[Bug] Autodetection of text/plain markdown files not working

[tkapias]

The issue

I checked Autodetection in Enlight, but the extension does not start by itself for markdown files even if the header was forced as text/plain.

Manually it works good, even Auto-Detect.

Context

I use Dufs file server and I forced nginx to rewrite content-type headers for .md files to text/plain; charset=UTF-8.

• File server: Dufs
• Nginx Proxy: nginx 1.18
• Firefox: 115.15.0esr
• Enlight: 2.7.11.0

On Firefox, the file opens inline as a text/plain with this header (same as with curl --include):

HTTP/2 200 OK
server: nginx/1.18.0
date: Wed, 02 Oct 2024 06:57:30 GMT
content-length: 11783
last-modified: Wed, 02 Oct 2024 04:31:32 GMT
etag: "1727843492695-11783"
content-disposition: inline; filename="readme.md"
accept-ranges: bytes
content-type: text/plain; charset=UTF-8
X-Firefox-Spdy: h2

I also tried without forcing charset=UTF-8.

[Qeole]

Thanks for the report! Would you have a Markdown file available somewhere, so I can reproduce, by any chance? Firefox tries to download and save the files when I host them on my own server, and I'd as well avoid playing with the nginx configuration if I could.

[Qeole]

My guess would be that forcing text/plain prevents Firefox from applying basic HTML mark-up to the text document? (To be confirmed). And that in that case, function checkBody() in the add-on would fail to find a PRE block, for example, and return false, thus blocking auto-detection.