LazyStruct and LazySequence don't use kwargs

[movermeyer]

When using the Python 3.6+ keyword argument syntax to build constructs, building lazy constructs fails.

Example Code:

from construct import Struct, LazyStruct, Byte

MY_STRUCT_NON_KW = Struct(
    "a" / Byte,
    "b" / Byte,
    "c" / Byte,
    "d" / Byte,
)

MY_LAZY_STRUCT_NON_KW = LazyStruct(
    "a" / Byte,
    "b" / Byte,
    "c" / Byte,
    "d" / Byte,
)

MY_STRUCT = Struct(a=Byte, b=Byte, c=Byte, d=Byte)
MY_LAZY_STRUCT = LazyStruct(a=Byte, b=Byte, c=Byte, d=Byte)

print("MY_STRUCT_NON_KW:", MY_STRUCT_NON_KW.parse(b'\x00\x00\x00\x00'))
print("MY_LAZY_STRUCT_NON_KW:", MY_LAZY_STRUCT_NON_KW.parse(b'\x00\x00\x00\x00'))
print("MY_STRUCT:", MY_STRUCT.parse(b'\x00\x00\x00\x00'))
print("MY_LAZY_STRUCT:", MY_LAZY_STRUCT.parse(b'\x00\x00\x00\x00'))

gives:

MY_STRUCT_NON_KW: Container:
    a = 0
    b = 0
    c = 0
    d = 0
MY_LAZY_STRUCT_NON_KW: <LazyContainer: 4 possible items, 0 cached>
MY_STRUCT: Container:
    a = 0
    b = 0
    c = 0
    d = 0
MY_LAZY_STRUCT: <LazyContainer: 0 possible items, 0 cached>

I would have expected the lazy versions to produce the same values as the non-lazy versions.
Especially since the docs say they are equivalent and take in kwargs:

construct.LazyStruct(*subcons, **kw):
Equivalent to Struct construct

This is a result of LazyStruct and LazySequence not using their kw parameters at all.

Credits: Discovered with help from pylint

[arekbulski]

Good find! Do you feel up to the challange to fix it? I trust your code.
Proper test case is also needed. You would need to turn prints into asserts.

[arekbulski]

I assume you found this in connection to previous topic #359

[movermeyer]

No, actually.

As part of my editor (Atom or VSCode), I run several static analysis tool/linters (pylint, flake8, autopep8) as plugins.

When I opened the construct code in my editor, the linters raised some interesting warnings, including this one: "Unused argument kw". I then investigated and found that it was actually a bug and reported it.

Most of the credit belongs to pylint.

[movermeyer]

As for this bug, I'll try to fix it.

[movermeyer]

I've fixed LazyStruct (very easy, see #368), but LazySequence is a more in-depth change.

Can I leave that to you?

[arekbulski]

Of course, thank you for all effort so far.

[arekbulski]

Fixed.