
('otto_king','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name otto_king embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Otto van Homberg: body has his full name (Otto van Homberg):
('otto_king','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name Otto van Homberg embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007')
]
def setUp(self):
self.mongoDb = MongoDB(dbName="unittest", collection="tinyForum")
# Fill the little MongoDB with test JSON lines
self.resetMongoTestDb()
self.mysqldb = MySQLDB(mySQLUser='******', db='unittest')
# Start with an empty result MySQL table for each test:
self.mysqldb.dropTable('contents')
# Fill the fake UserGrade table with records of course participants:
self.resetMySQLUserListDb()
# Instantiate a Forum scrubber without the
# name of a bson file that contains forum
# records. That 'None' for the bson file will
# make the class understand that it's being
# instantiated for a unit test.
self.forumScrubberAnonymized = EdxForumScrubber(None, mysqlDbObj=self.mysqldb, forumTableName='contents', allUsersTableName='unittest.UserGrade')
self.forumScrubberRelatable = EdxForumScrubber(None, mysqlDbObj=self.mysqldb, forumTableName='contents', allUsersTableName='unittest.UserGrade', allowAnonScreenName=True)
self.forumScrubberClear = EdxForumScrubber(None, mysqlDbObj=self.mysqldb, forumTableName='contents', allUsersTableName='unittest.UserGrade', anonymize=False)
def tearDown(self):
self.mysqldb.close()
@unittest.skipIf(not RUN_ALL_TESTS,
'Uncomment this decoration if RUN_ALL_TESTS is False, and you want to run just this test.')
def testAnonymized(self):
self.forumScrubberAnonymized.populateUserCache()
self.forumScrubberAnonymized.forumMongoToRelational(self.mongoDb, self.mysqldb, 'contents')
for rowNum, forumPost in enumerate(self.mysqldb.query('SELECT * FROM unittest.contents')):
# print(str(rowNum) + ':' + str(forumPost))
self.assertEqual(TestForumEtl.tinyForumGoldAnonymized[rowNum], forumPost)
@unittest.skipIf(not RUN_ALL_TESTS,
'Uncomment this decoration if RUN_ALL_TESTS is False, and you want to run just this test.')
def testNonAnonymizedRelatable(self):
self.forumScrubberRelatable.populateUserCache()
self.forumScrubberRelatable.forumMongoToRelational(self.mongoDb, self.mysqldb, 'contents')
for rowNum, forumPost in enumerate(self.mysqldb.query('SELECT * FROM unittest.contents')):
# print(str(rowNum) + ':' + str(forumPost))
self.assertEqual(TestForumEtl.tinyForumGoldRelatable[rowNum], forumPost)
@unittest.skipIf(not RUN_ALL_TESTS,
'Uncomment this decoration if RUN_ALL_TESTS is False, and you want to run just this test.')
def testNonAnonymized(self):
self.forumScrubberClear.populateUserCache()
self.forumScrubberClear.forumMongoToRelational(self.mongoDb, self.mysqldb, 'contents')
for rowNum, forumPost in enumerate(self.mysqldb.query('SELECT * FROM unittest.contents')):
# print(str(rowNum) + ':' + str(forumPost))
self.assertEqual(TestForumEtl.tinyForumGoldClear[rowNum], forumPost)
def resetMongoTestDb(self):
self.mongoDb.clearCollection()
# Use small, known forum collection:
currDir = os.path.dirname(__file__)
with open(os.path.join(currDir, 'data/tinyForum.json'), 'r') as jsonFd:
for line in jsonFd:
forumPost = json.loads(line)
self.mongoDb.insert(forumPost)
def resetMySQLUserListDb(self):
'''
Prepare a MySQL table that mimicks EdxPrivate.UserGrade.
