Use AST parser to validate alt text - Remark

[kendallgassner]

What

Closes #30

I updated the alt-text-bot to use an AST (remark) to parse images for bad alt text. Using an AST allows us to ignore any images defined in code blocks.

Why is this cool

• In the future, we can add additional linting rules
• The workflow is easier to read in javascript
• We can test more easily
• We can ensure that there aren't false positives
• We can provide more feedback. For example, it is easier to indicate what line the alt text is found on.

TODO

• - I was able to add codeblock tests and ensure all of our old test cases were passing. I have not had a chance to check if the action still runs properly.

Testing

If you find it useful I used AST Explorer to test remark output.

[khiga8]

Excited to see this! First question -- is there a reason you went with remark over markdownlint? I would personally prefer the latter because it already comes with the no alt text rule.

We also have additional rules in markdownlint-github like no-default-alt-text for flagging screenshots, and there are plans to add more rules soon! I would prefer if we align on markdownlint so we can use the existing rules!

[kendallgassner]

Remark:

• has a wider user base
• is a much small package (107.4kb vs 583.5kb), though I recognize that it's not a huge deal in the case of a github action.
• Documentation was a bit easier to follow for this specific scenario.
• It is hooked up to the AST explorer making the AST syntax more discoverable
• I don't think markdownlint api: lintInput allows you to test one rule I think they also include their default ruleset
• Also they have some accessibility focused plugins too!

I see the value in using a rule set GitHub has worked on though 🤔 ... Do we have more than the two rules?

cc. @khiga8