-
Notifications
You must be signed in to change notification settings - Fork 2
/
CLI.py
279 lines (261 loc) · 10.9 KB
/
CLI.py
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
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
from os import path, getcwd
from sys import argv
from multiprocessing import cpu_count
from argparse import ArgumentParser, ArgumentDefaultsHelpFormatter
from lib.utils import BOLD, CLR, ACTION
#######
# CLI #
#######
PARSER = ArgumentParser(description="[+] pcapAE wrapper", formatter_class=ArgumentDefaultsHelpFormatter)
PARSER.add_argument('-t',
'--train',
help=f"{BOLD}<path>{CLR} to dataset to learn",
metavar='',
default=None,
required=False)
PARSER.add_argument('-v',
'--vali',
metavar='',
default=None,
help=f"{BOLD}<path>{CLR} to dataset to validate",
required=False)
PARSER.add_argument('-f','--fit',
help=f"{BOLD}<path>{CLR} to data set to fit AD",
metavar='',
default='',
required=False)
PARSER.add_argument('-p','--predict',
help=f"{BOLD}<path>{CLR} to data to make a predict on",
metavar='',
default='',
required=False)
PARSER.add_argument('--eval',
help=f"{BOLD}<path>{CLR} to data to make a eval on",
metavar='',
default='',
required=False)
PARSER.add_argument('-m','--model',
help=f"{BOLD}<path>{CLR} to model to retrain or evaluate",
metavar='',
type=str,
default='',
required=False)
PARSER.add_argument('-b',
'--batch_size',
metavar='',
choices=[1]+[2**n for n in range(1,12)],
default=128,
type=int,
help=f'{BOLD}<number>{CLR} of samples per pass')
PARSER.add_argument('-lr',
'--learn_rate',
metavar='',
default=1e-3,
type=float,
help=f'starting learning {BOLD}<rate>{CLR} between | [1,0)')
PARSER.add_argument('-fi',
'--finput',
default=1,
metavar='',
type=int,
help=f'{BOLD}<number>{CLR} input frames')
PARSER.add_argument('-fo',
'--foutput',
metavar='',
default=0,
type=int,
help=f'{BOLD}<number>{CLR} predict frames. 0 - predict input')
PARSER.add_argument('-o',
'--optim',
default='adamW',
choices=['adam', 'adamW' ,'sgd'],
metavar='',
type=str,
help=f'gradient decent {BOLD}<strategy>{CLR} | [adam | adamW | sgd]')
PARSER.add_argument('-c',
'--clipping',
metavar='',
type=float,
default=10.0,
help=f"gradient clip {BOLD}<value>{CLR} | [0,10]",
required=False)
PARSER.add_argument('--fraction',
type=float,
default=1,
metavar='',
help=f"{BOLD}<fraction>{CLR} of data to process | (0, 1]",
required=False)
PARSER.add_argument('-w',
'--workers',
default=0,
metavar='',
type=int,
help=f'{BOLD}<number>{CLR} of data loader worker threads | [0, 8]')
PARSER.add_argument('--loss',
default='MSE',
metavar='',
choices=['MSE', 'BCE'],
type=str,
help=f'loss {BOLD}<criterion>{CLR} | [MSE | BCE]')
PARSER.add_argument('--scheduler',
default='cycle',
metavar='',
choices=['step', 'cycle', 'plateau'],
type=str,
help=f'learn rate {BOLD}<scheduler>{CLR} | [step | cycle | plateau]')
PARSER.add_argument('--epochs',
default=144,
type=int,
metavar='',
help=f'{BOLD}<number>{CLR} of epochs')
PARSER.add_argument('--cell',
default='GRU',
choices=['GRU', 'LSTM'],
metavar='',
type=str,
help=f'network cell {BOLD}<type>{CLR} | [GRU | LSTM]')
PARSER.add_argument('--no_bn',
default=False,
help='disable batch norm layers',
action='store_true')
PARSER.add_argument('--dropout',
type=float,
default=0,
metavar='',
help=f"{BOLD}<dropout>{CLR} value | [0, 1]",
required=False)
PARSER.add_argument('--seed',
metavar='',
help=f'{BOLD}<seed>{CLR} to fixing randomness',
default=1994,
type=int)
PARSER.add_argument('--noTensorboard',
action='store_true',
help="do not start tensorboard")
PARSER.add_argument('-d','--dir',
default=f"{getcwd()}/runs",
type=str,
help=f'experiment working directory {BOLD}<path>{CLR}')
PARSER.add_argument('--cuda',
'--CUDA','--GPU',
help='enable GPU support',
action='store_true')
PARSER.add_argument('-V',
'--verbose',
default=False,
help='print verbose messages',
action='store_true')
PARSER.add_argument('--noCache',
default=False,
help='disable caching',
action='store_true')
PARSER.add_argument('--retrain',
default=False,
help='retrain given model',
