def test_update(self):
simhash = Simhash()
assert simhash.tokens == []
simhash.update('This is for testing purpose \n It should work fine')
expected = [
'This', 'is', 'for', 'testing', 'purpose', 'It', 'should', 'work',
'fine'
]
assert simhash.tokens == expected
def test_hex_digest3(self):
simhash = Simhash()
test_file = self.get_test_loc('fingerprint/fingerprint-test3.py')
with open(test_file, 'r') as f:
hashable = f.read()
simhash.update(hashable)
result = simhash.hex_digest()
assert result == '7f43e1b18f9c0e705fcf28007bc41754'
def test_hex_digest2(self):
simhash = Simhash()
test_file = self.get_test_loc('fingerprint/fingerprint-test2.c')
with open(test_file, 'r') as f:
hashable = f.read()
simhash.update(hashable)
result = simhash.hex_digest()
assert result == 'baa2d1d169be06a306c1873afe6db4da'
def test_hex_digest1(self):
simhash = Simhash()
test_file = self.get_test_loc('fingerprint/fingerprint-test1.java')
with open(test_file, 'r') as f:
hashable = f.read()
simhash.update(hashable)
result = simhash.hex_digest()
assert result == '4e42d8c0ed6693654866425451210417'
def get_fingerprint(location, **kwargs):
"""
Return a mapping of fingerprint generated for the file at `location`.
"""
with open(location, 'r') as f:
hashable = f.read()
simhash = Simhash()
simhash.update(hashable)
result = simhash.hex_digest()
return dict(fingerprint=result)
def test_generate_fingerprint(self):
simhash = Simhash()
simhash.update('This should work')
expected = bitarray(
'11100010001110100111100010101000101110111111110000010011000110000110001110000000100000111011101110111100110001011011110001011100'
)
assert simhash.generate_fingerprint() == expected
simhash.update('this will get added too!')
expected = bitarray(
'00000010000000000011110010100000101000001111100000000001010110000110101110111000100000110101000000010100100000000010110011010010'
)
assert simhash.generate_fingerprint() == expected
def test_similarity_matching3(self):
simhash1 = Simhash()
simhash2 = Simhash()
test_file1 = self.get_test_loc('fingerprint/similarity_matching5.py')
test_file2 = self.get_test_loc('fingerprint/similarity_matching6.py')
with open(test_file1, 'r') as f:
hashable1 = f.read()
with open(test_file2, 'r') as f:
hashable2 = f.read()
simhash1.update(hashable1)
simhash2.update(hashable2)
distance = simhash1.hamming_distance(simhash1.generate_fingerprint(),
simhash2.generate_fingerprint())
assert distance == 13
def test_get_weighted_hash2(self):
simhash = Simhash()
test_file = self.get_test_loc(
'fingerprint/get_weighted_hash-test2.txt')
with open(test_file, 'r') as f:
hashable = f.read()
simhash.update(hashable)
result = simhash.get_weighted_hash()
expected = [
-1, -1, -1, -1, 1, 1, -1, -1, 1, 1, -1, -1, -1, -1, -1, 1, -1, 1,
1, 1, -1, 1, -1, 1, 1, -1, 1, 1, 1, -1, -1, 1, 1, 1, -1, -1, -1,
-1, -1, -1, 1, 1, 1, 1, -1, -1, -1, 1, 1, -1, 1, 1, -1, 1, 1, -1,
1, -1, 1, -1, 1, -1, -1, -1, -1, -1, 1, 1, -1, -1, -1, 1, 1, 1, -1,
-1, -1, -1, 1, 1, 1, -1, -1, 1, 1, -1, -1, 1, 1, 1, 1, -1, -1, -1,
1, -1, -1, 1, 1, -1, 1, -1, -1, 1, -1, 1, 1, 1, -1, 1, 1, 1, -1,
-1, 1, -1, -1, 1, 1, -1, -1, 1, 1, -1, -1, -1, -1, 1
]
assert result == expected