Delete cruft

[worg.git] / org-tutorials / org-latex-export.org

* Settings                                                         :noexport:
#+TITLE:   LaTeX Export for Org Mode < 8.0
#+AUTHOR:    Tom Dye
#+EMAIL:     tsd@tsdye2.com
#+DATE:      2010-04-24 Sat
#+DESCRIPTION:
#+KEYWORDS:
#+LANGUAGE:  en
#+OPTIONS:    H:3 num:t toc:t \n:nil ::t |:t ^:t -:t f:t *:t tex:t d:(HIDE) tags:not-in-toc
#+OPTIONS:   <:t d:nil todo:t pri:nil tags:not-in-toc
#+INFOJS_OPT: view:nil toc:nil ltoc:t mouse:underline buttons:0 path:https://orgmode.org/org-info.js
#+SELECT_TAGS: export
#+EXCLUDE_TAGS: noexport
#+LINK_UP:
#+LINK_HOME:

* Start up
Org-mode LaTeX export is easy.  Place the following code in your
.emacs and you're ready to go.

#+begin_src emacs-lisp
  (require 'ox-latex)
  (unless (boundp 'org-latex-classes)
    (setq org-latex-classes nil))
  (add-to-list 'org-latex-classes
               '("article"
                 "\\documentclass{article}"
                 ("\\section{%s}" . "\\section*{%s}")))
#+end_src

This tells the Org-mode LaTeX exporter to use the standard LaTeX
=article= document class by default and to map Org-mode headlines and
lists to LaTeX sectioning and list commands, as follows:

#+begin_src org
  ,* section
  ,** itemize
  ,*** itemize, etc.
  ,    - itemize
#+end_src

Now you can pick a LaTeX export command from the export menu, =C-c
C-e=, and expect to find a well-formed LaTeX document with the same
base name as your org-mode buffer.

It could hardly be easier!

* Why LaTeX Export?

As the manual says, "Org is a mode for keeping notes, maintaining TODO
lists, and doing project planning with a fast and effective plain-text
system."  LaTeX is a convenient interface to TeX, a typesetting system
that its author "intended for the creation of beautiful books — and
especially for books that contain a lot of mathematics."  Where do
these two activities--keeping notes and making beautiful books--overlap?

Org-mode is well supplied with a variety of other exporters.  The Org
ASCII exporter produces a readable and simple version of an Org file
for printing and sharing of notes.  The HTML exporter does such a good
job of preparing notes for the web that many people use it to create
entire web sites.  The XOXO exporter makes notes accessible to other
software applications and the DocBook exporter starts the journey to
many other formats, including LaTeX, using DocBook tools.

The niche for the LaTeX exporter is producing a version of an Org file
for printing and sharing notes that looks better than an ASCII
printout but that can be produced without the overhead required to
support DocBook.  All that is needed is a working LaTeX installation.
LaTeX distributions for all major platforms are today well-developed
and easy to use and maintain.

* More Highly-Structured Documents

You can prepare more highly structured LaTeX documents by adding a few
lines to =org-latex-classes= in =.emacs=.

#+begin_src emacs-lisp
  (add-to-list 'org-latex-classes
               '("article"
                 "\\documentclass{article}"
                 ("\\section{%s}" . "\\section*{%s}")
                 ("\\subsection{%s}" . "\\subsection*{%s}")
                 ("\\subsubsection{%s}" . "\\subsubsection*{%s}")
                 ("\\paragraph{%s}" . "\\paragraph*{%s}")
                 ("\\subparagraph{%s}" . "\\subparagraph*{%s}")))
#+end_src

This setup produces the following document structure:

#+begin_src org
  ,* section
  ,** subsection
  ,*** subsubsection
  ,**** paragraph
  ,***** subparagraph
  ,****** itemize, etc.
  ,       - itemize
#+end_src

* Exporting Other LaTeX Document Classes

The LaTeX exporter can be configured to produce LaTeX output for
books, reports, letters or any other document class, including custom
classes.  These can be useful if notes need to be formatted to a
certain specification, such as a company technical manual, or if the
notes are quite long, as might be the case when Org-mode is used as a
laboratory or project notebook.

Here is a standard setup for export to a LaTeX book class:

#+begin_src emacs-lisp
  (add-to-list 'org-latex-classes
               '("book"
                 "\\documentclass{book}"
                 ("\\part{%s}" . "\\part*{%s}")
                 ("\\chapter{%s}" . "\\chapter*{%s}")
                 ("\\section{%s}" . "\\section*{%s}")
                 ("\\subsection{%s}" . "\\subsection*{%s}")
                 ("\\subsubsection{%s}" . "\\subsubsection*{%s}"))
               )
#+end_src

Then, in the Org file with a book-full of notes, add this line:

#+begin_src org
  #+LaTeX_CLASS: book
#+end_src

<<koma-script>>
A useful set of alternatives is the [[http://www.komascript.de/][KOMA script]] classes.  These have a
somewhat more modern design than the standard LaTeX classes.

For example, the KOMA script article class can be configured in =.emacs=:

#+begin_src emacs-lisp
  (add-to-list 'org-latex-classes
            '("koma-article"
               "\\documentclass{scrartcl}"
               ("\\section{%s}" . "\\section*{%s}")
               ("\\subsection{%s}" . "\\subsection*{%s}")
               ("\\subsubsection{%s}" . "\\subsubsection*{%s}")
               ("\\paragraph{%s}" . "\\paragraph*{%s}")
               ("\\subparagraph{%s}" . "\\subparagraph*{%s}")))
#+end_src

Then, this class can be used by including the following line in the
Org-mode file.

#+begin_src org :exports code
  #+LaTeX_CLASS: koma-article
#+end_src

* Passing Options to Document Classes
The standard LaTeX document classes, =article=, =report=, =book=,
=slides=, and =letter= take options that, where applicable, select the
type size, paper size, orientation, whether to print one- or
two-sided, and a variety of formatting specifications.  Custom LaTeX
document classes can define their own options, as needed.

You can pass options to the LaTeX =\documentclass= macro by putting a
line like this in your Org-mode file:

#+begin_src org :exports code
  #+LaTeX_CLASS_OPTIONS: [a4paper,twoside,twocolumn]
#+end_src

A useful option with the [[koma-script][KOMA script]] classes typesets table captions
properly.  The standard LaTeX classes will incorrectly typeset table captions
exported by Org-mode above the table.  The following option to the
[[koma-script][KOMA script]] classes accomplishes this.

#+begin_src org
  #+LaTeX_CLASS_OPTIONS: [captions=tableheading]
#+end_src

* Using Custom Classes
If the user has custom LaTeX document classes that conflict with the
default packages or that only require a few of the default packages to
support all features of the LaTeX exporter, then this can be handled
in =.emacs= using [DEFAULT-PACKAGES], [NO-DEFAULT-PACKAGES],
[PACKAGES], [NO-PACKAGES], [EXTRA], [NO-EXTRA].

