[JA] Proofreading on GitHub Actions

[sfujiwara]

@lamberta
Japanese reviewers: @ohtaman, @masa-ita, @AseiSugiyama, @yukinagae, @nuka137, @chie8842, @kiszk

This PR is a proposal of #144

This PR allows us to automatically apply proofreading tool on GitHub Actions.
It would be a great help for Japanese document reviewers and translators.

I'm happy to hear your opinion about this PR!

Files and directories

.github/workflows/ja.yaml

This workflow runs only when directories and files below are changed:

on:
  push:
    paths:
      - ".github/workflows/ja.yaml"
      - "site/ja/**"
      - "tools/ja/**"

tools/ja/

This is a proofreading tool used in Japanese community.

Resulting outputs

We can see the result of proofreading on GitHub PR.
The CI would fail if there are too many errors:

We can find the detailed results in GitHub Actions log:

[lamberta]

Nice! I'll have a look through this week and will test some things.

[yashk2810]

Make the python version 3.6.

Google is at this version so we can;t test anything beyond 3.6.

[sfujiwara]

OK. I fixed it!

[lamberta]

Hi @sfujiwara , do you know if redpen has a list of supported languages somewhere? I recall it not working for Korean, are there any other restrictions?

You are using this to maintain a translation glossary, yes? (I'm just wondering if other communities can use this setup, as well)

[kiszk]

@lamberta This PR intentionally (and temporary) drop supporting Korean due to CI failure. There is an issue to support Korean again. But, it have not been closed yet.

[sfujiwara]

@lamberta RedPen has various kinds of validators for example sentence length, invalid words, and so on. Some validators does not support Korean. But, other validators supports all languages including Korean.

I guess RedPen and GitHub Actions are useful for other communities!
See the official document for details:
http://redpen.cc/docs/1.10/index.html#validator

[sfujiwara]

Hi, how is it going on this PR?

[lamberta]

Tested this locally and seems to work ... I used homebrew to install the redpen package, probably out of scope to provide detailed install instructions, but maybe mention the Java dependency---speaking of Java, will this run in the GitHub Actions VM?

There was a log of noise at the command-line when running this, is that usual? Any way to filter out all these deprecation warnings?

[lamberta]

Thanks for the bump on this @sfujiwara (been swamped in email lately).

Tested locally and over here: lamberta#6
and seems to work as advertised.

As long as this is restricted to the ja/ translations then seems fine. Long-term, I'll need your help to maintain this and update if there are structural changes to the docs (things move around). If that's ok with you, then we can merge.

[sfujiwara]

@lamberta OK. We will maintain and improve the proofreading tool to make our translation tasks efficient!

[sfujiwara]

Hi @lamberta

There was a log of noise at the command-line when running this, is that usual? Any way to filter out all these deprecation warnings?

I added a minor change for the above proposal.

• Separate RedPen logs to files and make it available as GitHub Actions artifacts
• Add logback.xml to set RedPen log level Warn on GitHub Actions

speaking of Java, will this run in the GitHub Actions VM?

Currently it works on my repository and GitHub Actions (I will delete this repository if this PR is merged):
https://github.com/sfujiwara/docs-l10n/actions/runs/107112341

[lamberta]

Thanks @sfujiwara for the updates.

@MarkDaoust and @yashk2810 if there's no objections, I'll merge this for the ja folks. If there are maintenance concerns or changes needed, we can loop in @sfujiwara

[MarkDaoust]

LGTM.

[lamberta]

Sounds good. We'll keep an eye on it and will loop you in if there ae any issues. Thanks!