fix #1940

[johnxguo]

Describe the Pull Request

fix #1940 

this algorithm can make sure that worse findings will result in lower score

Checklist for PR

• Run MobSF unit tests and lint tox -e lint,test
• Tested Working on Linux, Mac, Windows, and Docker
• Add unit test for any new Web API (Refer: StaticAnalyzer/tests.py)
• Make sure tests are passing on your PR

Additional Comments (if any)

DESCRIBE HERE

[ajinabraham]

I did a round of testing. I found that most apps with more than 4 high findings get a score of 0, which makes appscore comparison difficult. I was able to get around this by changing 30 to 70. What are your thoughts on that?

[johnxguo]

can i have a look at your test code? I tried value of (high, warn, sec) with (11, 7, 0) and get score 1, with (12, 9, 1) and get score 0, with (5, 21, 0) and get score 1.
But it is indeed difficult to achieve high scores with the current setting，If you want most apps to generate higher scores，you can change 30 to a higher value

----- here is my test code

import math

def get_secure_score(high, warn, sec):
      loss_score = high * 10 + warn * 5  - sec * 2
      normalize_reverse = 2 / (1 + pow(math.e, loss_score / 30))
      return int(min(normalize_reverse, 1) * 100)

def print_score(high, warn, sec):
      print('%5d | %4d %4d %3d' % (get_secure_score(high, warn, sec), high, warn, sec))
    
print("score | high warn sec")
print_score(4, 2, 0)
print_score(5, 3, 1)
print_score(5, 21, 0)
print_score(7, 4, 0)
print_score(8, 5, 1)
print_score(10, 6, 1)
print_score(11, 7, 0)
print_score(12, 9, 1)

----- output
score | high warn sec
31 | 4 2 0
21 | 5 3 1
1 | 5 21 0
9 | 7 4 0
6 | 8 5 1
2 | 10 6 1
1 | 11 7 0
0 | 12 9 1

[ajinabraham]

Let me test this with some real apps and get back.

[AmmeySaini]

Any update on this??