Here is a simple example that uses an experimental LaTeX class[fn:3]
that supports the Org-mode requirements and leaves open the
possibility of adding file specific packages:

#+begin_src emacs-lisp
  (add-to-list 'org-latex-classes
        '("org-article"
           "\\documentclass{org-article}
           [NO-DEFAULT-PACKAGES]
           [PACKAGES]
           [EXTRA]"
           ("\\section{%s}" . "\\section*{%s}")
           ("\\subsection{%s}" . "\\subsection*{%s}")
           ("\\subsubsection{%s}" . "\\subsubsection*{%s}")
           ("\\paragraph{%s}" . "\\paragraph*{%s}")
           ("\\subparagraph{%s}" . "\\subparagraph*{%s}")))
#+end_src

The Org-mode LaTeX exporter uses several packages to support special
characters used by =org-entities=.  One of these, =amssymb=, conflicts
with several [[font-section][LaTeX fonts]].  If you want finer control over which
packages are loaded, then it makes sense to define an export class
like this in =.emacs=:

#+begin_src emacs-lisp :tangle no
  (add-to-list 'org-latex-classes
            '("koma-article"
               "\\documentclass{scrartcl}
               [NO-DEFAULT-PACKAGES]
               [EXTRA]"
               ("\\section{%s}" . "\\section*{%s}")
               ("\\subsection{%s}" . "\\subsection*{%s}")
               ("\\subsubsection{%s}" . "\\subsubsection*{%s}")
               ("\\paragraph{%s}" . "\\paragraph*{%s}")
               ("\\subparagraph{%s}" . "\\subparagraph*{%s}")))
#+end_src

Alternatively, export classes can be defined on a per-file basis in a
source code block that is evaluated prior to LaTeX export.  This is
perhaps the most flexible way to set up LaTeX export.  Since the
=add-to-list= is a globally side-effecting action, this
example only calls it if it would add a LaTeX class that isn't
already on the =org-latex-classes= list:

#+begin_example
  #+name: setup
  #+begin_src emacs-lisp :results silent :exports none
 (unless (find "per-file-class" org-latex-classes :key 'car
           :test 'equal)
   (add-to-list 'org-latex-classes
            '("per-file-class"
               "\\documentclass{scrartcl}
               [NO-DEFAULT-PACKAGES]
               [EXTRA]"
               ("\\section{%s}" . "\\section*{%s}")
               ("\\subsection{%s}" . "\\subsection*{%s}")
               ("\\subsubsection{%s}" . "\\subsubsection*{%s}")
               ("\\paragraph{%s}" . "\\paragraph*{%s}")
               ("\\subparagraph{%s}" . "\\subparagraph*{%s}")))
  #+end_src
#+end_example

Then, add this line to the Org-mode file:

#+begin_src org :exports code
  #+LaTeX_CLASS: per-file-class
#+end_src

* Specifying LaTeX Packages
According to its author, the LaTeX macro package "represents a balance
between functionality and ease of use."  The LaTeX user who adds
functionality through the addition of packages necessarily makes the
software more difficult to use.  Like LaTeX itself, the Org-mode LaTeX
exporter has struck its own balance between functionality and ease of
use with the addition of several LaTeX packages.  These are written
out in the LaTeX header as LaTeX =\usepackage= commands.

Org-mode keeps the names of the LaTeX packages it uses in a data
structure designed for ease of maintenance as additional features are
added to the LaTeX exporter.  Packages in the default packages list,
=org-latex-default-packages-alist=, are required to support all
features of the LaTeX exporter.  This list is typically specified in
the Org-mode source code and its documentation contains a warning not
to modify it.  Packages not included on the default packages list that
the user needs consistently can be added to
=org-latex-packages-alist=.

If you want to specify particular packages, either in addition to
or in place of those used by Org-mode, then you can either place them
in a custom class definition, where they can be used by any Org-mode
document, or you can add them to the Org-mode document directly so
their effect is local to the Org-mode buffer.

An example custom class definition that adds the =graphicx= package
might look like this in =.emacs=:

#+name: custom-class-definition
#+begin_src emacs-lisp :exports code
  (add-to-list 'org-latex-classes
               '("per-file-class"
                 "\\documentclass{scrartcl}
                   \\usepackage{graphicx}
              [NO-DEFAULT-PACKAGES]
              [NO-PACKAGES]"
                 ("\\section{%s}" . "\\section*{%s}")
                 ("\\subsection{%s}" . "\\subsection*{%s}")
                 ("\\subsubsection{%s}" . "\\subsubsection*{%s}")
                 ("\\paragraph{%s}" . "\\paragraph*{%s}")
                 ("\\subparagraph{%s}" . "\\subparagraph*{%s}")))
#+end_src

Packages needed for a particular file can be specified by inserting a
line like this in the Org-mode buffer:
#+begin_src org :exports code
  ,#+LATEX_HEADER: \usepackage{xyz}
#+end_src

One reason for specifying LaTeX packages in the Org-mode buffer is
that highly configurable packages can be tailored for a particular
use.  Perhaps the best example among the packages in
=org-latex-default-packages-alist= is =hyperref=, which has an
elaborate list of keyval options.  PDF output destined for interactive
use might load =hyperref= with options to distinguish links with
attractive colors, for instance.  This might be achieved with
following lines, which load the =xcolor= package[fn:4] and then use
named colors to distinguish external blue links from internal red
links:

#+name: hyperref-header
#+begin_src org :exports code
  ,  #+LATEX_HEADER: \usepackage[hyperref,x11names]{xcolor}
  ,  #+LATEX_HEADER: \usepackage[colorlinks=true,urlcolor=SteelBlue4,linkcolor=Firebrick4]{hyperref}
#+end_src

In addition, you can pick up the encoding used in the Org-mode buffer and pass
this information on to LaTeX by loading the =inputenc= package as
follows:

#+name: inputenc
#+begin_src org :exports code
  #+LATEX_HEADER: \usepackage[AUTO]{inputenc}
#+end_src

* Creating PDF Output
The LaTeX exporter by default produces code ready for processing by
pdflatex.  pdflatex calls the pdfTeX program, a modern extension of
TeX that produces PDF output directly, using the standard LaTeX
macros.  pdfTeX is tightly integrated with PDF features such as
hypertext links and tables of contents, using LaTeX packages such as
hyperref, which is included in the default packages list.

Org-mode offers a command to produce a PDF file from the LaTeX export.
This is bound to =C-c C-e p=.  The command =C-c C-e d= does all this
*and* opens the PDF file in the default reader.

If you use a different TeX typesetting engine or would like to
customize how Org-mode produces the pdf file, then you will want to
modify the variable =org-latex-to-pdf-process=.  This is a list of
strings, each of which contains a call to one of the TeX typesetting
engines or to an auxiliary program, such as BibTeX, makeindex, etc.

For example, the shell script =texi2dvi= will run =pdflatex= as many
times as needed to build a pdf file successfully.  The following code
in =.emacs= will instruct Org-mode to use =texi2dvi= when making a pdf
file.  Note that you should check that =texi2dvi= is installed on your
system and that it works correctly before adding this code to your
=.emacs=.

#+name: texi2dvi
#+begin_src emacs-lisp :exports code
(setq org-latex-to-pdf-process '("texi2dvi --pdf --clean --verbose --batch %f"))
#+end_src

Note that =makeindex= and =bibtex= require a bit more effort to work
in this way because of path name issues.  Nick Dokos suggested
this fix, which specifies a path to one or more BibTeX =.bib= files:

#+begin_example
  ,#+begin_src sh :exports none
    BIBINPUTS=/path/to/bib/:$BIBINPUTS
    export BIBINPUTS
  ,#+end_src
#+end_example

If you have system-wide bibliography and index files, then it might be
useful to modify =BIBINPUTS= in =.profile= or similar.  The code
example above uses Babel to set the variable on a per-file basis.

An alternative to the default TeX engine is [[http://scripts.sil.org/cms/scripts/page.php?site_id%3Dnrsi&id%3Dxetex][XeTeX]], which merges TeX
with Unicode and modern font technologies.  The Emacs-Fu blog has an
[[http://emacs-fu.blogspot.com/2011/04/nice-looking-pdfs-with-org-mode-and.html][example XeTeX setup for Org-mode]].  There is also an entry in the
Org-Mode FAQ entitled [[https://orgmode.org/worg/org-faq.html#using-xelatex-for-pdf-export][How can I use XeLaTeX for LaTeX export instead
of pdfLaTeX?]]

* Exporting Parts of an Org-mode Buffer
Tags can be used to select which parts of an Org-mode buffer are sent
to the LaTeX exporter.  In the typical case, the
Org-mode buffer contains material for a single export file along with
material that shouldn't appear in the export; tags distinguish the
export parts from the non-export parts.  This is the single
export case.  It is also possible to use tags to specify multiple
export targets in a single Org-mode buffer.  In the multiple export
case, tags are resolved by a [[https://orgmode.org/org.html#Publishing][publishing management system]].

