def generate_rand_N(self, p, n_generate=20, max_length=MAX_LEN):
generate_list = []
samples = random.sample(p.passwords_string, n_generate)
for i in range(n_generate):
start_letter = samples[i][0]
input_tensor = P.passwordToInputTensor(start_letter)
with torch.no_grad():
hidden = torch.rand(self.layers, 1,
self.hidden_size).to(device)
output_password = start_letter
for c in range(max_length):
output, hidden = self.gru(self.embedding(input_tensor),
hidden)
output = self.h2o(output)
output = output.view(-1)
output = F.softmax(output, dim=0)
output = output.cpu().numpy()
index = np.random.choice(range(len(output)), p=output)
if index == CHARMAP_LEN - 1:
break
else:
letter = P.all_letters[index]
output_password += letter
input_tensor = P.passwordToInputTensor(letter).to(device)
generate_list.append(output_password)
return generate_list
def generate_N(self, p, n_generate=20, max_length=MAX_LEN):
generate_list = []
samples = random.sample(p.passwords_string, n_generate)
for i in range(n_generate):
start_letter = samples[i][0]
input_tensor = P.passwordToInputTensor(start_letter)
with torch.no_grad():
hidden = torch.rand(self.layers, 1,
self.hidden_size).to(device)
output_password = start_letter
for c in range(max_length):
output, hidden = self.gru(self.embedding(input_tensor),
hidden)
output = self.h2o(output)
output = output.view(1, -1)
topv, topi = output.topk(1)
topi = topi[0][0]
if topi == CHARMAP_LEN - 1:
break
else:
letter = P.all_letters[topi]
output_password += letter
input_tensor = P.passwordToInputTensor(letter).to(device)
generate_list.append(output_password)
return generate_list
def generate_from(self, start_letter, max_length=MAX_LEN):
input_tensor = P.passwordToInputTensor(start_letter)
with torch.no_grad():
hidden = torch.rand(self.layers, 1, self.hidden_size).to(device)
output_password = start_letter
for c in range(max_length):
output, hidden = self.gru(self.embedding(input_tensor), hidden)
output = self.h2o(output)
output = output.view(1, -1)
topv, topi = output.topk(1)
topi = topi[0][0]
if topi == CHARMAP_LEN - 1:
break
else:
letter = P.all_letters[topi]
output_password += letter
input_tensor = P.passwordToInputTensor(letter).to(device)
return output_password
def generatePassTensor(self, max_length = 18):
start_letter = p.passwords_string[random.randint(0,len(p.passwords_string) - 1)][0]
with torch.no_grad():
input_tensor = P.passwordToInputTensor(start_letter).to(device)
self.hidden = self.initHiddenZeros()
password = start_letter
for c in range(max_length):
output = self(input_tensor[0])
output = output.view(1,-1)
topv, topi = output.topk(1)
topi = topi[0][0]
if topi == P.n_letters - 1:
break
else:
letter = P.all_letters[topi]
password += letter
input_tensor = P.passwordToInputTensor(letter).to(device)
return P.passwordToInputTensor(password)