Update clipboard-win to v5 and replace winapi with windows-sys

[rhysd]

This PR does two things for Windows support:

1. Update clipboard-win to v5.0.0
   • I checked changes in v5.0.0 and the changes didn't affect any implementation in this crate
2. Replace winapi crate dependency with windows-rs
   • winapi is unmaintained. windows-rs is well-maintained by Microsoft officially
   • By this change, winapi can be removed completely from this crate
   • The replacement is somewhat mechanical translation from winapi to windows-rs. I read MSDN documents but my understanding of the Win32 APIs is not perfect. I'd like to have a detailed code review
   • Some constants used by this crate are not defined in windows-rs. I defined them directly in platform/windows.rs

I confirmed writing/reading clipboard with both text and image contents using hello_world and set_image/get_image examples on my Windows 10 machine.

With this PR, cargo dependencies will change to:

> cargo tree --edges no-dev
arboard v3.3.0
├── clipboard-win v5.0.0
│   └── error-code v3.0.0
├── log v0.4.20
├── thiserror v1.0.35
│   └── thiserror-impl v1.0.35 (proc-macro)
│       ├── proc-macro2 v1.0.43
│       │   └── unicode-ident v1.0.4
│       ├── quote v1.0.21
│       │   └── proc-macro2 v1.0.43 (*)
│       └── syn v1.0.100
│           ├── proc-macro2 v1.0.43 (*)
│           ├── quote v1.0.21 (*)
│           └── unicode-ident v1.0.4
└── windows v0.52.0
    ├── windows-core v0.52.0
    │   └── windows-targets v0.52.0
    │       └── windows_x86_64_msvc v0.52.0
    └── windows-targets v0.52.0 (*)

[complexspaces]

Thanks for the PR, I appreciate the effort. Would you consider modifying this to use windows-sys instead of the windows crate? The former has much less version churn and I'm not sure how much value arboard gets out of using windows because we don't utilize any COM or WinRT logic.

I would definitely be in favor of implementing a helper function to wrap some of the Global* calls to handle checking for errors internally though.

[rhysd]

It makes sense. I'll try windows-sys.

[rhysd]

I would definitely be in favor of implementing a helper function to wrap some of the Global* calls to handle checking for errors internally though.

I focused on replacing winapi with windows-rs with as small diff as possible. But I'll refactor the code more idiomtic.

[complexspaces]

No worries either way, I could refactor it later if you are busy. I generally don't mind a little bit of scope creep in PRs that touches the same area as long as its independently reviewable (its own commit, etc).

[rhysd]

No problem. I like refactorying so I'll try it by myself :)

[rhysd]

Separating this module as a new file may be a good idea.

[rhysd]

I added some refactorings:

• a833222 : Add Error::unknown factory function to avoid Error::Unknown { description: expression.into() } boilerplate
• ec5e1d8 : Separate the implementation details of image data handling into image_data module to avoid many repeats of #[cfg(feature = "image-data")] gates. Moving the module to a new file may be a good idea (I don't have a strong opinion here).
• ba5575b : Wrap Win32 functions which can return errors to leverage error handling with ? operator. This made add_cf_dibv5 and read_cf_dibv5 simpler
• ba9e7e8 : Move tests for image data handling to image_data module and added more test cases

[rhysd]

The dependency tree is now:

> cargo tree --edges no-dev
arboard v3.3.0
├── clipboard-win v5.0.0
│   └── error-code v3.0.0
├── log v0.4.20
├── thiserror v1.0.35
│   └── thiserror-impl v1.0.35 (proc-macro)
│       ├── proc-macro2 v1.0.43
│       │   └── unicode-ident v1.0.4
│       ├── quote v1.0.21
│       │   └── proc-macro2 v1.0.43 (*)
│       └── syn v1.0.100
│           ├── proc-macro2 v1.0.43 (*)
│           ├── quote v1.0.21 (*)
│           └── unicode-ident v1.0.4
└── windows-sys v0.52.0
    └── windows-targets v0.52.0
        └── windows_x86_64_msvc v0.52.0

[rhysd]

Thank you for your reviews. I'll have a look at them tonight.

[rhysd]

@complexspaces I addressed all review comments. Would you take a look again?

[complexspaces]

Thanks a lot for your work on this, everything looks good now!