ci: Add copyright-check workflow

[ko3n1g]

Description

Usage

# Add snippet demonstrating usage

Checklist

• I am familiar with the Contributing Guide.
• New or Existing tests cover these changes.
• The documentation is up to date with these changes.

[sarahyurick]

I have a few questions about how this is supposed to work. Some are a bit nitpicky but want to make sure:

• It looks like it is only supposed to check Python files. However, I see some places where init.py files are skipped. Is this intentional?
• Can we also have it check YAML/YML files and shell scripts?
• There are a couple Python files where the year is 2023 (see example) instead of 2024. Is there a way we can check for this too? I imagine when 2025 hits we will want to update too?
• There are other instances, such as here, where the copyright is even more out of date.
• Sometimes the files say Copyright (c) 2024, NVIDIA CORPORATION. All rights reserved. and sometimes they just say Copyright (c) 2024, NVIDIA CORPORATION.. Is there any way we can do an exact string match to ensure that all copyrights are identical?
• Should we make sure the Dockerfile has the copyright?

[ko3n1g]

I have a few questions about how this is supposed to work. Some are a bit nitpicky but want to make sure:

• It looks like it is only supposed to check Python files. However, I see some places where init.py files are skipped. Is this intentional?
• Can we also have it check YAML/YML files and shell scripts?
• There are a couple Python files where the year is 2023 (see example) instead of 2024. Is there a way we can check for this too? I imagine when 2025 hits we will want to update too?
• There are other instances, such as here, where the copyright is even more out of date.
• Sometimes the files say Copyright (c) 2024, NVIDIA CORPORATION. All rights reserved. and sometimes they just say Copyright (c) 2024, NVIDIA CORPORATION.. Is there any way we can do an exact string match to ensure that all copyrights are identical?
• Should we make sure the Dockerfile has the copyright?

Maybe best to sync with Pablo on this, he's in charge of the copyright script.

[ko3n1g]

Suggested change

	uses: NVIDIA/NeMo-FW-CI-templates/.github/workflows/_copyright_check.yml@v0.2.0
	uses: NVIDIA/NeMo-FW-CI-templates/.github/workflows/_copyright_check.yml@v0.14.0

This version allows to check other file types too.

[sarahyurick]

Thanks! From NVIDIA/NeMo-FW-CI-templates#57, I am thinking we should set additional-find-args to .yml, .yaml, and .sh.

[pablo-garay]

Sorry i was writing reply for this & got into something else & missed it

It looks like it is only supposed to check Python files. However, I see some places where init.py files are skipped. Is this intentional?
Yes, init.py files are empty so we skip them
Can we also have it check YAML/YML files and shell scripts?
Just created a PR to allow this
There are a couple Python files where the year is 2023 (see example) instead of 2024. Is there a way we can check for this too? I imagine when 2025 hits we will want to update too? There are other instances, such as here, where the copyright is even more out of date.
I am not sure about this actually . I had a few talks and we didnt see this a needed requirement so far
Sometimes the files say Copyright (c) 2024, NVIDIA CORPORATION. All rights reserved. and sometimes they just say Copyright (c) 2024, NVIDIA CORPORATION.. Is there any way we can do an exact string match to ensure that all copyrights are identical?
I guess we can do this / it's doable
Should we make sure the Dockerfile has the copyright?
Not sure, need to check on this / if this needed. And which exact Dockerfile are we talking about?

[sarahyurick]

Thanks @pablo-garay ! A couple more comments:

• A few of our __init__.py files are not empty. How should we handle this?
• Thanks for feat: Enable other file types (format) copyright checks NeMo-FW-CI-templates#57!
• I will check with the rest of the team about the year mismatches.
• I am referring to this Dockerfile: https://github.com/NVIDIA/NeMo-Curator/blob/main/Dockerfile. I think we should add the copyright, although since there is only 1 Dockerfile in this case I am okay with it not being part of the CI.