Updating existing content to Gutenberg doesn't honor existing attributes

[caraya]

Issue Overview

Steps to Reproduce (for bugs)

1. Open an old post (I use markdown and fenced code blocks with language-<name> to indicate the language Prism should use for highlight in my content). Prism is configured as a plugin and works correctly before opening the post in Gutenberg
2. Convert the classic block containing the post into blocks
3. The code block has a fixed default value wp-block-code for the class attribute
4. If I try to add a class attribute manually I get a validation error in Chrome dev tools
5. I have to manually add the class from the editor for every individual code block

Browsers tested with: Firefox Nightly (60.0a1 / 2/28/18), Chrome (64.0.3282) on MacOS
Gutenberg: 2.3.0 compiled from source on 3/3

Expected Behavior

If there is an existing class for an element that gets converted to a block, that class should be honored when converting the element to a block. I don't think this is related to Automattic/jetpack#7786 as the content has already been rendered and saved to the database before being opened in Gutenberg and the code displays properly before the conversion

Current Behavior

Code blocks lose the language- attribute and the code highlight.

Possible Solution

Screenshots / Video

Related Issues and/or PRs

Todos

• Tests
• Documentation

[youknowriad]

If I try to add a class attribute manually I get a validation error in Chrome dev tools

You can use the block inspector to add your className to avoid having these errors.

[caraya]

that's not the point, I indicated that I had tried it manually because the class had been removed. I'm also curious why I get a validation error when I add to an existing attribute in the element

The old post had a class to indicate the type of language I want to highlight. I shouldn't have to add it again just because Gutenberg removed it.

[youknowriad]

I understand that the converting mechanism should be able to add this extra className if detect, I just wanted to help with a workaround for now.

Also, the converting mechanism is not done automatically on purpose. Because there's no way to guarantee no data loss. That's why you have to do it on your own if it's not the result you hoped for you can still undo and continue using the classic block.

[youknowriad]

cc @iseulde @mcsf

[caraya]

If you use the classic or HTML blocks you lose a lot of the Gutenberg functionality. That's fine with me, I write a lot of HTML in my posts already, but it may not be for everyone upgrading content.

[ellatrix]

@caraya Could you provide the HTML for us to test and make a PR with?

[caraya]

I thought I had done this already, I've attached a file with the markdown doc and the converted HTML
Archive.zip

[danielbachhuber]

I understand that the converting mechanism should be able to add this extra className if detect, I just wanted to help with a workaround for now.

This was handled with #7538