** The Single Export Case
The tags used for the single export case are held in
two variables: =org-export-select-tags= is a list of tags, initially set
to =export=, that select a tree or sub-tree for export;
=org-export-exclude-tags= is a list of tags, initially set to
=noexport=, that exclude a tree or subtree for export.  The effect
they have on export is logical, but the logic isn't necessarily what
one might expect.  In particular, if both select tags and exclude tags
are to be used in the same buffer, then their use must follow certain
rules.  Also, the handling of unmarked trees and subtrees changes
depending on which tags are used and how they are used.

If neither select tags nor exclude tags are used, then all of the trees
and their subtrees are sent to the LaTeX exporter.  If, however, a
select tag is added to a tree as in the example below, then unmarked
trees will *not* be sent to the exporter.  Thus, the effect of a
select tag is not restricted to its tree; its effect extends to the
entire buffer.

#+begin_src org
  ,* Tree 1                                                             :export:
  ,   This is exported
  ,** Subtree 1
  ,   This is also exported
  ,* Tree 2
  ,  This is not exported
  ,** Subtree 2
  ,  This is not exported, either
#+end_src

Once the scope of the tag's effect is grasped, the primary rule of using
select and exclude tags is obvious: only one type of tag may be used
for the trees of a buffer.  If both types of tags are used for trees,
how can Org-mode decide what to do with the unmarked trees?

A corollary of this rule is that the other type of tag can only be
used in a subtree of the tagged tree in order to reverse the effect of
the tree-level tag, as in the following example.

#+begin_src org
  ,* Tree 1                                                             :export:
  ,   This is exported
  ,** Subtree 1                                                       :noexport:
  ,   This is not exported
  ,* Tree 2
  ,  This is not exported
  ,** Subtree 2
  ,  This is not exported, either
#+end_src


** The Multiple Export Case
In the multiple export case, tags used to select a tree or subtree for
export are defined in =.emacs= as part of the configuration needed to
specify the many properties of a publication project.  A tutorial
illustrates [[https://orgmode.org/worg/org-tutorials/org-publish-html-tutorial.html][the flexibility of the publishing mechanism]] using an HTML
example.  The intricacies of the publishing mechanism are beyond the
scope of of this LaTeX export tutorial.  Here, a working example[fn:1]
is described.

In the example, the file =index.org= holds material for two export
targets, one related to work items and the other related to home.  The
variable =org-publish-project-alist= has two entries, one for a
project named =work= and the other for a project named =home=.  Both
projects are based on the file =index.org= located in =~/notes/org=.

Both projects will create output files named =index.tex=, based on the
name of the Org-mode file used for import.  The two =index.tex= files
are kept separate by writing them to different directories, as
indicated by the keyword argument =:publishing-directory=.
#+begin_src emacs-lisp
  (setq org-publish-project-alist
        '(
          ("work"
           :base-directory "~/notes/org/"
           :base-extension "org"
           :publishing-directory "~/notes/export/work/"
           :publishing-function org-publish-org-to-latex
           :select-tags     ("@WORK")
           :title "Work Notes"
           :include ("index.org")
           :exclude "\\.org$"
           )
          ("home"
           :base-directory "~/notes/org/"
           :base-extension "org"
           :publishing-directory "~/notes/export/home/"
           :publishing-function org-publish-org-to-latex
           :select-tags     ("@HOME")
           :title "Home Phone"
           :include ("index.org")
           :exclude "\\.org$"
           )
          ))
#+end_src

The parts of =index.org= tagged =@WORK= can now be exported to
=~/notes/export/work/index.tex= with =C-c C-e X= and selecting the
=work= project.

Similarly, the parts of =index.org= tagged =@HOME= can now be exported to
=~/notes/export/home/index.tex= with =C-c C-e X= and selecting the
=home= project.

* Markup
Org-mode provides wiki-like markup for various display
characteristics.  This is often handy and it translates directly into
LaTeX, but the design philosophy of LaTeX is centered around semantic
markup, "what you say is what you mean" rather than "what you see is
what you get" (lampooned by zealous LaTeX users as "what you see is
all you get").  In practice, LaTeX users define common semantic
elements in a LaTeX class or style file and these are marked up in
text with commands often peculiar to the class or style file.
Clearly, there is no way that Org-mode can anticipate commands
peculiar to arbitrary class or style files.  Fortunately, Org-mode
provides facilities to define special characters and to specify inline
and block-level markup.

** Special Characters
The variable =org-entities-user= holds the information needed to
define special characters.  In response to a question posed by Rasmus
Pank Roulund, Lawrence Mitchell described how to use this variable to
define an escaped space, which is used in LaTeX to insert a single
space, instead of a double space, after a period that doesn't end a
sentence.

With this definition:

: (setq org-entities-user '(("space" "\\ " nil " " " " " " " ")))

then

: this is some text, e.g.\space foo bar

is exported as

: this is some text, e.g.\  foo bar

** Block-level Markup
For simple /block-level/ arbitrary markup, you can use the contributed
package [[file:../org-contrib/org-special-blocks.org][org-special-blocks]], which turns ~#+begin_foo~ into
~\begin{foo}~ upon LaTeX export. You can use ordinary Org markup
inside the block. For example, you can wrap an abstract in

: #+BEGIN_ABSTRACT
: This is an *abstract* with ordinary =Org-mode= /markup/.
: #+END_ABSTRACT

For more complex cases, where you need to pass parameters or process
block contents, but don't want to use literal LaTeX, you may want to
explore the possibilities of the contributed [[file:../org-contrib/org-exp-blocks.org][org-exp-blocks]] package.

An example of block-level markup for a results block that will be
typeset with a shaded background:

: #+LaTeX_HEADER: \usepackage{framed}
: #+LaTeX_HEADER: \usepackage{xcolor}
: #+LaTeX_HEADER: \definecolor{shadecolor}{gray}{.95}
: #+LaTeX_HEADER: \newenvironment{results}{\begin{shaded}}{\end{shaded}}


** Inline Markup
/Inline/ arbitrary semantic markup can be implemented by defining a new link type
in =.emacs=.[fn:2]   The following code block defines a new link type,
=latex=, whose =path= argument can hold the name of any LaTeX
command.  This one defines export markup for HTML and LaTeX.  A link
such as =[latex:proglang][Org-mode]= will export
=\proglang{Org-mode}= to the LaTeX file.  In this way, it is possible
to make the Org-mode LaTeX exporter conform to the semantic markup
defined in arbitrary style files.  Org-mode will even complete your
new link type!

#+name: semantic-markup
#+begin_src emacs-lisp :exports code
  (org-add-link-type
   "latex" nil
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "<span class=\"%s\">%s</span>" path desc))
      ((eq format 'latex)
       (format "\\%s{%s}" path desc)))))
#+end_src

Two examples of LaTeX commands for inline semantic markup:

: #+LaTeX_HEADER: \let\progstruct=\texttt
: #+LaTeX_HEADER: \newcommand{\progexample}[1]{{\ttfamily\small #1}}

** Captions
The LaTeX caption command is typically passed two arguments: the
required argument, which is typeset with the figure or table; and the
optional argument, which appears in the List of Figures or List of
Tables.  It is common nowadays, especially in the sciences, to have
long captions (actually captions plus legends) with figures and much
abbreviated versions, typically less than a line long, in the List of
Figures.  In addition, many styles require that figure captions end
with a period; the caption passed to the List of Figures should not
end in a period.

The Org-mode =#+CAPTION:= macro handles an optional argument.

This construct:

: #+CAPTION[Short caption]: Long caption.

exports:

: \caption[Short caption]{Long caption.}

* Styling the Frontmatter
The Org-mode LaTeX exporter requires configuration to gain full access
to the LaTeX frontmatter formatting capacity.

