In reality, we only want to model this as a `rest_framework.response.Response`, since our .qll modeling is more precise for rest-framework responses than if we also modeled it as a basic django http response. (specifically, that default mime-type handling is way different).
mostly removing of nodes from the graph.
One result lost:
```
check("submodule.submodule_attr", submodule.submodule_attr, "submodule_attr", globals()) #$ MISSING:prints=submodule_attr
```
TODO:
The member predicate `LibraryLambdaMethod::getACall` is
currently too permissive.
Ideally, we would have `libraryCallHasLambdaArg`
as in Ruby. But even a more precise
`libraryCall` predicate might be fine.
This provides variable capture in standard situations:
- nested functions
- lambdas
There are some deficiencies:
- we do not yet handle objects capturing variables.
- we do not handle variables captured via the `nonlocal` keyword.
This should be solved at the AST level, though, and then it
should "just work".
There are still inconsistencies in the case where
a `SynthesizedCaptureNode` has a comprehensions
as its enclosing callable. In this case,
`TFunction(cn.getEnclosingCallable())` is not
defined and so getEnclosingCallable does not exist
for the `CaptureNode`.
I fixed it in both predicates... I think we might still be able to remove
`newDirectAlias` -- but with it being better, it will allow us to better test if `newImportAlias` actually cover everything we need!
Due to the 'only model most specific spec' logic highlighted in previous
commit, I'm changing away from MethodView/View, and use Django view instead.
In practice this shouldn't matter at all, but for writing tests it would
have been a nice fix to only have the "same name but more specific"
logic apply when it's the same _definition_ location. We used to have
this information available, but right now we don't... so instead of
spending a lot of time rewriting the core library, I simply used a
different class :D :O :(
mostly removing of nodes from the graph.
One result lost:
```
check("submodule.submodule_attr", submodule.submodule_attr, "submodule_attr", globals()) #$ MISSING:prints=submodule_attr
```
Adds support for extraction filters as defined in
https://peps.python.org/pep-0706/
and implemented in Python 3.12.
By my reading, setting the filter to `'data'` or `'tar'` is probably
safe, whereas `'fully_trusted'` or the default (which is the same as
`None`) is not.
For now, I have just added this modelling to the tarslip query. We could
also share it with the modelling of `shutil.unpack_archive` (which has also
gained a `filter` argument), but it was unclear to me where we should put
this modelling in that case. Perhaps the best solution would be to merge
the experimental `py/tarslip-extended` query into the existing query (in
which case the current location is perhaps not too bad).
