strapdown2pdf

#!/usr/bin/env bash
# By: Lilian BESSON
# Email: Lilian.BESSON[AT]ens-cachan[DOT]fr
# Date: 23-11-2014
# Web: http://besson.qc.to/bin/strapdown2pdf
# Web2: http://lbesson.bitbucket.org/strapdown2pdf.html
#
# strapdown2pdf, a small script to simply convert a "StrapDown.js flavored"
# HTML file (used to write with a Markdown syntax) to a PDF document.
#
# More details on http://lbesson.bitbucket.org/strapdown2pdf.html
#
# Requirements:
#  + [lunamark](http://jgm.github.io/lunamark/lunamark.1.html)
#  + [autotex](http://besson.qc.to/bin/autotex) (from my webpage),
#  + [PDFCompress](http://besson.qc.to/bin/PDFCompress) (from my webpage),
#  + [pdflatex](http://besson.qc.to/bin/pdflatex) (from my webpage).
#
# References and more details:
#  + [StrapDown.js](http://lbesson.bitbucket.org/md/)
#
# Licence: [GPLv3](http://besson.qc.to/LICENCE.html)
#
# Bugs / FIXME:
#  1. I think it is way better now. Hack something to preserve LaTeX code verbatim between $ $ in the HTML initial document
#     Example: http://javis/a/maths.html -> http://javis/a/maths.pdf works very well,
#     only for MathJax flavored maths formulas, which are all copied in the .tex with escaped $ and { : \{ and \$ every where :(
#
#
version='0.8'
AUTOTEX="autotex batchmode"

quiet="false"
sign="false"
keep="false"
discrete="false"
htm="false"

scale="0.85"
policesize="11"

# $OPTARG can contain the argument of the option k (if specified with hvk: or hk:v for example)
# while getopts vhqkmdsi option; do
# FIXME: options with a for loop are better handled that getopts ? .. OK
# 	 case $option in
for arg in "$@"; do
	case "$arg" in
 	 	-v|-version|--version)
			echo -e "strapdown2pdf $version"
			exit 0
			;;
		-h|-help|--help)
			echo -e "${green}strapdown2pdf${white} -help | [options]"
			echo -e ""
			echo -e "Print a StrapDown-powered web-page to a PDF document, using lunamark and autotex."
			echo -e ""
			echo -e "Help:"
			echo -e "    ${yellow}-h${white}           to print this help message (and quit)."
			echo -e "    ${yellow}-v${white}           to print just the version of strapdown2pdf (and quit)."
			echo -e ""
			echo -e "Options:"
			echo -e "    ${yellow}-i|-interactive${white}	run PDFLaTeX with the interactive (errorstopmode) mode (default is batchmode)."
			echo -e "    ${yellow}-q|-quiet${white}		run strapdown2pdf in quiet mode (no output at all, everything is redirected to ${magenta}/tmp/strapdown2pdf.log${white})."
			echo -e "    ${yellow}-m|-htm${white}		run strapdown2pdf to produce a simple HTML file (which do not use StrapDown.js), written to a .htm file.\n\t\t\t For important document, producing a .htm autonomous file is a good idea."
			echo -e "    ${yellow}-d|-discrete${white}	run strapdown2pdf is discrete mode, without adding any creditentials in the produced document."
			echo -e "    ${yellow}-s[0-9][0-9]*%${white}	change the default scale used by autotex (default is 85%, '-70%', '-75%' or '-80%' are good also) ${cyan}"'New!'
			echo -e "    ${yellow}-[0-9][0-9]*pt${white}	change the default police size used by autotex (default is 11pt, '-10pt' or '-12pt' are good also) ${cyan}"'New!'
			echo -e "    ${yellow}-k|-keep${white}		keep the intermediate .tex file. ${cyan}"'New!'
			echo -e "    ${yellow}-s|-sign${white}		sign the produce PDF document with GnuPG (thanks to PDFCompress)."
			echo -e ""
			echo -e "strapdown2pdf v$version : Copyrights: (c) Lilian Besson 2011-2014."
			echo -e "Released under the term of the GPL v3 Licence (more details on http://besson.qc.to/LICENSE.html)."
			echo -e "In particular, strapdown2pdf is provided WITHOUT ANY WARANTY."
			exit 0
			;;
		-q|-quiet|--quiet)
			echo -e "${magenta}Running strapdown2pdf with option --quiet, to run silently.${white}"
			quiet="true"
			shift
			;;
		-k|-keep|--keep)
			echo -e "${magenta}Running strapdown2pdf with option --keep, to keep the intermediate .tex file.${white}"
			keep="true"
			shift
			;;
		-m|-htm|--htm)
			echo -e "${magenta}Running strapdown2pdf with option --htm, to produce a simple HTML file (which do not use StrapDown.js), written to a .htm file.${white}"
			htm="true"
			shift
			;;
		-d|-discrete|--discrete)
			echo -e "${magenta}Using raw document without adding any ${u}strapdown2pdf${U} and ${u}StrapDown${U}.js creditentials.${white}"
			discrete="true"
			shift
			;;
		-s|-sign|--sign)
			echo -e "${magenta}Running strapdown2pdf with option --sign, to sign the produced PDF (useless..).${white}"
			sign="true"
			shift
			;;
		-s[0-9][0-9]*%)
			scale="${arg//-s/}"
			scale=0."${scale//%/}"
			echo -e "${magenta}Running strapdown2pdf with option -s[0-9]*%, to choose the scale, which is now ${scale}.${white}"
			shift
			;;
		-[0-9][0-9]*pt)
			policesize="${arg//-/}"
			policesize="${policesize//pt/}"
			echo -e "${magenta}Running strapdown2pdf with option -[0-9][0-9]*pt, to choose the police size, which is now ${policesize}.${white}"
			shift
			;;
		-i|-interactive|--interactive)
			echo -e "${magenta}Using PDFLaTeX with '-interaction=errorstopmode' option (the compilation will pause if a problem occur).${white}"
			AUTOTEX="autotex errorstopmode"
			shift
			;;
	 esac
done

