1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
|
import joblib
from pandas.core.frame import DataFrame
import os
import re
import numpy as np
class PacketData(object):
"""
预测类：
构造函数1：传入msg和msg处理后的data部分pkt，模型所在的pkl位置
构造函数2：传入一个csv文件，用不到，也没测试过
传msg和pkt进去的时候，最好用try，格式不对会报错
主要方法：
predict：直接预测，返回结果
predict_npy：从默认npy文件处读取一个dict，dict的key是一个tuple，里面是ip对，判断当前包是否载dict里，更新dict的内容并预测
如果dict[ipSet]中结果已经是0（0表示恶意，1表示良性），或者已经判断了20个包，则不更新，直接返回之前的结果
update_helper：帮助更新的方法，不直接调用
"""
def __init__(self, msg, pkt, modelpath='model/lgbm_packetin_model.pkl', feature_dic={}):
if feature_dic == {}:
self.flow_list = [
int(re.search(r'dst_port=\d{1,5}', pkt).group().split('=')[1]), # 目的端口
int(re.search(r'src_port=\d{1,5}', pkt).group().split('=')[1]), # 源端口
int(re.search(r'flags=\d{1,3}', pkt).group().split('=')[1]), # flags
int(re.search(r'window_size=\d{1,10}', pkt).group().split('=')[1]) if 'tcp' in pkt else 0, # win_size
float(re.search(r'reason=\d{1,10}', msg).group().split('=')[1]), # reason
int(re.search(r'\'in_port\': \d{1,5}', msg).group().split(': ')[1]), # in_port
float(re.search(r'total_len=\d{1,5}', msg).group().split('=')[1]), # min_len
float(re.search(r'total_len=\d{1,5}', msg).group().split('=')[1]), # max_len
float(re.search(r'total_len=\d{1,5}', msg).group().split('=')[1]), # means_len
1, # 包数目
1 # 标签,默认是1
]
else: self.flow_list = feature_dic
self.model = joblib.load(modelpath)
self.ipSet = tuple(re.findall(r"\b(?:[0-9]{1,3}\.){3}[0-9]{1,3}\b", pkt))
self.line1 = msg
self.line2 = pkt
# 用不到的，从csv读取数据
@classmethod
def from_csv(cls, csv_path, modelpath='model/lgbm_packetin_model.pkl'):
feature_dic = {}
with open(csv_path, 'r') as f:
pcapNum = 0
while True:
pcapNum += 1
line1 = f.readline().strip()
line2 = f.readline().strip()
if not line2 or not line1: break
if pcapNum >= 20: break
ipSet = tuple(re.findall(r"\b(?:[0-9]{1,3}\.){3}[0-9]{1,3}\b", line2))
try:
if 'tcp' in line2:
window_size = re.search(r'window_size=\d{1,10}', line2).group().split('=')[1]
else:
window_size = 0
if ipSet not in feature_dic:
flow_list = [
re.search(r'dst_port=\d{1,5}', line2).group().split('=')[1], # 目的端口
re.search(r'src_port=\d{1,5}', line2).group().split('=')[1], # 源端口
re.search(r'flags=\d{1,3}', line2).group().split('=')[1], # flags
