feat: Added possibility for shared numbering of theorems, fixing issue #64

[Ljon4ik4]

This PR is an attempt to resolve issue #64 .
This is done by adding a configuration variable proof_uniform_numbering. When it is set to true all numbered proof environments share a counter.

under the hood I have split the type parameter of the node into who parameters, countertype and realtye.
In the default setting, those two agree.
When proof_uniform_numbering = True is set in conf.py, then the countertype of all objects is theorem, i.e. they all share the same counter.

pytest seems to pass, and the uniformly numbered version seems to work fine on the docs.
However, I don't really know how to write python tests, so I could not add tests for the new version.

[douden]

@Ljon4ik4 : Thank you for contributing.

If I read #64 correctly, this PR does it slightly different compared to what the original issue creator intended.

As far as I can deduce from the original issue, the request was that you could "group" the numbering of certain admonitions of this extension, where you could have several different groups sharing the numbering.

Am I correct that your suggestions groups them all together?

[Ljon4ik4]

@douden Oh yes, sorry - I did not read the issue precisely, you are absolutely right.
The only functionality added here is 'fully consecutive' numbering, not multiple separate ones.

[douden]

I looked at your code, and because of the manner you implemented it, I think it could be doable to have the grouped numbering.

Currently I would say that we could reuse the config value you introduced, and if it is a dictionary, you could map each given key to it's corresponding value, where key is the real type and the value is the counter type.

I think that line 42 is the location to expand/adapt for this.

Are you willing to add this functionality?

[Ljon4ik4]

Yes, thanks a lot - I think that should be possible!
I will try to do it (probably next week), and add it to the PR.

[douden]

Yes, thanks a lot - I think that should be possible! I will try to do it (probably next week), and add it to the PR.

Also thanks to you!

Let me know when you changed it, I will then review.

[Ljon4ik4]

@douden , in the end it was simpler to just pass a dictionary in any case, and just set the dictionary to have a constant value "theorem" for uniform numbering. I think it is built in now...

[douden]

Although I agree with the idea of only using a dictionary, the default behavior of the extension now would significantly change.

I think the best would be to

• Set the default value in docs/source/conf.py to {}
• In sphinx_proof/__init__.py to rename realtyp_to_countertyp to DEFAULT_REALTYP_TO_COUNTERTYP.
• Change line 103 to

app.add_config_value("prf_realtyp_to_countertyp", {}, "env")

• Change line 40 to

countertyp = env.config.prf_realtyp_to_countertyp[realtyp] or DEFAULT_REALTYP_TO_COUNTERTYP[realtyp]

This would allow a user in the conf.py/_config.yml to indicate only which directives should be using a different counter, while all others are untouched.

So if in the config a user sets

prf_realtyp_to_countertyp = {
    "lemma": "theorem",
    "conjecture": "theorem",
    "corollary": "theorem",
    "proposition": "theorem",
    "axiom": "definition",
    "assumption": "definition",
}

The directives lemma, conjecture, corollary and proposition will share the counter with theorem, while axiom and assumption will share the counter with definition. All other directives would use their original counter.

[Ljon4ik4]

Ah, that's a very good idea!
I implemented it, with a slight variation (the suggested line 40 did not work for me + I had to move the renamed DEFAULT_REALTYP_TO_COUNTERTYP to the file where it is used'')