fix or relation of sub-array

[TitouanT]

Currently the result of an or-array is forgotten if anything comes after it.

for example, this case fails with the relation and: [[false, false], true] As a logic expression it is expressed as (false || false) && true, which is false but the authentication process accepts it.

I generated multiple tests to find more example of error cases and see if any other existed. And added it to the test cases to prevent regression.

Checklist

• run npm run test && npm run benchmark --if-present
• tests and/or benchmarks are included
• documentation is changed or added
• commit message and code follows the Developer's Certification of Origin
  and the Code of conduct

[TitouanT]

Failing cases of the tests I added before the fix (shortened the output):

✖ failing tests:

test at test/logic-gate.test.js:28:5
✖ Relation and for expression [true,[false,false],true,true] (1.651285ms)
  AssertionError [ERR_ASSERTION]: true == false
  {
    generatedMessage: true,
    code: 'ERR_ASSERTION',
    actual: true,
    expected: false,
    operator: '==',
    diff: 'simple'
  }

test at test/logic-gate.test.js:28:5
✖ Relation and for expression [true,[false,false],true] (0.296815ms)
  AssertionError [ERR_ASSERTION]: true == false

test at test/logic-gate.test.js:28:5
✖ Relation and for expression [true,[false,false],[false,true]] (0.332723ms)
  AssertionError [ERR_ASSERTION]: true == false

test at test/logic-gate.test.js:28:5
✖ Relation and for expression [true,[false,false],[true,false]] (0.275936ms)
  AssertionError [ERR_ASSERTION]: true == false

test at test/logic-gate.test.js:28:5
✖ Relation and for expression [true,[false,false],[true,true]] (0.330319ms)
  AssertionError [ERR_ASSERTION]: true == false

test at test/logic-gate.test.js:28:5
✖ Relation and for expression [[false,false],true] (0.361428ms)
  AssertionError [ERR_ASSERTION]: true == false

test at test/logic-gate.test.js:28:5
✖ Relation and for expression [[false,false],[false,true]] (0.359024ms)
  AssertionError [ERR_ASSERTION]: true == false

test at test/logic-gate.test.js:28:5
✖ Relation and for expression [[false,false],[true,false]] (0.195352ms)
  AssertionError [ERR_ASSERTION]: true == false

test at test/logic-gate.test.js:28:5
✖ Relation and for expression [[false,false],[true,true]] (0.189361ms)
  AssertionError [ERR_ASSERTION]: true == false

test at test/logic-gate.test.js:28:5
✖ Relation and for expression [[false],true] (0.331511ms)
  AssertionError [ERR_ASSERTION]: true == false

They all fail only when the main relation is and, so the sub-array relation is or.
The expression are the following:

[true,[false,false],true,true]
[true,[false,false],true]
[true,[false,false],[false,true]]
[true,[false,false],[true,false]]
[true,[false,false],[true,true]]
[[false,false],true]
[[false,false],[false,true]]
[[false,false],[true,false]]
[[false,false],[true,true]]
[[false],true]

The test transforms a given boolean expression into a chain of auth functions that will pass if the boolean is true and fail if the boolean is false.

[TitouanT]

Apparaently someone else 'found' that bug and fixed it in 5.1.1
I would still recomend to add a fuzzer even if my strategy to add test was somewhat copied into handpicked tests. This way no subtle regression would pass through for the tested authent topologies.

[mcollina]

Sorry! I received a vulnerability and I published the advisory. I had sol generate the test and most of the fix. If you want to send a fuzzer PR, go ahead!