action='store_true')
PARSER.add_argument('--name',
default=None,
type=str,
help=f'experiment {BOLD}<name>{CLR} prefix')
PARSER.add_argument('--AD',
default=False,
help='use anomaly detection module',
action='store_true')
PARSER.add_argument('--baseline',
choices=[False, 'pcapAE' ,'noDL'],
default=False,
help='compute raw baseline')
PARSER.add_argument('--grid_search',
default=False,
help='do grid search',
action='store_true')
PARSER.add_argument('--n_jobs',
metavar='',
choices=range(1, cpu_count()-2),
default=1,
type=int,
help=f'{BOLD}<number>{CLR} of CPU cores to use [only for LOF & IF]')
PARSER.add_argument('-thr',
'--threshold',
default='',
metavar='',
type=str,
help=f'{BOLD}<min_thr-max_thr>{CLR} threshold for normal data | default None')
PARSER.add_argument('--no_banner', action='store_true')
PARSER.usage = f"""
{BOLD}# AE training {CLR}
python3 main.py --train <TRAIN_SET_PATH> --vali <VALI_SET_PATH> [--cuda]
{BOLD}# AE data compression (pcap -> _codes_) {CLR}
python3 main.py --model <PCAPAE_MODEL> --fit <FIT_SET_PATH> --predict <PREDICT_SET_PATH> [--cuda]
{BOLD}# shallow ML anomaly detection training {CLR}
python3 main.py --AD --model *.yaml --fit <REDU_FIT_SET_PATH> [--predict <REDU_PREDICT_SET_PATH>] [--grid_search]
{BOLD}# test training AD on new data {CLR}
python3 main.py --model <AD_MODLE_PATH> --predict <REDU_SET_PATH>
{BOLD}# only deep learning baseline {CLR}
python3 main.py --baseline pcapAE --model <PCAPAE_MODEL> --predict <PREDICT_SET_PATH>
{BOLD}# no deep learning baseline {CLR}
python3 main.py --baseline noDL --AD --model ../test/blueprints/base_if.yaml --fit <FIT_SET_PATH> --vali <VALI_SET_PATH> --predict <PREDICT_SET_PATH>
{BOLD}====={CLR}
"""
# verify input patterns
ARGS = PARSER.parse_args()
help_text = ""
from os import path
if len(argv) == 1:
help_text = "[!] no arguments passed! :<"
else:
if all([ARGS.retrain,\
ARGS.model == '']):
help_text = "[!] --retrain needs a --[m]odel <path>"
if all([ARGS.model,
ARGS.retrain is False,
ARGS.fit == '']) and all([x not in ARGS.model for x in ['save_model', 'AD']]):
help_text = "[!] --[m]odel option needs either --retrain flag or --[f]it data to prepare for AD"
if ARGS.baseline != 'noDL' and all([ARGS.train is None,\
ARGS.vali]):
if ARGS.eval is '':
help_text = "[!] no --[t]rain set specified"
if all([ARGS.predict == '',\
ARGS.fit == '',\
ARGS.train is None,\
ARGS.vali is None]):
help_text = "[!] no data provided"
if all([ARGS.train is None,\
ARGS.vali is None,\
ARGS.model == '',\
ARGS.AD is False,\
'redu_data' not in ARGS.predict]):
help_text = "[!] no --[t]rain set specified"
if all([ARGS.AD,\
ARGS.fit == '']):
help_text = "[!] no --[f]it data provided!"
if ARGS.model != '':
if ('yaml' not in ARGS.model) and ('.pth.tar' not in ARGS.model) and ('_checkpoint_' not in ARGS.model) and all([x not in ARGS.model for x in ['save_model', 'AD']]):
exit(f"[!] '{ARGS.model}' is not a valid path for a pytorch model")
if ARGS.AD and (ARGS.fit != '' or ARGS.predict != '') and ARGS.baseline != 'noDL':
from glob import glob
for ds in [ARGS.fit, ARGS.predict]:
test = (glob(ARGS.fit+'/*'))
for t in test:
if '.hdf5' in t:
help_text = "[!] provided data set is no suitable for anomaly detection! Transform the data first."
elif all([ARGS.model == '',\
'redu_data' not in ARGS.fit,\
'redu_data' not in ARGS.predict,\
ARGS.AD == True,\
ARGS.baseline == False]):
# TODO relax rule for baseline compression
help_text = "[!] --[f]it data must be in reduced form since no --model was provided"
if help_text != '':
PARSER.print_help()
print()
exit(help_text)
if not ARGS.no_banner:
from rich import print as rprint
banner = """ ______ ______ ______ ______ [cyan] ______ ______[/]
/\ __ \/\ ___\/\ __ \/\ __ \ [cyan]/\ __ \/\ ___\ [/]
\ \ __/\ \ \___\ \ __ \ \ __/[cyan] \ \ __ \ \ _\_ [/]
\ \_\ \ \_____\ \_\ \_\ \_\ [cyan] \ \_\ \_\ \_____\ [/]
\/_/ \/_____/\/_/\/_/\/_/ [cyan] \/_/\/_/\/_____/[/] [red]@13utters[/]
"""
rprint(f"[strike yellow]{' '*52}[/]\n[magenta bold]{banner}[/][strike yellow]{'_'*55}[/]")