Accessing the Original Value of Directives in remark-directive

[jrpedrianes]

I am developing an application that integrates remark-directive. In some cases, I need to access the original value of a directive, as the parsed version does not always meet my requirements. For example, given the following case:

::youtube{url="https://video.com" with="450px"}

Currently, I only get the parsed value, but I would like to access the original string as well to process it as needed.

Questions for the community:

• Is there a built-in way in remark-directive to retrieve the original string of a directive?
• Should I consider creating a custom extension to achieve this behavior?
• Are there any examples or recommended approaches for handling such cases?

Greetings

[JounQin]

I believe the solution is https://github.com/syntax-tree/unist-util-source

[jrpedrianes]

Thanks 🎉 ,

Using the suggested dependency we can resolve it 🚀

[ChristianMurphy]

I need to access the original value of a directive, as the parsed version does not always meet my requirements
Currently, I only get the parsed value, but I would like to access the original string as well to process it as needed.

This feels a lot like an XY problem. https://xyproblem.info/
What about the parsed version doesn't meet your requirements?
All of the information about the directive name, type, and attributes are in the AST.

Is there a built-in way in remark-directive to retrieve the original string of a directive?

You probably don't want or need to do this, but if you choose to see @JounQin's answer https://github.com/orgs/remarkjs/discussions/1398#discussioncomment-11835478

Should I consider creating a custom extension to achieve this behavior?
Are there any examples or recommended approaches for handling such cases?

What behavior do you want?
What case you trying to solve?

[jrpedrianes]

’m using Milkdown, which relies on remarkjs as a dependency. When configuring remark-directive, I encounter errors whenever a directive is not recognized. My goal is to create a fallback directive, similar to this one, but without losing any of the text entered by the user. This includes preserving attributes written with quotes and maintaining the order of attributes.

Thanks to the dependency suggested by @JounQin, I’m able to achieve this while ensuring that no text characters or attribute order are lost.

[ChristianMurphy]

When configuring remark-directive, I encounter errors whenever a directive is not recognized

What do you mean not recognized?
It sounds like a filter exists in milkdown? https://github.com/Milkdown/website/blob/63442482ad177809eb899e6edfee4f41f9be0542/docs/plugin/example-iframe-plugin.md

That isn't a restriction of remarks directive though.

You probably still want to rely on the AST rather than reparsing the source again.

no text characters or attribute order are lost.

What characters do you believe are lost? What order is lost and why does it need to be ordered?

[jrpedrianes]

In our case, users can input any text, and sometimes there isn’t a textDirective plugin available to support it. As a result, Milkdown throws an error because the corresponding node doesn’t have a properly registered schema that handle it.

It’s important for us to preserve all user input exactly as written, as we use internal marks that cannot be altered. Our users need to see the text in the markdown exactly as they wrote it.

Also, when directives are parsed by remark-directive, some information, like the order of attributes within a directive, can be lost. This is something we’d like to avoid to ensure the user’s input remains as close to the original as possible.

[ChristianMurphy]

In our case, users can input any text, and sometimes there isn’t a textDirective plugin available to support it. As a result, Milkdown throws an error because the corresponding node doesn’t have a properly registered schema that handle it.

You can offer a fallback without needing to reparse the source.
This plugin will have an AST node for anything directive like.
If you know what it is, replace it, otherwise represent a fallback.

It’s important for us to preserve all user input exactly as written, as we use internal marks that cannot be altered.
[...]
This is something we’d like to avoid to ensure the user’s input remains as close to the original as possible.

You may want to rethink that if you want to use milkdown and remark.
Both are AST based, meaning the care only about the structure, exact text is NEVER guaranteed.

I don't know what an "internal mark" is, but I'd recommend making it structural rather than needing to be exact text.