Parse brewfiles line by line

[mootari]

Currently the DSL seems to choke on invalid quotes, causing the whole Brewfile to fail with "missing formula" errors for each single line. I would suggest to either

• split by newline and parse each line on its own, or
• validate the whole Brewfile and throw an error if any invalid characters have been encountered.

I'm not a ruby guy and I've only briefly looked at the code, but my guess would be that the first option is actually easier to implement.

[MikeMcQuaid]

Could you provide a sample failing Brewfile? Thanks!

[mootari]

Sure! I've provided multiple samples here: https://gist.github.com/cisso/ccadc42b4133d810d212 The one I mentioned above is Brewfile_fails_for_all_1.

Edit: Assuming that "Dsl" stands for domain-specific language, shouldn't that class merely parse/tokenize the Brewfile and return a list of sanitized Entry objects, so that they can be processed by a different class? (again, not a ruby guy)

[xu-cheng]

This issue is related to the legacy codes here. Instead of raising the error when dealing with invalid Brewfile, it tries to treat each line as individual formula name. In the result, some weird messages get out.

Should we just drop these legacy codes? cc @andrew

[andrew]

Yeah I'm not attached to the legacy code at all!

[kevinSuttle]

So, to be clear, what was the reason for requiring quotes in formula names inside of Brewfiles?

[MikeMcQuaid]

@kevinSuttle It's a Ruby DSL which means we don't need to implement our own parser.

[kevinSuttle]

@mikemcquaid Ok. So, not wanting to derail here, but that does directly affect Homebrew formulae, especially Caskroom.

[MikeMcQuaid]

@kevinSuttle I doesn't affect anything outside of Brewfiles.