I might be wrong about curl - I use it just as a tool to make http calls but it might be more complicated than that - but why can't you build curl with python's http libraries? I guess there are use cases where you have to manipulate responses on the byte level?