tables as kwargs for Python functions about lupa HOT 3 CLOSED

I didn't even know that feature in Lua, but it seems useful. It's described here:

http://lua-users.org/wiki/NamedParameters

I agree that it has to be done explicitly to prevent exploding all tables being passed into Python functions into separate arguments. A decorator sounds like a good way to do it. It also allows calling it directly on a Python function before passing it into Lua (i.e. not using it as a decorator but as a normal function).

Although, if users can prevent the case of passing single table arguments by accident, a global LuaRuntime() option might also do the trick. But then the opposite decorator would be required to prevent table explosion in those undesired cases where it accidentally turns out that it cannot be avoided after all.

Note that Lua allows both positional and keyword arguments to be passed through tables, so simply dropping the table into a dict is not sufficient. Testing len(table) (or doing the equivalent Lua operation) should distinguish both cases.

from lupa.

Good catch about positional arguments. I'll add support for them and submit a PR (decorators.py file with 2 public functions, also imported from __init__.py).

Are table_arguments and table_arguments_method names OK, or do you have better names in mind? lua_named_parameters and lua_named_parameters_method? just lupa.named_parameters[_method]? lupa.allow_named_parameters[_method]? table_as_kwargs?

Also, would you mind adding py.test or nose as a test runner?

from lupa.

decorators.py file with 2 functions, also imported from __init__.py

I wouldn't mind adding them right in _lupa.pyx, should be faster.

Are table_arguments and table_arguments_method names OK, or do you
have better names in mind? lua_named_parameters and
lua_named_parameters_method? just lupa.named_parameters[_method]?
lupa.allow_named_parameters[_method]? table_as_kwargs?

"unpack_lua_table"? "arguments_from_lua_table"?

It's not just about kwargs or named arguments, even though that's the main
use case.

Also, would you mind adding py.test or nose as a test runner?

py.test, if any, but unless there is an actual need for it, I'm quite happy
that you can currently run the test suite without installing external
Python dependencies.

from lupa.