Import JRuby's strscan

[headius]

Finally getting around to addressing #15. Initial import has gone well but there are a few failures to look into, probably just feature/fix drift in our implementation.

I will fix the issues and make sure this packages properly as a gem, so we can start using it in JRuby 9.4 (equivalent to CRuby 3.0).

TODO:

• Import and build JRuby strscan impl
• Add license disclaimer or get sign-off from contributors to relicense
• Pass tests
• Add a CI run
• Package java gem

[kou]

The current CI failures are fixed by 3ceafa6.
Could you rebase on master?

And do you have a plan to add jobs for JRuby to the current CI?

[headius]

Could you rebase on master?

Thank you, yes I will rebase!

And do you have a plan to add jobs for JRuby to the current CI?

That is the primary reason this is a draft. We need to get all the tests green and then I will push a CI job for JRuby.

[headius]

Also needs some tweaks to get a -java gem released. I will add a checklist to the PR description.

[headius]

I have reviewed the current code and we would need confirmation from the following individuals to relicense.

Contributors listed below: please respond in a comment with the following, assuming you consent:

"I consent to having my contributions relicensed under the Ruby license and the BSD 2-clause license."

• @headius
• @lopex
• @vvs
• @enebo
• @kares
• @nicksieger
• @nahi
• @knawrocke
• @olabini / @olabiniV2

For me:

I consent to having my contributions relicensed under the Ruby license and the BSD 2-clause license.

[headius]

Silly me... I forgot that @knawrocke contributed the original code over email 15 years ago, so I added him to the list.

[knawrocke]

I consent to having my contributions relicensed under the Ruby license and the BSD 2-clause license.

[nicksieger]

Ohai 👋 !

I consent to having my contributions relicensed under the Ruby license and the BSD 2-clause license.

[headius]

Added @olabini / @olabiniV2 just to cover everyone who touched this code.

[enebo]

I consent to having my contributions relicensed under the Ruby license and the BSD 2-clause license.

[olabiniV2]

Hi everyone. I am Ola Bini - the owner of the @olabini account - which I'm currently locked out of. If any legal issues arise, I'm willing to provide offline government identification if necessary, to support this agreement.

I consent to my contributions being relicensed under the Ruby license and the BSD 2-clause license.

[lopex]

I consent to having my contributions relicensed under the Ruby license and the BSD 2-clause license.

[headius]

This should now be green on all tests and I am just tweaking build and CI.

Still looking for consent to relicense from @vvs@nahi@kares.

[kares]

👍 for the effort and

I consent to having my contributions relicensed under the Ruby license and the BSD 2-clause license.

[headius]

I have pinged @vvs and @nahi through other channels to try to get their consent to relicense. If anyone knows them personally, let them know I'm looking for them! 😀

[nahi]

I didn't know I did something for the file. Long history...
I consent to having my contributions relicensed under the Ruby license and the BSD 2-clause license.

[vvs]

Hi there, fellas, long time no see!

"I consent to having my contributions relicensed under the Ruby license and the BSD 2-clause license."

[headius]

Thank you @nahi and @vvs (and good to hear from you both!

With that, all contributors to this code have given consent to relicense!

[kou]

Great!!!

@headius Could you confirm the test failure only on Windows + JRuby?
https://github.com/ruby/strscan/runs/5120568110?check_suite_focus=true

================================================================================
Failure: test_AREF(TestStringScanner):
Expected Exception(IndexError) was raised, but the message doesn't match.
Expected /\u{30c6 30b9 30c8}/ to match "undefined name <\u00E3\u0192\u2020\u00E3\u201A\u00B9\u00E3\u0192\u02C6> reference".
<nil> is not true.
D:/a/strscan/strscan/test/lib/core_assertions.rb:489:in `assert'
D:/a/strscan/strscan/test/lib/core_assertions.rb:460:in `assert_raise_with_message'
D:/a/strscan/strscan/test/strscan/test_stringscanner.rb:468:in `test_AREF'
465: assert_equal 'foo', s['a']
466: assert_equal 'bar', s['b']
467: assert_raise(IndexError) { s['c'] }
=> 468: assert_raise_with_message(IndexError, /\u{30c6 30b9 30c8}/) { s["\u{30c6 30b9 30c8}"] }
469: end
470: 471: def test_pre_match
org/jruby/RubyKernel.java:1238:in `catch'
org/jruby/RubyKernel.java:1233:in `catch'
org/jruby/RubyKernel.java:1238:in `catch'
org/jruby/RubyKernel.java:1233:in `catch'
================================================================================

Note: "\u{30c6 30b9 30c8}" is "テスト" (Japanese) ("test" in English)

Is it a test problem? Or is it a JRuby on Windows problem?

[headius]

Is it a test problem? Or is it a JRuby on Windows problem?

@kou I believe this is a problem specific to Windows, and only affects the error message here (the error is still raised, but due to Windows + Java exception logic and character encoding, the message is encoded incorrectly).

I can open a JRuby issue to fix that but it will not be in a release for a little while. We could skip that test for now, or simplify it for JRuby.

[headius]

We could skip that test for now, or simplify it for JRuby.

I should also point out that this is the last test in test_AREF, and we pass everything else there. Perhaps the error message is not critical enough to block JRuby?

[headius]

jruby/jruby#7177 filed for the Windows error message encoding failure.

[kou]

OK.
I'll merge this. Thanks!

[headius]

@kou Can we get a preview gem for this one too please?

[kou]

Sure!
Done.

[headius]

I have updated JRuby master to use the strscan gem! That change will get picked up in the next nightly snapshot.

[kou]

Thanks!
I'll re-run CI tomorrow.

[kou]

CI sill works except jruby/jruby#7177 : https://github.com/ruby/strscan/actions/runs/2186790823

[eregon]

This is actually unfortunate, because it would just work on TruffleRuby otherwise, since the strscan.rb in stdlib would win over strscan.so.
Yet another reason for JRuby to support require 'strscan' to load the strscan.jar + the necessary setup.

[eregon]

Specifically, this causes:

bundle exec rake test
/home/eregon/.rubies/truffleruby-dev/bin/truffleruby run-test.rb
<internal:core> core/kernel.rb:236:in `gem_original_require': cannot load such file -- strscan.so (LoadError)
Did you mean? strscan
from /home/eregon/code/strscan/lib/strscan.rb:5:in `<top (required)>'
from <internal:core> core/kernel.rb:234:in `gem_original_require'
from /home/eregon/code/strscan/test/strscan/test_stringscanner.rb:7:in `<top (required)>'
from <internal:core> core/kernel.rb:234:in `gem_original_require'
from run-test.rb:8:in `block in <main>'
from <internal:core> core/dir.rb:261:in `each'
from <internal:core> core/dir.rb:261:in `glob'
from run-test.rb:7:in `<main>'
rake aborted!

while tests ran fine before.

[eregon]

#35 to fix it

[eregon]

Is this necessary or just RUBY_ENGINE == 'jruby' would be enough?
(I noticed this line when modifying this file and wondering whether it can be simplified)

[kou]

@headius What do you think about this?

[headius]

I think RUBY_ENGINE should be enough. It has been available for a long time.

[eregon]

#36