islem-esi

similarity = embed_model.wv.similarity('sorted', 'ord')
most_similar = embed_model.wv.most_similar('len(x)')
vector = embed_model.wv['for']

islem-esi

from gensim.models import FastText

embed_model = FastText(vector_size=meta_hyper['vector_size'], 
                       window=meta_hyper['window'], 
                       min_count = meta_hyper['min_count'], 
                       alpha= meta_hyper['alpha'], 
                       workers=meta_hyper['CPU'])

embed_model.build_vocab(tokenized_data)

islem-esi

meta_hyper = {
    "vector_size": 192, # size of embedding 
    "alpha": 0.025, # learning rate
    "window": 5,
    "min_freq" : 5,
    "epochs": 300, # number of training epochs
    "vocab_size": len(set([*td for td in tokenized_data])), #size of vocabulary
    "data_description": "Add your description here",
    "data_size": len(tokenized_data),
    "tokens_number": len([*td for td in tokenized_data]),

islem-esi

rootdir = "PATH TO THE DIRECTORY OF THE PROJECT YOU CLONED FROM GITHUB"
files_code = load_all_files(root_dir)

tokenized_data = []
for d in files_code:
    try:
        tokenized_data.append(tokenize_python(d))
    except:
        print('error')

islem-esi

from nltk.tokenize import word_tokenize
from tokenize import tokenize
from io import BytesIO

rename_globals = {}

def tokenize_python(code):
    g = tokenize(BytesIO(code.encode('utf-8')).readline)
    try:
        tokens = [c[1] for c in g if c[1]!='' and c[1]!='\n'][1:]

islem-esi

def load_all_files(rootdir):
  files_code = []
  for subdir, dirs, files in os.walk(rootdir):
    for file in files:
        if file.endswith('.py'):
            with open(os.path.join(subdir, file)) as pf:
                try:
                    code = pf.read()
                    files_code.append(code)
                except:

islem-esi

import sumy
from sumy.nlp.tokenizers import Tokenizer
from sumy.parsers.plaintext import PlaintextParser

#LSA algorithm
from sumy.summarizers.lsa import LsaSummarizer
#text: text to summarize
#no_sentences: number of sentences in your summary,
#lang: language of text
def lsa_summary(text, no_sentences, lang):

islem-esi

#do a lot of imports
from keras.models import Sequential
from keras.layers import LSTM, Dense
import numpy as np
from os import listdir
import os.path
import json
from sklearn.preprocessing import OneHotEncoder

islem-esi

inst_groups = {
        #Conditional Data Transfer
        'cdt': ['cmove', 'cmovz', 'cmovne', 'cmovnz', 'cmova', 'cmovnbe', 'cmovae', 'cmovnb', 'cmovb', 
                'cmovnae', 'cmovbe', 'cmovna', 'cmovg',
                'cmovnle', 'cmovge', 'cmovnl', 'cmovl', 'cmovnge', 'cmovle', 'cmovng',
                'cmovc', 'cmovnc', 'cmovo', 'cmovno', 'cmovs', 'cmovns', 'cmovp', 'cmovpe',
                'cmovnp', 'cmovpo',],
        #Unconditianl Data Transfer
        'udt': ['mov', 'xchg', 'bswap', 'movsx', 'movzx', 'movlps', 'movqda', 'lock xchg'],
        #Stack Data Transfer

islem-esi

#Change the parameters to whatever suits you
batch_size = 512 
epochs = 100
labels = [0 for _ in benign_images] + [1 for _ in malicious_images]
model.fit(benign_images+malicious_images, labels,
              batch_size = batch_size,
              epochs = epochs,
              validation_split = 0.25,
              shuffle = True)