chore(src/compiler/msvc): replace rust-encoding with encoding-rs

[liushuyu]

This pull request replaces rust-encoding with encoding-rs.

The former is unmaintained and uses ambiguous encoding for ISO 8859-1 (note that CP1252 aka "Latin-1" on Windows(R) is a superset of ISO 8859-1). This pull request also corrects this issue by using the WINDOWS-1252 encoding to match the encoding more closely.

[Xuanwo]

Please rebase with main branch to remove unrelated commits.

[liushuyu]

Please rebase with main branch to remove unrelated commits.

Done. Rebased on top of the main branch

[liushuyu]

No, there are some differences I failed to account for when switching the implementations.

[liushuyu]

Fixed. I also added one more test case to test the decoding logic more thoroughly.

[codecov-commenter]

Codecov Report

Attention: 15 lines in your changes are missing coverage. Please review.

Comparison is base (3d5d41f) 30.49% compared to head (65512a1) 30.95%.

Files	Patch %	Lines
src/compiler/msvc.rs	51.61%	3 Missing and 12 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2010      +/-   ##
==========================================
+ Coverage   30.49%   30.95%   +0.46%     
==========================================
  Files          51       51              
  Lines       19638    19617      -21     
  Branches     9437     9427      -10     
==========================================
+ Hits         5988     6072      +84     
+ Misses       8015     7874     -141     
- Partials     5635     5671      +36     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[liushuyu]

CI tests passed. The coverage misses are due to negative code-path not being reached (should I add tests to test invalid encodings?)

[xen0n]

CI tests passed. The coverage misses are due to negative code-path not being reached (should I add tests to test invalid encodings?)

In general, negative test cases are also, if not more, beneficial, as arguably the error-handling codepath is more valuable regarding UX and test coverage... and in this case I think we want to be robust against inputs that definitely can contain invalid encoding, and can be controlled by the user.

[sylvestre]

do you know if it fixes any issue ?

[liushuyu]

do you know if it fixes any issue ?

In theory, any valid Windows 1252 but invalid ISO 8859-1 code points (e.g. file names) should work now.