** Abstract, contents, and lists of figures and tables
In the default configuration, the Org-mode LaTeX exporter includes a
function that sandwiches the LaTeX =\tableofcontents= command between
a command that sets the depth of the headings that appear in the table
of contents (based on the number of headline levels that will be
exported as headings, rather than lists) and a command to add some
vertical space.  Neither of these additions to the =\tableofcontents=
command is especially desireable.  It is often the case that one wants
the table of contents depth to differ from the depth to which sections
are numbered.  Also, in the LaTeX world, the space between the end of one
element and the start of another is something that is specified within
a class or style file, rather than within the document itself.  Formatting with
the class or style file exclusively can give the finished document a pleasing
stylistic uniformity that is difficult to achieve in an ad hoc way.
Also, hardwiring the table of contents in this way always puts it
directly following the output of the LaTeX =maketitle= command.  In
practice, however, it is often useful to print an abstract or
executive summary between the title and the table of contents.
Fortunately, the LaTeX exporter is coded in such a way that it is
possible for the user to alter this behavior relatively easily.

#+name: format-no-toc
#+begin_src emacs-lisp :exports code
(defun org-export-latex-no-toc (depth)
    (when depth
      (format "%% Org-mode is exporting headings to %s levels.\n"
              depth)))
  (setq org-export-latex-format-toc-function 'org-export-latex-no-toc)
#+end_src

With this setup, place the abstract and =#+LATEX:= commands for
frontmatter before the first exported headline, e.g.,

: #+BEGIN_abstract
:   [Abstract here]
: #+END_abstract
: #+LATEX: \tableofcontents
: #+LATEX: \listoftables
: #+LATEX: \listoffigures
: * First Exported Headline


** Titles and Title Page Layout
The default title created by the LaTeX exporter is often just fine,
but in cases where you would like to include specific information in
the title, or create a custom title page, then perhaps the best way to
do this was posted to the Org-mode list by Nick Dokos:

#+begin_example
#+LATEX_HEADER: \input{mytitle}

 * Foo
 foo

 * Bar
 bar
#+end_example

where the file mytitle.tex looks like this:

#+begin_example
\renewcommand\maketitle{\begin{titlepage}%
FOO
\end{titlepage}%
}
#+end_example

* Exporting Listings
** Exporting Pseudo-Code
  The LaTeX exporter will fontify exported code blocks written in any
  language that has an associated Emacs mode for editing.  If you want
  to export pseudo-code, for which there is no corresponding Emacs
  mode, then one approach is to use =#+begin_latex ... #+end_latex=
  and write the pseudo-code directly in LaTeX.  This depends on the
  LaTeX [[http://www.ctan.org/tex-archive/macros/latex/contrib/listings/][listings package]], which is one of the default packages used by
  Org-mode.

  Dan Davison provided this example on the Org-mode list:

: #+begin_latex
: \begin{algorithm}
:  \caption{An algorithm}
:  \label{alg:myalg}
:  \begin{lstlisting}[mathescape,escapeinside='']
:    '\foreach individual $i$'
:        '\foreach group $k$'
:            $\gamma_{ik} \getsp Q_{k}\prod_{l}\prod_{a=1}^{2}P_{lkX_{ila}}$
:  \end{lstlisting}
: \end{algorithm}
: #+end_latex
** Typesetting and Fontifying Source Code
Org-mode supports two LaTeX packages for typesetting and fontifying
source code; listings and minted.  The listings package is a pure
LaTeX implementation that works reasonably well, but the package
appears to be orphaned and the latest documentation was written
in 2007.  In contrast, the minted package is not a pure LaTeX solution
and relies on an external Python script, pygmentize, to typeset and
fontify source code.  This package appears to be actively maintained.
Both packages are included in the TeXLive and MacTeX LaTeX
distributions.  If your distribution lacks one or the other, then
you'll need to check the documentation for instructions how to install
them.

The two packages are structured somewhat differently, but it is
possible to configure Org-mode so that these differences are mostly
smoothed over, making it possible to generate LaTeX code that will
work with either package.  The two main differences have to do with
the minted package's generation of language names by appending =code=
to the language name; this convention is not followed by the listings
package.  In addition, the minted package defines /styles/, by which
it means colorizing semantic elements of the programming language.
This facility is not implemented in the listings package.

The following sections show the basics of listings and minted setups
which, when exported to LaTeX, illustrate the ability of each package
to typeset and fontify source code on a per-language basis.

** Example minted setup

In this example, the minted package is specified in the
=#+LaTeX_HEADER:=.  This is followed by a command to use the minted
style =emacs=, which colors source code in a way familiar to emacs
users.  The final =#+LaTeX_HEADER:= line uses the =newminted= macro to
set the font size for code blocks of =common-lisp=.

The source code block =setup-minted= includes emacs-lisp code that
might typically appear in .emacs, but can be useful in a source code
block where it can be used to configure LaTeX export for an individual
document.  In the source code block, the variables that control LaTeX
export using the minted package are configured.  First,
=org-latex-listings= is set to use the minted package.  Then,
the variable =org-latex-custom-lang-environments= is used to
associate the emacs-lisp code of an Org-mode source code block with
the common-lisp language that the minted package knows how to parse.
Note that this is given as =common-lispcode= here, but as
=common-lisp= in the call to the =newminted= macro.  The minted
package appends =code= to language names by default.  The variable
=org-latex-minted-options= sets typesetting options that will
apply to all programming languages.  See the minted package
documentation for the very many options that can be set with key/val
pairs. The three entries in the association list call for source code
blocks to be framed with lines, set in a very small font, and
identified with line numbers.  In relation to these settings, the
emacs-lisp code will be set at a slightly larger font size.  Finally,
the variable =org-latex-to-pdf-process= is set using the
=-shell-escape= option so the external pygmentize program can be
called and its results incorporated into the pdf document.  Note that
using the =-shell-escape= option creates security holes.

A Python source code block at the end illustrates how the default
emacs-style code differs from the specially formatted emacs-lisp
source code.

#+begin_example
, #+LATEX_CLASS: article
, #+LaTeX_HEADER: \usepackage{minted}
, #+LaTeX_HEADER: \usemintedstyle{emacs}
, #+LaTeX_HEADER: \newminted{common-lisp}{fontsize=\footnotesize}

, #+name: setup-minted
, #+begin_src emacs-lisp :exports both :results silent
     (setq org-latex-listings 'minted)
     (setq org-latex-custom-lang-environments
           '(
            (emacs-lisp "common-lispcode")
             ))
     (setq org-latex-minted-options
           '(("frame" "lines")
             ("fontsize" "\\scriptsize")
             ("linenos" "")))
     (setq org-latex-to-pdf-process
           '("pdflatex -shell-escape -interaction nonstopmode -output-directory %o %f"
             "pdflatex -shell-escape -interaction nonstopmode -output-directory %o %f"
             "pdflatex -shell-escape -interaction nonstopmode -output-directory %o %f"))
, #+end_src

, #+name: another-listing
, #+begin_src python :exports code
   x = str(y)
, #+end_src
#+end_example

When this example is exported (=C-c C-e d=) the resulting pdf file
contains two source code blocks nicely typeset and colored in a way
common to many emacs setups.

** Example listings setup

In this example, the listings package is specified in the
=#+LaTeX_HEADER:=.  It is followed by three =#+LaTeX_HEADER:= lines
that construct a call to the =lstnewenvironment= command that
associates the language identifier =common-lispcode= with the =Lisp=
language that the listings package knows how to parse, and configures
typesetting options that will apply to code identified as
=common-lispcode=.  Note that =common-lispcode= is used here simply to
conform with the setup used by the minted package for typesetting
emacs-lisp code.  The final =#+LaTeX_HEADER:= line defines a new LaTeX
command that configures typesetting Python code inline.

