Thank you for sending this patch. I apologize for the delay since I was busy on other projects.

I reverted this patch because it is broken and it is not easily fixed in my side.

When replacing emoji markup with HTML element string, remark sanitizes the input.

So, for example,

import { remark } from 'remark';
import toHtml from 'remark-html';
import emoji from 'remark-emoji';

const processor = remark()
    .use(emoji, {
        customEmojis: {
            parrot: '<img src="https://cultofthepartyparrot.com/parrots/parrot.gif" style="width:auto;height:1em;">',
        },
    })
    .use(toHtml);

const file = await processor.process('Hello :parrot:!');
console.log(String(file));

The output is:

<p>Hello &#x3C;img src="https://cultofthepartyparrot.com/parrots/parrot.gif" style="width:auto;height:1em;">!</p>

< is replaced with <.

Please submit a new PR after fixing this issue. Before creating it,

• Please confirm your patch works in your local
• Please add tests to verify your modifications

The misunderstanding in this patch is that replaceEmoji returns a raw HTML string. It actually returns a text string, which will be sanitized as HTML text node.

Please read this API document for the details of return type: https://github.com/syntax-tree/mdast-util-find-and-replace?tab=readme-ov-file#returns-1

If we receive HTML fragments as configuration and replace emoji markups with them,

1. Parse the fragment strings into MDAST nodes in advance
2. replaceEmoji transforms the input with replacing the emoji markups with MDAST nodes.
   For example,

   ':parrot:'

   should be transformed to

   {'hName': 'img', ...}