I suspect that it's because the object for the message is { type: 'codeBlock', content: '\n' }.

I'm confused, wouldn't that trigger this base case?

Here's what I found looking into an empty codeBlock.

content is basically an empty string, so whilst typeof content === 'string is true, we're also checking if (content), and that is considered falsy in javascript. So it passes by the base case. We can see this by logging content on any iteration of the empty codeblock, and it's empty (similarly, the length of the string evaluates to 0).

Then, it obviously is not a spoiler, so we skip that check, and it similarly doesn't have any items. The check else if (content) on line 276 similarly evaluates to a falsy, so the error is therefore thrown.

Here's a sample log I setup to double check this.

function replaceSpoilersFromMarkdownAST(node: SingleASTNode): SingleASTNode {
  const { content, items, type } = node;
  console.log('1. node: ', node);
  if (content) {
    console.log('We hae content.');
  } else {
    console.log('We do not have content.');
  }

  ............ (rest of the function)

the output I get is:

[METRO]  LOG  1. node:  {"content": "", "type": "codeBlock"}
[METRO]  LOG  We do not have content.

I used the following sample message below. I think (?) there may be a disparity between web and native.

In the web console, this is what I see for that code block:

1. Node:  {type: 'codeBlock', content: '\n'}
We have content

Though in native, I'm seeing the following in my terminal

[METRO]  LOG  1. Node:  {"content": "", "type": "codeBlock"}
[METRO]  LOG  We don't have content

If this is the case, where web is detecting a \n, that would explain why I can open that problematic thread on the web version with no issues, but native is only seeing an empty string and therefore it reaches the error. Perhaps that's why the app crashes.

Sounds like we should update this check instead so that it's a proper base case. Is this what we should have?

Yeah, makes sense to update the base case. I'll do some more thorough testing again before putting up the revision. This should also catch "" and "\n"