0.2.2: Command checks fail due to color output and trying to check non-test files

[dvzrv]

Hi again 👋

After disabling the tests mentioned in #140 I was unfortunately met with further failing tests.

The below is a short excerpt, please have a look a the attached full logs for build and check for details.

     Running tests/command_check.rs (target/debug/deps/command_check-d1f8a22b4302851f)

running 8 tests
test check_stdin_with_file ... FAILED
test check_stdin ... FAILED
test check_empty_stdin_with_file ... FAILED
test check_exclusion ... ok
test check_quiet_with_config ... FAILED
test check_empty_stdin ... FAILED
test check ... FAILED
test check_quiet ... FAILED

mado-0.2.2-1-x86_64-build.log
mado-0.2.2-1-x86_64-check.log

From the logs I gather that some of the tests are somehow picking up the markdown files of dependencies in the target directory somewhere.
I don't think this is intentional and I guess the tests should instead specifically only target test files.

In other tests it appears as if the output is cross-checked but it looks like it fails on colored vs. plain output!

As further information (also for #140), this is how I'm building this project:

  export RUSTUP_TOOLCHAIN=stable
  export CARGO_TARGET_DIR=target
  cargo fetch --locked --target "$(rustc -vV | sed -n 's/host: //p')"
  cargo build --frozen --release --all-features
  cargo test --frozen --all-features

[akiomik]

Thank you for reporting too 😸

[dvzrv]

With 0.3.0 I ran into yet another test failure due to colorization:

running 2 tests
test generate_shell_completion_zsh ... ok
test generate_shell_completion_invalid ... FAILED

failures:

---- generate_shell_completion_invalid stdout ----

thread 'generate_shell_completion_invalid' panicked at /usr/src/debug/rust/rustc-1.87.0-src/library/core/src/ops/function.rs:250:5:
Unexpected stderr, failed diff original var
├── original: error: invalid value 'foo' for '<SHELL>'
│     [possible values: bash, elvish, fish, powershell, zsh]
│
│   For more information, try '--help'.
├── diff:
│   ---         orig
│   +++         var
│   @@ -1,2 +1,2 @@
│   -error: invalid value 'foo' for '<SHELL>'
│   -  [possible values: bash, elvish, fish, powershell, zsh]
│   +error: invalid value 'foo' for '<SHELL>'
│   +  [possible values: bash, elvish, fish, powershell, zsh]
│   @@ -4 +4 @@
│   -For more information, try '--help'.
│   +For more information, try '--help'.
└── var as str: error: invalid value 'foo' for '<SHELL>'
      [possible values: bash, elvish, fish, powershell, zsh]

    For more information, try '--help'.

command=`"/build/mado/src/mado-0.3.0/target/debug/mado" "generate-shell-completion" "foo"`
code=2
stdout=""
stderr=```
error: invalid value \'foo\' for \'<SHELL>\'
  [possible values: bash, elvish, fish, powershell, zsh]

For more information, try \'--help\'.
```


note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace


failures:
    generate_shell_completion_invalid

test result: FAILED. 1 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s

It would be great if these could be fixed, as disabling the tests introduces more maintainance overhead for downstreams.

(see below check logs for color escape codes)

mado-0.3.0-1-x86_64-build.log
mado-0.3.0-1-x86_64-check.log

[dvzrv]

FWIW, the unknown_command test is also broken now.