[Python] General issue: Wrong domination is reported in control flow analysis

[Ang9876]

Description of the issue
I have a python code and ql as follows:

def EXEC():
    raise Exception("Test exception")

def API():
    pass

def DUMMY():
    pass

try:
    EXEC()
    API()
except Exception as e:
    raise(e)
finally:
    DUMMY()

/**
 * @kind problem
 * @problem.severity warning
 * @id python/find-if
 */

import python

predicate isCall(Call call, string name) {
    exists (Name n | call.getFunc() = n and n.getId() = name)
}

from Call callAPI, Call callDummy
where
    isCall(callAPI, "API") and isCall(callDummy, "DUMMY") and
    not exists (ControlFlowNode nAPI, ControlFlowNode nDummy | 
        nAPI = callAPI.getAFlowNode() and nDummy = callDummy.getAFlowNode() and 
        nAPI.dominates(nDummy))
select callAPI, callAPI.toString()

The ql code simply check if the call to API() does not dominate the call to DUMMY()
Expected Behaviour: Call to API() is reported because API() does not dominate DUMMY().
Actual Result: nothing is reported

CodeQL version: CodeQL command-line toolchain release 2.23.1.

[jketema]

Hi @Ang9876,

I've asked someone who is familiar with the CodeQL Python code to have a look.