fix(4.x): migrate from optimist to yargs

[fabb]

Follow-up on #1662, added a few unit test on top of the original PR.

closes #1658
closes #1661

Note that there are not yet tests for all command line options. I know too little about handlebars yet (it's just a sub-sub-dependency of a package I use). If someone has good suggestions on what makes sense to test, fire away, I will update the PR.

Since I use npm audit to fail my build to not get vulnerable packages get released to production, I'm very interested into getting this merged.

Tested:

• -f
• --map
• -a
• -c
• -h
• -k
• -o
• -m
• -n
• -s
• -N
• -i
• -r
• -p
• -d
• -e
• -b
• -v
• --help

[aorinevo]

I recommend looking at yargs unit tests and maybe mimicking what is done there to test handlebars cli options.

@fabb, happy to help on this, can I get permissions to your fork?

[aorinevo]

Seems like the unit testing structure in place will make this some what difficult.

Will try to leverage the current unit testing structure to get this done.

[fabb]

@aorinevo I invited you to this repo, feel free to commit to this branch, thanks for offering! I‘ll probably only get to continue on this in 2 days. I‘ll then have a look at what you‘ve done in the mean time, and at the yargs tests as you suggested. I guess for coordination it should be enough when we write here in this PR when we start and stop making changes, what do you think?

[aorinevo]

I think that makes sense. I'll update with comments accordingly.

[aorinevo]

Tracking testing progress on my end via:

• -f
• --map
• -a
• -c
• -h
• -k
• -o
• -m
• -n
• -s
• -N
• -i
• -r
• -p
• -d
• -e
• -b
• -v
• --help

[aorinevo]

@fabb, please give me a heads up before merging any changes from your end on this branch. I'd like to squash my commits into one; it's easier if there are no interwoven commits from other contributors.

[aorinevo]

@fabb, unfortunately I squashed your commits down with mine.

If you have those commits locally, I can add them back. Just push them to a branch called faab-mty-commits and I'll take it from there.

[aorinevo]

Added back @AviVahl's commits.

[aorinevo]

Going to pause here for a bit.

Added unit tests for 8 options. Might pick this up again later.

[fabb]

@fabb, unfortunately I squashed your commits down with mine.
If you have those commits locally, I can add them back. Just push them to a branch called faab-mty-commits and I'll take it from there.

I'm not here for fame, so it's fine with me ;-)

[fabb]

maybe we can read out the dynamic value from package.json to make updating less annoying

[ErisDS]

That would really be preferable! It's possible to do e.g. require('path/to/package.json').version

[aorinevo]

Might require reworking the task runner but definitely doable and a better approach.

[ErisDS]

another approach might be just verifying that the output is /\d+\.\d+\.\d+/ but I wouldn't worry about missing a test for this one command if it's too fiddly.

Main thing is that having this fail after every release would be a major maintenance burden 😅

[aorinevo]

Like most things, this ended up being pretty straight forward 🥳.

[aorinevo]

This file can be removed now.

[ErisDS]

@fabb and @aorinevo Thanks so much for taking this on. I'll be supporting this PR through to release.

If anyone out there regularly uses the handlebars CLI and can verify this PR is working for them, that would be fantastic.

Otherwise I'll give this a spin once you're both happy it's done.

[aorinevo]

Thanks @ErisDS, I’ll try to get this wrapped up tonight :)

[aorinevo]

@fabb, picking this backup now. Let me know if you have any WIP.

[aorinevo]

@ErisDS, I'd like to recommend skipping a unit test for -b as use of BOM is not required or recommended (see https://en.wikipedia.org/wiki/Byte_order_mark).

Thoughts?

[aorinevo]

@ErisDS, assuming we skip -b, this PR is ready for review.

[aorinevo]

Cause I'm stubborn, I added the last remaining unit test for -b.

Unit test is valid as it runs against a file that includes a BOM (see image below).

[DeeDeeG]

Edit: In hindsight, I'm pretty sure the proper thing would have been to just wait for a review from someone who knows these things. Hiding my comment as off-topic for now. Un-hidden for context.

(Disclaimer: This is just something I tried in the volunteer spirit, while we wait for a proper review. I am not fluent with JS projects, so take this with a grain of salt/feel free to ignore if it is not helpful. Apologies in advance if that is the case.)

Should these tests pass with the old optimist-based code dropped in place of the new yargs code? I tried to do this (new tests, old code/dependencies) and it failed, but it may have been a mistake on my end.

(My Travis runs can be viewed here: https://travis-ci.com/github/DeeDeeG/handlebars.js/branches)

[aorinevo]

@DeeDeeG, I'm rather curious myself why that's the case. I'm going to take a look.

[aorinevo]

@DeeDeeG, still looking for the root cause but worth noting that all the tests pass with the exception of the unit test for --help when I use optimist.