Fix: Export PDF produces different HTML than Export HTML. #384

[lightrabbit]

This issue #384 is caused by the fact that cheerio automatically adds closing tags for unclosed opening tags. However, when processing HTML fragments with html_block, the input HTML may not always be a complete document.

This leads to situations where:

<div><img src="...">
# some title
some text
</div>

is processed into:

<div><img src="..."></div>
# some title
some text
// Here, the </div> is automatically removed by cheerio.

The issue arises because cheerio assumes that any unclosed tags should be properly closed, which can cause unexpected behavior when working with incomplete HTML fragments.

[Nat-Faeeria]

Hey,
Noticed the same problem, didn't think about checking the PR.
Since the goal is only to format the src of images, a simpler solution could be

if (type !== 'html') {
  md.renderer.rules.html_block = function (tokens, idx) {
    var html = tokens[idx].content;
    html = html.replace(/<img\s+([^>]*?)src=["']([^"']+)["']([^>]*)>/gi, function (match, before, src, after) {
        var href = convertImgPath(src, filename);
        return `<img ${before}src="${href}"${after}>`;
    });
    return html;
  };
}

(full disclosure : this code is ai generated)