don't log to stdout if not cli

[Dauaaa]

This PR is a proposal to solve #706. I understand there are a lot of ways to do this so I just tried a simple approach.

I did this by adding a static boolean that will be set to true in main.rs making the program output to stdout (current behavior). Otherwise (if boolean is not set) the old logs will use log::$level.

I used #[doc(hidden)] attribute to avoid flooding the API documentation with print related utilities.

---

Extra: removed once_cell dependency

[bee-san]

Running CI now, this PR looks cool 🔥

[Dauaaa]

@bee-san didn't catch the clippy warning because my clippy was not updated.

I fixed the error, please try running CI again :)

[bee-san]

I am also thinking about whether or not it's worth testing for this. We could write a test similar to https://github.com/RustScan/RustScan/blob/master/tests/timelimits.rs that runs rustscan as the api (2 files?) and if it logs to stdout, the test fails 🤔

[Dauaaa]

@bee-san I added a test to check if library outputs to stdout, I used an unstable feature for tests only.

The other options I thought of:

• build another binary that doesn't call set_cli_mode and do something similar to the timelimits test
• add third party dependencies that does something similar to what I did

[bee-san]

The ci is failing because Just is unhappy:

cargo nextest run
8
    Updating crates.io index
9
   Compiling rustscan v2.3.0 (/Users/runner/work/RustScan/RustScan)
10
error[E0554]: `#![feature]` may not be used on the stable release channel
11
 --> tests/log_mode.rs:4:1
12
  |
13
4 | #![feature(internal_output_capture)]
14
  | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

This is because the CI uses the stable release channel of Rust, may need to update the workflow :)

Thanks so much for the test!!

[Dauaaa]

@bee-san Because I can't use unstable rust, I created a rust example to test, I used example to avoid having binaries that could be confused for the actual program. Simple explanation about how examples work in rust.

I actually read the complete contributing.md and ran cargo test inside the docker container, it worked. So I'm pretty sure it won't error this time. I'm just not sure you're satisfied with the current approach for testing 😅.

Sorry for delay, I was a bit crunched this week so took me a while to respond.

[bee-san]

It's okay! It's open source, take as long as you need.

With that in mind, I'd need to read the docs on some of these functions. Give me a bit of time :)