window_size,
re.search(r'reason=\d{1,10}', line1).group().split('=')[1], # reason
re.search(r'\'in_port\': \d{1,5}', line1).group().split(': ')[1], # in_port
float(re.search(r'total_len=\d{1,5}', line1).group().split('=')[1]), # min_len
float(re.search(r'total_len=\d{1,5}', line1).group().split('=')[1]), # max_len
float(re.search(r'total_len=\d{1,5}', line1).group().split('=')[1]), # means_len
pcapNum, # 包数目
]
feature_dic[ipSet] = flow_list
else:
this_length = float(re.search(r'total_len=\d{1,5}', line1).group().split('=')[1])
feature_dic[ipSet][-2] += 1
feature_dic[ipSet][-3] = (feature_dic[ipSet][-3] * (feature_dic[ipSet][-2] - 1) + this_length) \
/ feature_dic[ipSet][-2]
if this_length > feature_dic[ipSet][-4]:
feature_dic[ipSet][-4] = this_length
else:
feature_dic[ipSet][-5] = this_length
except Exception as e:
pass
return cls('1', '2', feature_dic=feature_dic)
def predict(self):
previous_out = self.flow_list.pop(-1)
pre_out = self.model.predict(DataFrame(self.flow_list).T)
pre_out = pre_out.tolist()[0] # ndarray转int
self.flow_list.append(pre_out)
return pre_out
# 保存到一个dict,和update一样的
def predict_save(self, feature_hash):
return self.update_helper(feature_hash)
# 保存到npy文件
def predict_npy(self, npypath='model/feature_dict.npy', feature_hash=None):
if os.path.exists(npypath):
dict_load = np.load(npypath, allow_pickle=True)
feature_hash = dict_load.item()
pre_out, feature_dict = self.update_helper(feature_hash)
if pre_out == -1: # -1表示不需要写回保存的文件
return feature_dict[self.ipSet][-1]+2
else:
np.save(npypath, feature_dict)
return pre_out
else:
feature_hash = {self.ipSet: self.flow_list}
np.save('model/feature_dict.npy', feature_hash)
return self.predict()
def update_helper(self, feature_hash):
ipSet = self.ipSet
if self.ipSet in feature_hash:
# 如果存在且已至为0，或者num>=20则直接返回
if feature_hash[ipSet][-1] == 0 or feature_hash[ipSet][-2] >=20:
return -1, feature_hash
this_length = float(re.search(r'total_len=\d{1,5}', self.line1).group().split('=')[1])
feature_hash[ipSet][-2] += 1
feature_hash[ipSet][-3] = (feature_hash[ipSet][-3] * (feature_hash[ipSet][-2] - 1) + this_length) \
/ feature_hash[ipSet][-2]
if this_length > feature_hash[ipSet][-4]:
feature_hash[ipSet][-4] = this_length
else:
