There are a few solutions listed here: http://wikemacs.org/wiki/Markdown#Live_preview_as_you_type.
The pure-Emacs (nearly) solution and easy one, requiring no extra library from Python or Nodejs, is impatient-mode.
Impatient-mode
It's designed to work with html but the doc gives a trick to make it work with markdown. It also works like a charm but requires one configuration step:
- Install impatient-mode with
M-x package-install RET impatient-mode RET
, given you have configured package.el to use the melpa repository.
- Start an emacs' web server with
M-x httpd-start
.
- Start impatient mode in the buffers you're interested to live preview:
M-x impatient-mode
.
- Open your browser to localhost:8080/imp. You'll see the list of buffers with the mode enabled. Click on one: you see live rendering of the buffer.
To enable markdown conversion, we follow wikemacs:
Define this elisp function somewhere, like in your init file:
<!-- language: lang-lisp -->
(defun markdown-html (buffer)
(princ (with-current-buffer buffer
(format "<!DOCTYPE html><html><title>Impatient Markdown</title><xmp theme=\"united\" style=\"display:none;\"> %s </xmp><script src=\"http://ndossougbe.github.io/strapdown/dist/strapdown.js\"></script></html>" (buffer-substring-no-properties (point-min) (point-max))))
(current-buffer)))
Tell impatient mode to use it: M-x imp-set-user-filter RET markdown-html RET
.
Go back to your browser, it works!
livedown-mode (with npm)
https://github.com/shime/emacs-livedown requires the livedown
npm package. Also, this emacs package is not in MELPA, you have to clone it locally. Otherwise, it is a good and lightweight solution.
Vmd-mode (npm, Electron)
Another solution is vmd-mode, which works with the vmd
node package. This is not the most heavy-weight solution: vmd is based on Electron (!).
Grip-mode (Python, Github's rate limit)
Another one is grip-mode, that relies on a Python package:
pip install --user grip
M-x package-install grip-mode
Then run M-x grip-mode
in the markdown buffer. It opens a new tab in your browser.
Unfortunately, at the time of writing, it is limited by Github's rate limit. Indeed, to render content as precisely as Github, it calls its API. It doesn't render content locally. As such we are limited to 60 calls an hour, which is very few. See this issue: https://github.com/joeyespo/grip/issues/35