emacs-jupyter connects to an in-process jupyter kernel, executes notebook code, and displays the results automatically in a LaTeX or Markdown buffer.

First install Pymacs - https://github.com/pinard/Pymacs. Build, install. In your .emacs

(load-file "[PYMACS DIR]/pymacs.el")

Then pip install

ipython
matplotlib

In your .emacs add the snippet below. The code can seem excessive, but the extra stuff was needed to be able to detect if I am in an .md or .tex buffer so we can load the right module, also set the right tempo template functions.

(defun reload-pymacs()
  (interactive)
  (if (buffer-live-p (get-buffer "*Pymacs*" ))
      (kill-buffer (get-buffer "*Pymacs*")))
  (message (buffer-file-name (current-buffer)))
  ;;
  ;; load tex or md mode based on the extension
  (if (equal (file-name-extension (buffer-file-name (current-buffer))) "tex")
      (progn 
	(pymacs-load "[INSTALL DIR]/ipython-tex")
	(global-set-key "\M-," 'ipython-tex-run-py-code)
	(global-set-key [f5] 'ipython-tex-complete-py)
	(tempo-define-template 
	 "tex-listings-python" 
	 '("\\begin{minted}[fontsize=\\footnotesize]{python}\n"
	   (s)
	   "\n\\end{minted}\n"
	   )
	 "")	
	))
  (if (equal (file-name-extension (buffer-file-name (current-buffer))) "md")
      (progn 
	(pymacs-load "[INSTALL DIR]/ipython-md")
	(global-set-key "\M-," 'ipython-md-run-py-code)
	(global-set-key [f5] 'ipython-md-complete-py)
	(tempo-define-template 
	 "tex-listings-python" 
	 '("```python\n"
	   (s)
	   "\n```\n"
	   )
	 "")	
	))

  )

(global-set-key [f1] 'tempo-template-tex-listings-python)
(global-set-key [f11] 'reload-pymacs)

For better management create a virtualenv environment, pip install the necessary packages from there, install pymacs on it, then start emacs with the bash script below,

#!/bin/bash
source /some/dir/yourenv/bin/activate
/usr/bin/emacs24

• When you are in \begin{minted} and \end{minted} blocks, or its equivalent in markdown, simply click on M-, or call 'ipython-tex-run-py-code, and all code in that block will be sent to a ipython kernel and the result will be displayed underneath. Results will be placed in \begin{verbatim}, \end{verbatim} blocks right next to the code, with one space in between. If a verbatim blocks already exists there, it will be refreshed. If not, it will be added.

• If on \inputminted{python}{file.py} block, code will be loaded from script filename between the curly braces. Results outputted as described above.

• If plt.show() is detected in code block, it will be replaced with plt.savefig(..) command and a \includegraphics LaTeX command will be added in the TeX buffer, same for Markdown. How do we determine which image file name to generate? All previous image files with png extension are searched, e.g. if test_01.png, test_02.png exist, the new file will be the last one in the list plus 1, so in this case show() will be replaced with plt.savefig('test_03.png'),

• After entering any expression, hit F-5 or call 'ipython-tex-complete-py, and emacs-ipython will show a list of possible completions in a *pytexipy* buffer. This list comes directly from ipython, hence it reflects the cumulation of runtime code that has been executed and brought into memory thusfar through multiple 'ipython-tex-run-py-code calls.

• For now, there is one kernel per Emacs process.

• At the start, I have to hit f11 or 'reload-pymacs to force initialization. However this command is useful for other purposes as well such as wanting to clean up ipyton memory and starting from scratch, anytime.

• Markdown and LaTeX versions of the code can use some refactoring.

Take a look at our other sites

IT

Science, Mathematics

Emacs Jupyter is licensed under GPL v3. See COPYING for details.