'''
userGradeColSpecs = OrderedDict(
{
'name' : 'varchar(255)',
'screen_name' : 'varchar(255)',
'grade' : 'int',
'course_id' : 'varchar(255)',
'distinction' : 'tinyint',
'status' : 'varchar(50)',
'user_int_id' : 'int(11)',
'anon_screen_name' : 'varchar(40)'
})
self.mysqldb.dropTable('UserGrade')
self.mysqldb.createTable('UserGrade', userGradeColSpecs)
self.mysqldb.bulkInsert('UserGrade',
('name','screen_name','grade','course_id','distinction','status','user_int_id','anon_screen_name'),
[
('Otto van Homberg','otto_king',5,'oldCourse',0,'notpassing',5,'abc'),
('Andreas Fritz','fritzL',2,'newCourse',0,'notpassing',7,'def'),
('Bebe Winter', 'bebeW',10,'History of Baking',1,'passing',10,'ghi')
])
class TestForumEtl(unittest.TestCase):
# Forum rows have the following columns:
# type, anonymous, anonymous_to_peers, at_position_list, user_int_id, body, course_display_name, created_at, votes, count, down_count, up_count, up, down, comment_thread_id, parent_id, parent_ids, sk
# Correct result for relationization of tinyForum.json
# (in <projDir>/src/forum_etl/data). This result is anonymized and not relatable,
# i.e. poster name UIDs use integers, while other tables use hashes:
tinyForumGoldAnonymized = \
[
# poster Otto van Homberg: body is clean to start with:
('anon_screen_name_redacted','CommentThread', 'False', 'False', '[]', 5L, 'Harmless body', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 20), "{u'count': 10, u'point': -6, u'down_count': 8, u'up': [u'2', u'10'], u'down': [u'1', u'3', u'4', u'5', u'6', u'7', u'8', u'9'], u'up_count': 2}", 10L, 8L, 2L, "['2', '10']", "['1', '3', '4', '5', '6', '7', '8', '9']", None, None, None, None),
# poster Andreas Fritz: body has someone's email:
('anon_screen_name_redacted','Comment', 'False', 'False', '[]', 7L, ' Body with <emailRedac> email.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 10, u'point': -4, u'down_count': 7, u'up': [u'6', u'8', u'10'], u'down': [u'1', u'2', u'3', u'4', u'5', u'7', u'9'], u'up_count': 3}", 10L, 7L, 3L, "['6', '8', '10']", "['1', '2', '3', '4', '5', '7', '9']", '519461545924670200000001', None, '[]', '519461555924670200000006'),
# poster Otto van Homberg: body has 'Otto':
('anon_screen_name_redacted','Comment', 'False', 'False', '[]', 5L, 'Body with poster name <nameRedac_anon_screen_name_redacted> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Andreas Fritz: body has a phone number:
('anon_screen_name_redacted','Comment', 'False', 'False', '[]', 10L, 'Body with <phoneRedac> a phone number.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461545924670200000005', "[u'519461545924670200000005']", '519461545924670200000005-519461555924670200000008'),
# poster Otto van Homberg: body has his screen name (otto_king):
('anon_screen_name_redacted','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name <nameRedac_anon_screen_name_redacted> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Otto van Homberg: body has his full name (Otto van Homberg):
('anon_screen_name_redacted','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name <nameRedac_anon_screen_name_redacted> <nameRedac_anon_screen_name_redacted> <nameRedac_anon_screen_name_redacted> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007')
]
# Gold result for anonymization that allows relating to other tables (i.e. hashes are constant)
tinyForumGoldRelatable = \
[
# poster Otto van Homberg: body is clean to start with:
('abc','CommentThread', 'False', 'False', '[]', 5L, 'Harmless body', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 20), "{u'count': 10, u'point': -6, u'down_count': 8, u'up': [u'2', u'10'], u'down': [u'1', u'3', u'4', u'5', u'6', u'7', u'8', u'9'], u'up_count': 2}", 10L, 8L, 2L, "['2', '10']", "['1', '3', '4', '5', '6', '7', '8', '9']", None, None, None, None),
# poster Andreas Fritz: body has someone's email:
('def','Comment', 'False', 'False', '[]', 7L, ' Body with <emailRedac> email.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 10, u'point': -4, u'down_count': 7, u'up': [u'6', u'8', u'10'], u'down': [u'1', u'2', u'3', u'4', u'5', u'7', u'9'], u'up_count': 3}", 10L, 7L, 3L, "['6', '8', '10']", "['1', '2', '3', '4', '5', '7', '9']", '519461545924670200000001', None, '[]', '519461555924670200000006'),
