Exporting unicode characters to pdf using latex from org mode

Question

Hello Good people of Emacs!

I'm having trouble exporting unicode math symbols from buffer (org-mode) to pdf file.

1. Problem Description:

Symbols are inserted to the buffer as unicode characters (via TEX input method or company-math)

Here is source code demonstration:

#+TITLE: Unicode characters export test 
#+AUThor: 
#+date:
Unicode characters:

ℝ ℤ ℕ ⇒ ∈ ∀ 

Same symbols in latex format:

$$\Bbb{R} \Bbb{Z} \Bbb{N} \Rightarrow \in \forall$$

Produced .tex file includes unicode symbols:

% Created 2016-03-04 Pá 21:01
\documentclass[11pt]{article}
\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}
\usepackage{fixltx2e}
\usepackage{graphicx}
\usepackage{longtable}
\usepackage{float}
\usepackage{wrapfig}
\usepackage{rotating}
\usepackage[normalem]{ulem}
\usepackage{amsmath}
\usepackage{textcomp}
\usepackage{marvosym}
\usepackage{wasysym}
\usepackage{amssymb}
\usepackage{hyperref}
\tolerance=1000
\date{}
\title{Unicode characters export test}
\hypersetup{
  pdfkeywords={},
  pdfsubject={},
  pdfcreator={Emacs 24.4.1 (Org mode 8.2.10)}}
\begin{document}

\maketitle
\tableofcontents

Unicode characters:

ℝ ℤ ℕ ⇒ ∈ ∀ 

Same symbols in latex format:

$$\Bbb{R} \Bbb{Z} \Bbb{N} \Rightarrow \in \forall$$
% Emacs 24.4.1 (Org mode 8.2.10)
\end{document}

Pdf file does not :

2. Things I have tried so far:

Xelatex and unicode-math: This is included in answer from Rasmus. Here I have to admit: I'm not using development version of org-mode he's mentioning. (failed to install it) I've tried xelatex adn unicode-math anyway. My version of Org mode is 8.2.10.

#+latex_compiler: xelatex
#+latex_header: \usepackage{libertine}
#+latex_header: \usepackage{unicode-math}

Including this in file introduces not so lovely message instead pdf file.

org-latex-compile: PDF file ./unicode_export_test.pdf wasn't produced: [package error]

I have checked for correct unicode-math installation:

~ $ kpsewhich unicode-math.sty
/usr/share/texlive/texmf-dist/tex/latex/unicode-math/unicode-math.sty

Installation of Xelatex is the newest version:

~ $ sudo apt-get install texlive-xetex
Reading package lists... Done
Building dependency tree       
Reading state information... Done
texlive-xetex is already the newest version.
0 upgraded, 0 newly installed, 0 to remove and 16 not upgraded.

install development version of org-mode: (failed) Following How do I keep current with bleeding edge development? - tutorial I have been able (I think) download it and compile it. Since I have no idea how to tell to Emacs run this version, Emacs runs Org-mode version 8.2.10 (release_8.2.10 @ /usr/local/share/emacs/24.4/lisp/org/). I have tried to add following lines to my config.

(add-to-list 'load-path (expand-file-name "~/elisp/org-mode/lisp"))
(remove 'load-path (expand-file-name "/usr/local/share/emacs/24.4/lisp/org/"))
(add-to-list 'auto-mode-alist '("\\.\\(org\\|org_archive\\|txt\\)$" . org-mode))
(require 'org)

Without pleasing result.

Searching internet: Solution was not found.

Beating my self for being stupid: Didn't help.

3. Question: How to export unicode characters from org-mode to pdf?

• Do I have to configure some org-mode variable?
• Compile with different Latex interpreter?
• Any other ideas?
• If development version fixes this.... How can I install it?

Answer 1

You need the latex file to use \usepackage[utf8x]{inputenc} rather than the default \usepackage[utf8]{inputenc}. This works with standard (pdf)latex.

For org export to generate this you have to set the variable org-latex-inputenc-alist accordingly. For example:

(setq org-latex-inputenc-alist '(("utf8" . "utf8x")))

For the variable to be defined, you may need to (require 'ox-latex) first.

Unfortunately, this does support only a limited set of math letters. Much more can be supported by including \usepackage[mathletters]{ucs}. This has to appear before the inputenc stuff, so you can't just use a #+latex_header:. A way to do that is to add it at head of org-latex-default-packages-alist.

(setq org-latex-default-packages-alist (cons '("mathletters" "ucs" nil) org-latex-default-packages-alist))

Note that ucs documentation (usually available through texdoc ucs) warns about the mathletters option:

This option is disabled by default, because using math greek in a normal text does not look good.

If you want those settings to apply only to the current file rather than globally, you can define them as file local variables. Add the following near the end of the file:

# Local Variables:
# org-latex-inputenc-alist: (("utf8" . "utf8x"))
# eval: (setq org-latex-default-packages-alist (cons '("mathletters" "ucs" nil) org-latex-default-packages-alist))
# End:

eval is a pseudo variable allowing to execute lisp code when the file is loaded.

Answer 2

You'd just need to load proper unicode support in LaTeX via unicode-math and fontspec. Then you'd need to compile with xelatex or lualatex.

See the docstring of the following variables: org-latex-packages-alist, org-latex-pdf-process and org-latex-compiler (the latter is introduced in Org 9).

In the development version of Org you could do something like this:

#+latex_compiler: xelatex
#+latex_header: \usepackage{libertine}
#+latex_header: \usepackage{unicode-math}

* Unicode symbols
 $z$ ∈ ℝ ∖ ℤ
\begin{equation}
\label{eq:1}
X ≔ \left\{ z\colon z ∈ ℝ ∖ ℤ \right\}
\end{equation}

Answer 3

Two ways to go:

1) Either write proper TeX expressions (instead of unicode) and let org just display them as utf-8 characters (with C-c C-x \) . http://orgmode.org/manual/Special-symbols.html

Or 2) Load a TeX package which translates your utf-8 characters into the proper math expressions. For example https://www.ctan.org/pkg/unicode-math?lang=en.