PyCoq is a set of bindings and libraries allowing to interact with the Coq interactive proof assistant from inside Python 3.
This way, typical operations such as parsing Coq documents, and checking them, can be performed programatically using Python.
This project is extremely experimental, and we don't provide a release yet. Use at your own risk.
Help is more than welcome; don't hesitate to open issues, suggest improvements, or help with documentation
We have a contributing guide and a Code of Conduct
To build PyCoq, you need Python 3 and an OCaml environment able to build SerAPI; usually, this can be achieved using the OPAM package manager and doing:
$ opam install --deps-only coq-serapi
$ opam install pythonlib
for the 0.13 version that targets Coq v8.13; note that OCaml >= 4.11.0 is recommended, and >= 4.08.0 required by the dependencies.
If that's your first checkout, you'll also have to update the git
submodules, usually using git submodule update --init --recursive.
For the moment, PyCoq requires a special version of ppx_python, while the situation stabilizes
in the OPAM repository we have vendored it in the SerAPI branch we use.
Once the right dependencies have been installed, you can do:
$ make install && dune build examples/test.py && dune exec -- python3 _build/default/examples/test.py
If you want an interactive environment, use:
$ make install && dune exec -- python3
>>> import os
>>> os.chdir('_build/default')
>>> import pycoq, coq
Our continuous integration should contain up-to-date build instructions.
As of today, the only documentation we have is the small example below.
Note however, that if you are familiar with the SerAPI protocol, you will quickly get used to the current API as it basically mirrors the one there [and it is derived automatically, so SerAPI's documentation could be considered as canonical for pyCoq for now]
import pycoq, coq
# Some common definitions we'll use
ppopts = { "pp_format": ("PpSer",None), "pp_depth": 1000, "pp_elide": "...", "pp_margin": 90}
# Add a full document chunk [mainly parsing]
opts = {"newtip": None, "ontop": 1, "lim": 20 }
res = coq.add("Lemma addnC n m : n + m = m + n. Proof. now induction n; simpl; auto; rewrite IHn. Qed.", opts)
# Check the proof.
res = coq.exec(5)
# We can show the goals at the tactic point
opts = {"limit": 100, "preds": [], "sid": 3, "pp": ppopts, "route": 0}
cmd = ('Goals', None)
res = coq.query(opts, cmd)
print(res)
# We can also output the ast for example, of the main tactic in our
# document
opts = {"limit": 100, "preds": [], "sid": 3, "pp": ppopts, "route": 0}
cmd = ('Ast', None)
res_ast = coq.query(opts, cmd)
print(res)
# Such AST can be actually pretty-printed
obj = res[0][1][0][0]
opts = { "pp": { "pp_format": ("PpStr",None) } }
res = coq.print(obj, opts)
print(res)As a nascent project, the roadmap is still in flux; issues should track the more concrete items, but indeed a lot more work is needed in order for PyCoq to be useful.
For a start, we need handle errors correctly and to adhere to Python conventions. A large amount of feedback is required from Python experts as I am not familiar with Python myself at all.
ppx_pythonandpythonlibdevelopers were very helpful and resolved our issues quickly, thanks!- Current setup was bootstrapped from https://github.com/jonathan-laurent/pyml_example , thanks a lot!
- Emilio Jesus Gallego Arias, Inria, Université de Paris, IRIF, Paris
- Thierry Martinez, Inria Paris