# poster Otto van Homberg: body has 'Otto':
('abc','Comment', 'False', 'False', '[]', 5L, 'Body with poster name <nameRedac_abc> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Andreas Fritz: body has a phone number:
('ghi','Comment', 'False', 'False', '[]', 10L, 'Body with <phoneRedac> a phone number.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461545924670200000005', "[u'519461545924670200000005']", '519461545924670200000005-519461555924670200000008'),
# poster Otto van Homberg: body has his screen name (otto_king):
('abc','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name <nameRedac_abc> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Otto van Homberg: body has his full name (Otto van Homberg):
('abc','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name <nameRedac_abc> <nameRedac_abc> <nameRedac_abc> embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007')
]
# Gold result for non-anonymized forum:
tinyForumGoldClear = \
[
# poster Otto van Homberg: body is clean to start with:
('otto_king','CommentThread', 'False', 'False', '[]', 5L, 'Harmless body', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 20), "{u'count': 10, u'point': -6, u'down_count': 8, u'up': [u'2', u'10'], u'down': [u'1', u'3', u'4', u'5', u'6', u'7', u'8', u'9'], u'up_count': 2}", 10L, 8L, 2L, "['2', '10']", "['1', '3', '4', '5', '6', '7', '8', '9']", None, None, None, None),
# poster Andreas Fritz: body has someone's email:
('fritzL','Comment', 'False', 'False', '[]', 7L, ' Body with [email protected] email.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 10, u'point': -4, u'down_count': 7, u'up': [u'6', u'8', u'10'], u'down': [u'1', u'2', u'3', u'4', u'5', u'7', u'9'], u'up_count': 3}", 10L, 7L, 3L, "['6', '8', '10']", "['1', '2', '3', '4', '5', '7', '9']", '519461545924670200000001', None, '[]', '519461555924670200000006'),
# poster Otto van Homberg: body has 'Otto':
('otto_king','Comment', 'False', 'False', '[]', 5L, 'Body with poster name Otto embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Andreas Fritz: body has a phone number:
('bebeW','Comment', 'False', 'False', '[]', 10L, 'Body with 650-333-4567 a phone number.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461545924670200000005', "[u'519461545924670200000005']", '519461545924670200000005-519461555924670200000008'),
# poster Otto van Homberg: body has his screen name (otto_king):
('otto_king','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name otto_king embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007'),
# poster Otto van Homberg: body has his full name (Otto van Homberg):
('otto_king','Comment', 'False', 'False', '[]', 5L, 'Body with poster screen name Otto van Homberg embedded.', 'MITx/6.002x/2012_Fall', datetime.datetime(2013, 5, 16, 4, 32, 21), "{u'count': 0, u'point': 0, u'down_count': 0, u'up': [], u'down': [], u'up_count': 0}", 0L, 0L, 0L, '[]', '[]', '519461545924670200000001', '519461555924670200000006', "[u'519461555924670200000006']", '519461555924670200000006-519461555924670200000007')
]
def setUp(self):
self.mongoDb = MongoDB(dbName="unittest", collection="tinyForum")
# Fill the little MongoDB with test JSON lines
self.resetMongoTestDb()
self.mysqldb = MySQLDB(user='******', db='unittest')
# Start with an empty result MySQL table for each test:
self.mysqldb.dropTable('contents')
# Fill the fake UserGrade table with records of course participants:
self.resetMySQLUserListDb()
# Instantiate a Forum scrubber without the
# name of a bson file that contains forum
# records. That 'None' for the bson file will
# make the class understand that it's being
# instantiated for a unit test.
self.forumScrubberAnonymized = EdxForumScrubber(None, mysqlDbObj=self.mysqldb, forumTableName='contents', allUsersTableName='unittest.UserGrade')
self.forumScrubberRelatable = EdxForumScrubber(None, mysqlDbObj=self.mysqldb, forumTableName='contents', allUsersTableName='unittest.UserGrade', allowAnonScreenName=True)
self.forumScrubberClear = EdxForumScrubber(None, mysqlDbObj=self.mysqldb, forumTableName='contents', allUsersTableName='unittest.UserGrade', anonymize=False)
def tearDown(self):
self.mysqldb.close()
@unittest.skipIf(not RUN_ALL_TESTS,
'Uncomment this decoration if RUN_ALL_TESTS is False, and you want to run just this test.')
