Could the markdown conversion convert unsupported attributes to plain text insead of throwing an error?

[maelchiotti]

Is your feature request related to a problem? Please describe.

When using text formatting in the editor that isn't supported in markdown (such as underlining), it breaks the conversion to markdown since the attribute cannot be handled.

Describe the solution you'd like

If you want to keep this behavior, maybe an option could be added to just treat those attributes as plain text? It would allow me to export text even if it has some unsupported markdown in it. This option would be opt-in, and I would handle warning my users.

Describe alternatives you've considered

I don't really know, I could just skip all the notes in my app that cannot be fully converted to markdown, but skipping a huge note just for one underlined word is a shame 😕

[amantoux]

@maelchiotti Do you want to propose a PR for this?
I can take a look otherwise

[maelchiotti]

I can make a PR if you think my proposal is good? Something like adding a encodeUnsupportedAttributesAsPlainText parameter to the encode() function.

[amantoux]

Yes I find it a bit harsh to fail when something is not supported
Regarding the opt-in parameter maybe something simpler like strict that defaults to false?

[amantoux]

Also, we need to decide on a way to handle embeds, maybe simply convert them to object, WDYT?

[maelchiotti]

What do you mean about "converting them to object"?

[amantoux]

I mean when the codec finds an embed, it replaces the embed by the string 'object' or '[object]'

This PR could be interesting to support embeds later on #371

[maelchiotti]

Oh yeah I see. [object] makes sense I think. Also, horizontal rules could be supported.

[maelchiotti]

Hi @amantoux, I did some work in #418.

I added support for horizontal lines.

I added the strict parameter. I just don't really know how to detect when to set the close parameter to true in _writeAttribute() in order to put a </u> after the text and not a <u> again?

However I didn't handle objects because I can't test that easily. But I guess in convert(), instead of returning when I see a block embed, that's when I should write [object] to the buffer?

void handleLine(LineNode node) {
  if (node.hasBlockEmbed) {
    if (node.embedNode.value == BlockEmbed.horizontalRule) {
      _writeHorizontalLineTag(buffer);
    } else {
      buffer.write('[object]');
    }
  }

  ...
}