@mre I've got three questions:

1. The test_readme_usage_up_to_date test was annoying me a bit sometimes because all my editors trim empty lines when saving README.md. I've updated the test so that lines containing only whitespace are trimmed, so that empty lines are expected. Is this okay?
2. For now --gitignore is required so that ignored files are skipped. Would it be okay if we switched it around and update the flag to --no-gitignore? This would be a more sensible default in my opinion but also a breaking change.
3. Currently there is no way to include hidden files. (hardcoded previously as skip_hidden(true) now hidden(true)) Should I add an option (lychee-lib) and flag (lychee-bin) to include hidden files?

Very nice!

1. The test_readme_usage_up_to_date test was annoying me a bit sometimes because all my editors trim empty lines when saving README.md. I've updated the test so that lines containing only whitespace are trimmed, so that empty lines are expected. Is this okay?

Yes, same. Sounds very reasonable to me. Thanks!

2. For now --gitignore is required so that ignored files are skipped. Would it be okay if we switched it around and update the flag to --no-gitignore? This would be a more sensible default in my opinion but also a breaking change.

Absolutely. I'm all for it. Please go forward and change it if you like. Maybe we could use the --include prefix to align it with the other options. Maybe --include-gitignore or --include-hidden?

3. Currently there is no way to include hidden files. (hardcoded previously as skip_hidden(true) now hidden(true)) Should I add an option (lychee-lib) and flag (lychee-bin) to include hidden files?

Yes, that sounds sensible to me. 👍

@mre Thanks for the answers. The PR is now ready for review.

Just one last thing; clippy complains about too many booleans in the Collector struct. Should I extract the three skip booleans into a separate struct? How would you name that struct?

pub struct Collector {
    basic_auth_extractor: Option<BasicAuthExtractor>,
    skip_missing_inputs: bool,
    skip_ignored: bool,
    skip_hidden: bool,
    include_verbatim: bool,
    use_html5ever: bool,
    base: Option<Base>,
}

Awesome. I don't like that lint too much. Maybe we can ignore it for now.

Coincidentally, it's the most ignored lint: rust-lang/rust-clippy#5418
(Except for cognitive_complexity, which got removed.)

Yeah I also felt like this is very picky and currently there are 292 results which is quite a lot in comparison to other lints. Began extracting the three bools but it didn't feel like a benefit, only like more code. So I also prefer disabling the lint here.

I don't know why the publish-check is failing. It seems like that's unrelated and not a blocker.

For consistency, I'd propose to rename the options:

• --no-ignore → --include-ignored or --include-gitignore
• --hidden → --include-hidden

Here's the other options so far:

❯❯❯ lychee --help | rg 'exclude|include'                                                                                                                                                         
      --include <INCLUDE>
          URLs to check (supports regex). Has preference over all excludes
      --exclude <EXCLUDE>
      --exclude-file <EXCLUDE_FILE>
          Deprecated; use `--exclude-path` instead
      --exclude-path <EXCLUDE_PATH>
  -E, --exclude-all-private
          Equivalent to `--exclude-private --exclude-link-local --exclude-loopback`
      --exclude-private
      --exclude-link-local
      --exclude-loopback
      --exclude-mail
          Exclude all mail addresses from checking (deprecated; excluded by default)
      --include-mail
      --include-fragments
      --include-verbatim

I'm assuming you probably modeled the existing flags after ripgrep, which is a fine choice. Looking through the list of existing options above, I realize that they all have to do with links, not inputs (with the exception of --exclude-path). So I have no strong opinion on this. We could merge as-is, but I'd like to document our thought process here, so happy for any feedback.

@mre Yes exactly, I used the same names as ripgrep does as I find them very good.

--include-gitignore too me sounds like I'd include the .gitignore files themself for checking. But that's not the case. As you have mentioned all flags with include or exclude have to do with individual links not files (with one exception). So I prefer the shorter and clearer flags that align with ripgrep.

Ah, that makes sense. Thanks for the clarification and the awesome PR.