feature_hash[ipSet][-5] = this_length
feature_hash[ipSet].pop(-1)
pre_out = self.model.predict(DataFrame(feature_hash[ipSet]).T)
pre_out = pre_out.tolist()[0]
feature_hash[ipSet].append(pre_out)
else:
# 如果不存在
self.flow_list.pop(-1) # 先删了默认的tag
pre_out = self.model.predict(DataFrame(self.flow_list).T)
pre_out = pre_out.tolist()[0]
self.flow_list.append(pre_out) # 再加回去
feature_hash[ipSet] = self.flow_list
return pre_out, feature_hash
def predict_mysql(self):
pass
if __name__ == '__main__':
# 来自恶意样本Shifu
line1 = "version=0x4,msg_type=0xa,msg_len=0x6c,xid=0x0,OFPPacketIn(buffer_id=343,cookie=0,data=""RT\x00\x125\x02\x08\x00'\xc1v\xc3\x08\x00E\x00\x004\x002@\x00\x80\x063\xbb\n\x00\x02k\xbc*\xfeA\xc0\x0e\x01\xbb\xf1\x1e=2\x00\x00\x00\x00\x80\x02 \x00\x98$\x00\x00\x02\x04\x05\xb4\x01\x03\x03\x02\x01\x01\x04\x02"",match=OFPMatch(oxm_fields={'in_port': 1}),reason=0,table_id=0,total_len=66)"
line2 = "ethernet(dst='52:54:00:12:35:02',ethertype=2048,src='08:00:27:c1:76:c3'), ipv4(csum=13243,dst='188.42.254.65',flags=2,header_length=5,identification=50,offset=0,option=None,proto=6,src='10.0.2.107',tos=0,total_length=52,ttl=128,version=4), tcp(ack=0,bits=2,csum=38948,dst_port=443,offset=8,option=[TCPOptionMaximumSegmentSize(kind=2,length=4,max_seg_size=1460), TCPOptionNoOperation(kind=1,length=1), TCPOptionWindowScale(kind=3,length=3,shift_cnt=2), TCPOptionNoOperation(kind=1,length=1), TCPOptionNoOperation(kind=1,length=1), TCPOptionSACKPermitted(kind=4,length=2)],seq=4045290802,src_port=49166,urgent=0,window_size=8192)"
thispacket = PacketData(line1, line2)
print('测试1：', thispacket.predict())
print(thispacket.predict_npy())
# 来自良性样本weibo
line1 = "version=0x4,msg_type=0xa,msg_len=0x618,xid=0x0,OFPPacketIn(buffer_id=4294967295,cookie=0,data='\x02\x1a\xc5\x01\x00\x00\x02\x1a\xc5\x02\x00\x00\x08\x00E\x00\x05\xdc\x7f\xf3@\x00 \x06\xcdH\x01\x02\xf3i\x01\x01\x12t\x00PL$\xee\xe1\xd1\x1d\x86\xf8\xc8\xe8\x80\x10!\xf0\x9f\x19\x00\x00\x01\x01\x08\n\x1d\x16\xdc\xd9\n\xf7\xf8\xd7ke\xda\x92^\xcag\x94\x8fB\xed\x92}\xbd)8>\x83[\xdd\x9f""|A\xd6/|O\xe3\xfb\xad#M_\x1e\xc0n\xef\xe0\xd3e\xba\xb0E\x86\xce\xc0DD\x8d2\x16\xea\xaf\x9c1\xef\x83^\xaf\xe2D\xb5\xbb\xb1\x92\xca\xf1V\xee\xdeX\xcc2$\x83p\x90\x11\x83\x9f\xad{\xac\x9f\x08-%\x18:\x95\xc0\xfa*\xd6e\xcf\xc0M>\xe5\xf7\x9dR\xe5\x089\x1f""\xd62\xa7.