mirror of
https://github.com/github/codeql.git
synced 2025-12-17 09:13:20 +01:00
A slightly complicated test setup. I wanted to both make sure I captured the semantics of Python and also the fact that the kinds of global flow we expect to see are indeed present. The code is executable, and prints out both when the execution reaches certain files, and also what values are assigned to the various attributes that are referenced throughout the program. These values are validated in the test as well. My original version used introspection to avoid referencing attributes directly (thus enabling better error diagnostics), but unfortunately that made it so that the model couldn't follow what was going on. The current setup is a bit clunky (and Python's scoping rules makes it especially so -- cf. the explicit calls to `globals` and `locals`), but I think it does the job okay.