Interesting. I love requests. SQL isn't my area of expertise so im genuinely surprised in 2016 we're still building libraries like this for Python; are existing libraries really that bad?
I would say 'no'. SQLAlchemy is very good, but extremely complex and powerful. I think people don't understand that you can use the core features without any of the ORM overhead, if you don't want it.
If you want to "Just write SQL" you can do that, easily: http://docs.sqlalchemy.org/en/latest/core/tutorial.html#usin...
I think jumping into that example directly is probably kind of disorienting. The SQL in that string is a little bit weird/contorted just to illustrate, "it's text, do whatever you want!". I also wrote that example like ten years ago.
I spent a couple of hours looking for exactly this yesterday for a new Tornado based project and ended up using queries: https://github.com/gmr/queries which seems on the same level for my purposes of wanting to cleanly pass raw SQL and get python data structures in return. Queries also provides asynchronous API interaction with Tornado which will be nice for me.
Coincidently queries was inspired by Kenneth Reitz's work on requests.
I think it is often tempting to write something like this as it is quite a light thing and can reacquaint someone with SQL if they haven't used it for a while. I know I was tempted after pouring through ORMs.
pyscopg2 is pretty great. Every lib I've seen is based on it. Many apps use it directly. It is the lowest, comfortable layer you'd want. What you need on top of that is Either;
1) So minor, most just write a few helper functions/classes
2) So elaborate and opinionated it needs years of use/work to get right. SQLAlchmey, Django ORM.
3) So specific, customized to your requirements that general purpose libs/frameworks don't fit.
Edit: actually it just works on postgres. I don't know if there are plans to make it work on other SQL databases.
No, but they're meant for other use-cases. This fills in a gap for "simple, but powerful", as advertised. :)