Add a contrib module to generate type hints for capnp schemas

[gjcarneiro]

Hello. I wrote a python module to generate python type hints for capnp schema modules.

I have done some testing on it, with a private code base. It probably isn't complete. Most notably, I haven't written support for interfaces (have not needed them yet).

I thought you could include it as a contrib module, to be invoked thus: python -m capnp.contrib.genpyi ....

Even if not complete, I think it can be useful already. At least I am finding it useful on my code bases...

[lgtm-com]

This pull request introduces 5 alerts when merging 6cb558c into 581332f - view on LGTM.com

new alerts:

• 3 for Unused import
• 2 for Unused local variable

[lgtm-com]

This pull request introduces 1 alert when merging 9166a49 into 581332f - view on LGTM.com

new alerts:

• 1 for Unused import

[jayvdb]

> pyflakes capnp/contrib/genpyi.py
capnp/contrib/genpyi.py:98:23 undefined name 'Set'
capnp/contrib/genpyi.py:99:30 undefined name 'Set'
capnp/contrib/genpyi.py:278:13 import 'field' from line 24 shadowed by loop variable
capnp/contrib/genpyi.py:322:21 local variable 'ex' is assigned to but never used
capnp/contrib/genpyi.py:329:21 local variable 'ex' is assigned to but never used
capnp/contrib/genpyi.py:360:17 local variable 'ex' is assigned to but never used
capnp/contrib/genpyi.py:400:20 f-string is missing placeholders
capnp/contrib/genpyi.py:402:20 f-string is missing placeholders

[haata]

I'm definitely interested in including this, but please fix the warnings and errors first.

[lgtm-com]

This pull request introduces 1 alert when merging 574eb05 into 5dade41 - view on LGTM.com

new alerts:

• 1 for Unused import

[haata]

Closer, but still seeing some linting errors

 ./capnp/contrib/genpyi.py:133:52: E203 whitespace before ':'
                schema.node.displayNamePrefixLength :
                                                   ^
./capnp/contrib/genpyi.py:187:5: C901 'Writer.type_ref' is too complex (11)
    def type_ref(self, type_node) -> str:
    ^
./capnp/contrib/genpyi.py:227:17: W503 line break before binary operator
                + ", ".join(imp for imp in sorted(self.typing_imports))
                ^
./capnp/contrib/genpyi.py:242:71: E203 whitespace before ':'
    name = schema.node.displayName[schema.node.displayNamePrefixLength :]
                                                                      ^
./capnp/contrib/genpyi.py:252:71: E203 whitespace before ':'
    name = schema.node.displayName[schema.node.displayNamePrefixLength :]
                                                                      ^
./capnp/contrib/genpyi.py:265:1: C901 'gen_struct' is too complex (35)
def gen_struct(schema, writer, name: str = ""):
^
./capnp/contrib/genpyi.py:272:75: E203 whitespace before ':'
        name = schema.node.displayName[schema.node.displayNamePrefixLength :]
                                                                          ^
./capnp/contrib/genpyi.py:283:17: W503 line break before binary operator
                and field.slot.type.anyPointer.which() == "parameter"
                ^
2     C901 'Writer.type_ref' is too complex (11)
4     E203 whitespace before ':'
2     W503 line break before binary operator

[gjcarneiro]

Hm, can you tell me the command to check those?

To be honest, I just format with black. How quaint that this project wants a formatting style not compatible with black..

[haata]

I've started using black for some other projects (I quite like it actually). I'll do some experimentation to see how much it changes pycapnp (losing git blame history is a bit of a pain).

flake8 . --filename '*.py,*.pyx,*.pxd' --count --max-complexity=10 --max-line-length=120 --ignore=E211,E225,E226,E227,E231,E251,E261,E262,E265,E402,E999 --show-source --statistics --exclude benchmark,build,capnp/templates/module.pyx

flake8 . --count --max-complexity=10 --max-line-length=120 --show-source --statistics --exclude benchmark,build

[haata]

I think I've found the issue, I'll be updating the flake8 configuration shortly.

[haata]

@gjcarneiro can you rebase your commit to master? It should hopefully pass now.

[lgtm-com]

This pull request introduces 1 alert when merging b310447 into 6e7fffd - view on LGTM.com

new alerts:

• 1 for Unused import