\x88\xd1K\xb9\xf1\x87\xc3M\x1b\xfe\x10\xebI\xb4%\x99\xa7\xb4\xd3\xaee\xb7\xb6\'\x92\xb0\x86\xca.{\xed\x07\x1f\x85z\xbe\x96\xdb\x9e28\x07\x91^\x8b\xa4~\xc8\x9av\x99\xbc\xc9\xe2mB\xe6i$y^G\x89\x06K\x12k\xa8\xb3\xfd\x9d\xb4\xeb0\xbbu\x8b\xa6#\xa6Qk\xcbx:\xb2\x9b\x95\x8feb\xe9F\x11\x8a{\x1c>\x9cCF=\xaa\xcb\r\xc4\xf3\xc5zD?\x06\xac\xe2\x00\rJ\xe0\xff\x00\xc0\x16\xa5\x1f\x08\xac\xc0#\xfbBs\xf5QW\x1c-U\xd0\xceX\xaaL\xf2\xc5\xf9%\xc7\x1fZ\xb0b\xde\xb9\x07\x18\xafF\x1f\x06\xed\x03n\xfe\xd1\x9c\xe7\xb1E\xa9G\xc2K@1\xf6\xf9\xbf\xef\x85\xae\x8faW\xb1\x9a\xc4Au<\xde y\xe7""\xa5M\xea\x8c\xc4c\x8a\xf4t\xf8Sh\x83\x1fo\x9c\xff\x00\xc0EK\xff\x00\n\xc2\xd4\xa9_\xb6\xcd\xcf\xfb""\xab\xd8N\xfb\x12\xf1\x14\xdfS\xc7\xe7\xba\x0cOPEb\xdc\xcc\x1eL\xee?\x8d{|\x9f\x07\xacd\x07\x17\xf3\x8c\xff\x00\xb0\xb5JO\x81\x9a|\x9c\xb6\xa3?\xfd\xf0\xb4K\r6\xcd#\x8b\xa6\x91\xe2s<B""N\x07\xbda\xdd\xce-\xe2wS\x9c\x8cs\xde\xbe\x80\x9f\xf6}\xd3\xa6\x18:\xad\xd6=\x91j\x8d\xd7\xec\xd3\xa6\xdc\xc7\xb0\xebW\x8a=\xa3Z!BQz\x98\xd4\xc4)\xad\x0f\x97t}3\xe1\xef\xc2\x1d/T\xf1\x06\xa7mo\xa7\xc0\xf3\xfd\xb1\xdd\xfeg\x92\xe0\xe0|\x80\xe4\x96\xf4\x02\xad\xcf\xe1\xddg\xe3-\xee\x9b\xe2;\x9d[\xfe\x10m\x1e\x02\xed\xa7Evs6\x0eG\xda\n\x12\x06\xe0\xb8#<\x8a\xf5\xa6\xfd\x84t\x1dO\xc7\xd6\xfe%\xd7\xbcQ\xa9\xeb\xd1\xd9\x90\xd6:]\xcc1\x8b[f\x07!\x82\x81\xf37\xb9\xae\xef\xc7\x1f\xb2\xd7\x85\xbe""Ae\x0e\xb9\xba\xf6+M\xe65\x95\x01\x03p\xc1\xe8Ez\x91\xd1jyrM\xecxM\x87\xec\x9f\xa2\xcc\xad}s\xac\xaf\x8ed\x9dL\xa5\xfcD^Uf\xc0\xc6\xd6F\xc0\x04\xf3\xc0\xef]\x17\x84~\x1fx^\xdbQ\xfb&\x9be?\x81|Ue\x1b\xb7\x91\xa7\xce\xc6\x19\x14\xa8\x01\xc2\x9f\x96T\xcfb85\xf4\x0f\x86>\x12\xe9\x9e\x13\xd0\xac\xb4\x9d:W\x86\xc6\xd2%\x86(\xd5F\x02\x8e\x94\xcd_\xe1>\x9f\xab\xdc\xd9\xdd}\xaeX.\xed$\xf3""\x9e5]\xc0te>\xa0\x8e\rn\xa6\xb61T\xda>i\xf8\x8bc\xaax\xb2\xfa\xebG\xd4c\xb7Y\xad\xe2Kx\xa5\x132-\xe3\x1f\x98\xb9P>PWw\x198 \xd7\xa3\xf8gD\xb0\xd0\xfc;kk\x13\xed\x10\xa0DfbO\x07\x1f\x96+\xac\xb8\xfd\x9e\xad.\xbcYw\xae\\k\xb7\xb3K3\x83\x1c\x0c\x8b\xb2\x15\x08\x17j\xfe\xa7\xf15v\xe7\xe0\x7f\x9byk,\x1e \xbb\xb5\x86\x16\xdd$)\n\x11(\xf4$\xf2?