def test_request_similars(self):
"""
Check basic operation of DBEndpoint
"""
mfe = MockFrameEndpoint()
dbe = DBEndpoint(mfe,'my_db_name')
# We expect the CHOOSE_DB message to be here:
self.assertEqual(len(mfe.out_list),1)
frame = mfe.out_list.pop(0)
assert 'my_db_name' in frame
func_data = '029348509238459kldfjaklsdjflkasjdfklasdf'
num_similars = 8
# Prepare a response for the REQUEST_SIMILARS message ahead of time:
# Two results:
name1 = 'name1'
comment1 = 'comment1'
sim_grade1 = 7
name2 = 'name1'
comment2 = 'comment1'
sim_grade2 = 7
msg = ""
# Two records:
msg += struct.pack('I',2)
# First record:
msg += struct.pack('I',len(name1))
msg += name1
msg += struct.pack('I',len(comment1))
msg += comment1
msg += struct.pack('I',sim_grade1)
# Second record:
msg += struct.pack('I',len(name2))
msg += name2
msg += struct.pack('I',len(comment2))
msg += comment2
msg += struct.pack('I',sim_grade2)
# Add message type:
frame = dword_pack(MsgTypes.RESPONSE_SIMILARS,msg)
mfe.in_list.append(frame)
dbe.request_similars(func_data,num_similars)
similars = dbe.response_similars()
self.assertEqual(len(mfe.out_list),1)
frame = mfe.out_list.pop(0)
assert func_data in frame
# We expect two results:
self.assertEqual(len(similars),2)
self.assertIsInstance(similars[0],FSimilar)
dbe.close()
def test_basic_db_function(self):
# Get a random db name:
db_name = rand_db_name()
frame_endpoint = TCPFrameClient(remote)
dbe = DBEndpoint(frame_endpoint,db_name)
# A three somewhat similar functions:
func_name1 = 'func_name1'
func_comment1 = 'func_comment1'
func_data1 = '230948509238459238459283409582309458230945'
func_name2 = 'func_name2'
func_comment2 = 'func_comment2'
func_data2 = '230948509218459238459223409582309458230945'
func_name3 = 'func_name3'
func_comment3 = 'func_comment3'
func_data3 = '230948509018459238459223409280309458030945'
# A very different function:
func_name4 = 'func_name4'
func_comment4 = 'func_comment4'
func_data4 = 'kasjflkasjfdlkasjdfoiuweoriuqwioreuwqioekaskldfjaslk'
dbe.add_function(func_name1,func_comment1,func_data1)
dbe.add_function(func_name2,func_comment2,func_data2)
dbe.add_function(func_name3,func_comment3,func_data3)
dbe.add_function(func_name4,func_comment4,func_data4)
# Check if the amount of returned functions is reasonable:
dbe.request_similars(func_data1,1)
similars = dbe.response_similars()
self.assertEqual(len(similars),1)
dbe.request_similars(func_data1,2)
similars = dbe.response_similars()
self.assertEqual(len(similars),2)
dbe.request_similars(func_data1,3)
similars = dbe.response_similars()
self.assertEqual(len(similars),3)
dbe.request_similars(func_data1,4)
similars = dbe.response_similars()
self.assertEqual(len(similars),3)
self.assertEqual(similars[0].name,func_name1)
self.assertEqual(similars[0].comment,func_comment1)
self.assertEqual(similars[0].sim_grade,NUM_HASHES)
# Function 2 is second place with respect to similarity to function 1:
self.assertEqual(similars[1].name,func_name2)
self.assertLess(similars[1].sim_grade,NUM_HASHES)
# Function 3 is third place:
self.assertEqual(similars[2].name,func_name3)
self.assertLess(similars[2].sim_grade,NUM_HASHES)
# function 4 is the only function that looks like function 4 in this
# dataset:
dbe.request_similars(func_data4,3)
similars = dbe.response_similars()
self.assertEqual(len(similars),1)
self.assertEqual(similars[0].name,func_name4)
dbe.close()
# Check persistency of the database by opening the same one again and
# running a query:
frame_endpoint = TCPFrameClient(remote)
dbe = DBEndpoint(frame_endpoint,db_name)
dbe.request_similars(func_data1,4)
similars = dbe.response_similars()
self.assertEqual(len(similars),3)
self.assertEqual(similars[0].name,func_name1)
self.assertEqual(similars[0].comment,func_comment1)
self.assertEqual(similars[0].sim_grade,NUM_HASHES)
dbe.close()
