def is_similar_page(res1, res2, radio=0.85):
if res1 is None or res2 is None:
return False
body1 = res1.text
body2 = res2.text
simhash1 = Simhash(body1.split())
simhash2 = Simhash(body2.split())
calc_radio = simhash1.similarity(simhash2)
if calc_radio > radio:
return True
else:
return False
def IsSimilarPage(res1, res2, radio):
'''
计算页面相似度函数
'''
if res1 is None or res2 is None:
return False
simhash1 = Simhash(str(res1))
simhash2 = Simhash(str(res2))
calc_radio = simhash1.similarity(simhash2)
# print("两个页面的相似度为:%s" % (calc_radio))
if calc_radio >= radio and calc_radio < 0.99:
return True
else:
return False
def print_guesses(self, name):
""" Compares plugins and their similarity hashes for a given connection name. """
plugin_found = False
connection_hash = self.get_hash(name)
for plugin in self.core.plugin_manager.plugins:
plugin_hash = int(plugin.config['target']['sim_hash'])
hash_object = Simhash('', hashbits=64)
hash_object.hash = plugin_hash
similarity = connection_hash.similarity(hash_object)
if similarity > 0.75:
plugin_found = True
print("Found plugin {} with {:.1f}% similarity.".format(
plugin.config['name'], similarity * 100))
if not plugin_found:
print("No matching plugin found")
def is_similar_page(res1, res2, radio):
'''
计算页面相似度函数
'''
if res1 is None or res2 is None:
return False
# body1 = res1.text
# body2 = res2.text
simhash1 = Simhash(str(res1))
simhash2 = Simhash(str(res2))
calc_radio = simhash1.similarity(simhash2)
if calc_radio >= float(radio):
return True
else:
return False
def is_similar_page(res1, res2, radio):
'''
计算页面相似度函数
'''
if res1 is None or res2 is None:
return False
# body1 = res1.text
# body2 = res2.text
simhash1 = Simhash(str(res1))
simhash2 = Simhash(str(res2))
calc_radio = simhash1.similarity(simhash2)
# print("两个页面的相似度为:%s" % (calc_radio))
if calc_radio >= radio:
return True
else:
return False
def get_hash(self, name):
""" Generates a similarity hash for a given connection name. """
serialized_database = self.serialize_database(name)
return Simhash(serialized_database, hashbits=64)