\n\x1c\xd3\xd8\xc7\xd9J\xe7\x95|Q\x97AO\x08\xca\xba\xca\xc1>\x92\xe0\xa5\xcaO\xbb\r\xc7\x00\xe0\x1c\x0e\xbc\xd7\x8a^i\x7f\x03\xbe\x1cxwG\xd2\xc5\xc0\xd2$\xd5e\xfbm\x8c\xe1\xc9*\xfb\xc6>n\xc3\x04\xd7\xdaR|\x1e\xb0\x9a\x06\x8ak\xb9&\x8d\x86\nI\x1a\x90\x7f\n\xe2|]\xfb\x1b|=\xf1\xd5\xc5\x9c\xfa\xce\x9a\x97r\xd9\x8cBLJ\x02\x0c\xe7\x00\x0e\xde\xd4)E\x15\xec\xe4|\xf5\xf03\xc3gQ\xf8\x8f\xe2\x7f\x16]\xea\xeb}{\xbc\xd9\x08\xa29O(\x1f\xdd\xb8\xf7+\x8a\xfa\x1f\xe0\xfe\xbe\xda\x97\xc5\x9dsOG\x06\x0b\x1d60\xc3\xfd\xb6\x93?\xca\xafx\x7f\xf6c\xf0\xff\x00\x855\xeb\x9dKJ\xbb\x9a\xc5nc\x11\xcbm\x12(C\x8e\x87\xd8\xe2\xbao\x00\xfc!\xb2\xf0\x17\x8a5\xcdn\x0b\xd9\xae\xeeu]\xa1\xd6U\x00F\x14\x0c\x01\x8f\xa5\x13\x94^\xc2\xa7NQ\x95\xd9\xe8\x94QEdw\x88zW\x19\xf1""k\x8b=\x1e\x1b\xbf\xf8I\x13\xc2\xfa]\xbc\xdef\xa1{\xb1ZV\x8bi\xda\x91\x96\x04\x02\\\xafbH\xc8\x1c\x9a\xec\xcfJ\xe4\xbco\xf0\xfe\xc3\xc7?\xd9f\xf6\xea\xf6\xd8\xe9\xd7B\xf2\x0f\xb3H\xa0y\xa0\x10\xac\xca\xca\xca\xd8\xc9##\x83\xcfZ\x00\xf2/\r\xf8\x8b\xc7\x9e*\xb6\x96\xca\xe7\xc4\x17Z,\xb6\x1aU\xc6\xa5\x14\xf2ZD\x97\x17h\xd7\x12\xad\xab\xcc\x85p\xbf\xba\x8dK(\x00\xe5\xb9\xc1\xe2\xb6\xed<u\xe3\x1dS\xc4\x1f\r\xee\x9ak+-\x0fZ\x81\x85\xc41\xa1ig\x9b\xec\xcc\xe4\xe4\xfd\xc4V\x1c\x01\xc9\xcdu\xda\xc7\xc2m;[x\xa6\xbb\xd6usz {Io#\xb8H\xe5\xb9\xb7v\xcbC!T\x00\xaes\x82\x00a\xce\rm\\\xf8+C\xb8:\x101*G\xa3n\xfb\x1cH\xf8U\x063\x19\x04w\x1bN(\x15\xcf\x1e\xd4<e\xe3o\x07\xf8\x95\xb4\xd3\xac\x1dF\xf2]""\xf2\xe4\xbe\xb5n\xb6\xb6\x12\xdd!C\x1f\xd9]W%UY\xb7)<\x80\xbd\xc95\xc4|?\xf8\x89\xe3;\xfdkE\xd2\xed\xfcY\xab\xddiw\x1a\x14z\xb2\xcfu\xfd\x99\xe7\xb9yG\xdef\x1c.\x1b\xee\x9f\x98~\x15\xee\xb6?\x05\xbc3mql\xd7s\xdekV\xf6q<\x166Z\xac\xe2x,\xd1\xf1\xb8""\xe0\x13\x9c\x01\x97,@\x18\x04TK\xf0\'\xc1+\xe3\x16\xd7\x86\x91\xa7\x05\xfe\xcf\xfb\x07\xd8>\xcd\x1f\x93\x8f0?\x99\xd3;\xb8\xc7\xd2\x81\x9c\x15\xcf\xc4\x14\x08I\xe2',match=OFPMatch(oxm_fields={'in_port': 1}),reason=0,table_id=0,total_len=1518)"
line2 = "ethernet(dst='02:1a:c5:01:00:00',ethertype=2048,src='02:1a:c5:02:00:00'), ipv4(csum=52552,dst='1.1.18.116',flags=2,header_length=5,identification=32755,offset=0,option=None,proto=6,src='1.2.243.105',tos=0,total_length=1500,ttl=32,version=4), tcp(ack=2264451304,bits=16,csum=40729,dst_port=19492,offset=8,option=[TCPOptionNoOperation(kind=1,length=1), TCPOptionNoOperation(kind=1,length=1), TCPOptionTimestamps(kind=8,length=10,ts_ecr=184023255,ts_val=488037593)],seq=4007776541,src_port=80,urgent=0,window_size=8688), 