chore: upgrade to micromatch 3

[aldarund]

Summary

Fixes #6546 .
From micromatch docs ->

when forward slashes are defined in a glob pattern, the resulting regular expression will match windows or POSIX path separators just fine.

Test plan

Same config as in referenced issue work fine with this fix

yarn test
yarn run v1.8.0-20180601.1611
$ jest
 PASS  tests/unit/components/passwordValidation.spec.js
 PASS  tests/unit/components/paymentForm.spec.js
 PASS  tests/unit/pages/index.spec.js
 PASS  tests/unit/components/loginForm.spec.js
 PASS  tests/unit/pages/signup.spec.js (5.426s)

Test Suites: 5 passed, 5 total
Tests:       14 passed, 14 total
Snapshots:   0 total
Time:        8.644s
Ran all test suites.
Done in 10.40s.

[SimenB]

Thanks for the PR! Mind adding a test as well?

[thymikee]

We have a helper doing something similar in jest-config:
https://github.com/facebook/jest/blob/c6512ad1b32a5d22aab9937300aa61aa87f76a27/packages/jest-config/src/utils.js#L58-L59

[aldarund]

@SimenB there is already tests that was able to catch this problem, but they are disabled on windows
https://github.com/facebook/jest/blob/master/packages/jest-cli/src/__tests__/search_source.test.js#L28
I have checked it and there only 1 test fails on windows ( last one with /foo/bar/prefix path ). But there other test that fail without this fix and work with it. So tests exists but they all disabled due to one test.

@thymikee ye, i saw, and there also similar thing in jest-regex-util, but not exactly same.

[thymikee]

@aldarund feel free to wrap this test single test in process.platform === 'win32' check

[SimenB]

This is great, thank you so much or tackling it!

Mind adding an entry to the changelog?

Might be a good idea to rebase first, as you'll probably get a conflict

[codecov-io]

Codecov Report

Merging #6650 into master will decrease coverage by <.01%.
The diff coverage is 62.5%.

@@            Coverage Diff             @@
##           master    #6650      +/-   ##
==========================================
- Coverage   68.28%   68.27%   -0.01%     
==========================================
  Files         249      250       +1     
  Lines        9628     9633       +5     
  Branches        5        6       +1     
==========================================
+ Hits         6574     6577       +3     
- Misses       3052     3054       +2     
  Partials        2        2

Impacted Files	Coverage Δ
packages/jest-cli/src/runJest.js	75% <ø> (ø)	⬆️
packages/jest-runtime/src/shouldInstrument.js	100% <ø> (ø)	⬆️
packages/jest-message-util/src/index.js	82.3% <ø> (ø)	⬆️
packages/jest-util/src/index.js	0% <ø> (ø)	⬆️
packages/jest-haste-map/src/HasteFS.js	50% <0%> (ø)	⬆️
packages/jest-util/src/replacePathSepForGlob.js	0% <0%> (ø)
packages/jest-cli/src/SearchSource.js	58.75% <100%> (+0.52%)	⬆️
packages/jest-config/src/normalize.js	83.01% <75%> (-0.19%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update cfc1782...9e28572. Read the comment docs.

[aldarund]

@SimenB done

[SimenB]

@mjesun this might be worth a patch release

[thymikee]

We've switched back to micromatch 2. Is this still necessary?

[SimenB]

Not sure about the last commit, although it seems correct: https://www.npmjs.com/package/micromatch#backslashes

[SimenB]

Finally green. The normalization is super ugly, but I think we should clean that up in jest 25 with #7185

The issue here is that globs need forward slash as separator, which doesn't work for regexes (at least not how we use them currently). I think it'll be easier to normalize correctly when it's all globs (or we'll at least be forced to go through where it's used and fix it properly)

[thymikee]

Finally green. The normalization is super ugly, but I think we should clean that up in jest 25 with #7185

Agreed. Btw, how about placing replacePathSepForGlob inside jest-util and use it consistently for all globs?

[thymikee]

Looks like we're pretty consistent now in terms of escaping globs for micromatch. Would be even better to extract a helper for matching globs that uses micromatch as its implementation detail, which seems like a nice followup refactor to this PR :)

[SimenB]

I can try to PR replacePathSepForGlob to micromatch - seems like a nice utility function it should export?

[thymikee]

Would be awesome if they agreed! Maybe it's better to ask first

[SimenB]

Done: micromatch/micromatch#142

[SimenB]

🤞

hopefully it'll work out 🙂

[github-actions]

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.
Please note this issue tracker is not a help forum. We recommend using StackOverflow or our discord channel for questions.