The source code block =setup-listings= instructs Org-mode to use the
listings package.  It sets the variable
=org-latex-custom-lang-environments= to associate =emacs-lisp=
source code in Org-mode with =common-lispcode=, as defined by
=lstnewenvironment=.  Then it configures default options in the
variable =org-latex-listings-options= that place a frame around
source code blocks, set code in a fairly small font, and number lines
at the left with tiny numbers.  The variable
=org-latex-to-pdf-process= doesn't need to use the =-shell-escape=
option in the call to pdflatex because the listings package is pure
LaTeX and doesn't rely on the output of external programs.  A new link
type, =latex=, is defined.  This can be used to markup inline code
snippets.  This is followed by a short Python source code block that
illustrates the difference between the default typesetting options and
the options specified for =emacs-lisp= source code blocks.  Finally,
the =latex= link type is used to call the new =python= command to
typeset an inline code snippet.

#+begin_example
, #+LATEX_CLASS: article
, #+LaTeX_HEADER: \usepackage{listings}
, #+LaTeX_HEADER: \lstnewenvironment{common-lispcode}
, #+LaTeX_HEADER: {\lstset{language={Lisp},basicstyle={\ttfamily\footnotesize},frame=single,breaklines=true}}
, #+LaTeX_HEADER: {}
, #+LaTeX_HEADER: \newcommand{\python}[1]{\lstset{language={Python},basicstyle={\ttfamily\small}}\lstinline{#1}}

, #+name: setup-listings
, #+begin_src emacs-lisp :exports both :results silent
  (setq org-latex-listings 'listings)
  (setq org-latex-custom-lang-environments
        '((emacs-lisp "common-lispcode")))
  (setq org-latex-listings-options
        '(("frame" "lines")
          ("basicstyle" "\\footnotesize")
          ("numbers" "left")
          ("numberstyle" "\\tiny")))
  (setq org-latex-to-pdf-process
        '("pdflatex -interaction nonstopmode -output-directory %o %f"
        "pdflatex -interaction nonstopmode -output-directory %o %f"
        "pdflatex -interaction nonstopmode -output-directory %o %f"))
  (org-add-link-type
   "latex" nil
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "<span class=\"%s\">%s</span>" path desc))
      ((eq format 'latex)
       (format "\\%s{%s}" path desc)))))
, #+end_src

, #+name: another-listing
, #+begin_src python :exports code
  x = str(y)
, #+end_src

, This is an inline snippet of Python: [[latex:python][x = str(y)]].
#+end_example

When this example is exported, both source code blocks are typeset
without any semantic markup, which would be specified by element,
rather than with a style as with the minted package.
* Exporting Tables
There are several ways to prepare a table in Org-mode for export to
LaTeX.  Perhaps the easiest is to supply the Org-mode table with a
caption and a label:

: #+CAPTION: [Short caption]{Long caption}
: #+LABEL: tab:my-table

This has the great advantage of exporting cleanly to HTML, ASCII, and
other backends, but the amount of control over the results is somewhat
limited.

At the other extreme, you can create an arbitrarily complex LaTeX
table and wrap it in =#+BEGIN_LATEX:= ... =#+END_LATEX:=.  This will
give you complete control over the LaTeX export, but leave you high
and dry for export to other backends and deprive you of the pleasure
of building a table with the facilities provided by Org-mode.

A middle ground passes one or more Org-mode tables to a Babel source
block, which uses =orgtbl-to-generic= to set the table.  Currently,
the entries in the Library of Babel convert only to LaTeX.  This
approach lets you compose tables in Org-mode but currently requires
entries in the table to be marked up with LaTeX, rather than Org-mode
conventions.

** How LaTeX typesets tables
LaTeX divides table handling into two parts: typesetting the actual
table, and placing the table on the page along with a caption and
cross-reference information.

LaTeX provides several environments for typesetting tables.  The LaTeX
exporter currently supports a well-rounded subset of these:

  - tabular :: table doesn't break across pages, user responsible for
               determining column widths
  - tabularx :: table doesn't break across pages, width of table
                specified by user, automatic calculation of column widths
  - tabulary :: like tabularx, but tries harder to calculate optimal
                column widths
  - longtable :: table breaks across pages, can't be used in a
                 multicolumn page layout

Because the various tabular envionments don't break across pages, they
are typically "floated" by wrapping them in a =table= environment, or
for a table that spans columns in a multi-column page layout, a
=table*= environment.  LaTeX will "float" the typeset table to an
appropriate place in the output, ensuring that the table doesn't run
off the end of the page (unless it is taller than the text height).
Note that the LaTeX environments responsible for handling the
information specified by =#+CAPTION:= and =#+LABEL:= are =table=,
=table*=, and =longtable=.  If =tabular=, =tabularx=, and =tabulary=
are used by themselves, then they won't support captions and
cross-references.

** Table Rules
The tables in many high-quality publications use rules of different
widths for the different "lines" in a table.  The horizontal rules at
the top and bottom of a table are heavier than the rule separating the
column heads from the table body.  In general, good table design
discourages the use of vertical rules; they are superfluous in an
otherwise well-designed table.

The LaTeX exporter currently uses rules of the same width everywhere
in a table.  It is possible to get book quality tables.

The following steps assume that the =org-babel-load-languages=
variable has an entry =(latex . t)= and that functions in the Library
of Babel are available.  I have this in =.emacs= so that Org-mode
knows where to find the Library of Babel (but don't know if it
is strictly necessary):
:  (org-babel-lob-ingest "/path/to/library-of-babel.org")


First, load the =booktabs= package:

: #+LATEX_HEADER: \usepackage{booktabs}

Then, give your table a name, e.g., =#+tblname: test-table= and put it
somewhere the LaTeX exporter won't see, perhaps in a sub-tree tagged with
=:noexport:=.

Finally, create a LaTeX source block something like this:

: #+name: tabularx-export
: #+begin_src latex :exports results :results latex :noweb yes
:   \begin{table}[htb!]
:   \centering
:   \footnotesize
:   \caption{A table to test booktabs}
:   \label{tab:test-table}
:   <<booktabs(table=test-table,align="lrX",env="tabularx",width="\\textwidth")>>
:   \end{table}
:  #+end_src

When you export the file to LaTeX, this code block is expanded in a
=#+results:= block and wrapped in =#+BEGIN_LaTeX= ... =#+END_LaTeX=.
If the table has column heads, then the rule beneath them should be
finer than the rules at the top and bottom of the table.

** Notes in tables

In a better world, the LaTeX =footnote= command would work inside a
=tabular= environment, setting the notes at the end of the table. The
current situation is described in the
[[http://www.tex.ac.uk/cgi-bin/texfaq2html?label%3Dfootintab][TeX
FAQ]].  A manual solution is possible, and can be implemented in
Org-mode as follows.

First, define a counter that uses table note marks (asterisk, dagger,
pilcrow, etc.) and a command to use it.  Here is an example:

: #+LaTeX_HEADER: \newcounter{my@fn}
: #+LaTeX_HEADER: \newcommand{\fn}[1]{\setcounter{my@fn}{#1}\textsuperscript{\fnsymbol{my@fn}}}

Then use a LaTeX source block that calls =booktabs-notes= from the
Library of Babel.

: #+begin_src latex :exports results :results latex
:   \begin{table}[htb!]
:     \centering
:     \footnotesize
:     \caption{The table caption without Org-mode markup}
:     \label{tab:dates}
:     <<booktabs-notes(table=dates,notes=dates-fn,lspace=t,align="rcr",env="tabular")>>
:   \end{table}
: #+end_src

This works like =booktabs= (see [[Table%20Rules][Table Rules]]),
except it accepts a second Org-mode table that contains the table
notes.  Remember that both of these Org-mode tables are sent to
=orgtbl-to-generic= so Org-mode markup within them won't give the
results you might expect.  Instead, the table entries use LaTeX
markup.