'ke\xda\x92^\xcag\x94\x8fB\xed\x92}\xbd)8>\x83[\xdd\x9f""|A\xd6/|O\xe3\xfb\xad#M_\x1e\xc0n\xef\xe0\xd3e\xba\xb0E\x86\xce\xc0DD\x8d2\x16\xea\xaf\x9c1\xef\x83^\xaf\xe2D\xb5\xbb\xb1\x92\xca\xf1V\xee\xdeX\xcc2$\x83p\x90\x11\x83\x9f\xad{\xac\x9f\x08-%\x18:\x95\xc0\xfa*\xd6e\xcf\xc0M>\xe5\xf7\x9dR\xe5\x089\x1f""\xd62\xa7.\x88\xd1K\xb9\xf1\x87\xc3M\x1b\xfe\x10\xebI\xb4%\x99\xa7\xb4\xd3\xaee\xb7\xb6\'\x92\xb0\x86\xca.{\xed\x07\x1f\x85z\xbe\x96\xdb\x9e28\x07\x91^\x8b\xa4~\xc8\x9av\x99\xbc\xc9\xe2mB\xe6i$y^G\x89\x06K\x12k\xa8\xb3\xfd\x9d\xb4\xeb0\xbbu\x8b\xa6#\xa6Qk\xcbx:\xb2\x9b\x95\x8feb\xe9F\x11\x8a{\x1c>\x9cCF=\xaa\xcb\r\xc4\xf3\xc5zD?\x06\xac\xe2\x00\rJ\xe0\xff\x00\xc0\x16\xa5\x1f\x08\xac\xc0#\xfbBs\xf5QW\x1c-U\xd0\xceX\xaaL\xf2\xc5\xf9%\xc7\x1fZ\xb0b\xde\xb9\x07\x18\xafF\x1f\x06\xed\x03n\xfe\xd1\x9c\xe7\xb1E\xa9G\xc2K@1\xf6\xf9\xbf\xef\x85\xae\x8faW\xb1\x9a\xc4Au<\xde y\xe7""\xa5M\xea\x8c\xc4c\x8a\xf4t\xf8Sh\x83\x1fo\x9c\xff\x00\xc0EK\xff\x00\n\xc2\xd4\xa9_\xb6\xcd\xcf\xfb""\xab\xd8N\xfb\x12\xf1\x14\xdfS\xc7\xe7\xba\x0cOPEb\xdc\xcc\x1eL\xee?\x8d{|\x9f\x07\xacd\x07\x17\xf3\x8c\xff\x00\xb0\xb5JO\x81\x9a|\x9c\xb6\xa3?\xfd\xf0\xb4K\r6\xcd#\x8b\xa6\x91\xe2s<B""N\x07\xbda\xdd\xce-\xe2wS\x9c\x8cs\xde\xbe\x80\x9f\xf6}\xd3\xa6\x18:\xad\xd6=\x91j\x8d\xd7\xec\xd3\xa6\xdc\xc7\xb0\xebW\x8a=\xa3Z!BQz\x98\xd4\xc4)\xad\x0f\x97t}3\xe1\xef\xc2\x1d/T\xf1\x06\xa7mo\xa7\xc0\xf3\xfd\xb1\xdd\xfeg\x92\xe0\xe0|\x80\xe4\x96\xf4\x02\xad\xcf\xe1\xddg\xe3-\xee\x9b\xe2;\x9d[\xfe\x10m\x1e\x02\xed\xa7Evs6\x0eG\xda\n\x12\x06\xe0\xb8#<\x8a\xf5\xa6\xfd\x84t\x1dO\xc7\xd6\xfe%\xd7\xbcQ\xa9\xeb\xd1\xd9\x90\xd6:]\xcc1\x8b[f\x07!\x82\x81\xf37\xb9\xae\xef\xc7\x1f\xb2\xd7\x85\xbe""Ae\x0e\xb9\xba\xf6+M\xe65\x95\x01\x03p\xc1\xe8Ez\x91\xd1jyrM\xecxM\x87\xec\x9f\xa2\xcc\xad}s\xac\xaf\x8ed\x9dL\xa5\xfcD^Uf\xc0\xc6\xd6F\xc0\x04\xf3\xc0\xef]\x17\x84~\x1fx^\xdbQ\xfb&\x9be?\x81|Ue\x1b\xb7\x91\xa7\xce\xc6\x19\x14\xa8\x01\xc2\x9f\x96T\xcfb85\xf4\x0f\x86>\x12\xe9\x9e\x13\xd0\xac\xb4\x9d:W\x86\xc6\xd2%\x86(\xd5F\x02\x8e\x94\xcd_\xe1>\x9f\xab\xdc\xd9\xdd}\xaeX.