pd: a wrapper around pandoc that allows setting up some default options

pd

#!/usr/bin/env bash
#
# AUTHOR:: David Sanson
# URL: https://gist.github.com/857619
# LICENSE: http://creativecommons.org/licenses/MIT/
# VERSION: 0.2
#
# IN PROGRESS:
#    Support for "profiles".
#    Removed $defaultopts_markdown2pdf---just use formatopts[pdf] instead.
#
#
##########################################
# USER SETTINGS
##########################################
#
# Path to Bibliography
# ====================
#
# I recommend setting $PANDOC_BIB in your .bashrc or placing
# default.bib in $HOME/.pandoc. But you can also set the 
# path here by uncommenting the line below:
#
# bibfile=/path/to/my/bib
#
# Default CSL
# ===========
# By default, pandoc will use the chicago-author-date style.
# If you want something else, set it here.
#
# csl=$HOME/.csl/apa.csl
#
# Default Options
# ===============
#
# If you set $defaultopts, its value will be passed to pandoc
# (except for PDF output, since markdown2pdf doesn't accept
# all the options pandoc accepts: set PDF options using 
# $defaultopts[pdf] below).
#
	defaultopts="-sS"
#
# Launch Command
# ==============
#
# The commandline tool that will be used to open the generated
# file when requested. Linux users might set this to 'xdg-open'.
# Windows users might set it to 'start'.
#
	launcher="open"
#
#
# Format-specific defaults
# ========================
# 
# Here you can specify default options for specific output formats.
# These will be appended to the command line *after* the $defaultopts
# and so will override any conflicting values in $defaultops.
#
# NOTE: formatopts requires bash 4 or higher. If you want
# to use this script with generic sh or bash < 4, Comment out
# all lines with ending with #BASH4, and change the shebang on the
# first line to something like /bin/sh.
#
	declare -A formatopts	 				#BASH4 
 	formatopts[pdf]=						#BASH4
 	formatopts[native]=						#BASH4
 	formatopts[json]=						#BASH4
	formatopts[html]="--mathjax='http://cdn.mathjax.org/mathjax/latest/MathJax.js?config=Accessible'"				#BASH4 
 	formatopts[html+lhs]=					#BASH4
 	formatopts[s5]=							#BASH4
 	formatopts[slidy]=						#BASH4
 	formatopts[docbook]=					#BASH4
 	formatopts[opendocument]=				#BASH4
 	formatopts[latex]=						#BASH4
 	formatopts[latex+lhs]=					#BASH4
 	formatopts[context]=					#BASH4
 	formatopts[texinfo]=					#BASH4
 	formatopts[man]=						#BASH4
	formatopts[markdown]=	 				#BASH4 
 	formatopts[markdown+lhs]=				#BASH4
 	formatopts[plain]=						#BASH4
 	formatopts[rst]=						#BASH4
 	formatopts[rst+lhs]=					#BASH4
 	formatopts[mediawiki]=					#BASH4
 	formatopts[textile]=					#BASH4
 	formatopts[rtf]=						#BASH4
 	formatopts[org]=						#BASH4
 	formatopts[odt]=						#BASH4
 	formatopts[epub]=						#BASH4

# Profiles
# ========
# 
# Specify the path to a file containing "profiles".
#
    profiles="$HOME/.pandoc/profiles"
#
#
##########################################
# END OF USER SETTINGS
##########################################

helpme() { 
cat <<EOH

$commandname [$commandname-commands] [OPTIONS] [FILES]

$commandname is a general purpose wrapper around pandoc. It passes OPTIONS and
FILES untouched to pandoc. But before it does that, it: 

1. Infers an output filename, using the same rules
   used by markdown2pdf.

2. Parses special $commandname-commands.

$commandname-commands
===========

$commandname-commands must come *before* any pandoc OPTIONS.

  bib => --bibliography=/path/to/my/bib

    Set the path in the USER SETTINGS section of the script.

  <format> => -t <format>

    <format> can be an output format that pandoc supports. If <format> is
   "pdf", then markdown2pdf will be called instead of pandoc.

  <profile>
   	
    <profile> is a key from $profiles. The format of $profiles should be

	   key1: options
       key2: options
	   etc.
	For example,
	
	   cv: --template=/path/to/cv.template --xetex
	   exam: --template=/path/to/exam.template --xetex
	   
  open => pandoc -o <output> && <\$launcher> <output>

    Set the value of \$launcher(=$launcher) in USER SETTINGS.

  no-out 

    Don't try to infer an output filename. Use this when you want to pipe
    the output to STDOUT. This overrides "open".

  dry

    Print the expanded command but don't execute it.

  help

    Display this help.

USER SETTINGS
=============

Before running $commandname for the first time, you will want to edit the
default values in the USER SETTINGS section of the script. These include the
path to your bibliography file, the command to use for launching files, 
global default options to be passed to pandoc, and format specific default
options. Some of these current values are:

	\$bibfile=$bibfile
	\$defaultopts=$defaultopts
	\$defaultopts_markdown2pdf=$defaultopts_markdown2pdf
	\$launcher=$launcher
  
Examples
========

  $commandname html example.markdown

expands to

  pandoc -o example.html -t html $defaultopts ${formatopts[html]} example.markdown

The "-o example.html" is the result of inferring an output filename. "-t html" is due to
the "html" $commandname-command. Everything between that and "example.markdown" comes from
the values of \$defaultopts or \$formatopts[html].

  $commandname html bib example.markdown

expands to

  pandoc --bibliography=<\$bibfile> -o example.html -t html $defaultopts ${formatopts[html]} example.markdown

Where \$bibfile is your bibliography file.

