def test_getNearestDocument_usingANearbyKeyInTrie(self):
digitReplacement = {'0': '1', '1': '0'}
newDocWithANearbySignature = Document(3, VectorGenerator.getRandomGaussianUnitVector(dimension=self.dimension, mu=0, sigma=1))
exactSignature = self.doc1.signature.to01()
newDocWithANearbySignature.signature = Signature(exactSignature[:-1]+digitReplacement[exactSignature[-1]])
self.assertNotEquals(self.doc1.signature.to01(), newDocWithANearbySignature.signature.to01())
self.assertEqual(self.pm.getNearestDocuments(newDocWithANearbySignature), set([1])) # This assertion can sometimes fail because of randomization. Run the tests again. It's OK!
def test_removeDocument_documents(self):
newDocModifiedWithExistingSignature = Document(3, VectorGenerator.getRandomGaussianUnitVector(dimension=self.dimension, mu=0, sigma=1))
newDocModifiedWithExistingSignature.signature = Signature(self.doc1.signature.to01())
self.pm.addDocument(newDocModifiedWithExistingSignature)
self.assertEqual(self.pm.signatureTrie[self.doc1.signature.permutate(self.pm).to01()], set([1, 3]))
self.pm.removeDocument(newDocModifiedWithExistingSignature)
self.assertEqual(self.pm.signatureTrie[self.doc1.signature.permutate(self.pm).to01()], set([1]))
self.pm.removeDocument(self.doc1)
self.assertEqual(None, self.pm.signatureTrie.get(self.doc1.signature.permutate(self.pm).to01()))
def test_addDocument_existingKey(self):
newDocModifiedWithExistingSignature = Document(
3,
VectorGenerator.getRandomGaussianUnitVector(
dimension=self.dimension, mu=0, sigma=1))
newDocModifiedWithExistingSignature.signature = Signature(
self.doc1.signature.to01())
self.pm.addDocument(newDocModifiedWithExistingSignature)
self.assertEqual(
self.pm.signatureTrie[self.doc1.signature.permutate(
self.pm).to01()], set([1, 3]))
def test_getNearestDocument_usingANearbyKeyInTrie(self):
digitReplacement = {'0': '1', '1': '0'}
newDocWithANearbySignature = Document(
3,
VectorGenerator.getRandomGaussianUnitVector(
dimension=self.dimension, mu=0, sigma=1))
exactSignature = self.doc1.signature.to01()
newDocWithANearbySignature.signature = Signature(
exactSignature[:-1] + digitReplacement[exactSignature[-1]])
self.assertNotEquals(self.doc1.signature.to01(),
newDocWithANearbySignature.signature.to01())
self.assertEqual(
self.pm.getNearestDocuments(newDocWithANearbySignature), set([1])
) # This assertion can sometimes fail because of randomization. Run the tests again. It's OK!
