feat: add actions to export and import redactions to/from JSON #214

arthaud commented 2026年06月27日 23:58:20 +02:00

• Redactions are saved as a list in JSON object
• JSON object saves a version number for breaking changes
• Closes #8

DeepSeek-V4-Flash was used to write some of the code, but none of the comments or commit messages

- Redactions are saved as a list in JSON object - JSON object saves a version number for breaking changes - Closes #8 DeepSeek-V4-Flash was used to write some of the code, but none of the comments or commit messages

mahlzahn commented 2026年06月29日 11:48:02 +02:00

Dear @arthaud, thanks a lot for your contributions and inputs to the project! I‘d like to ask you for some clarifications.

1. What is the problem you want to solve with this feature? Could you please describe your usage and the limits you encountered?

2. Thanks for being honest in noting that LLMs were used in writing code. I am afraid that I cannot accept these contributions, because how can you demonstrate that the 1. contributed code can be submitted under the GPL license and 2. it doesn’t violate copyright of LLM training material? See, e.g.,

   • https://www.techtarget.com/searchEnterpriseAI/tip/Examining-the-future-of-AI-and-open-source-software
   • https://www.recordinglaw.com/us-laws/ai-copyright-laws/ai-generated-code-copyright-infringement/

   Also, the AI agreement of Forgejo (which is the base of the one from Censor) was recently amended to clarify this issue: forgejo/governance@57bf0779be (and I likely should update Censor’s AI agreement, too).

I personally use LLMs as "better search machine" in development sometimes, too, ensuring that all code I write is copyrightable by myself.

Dear @arthaud, thanks a lot for your contributions and inputs to the project! I‘d like to ask you for some clarifications. 1. What is the problem you want to solve with this feature? Could you please describe your usage and the limits you encountered? 2. Thanks for being honest in noting that LLMs were used in writing code. I am afraid that I cannot accept these contributions, because how can you demonstrate that the 1. contributed code can be submitted under the GPL license and 2. it doesn’t violate copyright of LLM training material? See, e.g., - https://www.techtarget.com/searchEnterpriseAI/tip/Examining-the-future-of-AI-and-open-source-software - https://www.recordinglaw.com/us-laws/ai-copyright-laws/ai-generated-code-copyright-infringement/ Also, the AI agreement of Forgejo (which is the base of the one from Censor) was recently amended to clarify this issue: https://codeberg.org/forgejo/governance/commit/57bf0779bec61e2facd1679efc9bc5839e631d40 (and I likely should update Censor’s AI agreement, too). I personally use LLMs as "better search machine" in development sometimes, too, ensuring that all code I write is copyrightable by myself.

arthaud commented 2026年07月01日 05:53:59 +02:00

Hi, thank you for your message.

1. The goal is to implement the changes from #8, i.e. allow users to export redactions and re-import them. This improves QOL greatly because it means you can save your progress as you redact a huge document, and if you render it (by saving) and later realize you wanted to redact less —or any other issue —then you can recover your progress.
2. I rewrote the LLM-generated parts of the code without using any AI on this PR. I will do the same on the other two shortly.

Hi, thank you for your message. 1. The goal is to implement the changes from #8, i.e. allow users to export redactions and re-import them. This improves QOL greatly because it means you can save your progress as you redact a huge document, and if you render it (by saving) and later realize you wanted to redact less —or any other issue —then you can recover your progress. 2. I rewrote the LLM-generated parts of the code without using any AI on this PR. I will do the same on the other two shortly.

This pull request doesn't have enough approvals yet. 0 of 1 approvals granted.

You are not authorized to merge this pull request.