I think that the separation between Status::Failed and Status::Error is a
bit misleading. It was easier to implement, but from a user's perspective
they are more or less the same: something unexpected happened.
So I merged both into one: Status::Error. Still not 100% happy with the
semantics, but it's an improvement I'd say.
If an error occurs during link checking,
it is important to know where the error occured.
Therefore the request and response objects now contain a the input
source as a field. This makes error tracking easier.
For now we only support JSON.
I honestly don't know if it makes sense to include other formats.
For example, MD and HTML are not really
machine-readable. YAML is not
a great standard format for this use-case. Open for discussions, though.
This splits up the code into a `lib` and a `bin`
to make the runtime usable from other crates.
Co-authored-by: Paweł Romanowski <pawroman@pawroman.dev>
This implements a basic builder for the Checker struct as discussed in #12.
It is using derive_builder and uses a custom build method to instantiate the more elaborate fields like reqwest::Client.
It also adds deadpool and tokio::mpsc as dependencies to handle a pool of clients to query websites.
* Make GITHUB_TOKEN optional
This also makes the token possible to pass in from CLI args.
* Add missing test fixture file
* Normalize exit codes and GitHub checking behavior
The exit code is now defined as 1 for unexpected or config errors,
and 2 for link check failures.
GitHub checking behavior has been tweaked to generate errors if
a GitHub-specific check cannot be performed because of a missing
token.
* Remove short flag for github token