Pandoc options will be passed *after* any options generated by pd. So

  $commandname html --no-wrap example.markdown

expands to

  pandoc -o example.html -t html $defaultopts ${formatopts[html]} --no-wrap example.markdown

If you supply <format> commands, the last wins:

  $commandname html odt pdf rtf example.markdown

expands to

  pandoc -o example.rtf -t rtf $defaultopts ${formatopts[rtf]} example.markdown

Note that $commandname makes no attempt to parse pandoc options. This can lead to 
weird expansions, e.g., (ignoring \$defaultopts and \$formatopts)

  $commandname html -t rtf example.markdown

expands to

  pandoc -o example.html -t html -t rtf example.markdown

$commandname stops looking for $commandname commands when it encounters
an argument that it doesn't recognize: that and all subsequent arguments are
passed on to pandoc untouched. So, for example,

  $commandname rtf blah bib example.markdown

expands to

  pandoc -o blah.rtf -t rtf blah bib example.markdown

which is probably not what you want!

If you run into problems, you can use the "dry" command to print out the expansion
without executing it.

Output Filename
===============

$commandname uses the same logic and markdown2pdf to guess at reasonable 
filename roots, and appends the extension given by the <format> command. If
no <format> command is given, the ".html" extension is used. If the format 
is 'plain', then the extension 'txt' is used.

The "no-out" command disables this behavior. So something like

   cat file.markdown | $commandname no-out html | tidy

will do what you expect. The no-out option disables the "open" command, since 
the "open" command uses the infered filename.

EOH
}

##########################################
# Find path to bibliography
##########################################

getbibpath() {
	if [ -n "$PANDOC_BIB" ]; then
		bibfile="$PANDOC_BIB"
	elif [ -f "$HOME/.pandoc/default.bib" ]; then
		bibfile="$HOME/.pandoc/default.bib"
	else
		echo "You need to specify a bibliographic database."
		echo "Either set \$PANDOC_BIB or place a file named"
		echo "default.bib in $HOME/.pandoc or set the path"
		echo "manually by editing $0."
		exit
	fi
}

commandname=`basename $0`

##########################################
# Parse Profiles File
##########################################

declare -A profile
while read line; do
	key=${line%%:*}
    profile[$key]=${line##*:}
done < "$profiles"


##########################################
# Parse pd commands
##########################################

profileopts=""
openfile=false
dryrun=false
outfile=true
bib=false
checking=true
while $checking
	do
		case $1 in
			pdf | native | json | html | html+lhs | s5 | slidy | docbook | opendocument | latex | latex+lhs | context | texinfo | man | markdown | markdown+lhs | plain | rst | rst+lhs | mediawiki | textile | rtf | org | odt | epub)
				format=$1
				if [ $format == "plain" ]; then
					extension="txt"
				elif [ $format == "latex" ]; then
					extension="tex"
				else
					extension=$format
				fi
				shift 1
			;;
			bib)
				bib=true
			shift 1
			;;
			open)
				openfile=true
				shift 1
			;;
			no-out)
				outfile=false
				shift 1
			;;
			dry)
				dryrun=true
				shift 1
			;;
			help)
				helpme
				exit
			;;
			*)
				itsaprofile=false
				for k in "${!profile[@]}"; do
    				if [ "$1" == "$k" ]; then
						echo "match: $k"
						profileopts="${profile[$k]}"
						itsaprofile=true
					fi
				done
				if $itsaprofile; then
					shift 1
				else
					checking=false
				fi
			;;
		esac
	done
	echo "$profileopts"

###################################################
# Find output file (cribbed from old copies of markdown2pdf)
###################################################

SYNOPSIS="Or try \`$commandname help\` for help with the wrapper script."

THIS=${0##*/}

err ()  { echo "$*"   | fold -s -w ${COLUMNS:-110} >&2; }

CONF=$(pandoc --dump-args "$@" 2>&1) || {
    errcode=$?
    echo "$CONF" | sed -e '/^pandoc \[OPTIONS\] \[FILES\]/,$d' >&2
    [ $errcode -eq 2 ] && err "$SYNOPSIS"
    exit $errcode
}

# Supply an output file name if none was supplied on
# command line.

OUTPUT=$(echo "$CONF" | sed -ne '1p')
ARGS=$(echo "$CONF" | sed -e '1d')

if [ "$OUTPUT" = "-" ]; then
    firstinfile="$(echo "$ARGS" | sed -ne '1p')"
    firstinfilebase="${firstinfile%.*}"
    destname="${firstinfilebase:-stdin}.${extension:-html}"
else
    destname="$OUTPUT"
fi


##########################################
# Generate command line options
##########################################

cliopts="$*"
if $outfile; then
	opts=" -o $destname " # add output command
fi
command="pandoc"

if [ -n "$format" ]; then

	######################################
    # if format is pdf, run markdown2pdf
    ######################################    
	if [ "$format" == "pdf" ]; then
		opts=""
		command="markdown2pdf"
	else
		opts="$opts $defaultopts -t $format"
	fi
	######################################
    # Apply formatopts, if there are any
    ######################################
	
	opts="$opts ${formatopts[$format]}"		 #BASH4
 	
	######################################
    # Apply profile opts, if any
    ######################################
 	
	opts="$opts $profileopts"
fi

######################################
# Add bibliography option
######################################
if $bib; then
	if [ -z "$bibfile" ]; then
		getbibpath
	fi
	opts="--bibliography=$bibfile --csl=$csl $opts"
fi

##########################################
# Run command
##########################################

# Is this a dry run?

if $dryrun; then
	command="echo $command"
	launcher="echo $launcher"
fi

$command $opts $cliopts &&
	if $openfile && $outfile; then
		$launcher $destname
	else
		if [ $format != "pdf" ] && $outfile && $dryrun; then
			echo "Would have created $destname"
		else
			echo "Created $destname"
		fi
	fi