def testAnonymized(self):
self.forumScrubberAnonymized.populateUserCache()
self.forumScrubberAnonymized.forumMongoToRelational(self.mongoDb, self.mysqldb, 'contents')
for rowNum, forumPost in enumerate(self.mysqldb.query('SELECT * FROM unittest.contents')):
# print(str(rowNum) + ':' + str(forumPost))
self.assertEqual(TestForumEtl.tinyForumGoldAnonymized[rowNum], forumPost)
@unittest.skipIf(not RUN_ALL_TESTS,
'Uncomment this decoration if RUN_ALL_TESTS is False, and you want to run just this test.')
def testNonAnonymizedRelatable(self):
self.forumScrubberRelatable.populateUserCache()
self.forumScrubberRelatable.forumMongoToRelational(self.mongoDb, self.mysqldb, 'contents')
for rowNum, forumPost in enumerate(self.mysqldb.query('SELECT * FROM unittest.contents')):
# print(str(rowNum) + ':' + str(forumPost))
self.assertEqual(TestForumEtl.tinyForumGoldRelatable[rowNum], forumPost)
@unittest.skipIf(not RUN_ALL_TESTS,
'Uncomment this decoration if RUN_ALL_TESTS is False, and you want to run just this test.')
def testNonAnonymized(self):
self.forumScrubberClear.populateUserCache()
self.forumScrubberClear.forumMongoToRelational(self.mongoDb, self.mysqldb, 'contents')
for rowNum, forumPost in enumerate(self.mysqldb.query('SELECT * FROM unittest.contents')):
# print(str(rowNum) + ':' + str(forumPost))
self.assertEqual(TestForumEtl.tinyForumGoldClear[rowNum], forumPost)
def resetMongoTestDb(self):
self.mongoDb.clearCollection()
# Use small, known forum collection:
currDir = os.path.dirname(__file__)
with open(os.path.join(currDir, 'data/tinyForum.json'), 'r') as jsonFd:
for line in jsonFd:
forumPost = json.loads(line)
self.mongoDb.insert(forumPost)
def resetMySQLUserListDb(self):
'''
Prepare a MySQL table that mimicks EdxPrivate.UserGrade.
'''
userGradeColSpecs = OrderedDict(
{
'name' : 'varchar(255)',
'screen_name' : 'varchar(255)',
'grade' : 'int',
'course_id' : 'varchar(255)',
'distinction' : 'tinyint',
'status' : 'varchar(50)',
'user_int_id' : 'int(11)',
'anon_screen_name' : 'varchar(40)'
})
self.mysqldb.dropTable('UserGrade')
self.mysqldb.createTable('UserGrade', userGradeColSpecs)
self.mysqldb.bulkInsert('UserGrade',
('name','screen_name','grade','course_id','distinction','status','user_int_id','anon_screen_name'),
[
('Otto van Homberg','otto_king',5,'oldCourse',0,'notpassing',5,'abc'),
('Andreas Fritz','fritzL',2,'newCourse',0,'notpassing',7,'def'),
('Bebe Winter', 'bebeW',10,'History of Baking',1,'passing',10,'ghi')
])
class MongoTest(unittest.TestCase):
'''
Test the mongodb.py module. Uses a library that fakes
a MongoDB server. See https://pypi.python.org/pypi/mongomock/1.0.1
'''
def setUp(self):
self.objs = [{"fname" : "Franco", "lname" : "Corelli"},
{"fname" : "Leonardo", "lname" : "DaVinci", "age" : 300},
{"fname" : "Franco", "lname" : "Gandolpho"}]
self.mongodb = MongoDB(dbName='unittest', collection='unittest')
self.mongodb.clearCollection(collection="unittest")
self.mongodb.clearCollection(collection="new_coll")
self.mongodb.setCollection("unittest")
def tearDown(self):
self.mongodb.dropCollection(collection='unittest')
self.mongodb.dropCollection(collection='new_coll')
self.mongodb.close()