: #+tblname: dates
: | One\fn{1} | Two\fn{2} | Three\fn{3} |
: |-----------+------------+-------------|
: | A         | B          | C           |
: | D         | E          | F           |

: #+tblname: dates-fn
: | \multicolumn{3}{l}{\fn{1} A table note.}               |
: | \multicolumn{3}{l}{\fn{2} Another table note.}         |
: | \multicolumn{3}{l}{\fn{3} The \emph{last} table note.} |


There is also a switch, =lspace=, which if set non-nil calls the
=addlinespace= command from the =booktabs= package after the
=bottomrule= that separates the table body from the notes.

** Changing font size
Tables often benefit from a smaller font size than the body text of a
document.  However, Org-mode doesn't offer a way to change the font
size in a table.  A discussion on the Org-mode list yielded two solutions,
one of which makes a setting for all the tables in the Org-mode
document and another that is used on individual tables.

The first solution, offered by Nick Dokos, uses a LaTeX style file,
illustrated here with =scripttab.sty=.  This style file sets all the
tables in the document with the same size font.  The command
=\scriptsize= yields a very small font; the particular size is
determined in LaTeX by the font size of =\normalsize=.  Note that
=\scriptsize= can be replaced by another legal LaTeX font size, such
as =\footnotesize= or =\small=.

#+name: scripttab
#+begin_src latex :exports code
\makeatletter
\def \@floatboxreset {%
        \reset@font
        \scriptsize %\footnotesize %\small
        \@setminipage
}
\makeatother
#+end_src

Put =scriptab.sty= where LaTeX can find it and then add this line to
the Org-mode file:

: #+LaTeX_HEADER: \usepackage{scripttab}

A second approach, which sets the font size for an individual table
with the =#+ATTR_LaTeX= line, was proposed by Suvayu Ali. This
solution is documented on
[[https://orgmode.org/worg/org-hacks.html#latex-command-for-floats][Org
ad hoc code, quick hacks and workarounds]]. Any LaTeX command that
works inside the table environment can be passed in this way.

* Exporting Lists
LaTeX typically sets lists as displayed material, outside of normal
paragraphs.  The three LaTeX list environments --- =itemize=,
=enumerate=, and =description= --- are all supported by Org-mode.
Additionally, Org-mode includes code to typeset its checkboxes.

** Vertical White Space
In standard LaTeX, lists are set with a lot of vertical white space
and it is sometimes nicer to set them more compactly.  This can be
accomplished with the =paralist= package.

: #+LaTeX_HEADER: \usepackage{paralist}

You can redefine the standard LaTeX list environments with their
compact =paralist= equivalents:

: #+LaTeX_HEADER: \let\itemize\compactitem
: #+LaTeX_HEADER: \let\description\compactdesc
: #+LaTeX_HEADER: \let\enumerate\compactenum

With these settings, all the lists in the Org-mode document will be
exported in compact form.

** Set Enumerated Lists in a Paragraph

The =paralist= package has a facility for setting lists in paragraphs,
rather than displayed separately.  Typically the lists in a paragraph
are enumerated and it is possible to set Org-mode's enumerated lists
in paragraph, while displaying itemized and description lists.

: #+LaTeX_HEADER: \let\itemize\compactitem
: #+LaTeX_HEADER: \let\description\compactdesc
: #+LaTeX_HEADER: \let\enumerate\inparaenum


** Specify the Enumerator
With the =paralist= package you can also specify how to enumerate the
list.  These two examples use lowercase Roman numerals in parentheses
and lowercase letters in parentheses, either of which might be used
when an enumerated list is set in a paragraph.

: #+LaTeX_HEADER: \renewenvironment{enumerate}{\begin{inparaenum}[(i)]}{\end{inparaenum}}
: #+LaTeX_HEADER: \renewenvironment{enumerate}{\begin{inparaenum}[(a)]}{\end{inparaenum}}

** Checkboxes
Checkboxes aren't supported natively in LaTeX, so Org-mode rolls its
own.  The variable =org-export-latex-list-parameters= controls how
this is done.  Skip Collins and Nick Dokos discussed how to modify
these parameters.  Their discussion resulted in a proposal for better
looking checkboxes.

: #LATEX_HEADER: \setbox0=\hbox{\large$\square$}

: #+BIND: org-export-latex-list-parameters (:cbon "[{\\parbox[][][c]{\\wd0}{\\large$\\boxtimes$}}]" :cboff "[{\\parbox[][][c]{\\wd0}{\\large$\\square$}}]")

* Fonts
<<font-section>>
The default LaTeX font is Computer Modern, which was designed by
Donald Knuth.  Computer Modern is perfectly serviceable, but not as
nice as other fonts available for LaTeX.  Alternative font sets are
suggested below.

There are many choices and your choice of which ones to use will be
guided, in part, by whether or not your document uses math.  If so,
then you'll want to choose a font with math support.

Another consideration will be whether or not the font clashes with the
=amssymb= package that Org-mode loads by default.

LaTeX documents might need three text fonts, one for the serif
typeface used for text, the sans-serif typeface often used for heads
and sub-heads, and the monospace typewriter typeface typically used to
set code examples and the like.  Choosing fonts that look good with
one another is part of the typesetter's art; some care has been taken
to match fonts in each set.

** Bera
<<bera-font>>

Bera is the LaTeX version of the Bitstream's Vera family of fonts.
The family includes serif, sans-serif, and monospace fonts designed to
work well with one another.

: #+LaTeX_HEADER: \usepackage[T1]{fontenc}
: #+LaTeX_HEADER: \usepackage[scaled]{beraserif}
: #+LaTeX_HEADER: \usepackage[scaled]{berasans}
: #+LaTeX_HEADER: \usepackage[scaled]{beramono}

** Charter
<<charter-font>>

[[https://en.wikipedia.org/wiki/Bitstream_Charter][Charter]] was designed to reproduce well on low-resolution 300 dpi
printers.  It is paired here with Helvetica and Courier, like [[times-font][Times]],
for which it is an alternative.  Helvetica is set a bit smaller to
match the shape of the Charter font.  These fonts conflict with the
=amssymb== package.

: #+LaTeX_HEADER: \usepackage[T1]{fontenc}
: #+LaTeX_HEADER: \usepackage[bitstream-charter]{mathdesign}
: #+LaTeX_HEADER: \usepackage[scaled=.9]{helvet}
: #+LaTeX_HEADER: \usepackage{courier} % tt

** Garamond
<<garamond-font>>

[[https://en.wikipedia.org/wiki/Garamond][Garamond]] refers to a group of old-style serif typefaces and is named
after the sixteenth-century type designer, Claude Garamond.  It is an
elegant typeface.  Garamond requires a bit more leading than normal.
The sans-serif font is Latin Modern and the typewriter font is
Courier.  Both were chosen to match the shape and stroke weight of
Garamond.

: #+LaTeX_HEADER: \usepackage[T1]{fontenc}
: #+LaTeX_HEADER: \usepackage[urw-garamond]{mathdesign}
: #+LaTeX_HEADER: \usepackage{lmodern}
: #+LaTeX_HEADER: \usepackage{courier}
: #+LaTeX_HEADER: \linespread{1.0609}

** KP family
<<kp-font>>

The [[http://tug.ctan.org/pkg/kpfonts][KP font family]] is produced by Christophe Caignaert for the
Johannes Kepler project.  The family supports math.

: #+LaTeX_HEADER: \usepackage[T1]{fontenc}
: #+LaTeX_HEADER: \usepackage{kpfonts}

** Libertine
<<libertine-font>>

The [[http://www.linuxlibertine.org/][Linux Libertine Project]] produces OpenSource fonts.  Libertine is a
replacement for Times New Roman and includes a companion sans-serif
font.  It was used to typeset the Wikipedia logo.  The monospace
typewriter font is Latin Modern.