\xed$\xf3""\x9e5]\xc0te>\xa0\x8e\rn\xa6\xb61T\xda>i\xf8\x8bc\xaax\xb2\xfa\xebG\xd4c\xb7Y\xad\xe2Kx\xa5\x132-\xe3\x1f\x98\xb9P>PWw\x198 \xd7\xa3\xf8gD\xb0\xd0\xfc;kk\x13\xed\x10\xa0DfbO\x07\x1f\x96+\xac\xb8\xfd\x9e\xad.\xbcYw\xae\\k\xb7\xb3K3\x83\x1c\x0c\x8b\xb2\x15\x08\x17j\xfe\xa7\xf15v\xe7\xe0\x7f\x9byk,\x1e \xbb\xb5\x86\x16\xdd$)\n\x11(\xf4$\xf2?\n\x1c\xd3\xd8\xc7\xd9J\xe7\x95|Q\x97AO\x08\xca\xba\xca\xc1>\x92\xe0\xa5\xcaO\xbb\r\xc7\x00\xe0\x1c\x0e\xbc\xd7\x8a^i\x7f\x03\xbe\x1cxwG\xd2\xc5\xc0\xd2$\xd5e\xfbm\x8c\xe1\xc9*\xfb\xc6>n\xc3\x04\xd7\xdaR|\x1e\xb0\x9a\x06\x8ak\xb9&\x8d\x86\nI\x1a\x90\x7f\n\xe2|]\xfb\x1b|=\xf1\xd5\xc5\x9c\xfa\xce\x9a\x97r\xd9\x8cBLJ\x02\x0c\xe7\x00\x0e\xde\xd4)E\x15\xec\xe4|\xf5\xf03\xc3gQ\xf8\x8f\xe2\x7f\x16]\xea\xeb}{\xbc\xd9\x08\xa29O(\x1f\xdd\xb8\xf7+\x8a\xfa\x1f\xe0\xfe\xbe\xda\x97\xc5\x9dsOG\x06\x0b\x1d60\xc3\xfd\xb6\x93?\xca\xafx\x7f\xf6c\xf0\xff\x00\x855\xeb\x9dKJ\xbb\x9a\xc5nc\x11\xcbm\x12(C\x8e\x87\xd8\xe2\xbao\x00\xfc!\xb2\xf0\x17\x8a5\xcdn\x0b\xd9\xae\xeeu]\xa1\xd6U\x00F\x14\x0c\x01\x8f\xa5\x13\x94^\xc2\xa7NQ\x95\xd9\xe8\x94QEdw\x88zW\x19\xf1""k\x8b=\x1e\x1b\xbf\xf8I\x13\xc2\xfa]\xbc\xdef\xa1{\xb1ZV\x8bi\xda\x91\x96\x04\x02\\\xafbH\xc8\x1c\x9a\xec\xcfJ\xe4\xbco\xf0\xfe\xc3\xc7?\xd9f\xf6\xea\xf6\xd8\xe9\xd7B\xf2\x0f\xb3H\xa0y\xa0\x10\xac\xca\xca\xca\xd8\xc9##\x83\xcfZ\x00\xf2/\r\xf8\x8b\xc7\x9e*\xb6\x96\xca\xe7\xc4\x17Z,\xb6\x1aU\xc6\xa5\x14\xf2ZD\x97\x17h\xd7\x12\xad\xab\xcc\x85p\xbf\xba\x8dK(\x00\xe5\xb9\xc1\xe2\xb6\xed<u\xe3\x1dS\xc4\x1f\r\xee\x9ak+-\x0fZ\x81\x85\xc41\xa1ig\x9b\xec\xcc\xe4\xe4\xfd\xc4V\x1c\x01\xc9\xcdu\xda\xc7\xc2m;[x\xa6\xbb\xd6usz {Io#\xb8H\xe5\xb9\xb7v\xcbC!T\x00\xaes\x82\x00a\xce\rm\\\xf8+C\xb8:\x101*G\xa3n\xfb\x1cH\xf8U\x063\x19\x04w\x1bN(\x15\xcf\x1e\xd4<e\xe3o\x07\xf8\x95\xb4\xd3\xac\x1dF\xf2]""\xf2\xe4\xbe\xb5n\xb6\xb6\x12\xdd!C\x1f\xd9]W%UY\xb7)<\x80\xbd\xc95\xc4|?\xf8\x89\xe3;\xfdkE\xd2\xed\xfcY\xab\xddiw\x1a\x14z\xb2\xcfu\xfd\x99\xe7\xb9yG\xdef\x1c.\x1b\xee\x9f\x98~\x15\xee\xb6?\x05\xbc3mql\xd7s\xdekV\xf6q<\x166Z\xac\xe2x,\xd1\xf1\xb8""\xe0\x13\x9c\x01\x97,@\x18\x04TK\xf0\'\xc1+\xe3\x16\xd7\x86\x91\xa7\x05\xfe\xcf\xfb\x07\xd8>\xcd\x1f\x93\x8f0?\x99\xd3;\xb8\xc7\xd2\x81\x9c\x15\xcf\xc4'"
thispacket1 = PacketData(line1, line2)
print('测试1：', thispacket1.predict())
print(thispacket1.predict_npy())
|