@unittest.skipIf(not TEST_ALL, "Skipping")
def test_update_and_find_one(self):
self.mongodb.insert(self.objs[0])
# Get a generator for the results:
resGen = self.mongodb.query({"fname" : "Franco"}, limit=1, collection="unittest")
res = resGen.next()
self.assertEqual('Corelli', res['lname'], "Failed retrieval of single obj; expected '%s' but got '%s'" % ('Corelli', res['lname']))
@unittest.skipIf(not TEST_ALL, "Skipping")
def test_set_coll_use_different_coll(self):
# Insert into unittest:
self.mongodb.insert(self.objs[0])
# Switch to new_coll:
self.mongodb.setCollection('new_coll')
self.mongodb.insert({"recommendation" : "Hawaii"})
# We're in new_coll; the following should be empty result:
self.mongodb.query({"fname" : "Franco"}, limit=1)
resCount = self.mongodb.resultCount({"fname" : "Franco"})
self.assertIsNone(resCount, "Got non-null result that should be null: %s" % resCount)
# But this search is within new_coll, and should succeed:
resGen = self.mongodb.query({"recommendation" : {'$regex' : '.*'}}, limit=1)
res = resGen.next()
self.assertEqual('Hawaii', res['recommendation'], "Failed retrieval of single obj; expected '%s' but got '%s'" % ('Hawaii', res['recommendation']))
# Try inline collection switch:
resGen = self.mongodb.query({"fname" : "Franco"}, limit=1, collection="unittest")
res = resGen.next()
self.assertEqual('Corelli', res['lname'], "Failed retrieval of single obj; expected '%s' but got '%s'" % ('Corelli', res['lname']))
# But the default collection should still be new_coll,
# so a search with unspecified coll should be in new_coll:
resGen = self.mongodb.query({"recommendation" : {'$regex' : '.*'}}, limit=1)
res = resGen.next()
self.assertEqual('Hawaii', res['recommendation'], "Failed retrieval of single obj; expected '%s' but got '%s'" % ('Hawaii', res['recommendation']))
@unittest.skipIf(not TEST_ALL, "Skipping")
def test_multi_result(self):
# Insert two docs with fname == Franco:
self.mongodb.insert(self.objs[0])
self.mongodb.insert(self.objs[2])
resGen = self.mongodb.query({"fname" : "Franco"})
# To get result count, must retrieve at least one result first:
resGen.next()
resCount = self.mongodb.resultCount({"fname" : "Franco"})
if resCount != 2:
self.fail("Added two Franco objects, but only %s are found." % str(resCount))
@unittest.skipIf(not TEST_ALL, "Skipping")
def test_clear_collection(self):
self.mongodb.insert({"foo" : 10})
resGen = self.mongodb.query({"foo" : 10}, limit=1)
res = resGen.next()
self.assertIsNotNone(res, "Did not find document that was just inserted.")
self.mongodb.clearCollection()
resGen = self.mongodb.query({"foo" : 10}, limit=1)
self.assertRaises(StopIteration, resGen.next)
@unittest.skipIf(not TEST_ALL, "Skipping")
def test_only_some_return_columns(self):
# Also tests the suppression of _id col when desired:
self.mongodb.insert(self.objs[0])
self.mongodb.insert(self.objs[1])
resGen = self.mongodb.query({}, ("lname"))
names = []
for lnameDict in resGen:
resCount = self.mongodb.resultCount({})
self.assertEqual(2, resCount)
names.append(lnameDict['lname'])
self.assertItemsEqual(['Corelli','DaVinci'], names, "Did not receive expected lnames: %s" % str(names))