: #+LaTeX_HEADER: \usepackage[T1]{fontenc}
: #+LaTeX_HEADER: \usepackage{libertine}
: #+LaTeX_HEADER: \renewcommand*\oldstylenums[1]{{\fontfamily{fxlj}\selectfont #1}}
: #+LaTeX_HEADER: \usepackage{lmodern}
** Nimbus
<<nimbus-font>>

The Nimbus font set uses fonts from the [[http://www.gust.org.pl/projects/e-foundry/tex-gyre/][Tex-Gyre]] distribution, which
provides a rich collection of diacritical characters in the attempt to
cover as many Latin-based scripts as possible.  The serif font is
Termes, which is a replacement for Times Roman.  The sans-serif font
is Heros, which is a replacement for Helvetica.  The monospace font is
Cursor, which is a Courier replacement.

: #+LaTeX_HEADER: \usepackage[T1]{fontenc}
: #+LaTeX_HEADER: \usepackage{tgtermes}
: #+LaTeX_HEADER: \usepackage[scale=.85]{tgheros}
: #+LaTeX_HEADER: \usepackage{tgcursor}

** Palatino
<<palatino-font>>

The beautiful, old-style serif font, [[https://en.wikipedia.org/wiki/Palatino][Palatino]], was designed by [[https://en.wikipedia.org/wiki/Herman_Zapf][Herman
Zapf]].  It is somewhat heavier and easier to read than [[garamond-font][Garamond]].
Palatino gets a bit more leading than normal.  It is paired here with
Helvetica and Courier, as is [[times-font][Times]], for which it is an alternative.

: #+LaTeX_HEADER: \usepackage[T1]{fontenc}
: #+LaTeX_HEADER: \usepackage{mathpazo}
: #+LaTeX_HEADER: \linespread{1.05}
: #+LaTeX_HEADER: \usepackage[scaled]{helvet}
: #+LaTeX_HEADER: \usepackage{courier}

** Times
<<times-font>>

The =times= option uses URW Nimbus Roman, a Times clone, for the serif
font, URW Nimbus Sans, a Helvetica clone, for the sans-serif font,
and URW Nimbus Mono, a Courier clone, for the typewriter font.  This
is a standard set of common typefaces typically used in scientific
publications.  All of the fonts should be included in a typical LaTeX
distribution.

[[https://en.wikipedia.org/wiki/Times_Roman][Times New Roman]] was designed by [[https://en.wikipedia.org/wiki/Stanley_Morison][Stanley Morison]] for /The Times/ of
London during a redesign of the newspaper prompted, in part, by
Morison's criticism of its typography in 1929.  [[https://en.wikipedia.org/wiki/Helvetica][Helvetica]] was
developed in 1957 by [[https://en.wikipedia.org/wiki/Max_Miedinger][Max Miedinger]].  Helvetica looks better with Times
if it is set slightly smaller than the serif font.  [[https://en.wikipedia.org/wiki/Courier_(typeface)][Courier]] was
designed by Howard Kettler in 1955 for use in IBM typewriters.

: #+LaTeX_HEADER: \usepackage[T1]{fontenc}
: #+LaTeX_HEADER: \usepackage{mathptmx}
: #+LaTeX_HEADER: \usepackage[scaled=.90]{helvet}
: #+LaTeX_HEADER: \usepackage{courier}

** Utopia
<<utopia-font>>

[[https://en.wikipedia.org/wiki/Utopia_(typeface)][Utopia]] is a transitional serif font designed by [[https://en.wikipedia.org/wiki/Robert_Slimbach][Robert Slimbach]] for
Adobe in 1989.  It became free software in 2006.  It is paired here
with the Bera sans serif and monospaced fonts.  Note that the Utopia
font clashes with the =amssymb= package.

: #+LaTeX_HEADER: \usepackage[T1]{fontenc}
: #+LaTeX_HEADER: \usepackage[adobe-utopia]{mathdesign}
: #+LaTeX_HEADER: \usepackage[scaled]{berasans}
: #+LaTeX_HEADER: \usepackage[scaled]{beramono}

* Cross-references
The Org-mode LaTeX exporter assigns its own labels to section
headings.  In response to a question from Rasmus Pank Roulund on how
to resolve a label when cross-referencing a section heading, Lawrence
Mitchell pointed out that with the following setting:

: (setq org-export-latex-hyperref-format "\\ref{%s}")

a link to a headline will cross reference properly during LaTeX
export.

For example,

: * My Headline
: In section [[My Headline]] we discuss ...

exports to LaTeX correctly.  Note that the link must match the
headline exactly for this to work.

* Bibliography
<<bibliography section>>

LaTeX has its own bibliography management system based on the BibTeX
software.  The BibTeX software uses an external database of
bibliographic entries identified by keys.  The user passes the
bibliographic key to any one of several citation commands at an
appropriate place in the text and BibTeX replaces it in the output
with a formatted in-text citation.  It also compiles and typesets a
list of cited works, ensuring that every work cited in the text
appears in the list, and that every work that appears in the list is
cited in the text.

In its standard configuration Org-mode uses the BibTeX software.
Citation commands can be inserted directly into the Org-mode file,
e.g.,

: \cite{dominick_10}

and the list of cited works inserted at an appropriate place, e.g.,

: \bibliographystyle{plain}
: \bibliography{my-bib-db}

where =my-bib-db= is the name of an external bibliographic database,
and =plain= is a BibTeX style.

In-text citations can be entered by hand, but in practice it quickly
becomes cumbersome to remember bibliographic keys and to type them
correctly.  Authors writing in LaTeX typically use the RefTeX
software, originally written by Carsten Dominick, to select the
bibliographic key from the external database and to insert it into the
text with a citation command.  RefTeX can also be used to insert
citations into the Org-mode document, as detailed in [[https://orgmode.org/worg/org-faq.html#using-reftex-in-org-mode][this FAQ.]]  A more
detailed setup that associates reading notes with BibTeX
bibliographic entries is described in this [[http://article.gmane.org/gmane.emacs.orgmode/2406/match%3Dbibliography][post to the Org-mode list.]]

The =natbib= package extends BibTeX with several convenient in-text
citation commands.  The commands all start with =cite= and end with
one or more letters that indicate a style.  The most commonly used
commands are:
  1. =\citep= for parenthetical citations, e.g., (Jones, 1958);
  2. =\citet= for textual citations, e.g., Jones (1958);
  3. =\citealt= for textual citations without parentheses, e.g. Jones 1958.
There are starred versions of most commands that will output the full
author lists rather than use the abbreviation =et. al.=

To use =natbib= place this line in the Org-mode file:

: LATEX_HEADER: \usepackage{natbib}

** Biblatex

The =biblatex= package, which recently emerged from beta status, offers
support for the full range of citation styles, including the numeric
styles favored by scientists, the author-date styles used by social
scientists, and the footnote styles popular in the humanities.

You'll need two additions to the LaTeX preamble, one to load biblatex
and the other to specify the =*.bib= files that hold the BibTeX
database.  Loading biblatex without options defaults to a numeric
citation style typical in the sciences.

: #+LaTeX_HEADER: \usepackage{biblatex}
: #+LaTeX_HEADER: \bibliography{my-bib,my-other-bib}

There are a number of citation and bibliography styles that ship with
biblatex and there is a growing body of contributed styles.  These can
be selected by passing options to biblatex.  The author/date style
used in the social sciences can be selected:

: #+LaTeX_HEADER: \usepackage[style=authordate]{biblatex}

Or a footnote style used in the humanities:

: #+LaTeX_HEADER: \usepackage[style=verbose]{biblatex}


Biblatex supports the various citation commands implemented by
=natbib=.

: #+LaTeX_HEADER: \usepackage[natbib]{biblatex}

The list of cited works can be placed in the document by adding this
line to the appropriate place in the Org-mode file:

: \printbibliography

** Using Ebib and Extended Link Syntax

Thanks to work by Ali Tofigh and Joost Kremers, it is possible to
manage citations using the [[http://ebib.sourceforge.net/][ebib]] program for managing BibTeX databases.
Using ebib with Org-mode was the idea of Ali Tofigh, who sent this
message to the Org-mode list:

A while ago I asked on this list about connecting org-mode with ebib,
which is a bibtex database manager for emacs. Thanks to Joost Kremers,
there is now a solution.

I asked the developer of ebib, Joost Kremers, if he could write a
function that would start ebib on a given bibtex entry. He kindly
added this functionality to the 'ebib' function (which starts ebib in
emacs) and it is now available in the ebib git repository (see
http://ebib.sourceforge.net). If you are using ebib and would like to
get org-mode to open bibtex entries do the following:

1) Install the latest development version of ebib.

2) make sure =ebib-preload-bib-files= is set properly so that your =.bib= file is loaded by ebib when ebib starts

3) add the following line to your =.emacs=:
: (org-add-link-type "ebib" 'ebib)

Now you can insert ebib links in your documents like this:
=[ebib:Jones1998][some paper title]=. Opening this link should now
result in ebib starting, loading your default bibtex database, and
highlighting the bibtex entry Jones1998. Alternatively, if you have
already started ebib, then opening the link will get you to the bibtex
entry in your opened ebib database.

*** Natbib citation commands
The link types defined for use with ebib can also format output for
the LaTeX exporter.  The following link types insert the natbib
citation commands, using an optional command if the citation link
includes a description.

#+name: define-citep-link
#+begin_src emacs-lisp :results silent
  (org-add-link-type
   "citep" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "citep:" desc)))
             (format "\\citep{%s}" path)
             (format "\\citep[%s]{%s}" desc path)
)))))
#+end_src

#+name: define-citet-link
#+begin_src emacs-lisp :results silent
  (org-add-link-type
   "citet" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "citet:" desc)))
             (format "\\citet{%s}" path)
             (format "\\citet[%s]{%s}" desc path)
)))))
#+end_src

#+name: define-citealt-link
#+begin_src emacs-lisp :results silent
  (org-add-link-type
   "citealt" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "citealt:" desc)))
             (format "\\citealt{%s}" path)
             (format "\\citealt[%s]{%s}" desc path)
)))))
#+end_src

#+name: define-citealp-link
#+begin_src emacs-lisp
  (org-add-link-type
   "citealp" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "citealp:" desc)))
           (format "\\citealp{%s}" path)
         (format "\\citealp[%s]{%s}" desc path)
         )))))
#+end_src

#+name: define-citealtstar-link
#+begin_src emacs-lisp
  (org-add-link-type
   "citealt*" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "citealt*:" desc)))
           (format "\\citealt*{%s}" path)
         (format "\\citealt*[%s]{%s}" desc path)
         )))))
#+end_src

#+name: define-citealpstar-link
#+begin_src emacs-lisp
  (org-add-link-type
   "citealp*" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "citealp*:" desc)))
           (format "\\citealp*{%s}" path)
         (format "\\citealp*[%s]{%s}" desc path)
         )))))
#+end_src

#+name: define-citepstar-link
#+begin_src emacs-lisp
  (org-add-link-type
   "citep*" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "citep*:" desc)))
           (format "\\citep*{%s}" path)
         (format "\\citep*[%s]{%s}" desc path)
         )))))
#+end_src

#+name: define-citetstar-link
#+begin_src emacs-lisp
  (org-add-link-type
   "citet*" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "citet*:" desc)))
             (format "\\citet*{%s}" path)
             (format "\\citet*[%s]{%s}" desc path)
)))))
#+end_src


*** Standard biblatex citation commands

The standard biblatex citation commands have the following syntax:
\command[⟨prenote⟩][⟨postnote⟩]{⟨keys⟩}

They have been implemented by parsing the description part of the link
on a semicolon, so that, e.g.,
=[[cite:foo][postnote;prenote]]= becomes
=\cite[prenote][postnote]{foo}=.  Note that =[[cite:foo]]= and
=[[cite:foo][;]]= are functionally equivalent.

#+name: define-standard-biblatex-commands
#+begin_src emacs-lisp :noweb yes :results silent :exports code
  <<define-biblatex-cite-link>>
  <<define-biblatex-cap-cite-link>>
  <<define-biblatex-parencite-link>>
  <<define-biblatex-cap-parencite-link>>
  <<define-biblatex-footcite-link>>
  <<define-biblatex-footcitetext-link>>

#+end_src

#+name: define-biblatex-cite-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "cite" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "cite:" desc)))
           (format "\\cite{%s}" path)
         (format "\\cite[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-cap-cite-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "Cite" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "Cite:" desc)))
           (format "\\Cite{%s}" path)
         (format "\\Cite[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-parencite-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "parencite" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "parencite:" desc)))
           (format "\\parencite{%s}" path)
         (format "\\parencite[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-cap-parencite-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "Parencite" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "Parencite:" desc)))
           (format "\\Parencite{%s}" path)
         (format "\\Parencite[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-footcite-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "footcite" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "footcite:" desc)))
           (format "\\footcite{%s}" path)
         (format "\\footcite[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-footcitetext-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "footcitetext" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "footcitetext:" desc)))
           (format "\\footcitetext{%s}" path)
         (format "\\footcitetext[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

*** Style-specific biblatex commands

These commands can only be used by some of the citation styles that
ship with biblatex.

#+name: define-style-specific-biblatex-commands
#+begin_src emacs-lisp :noweb yes :results silent :exports code
  <<define-biblatex-textcite-link>>
  <<define-biblatex-cap-textcite-link>>
  <<define-biblatex-smartcite-link>>
  <<define-biblatex-cap-smartcite-link>>
  <<define-biblatex-citestar-link>>
  <<define-biblatex-parencitestar-link>>
  <<define-biblatex-supercite-link>>
#+end_src

#+name: define-biblatex-textcite-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "textcite" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "textcite:" desc)))
           (format "\\textcite{%s}" path)
         (format "\\textcite[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-cap-textcite-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "Textcite" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "Textcite:" desc)))
           (format "\\Textcite{%s}" path)
         (format "\\Textcite[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-smartcite-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "smartcite" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "smartcite:" desc)))
           (format "\\smartcite{%s}" path)
         (format "\\smartcite[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-cap-smartcite-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "Smartcite" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "Smartcite:" desc)))
           (format "\\Smartcite{%s}" path)
         (format "\\Smartcite[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-citestar-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "cite*" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "cite*:" desc)))
           (format "\\cite*{%s}" path)
         (format "\\cite*[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-parencitestar-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "parencite*" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (if (or (not desc) (equal 0 (search "parencite*:" desc)))
           (format "\\parencite*{%s}" path)
         (format "\\parencite*[%s][%s]{%s}"
                 (cadr (split-string desc ";"))
                 (car (split-string desc ";"))  path))))))
#+end_src

#+name: define-biblatex-supercite-link
#+begin_src emacs-lisp :results silent :exports code
  (org-add-link-type
   "supercite" 'ebib
   (lambda (path desc format)
     (cond
      ((eq format 'html)
       (format "(<cite>%s</cite>)" path))
      ((eq format 'latex)
       (format "\\cite*{%s}" path)))))
#+end_src

* Footnotes

[fn:1] Based on posts to the mailing list by Karl Marino and Carsten
Dominik.  Thanks to Bernt Hansen and Nick Dokos for help debugging a
problem implementing the publishing project.
[fn:2] This non-obvious use of the Org-mode link syntax appeared on the
Org-mode mailing list under the heading =text color + highlight= (!).
There was a lively discussion there, to which the ideas of
Samuel Wales, Christian Moe and Eric Schulte contributed directly to
this implementation.
[fn:3] The results of this experiment are now included in this
document.
[fn:4] The =xcolor= manual is an education in color management and is
highly recommended reading.