StrapDown2PDF() {
	input="$(basename "$1")"
	name="${input%.html}"

	# Be sure we have a nice HTML or Markdown file and not something else :)
	if [ "${name}.html" != "${input}" ]; then
		echo -e "${red}WARNING${white} The input file ${input} seems to not be a valid HTML file."
		if [ "${input%.md}.md" != "${input}" ]; then
			echo -e "${red}WARNING${white} The input file ${input} seems to not be a Markdown either."
			echo -e "${red}I prefer to quit NOW.${white}"
			exit 5
		else
			echo -e "${green}COOL${white} The input file ${input} seems to be a valid MarkDown file : ${blue}good :)${white}."
			name="${input%.md}"
		fi
	else
		echo -e "${green}COOL${white} The input file ${input} seems to be a valid MarkDown file or StrapDown-powered HTML file : ${blue}good :)${white}."
	fi
	# To be even more paranoid, I could search the HTML file and be sure that StrapDown.js is indeed used, but pfiou I'm lazy.

	p="$(pwd)"

	echo -e "Working with $u$input$U on $blue$p${white}." | tee -a /tmp/strapdown2pdf.log

	echo -e "${magenta}The following lines will be removed :${white}"
	grep -n "^<" "$input"

	# Remove HTML only lines (typically, the first one and the 4 last ones),
	grep -v "^<" "$input" \
		| sed s/'\\\\'/'\\\\\\\\'/g \
		| sed s/'\\_'/'\\\\\\_'/g \
		| sed s/'`_`'/'`\\\\_`'/g \
		| sed s/'`\\`'/'`\\\\`'/g \
		| sed s/'`\\\\`'/'`\\\\\\\\`'/g \
		| sed s/'`\$`'/'`\\\\$`'/g \
		| sed s/'`\$\$`'/'`\\\\$$`'/g \
		| sed s/'`\*`'/'`\\*`'/g \
		| sed s/'`\*\*`'/'`\\*\\*`'/g \
		> /tmp/"${name}".md
	# Now we have a pure Markdown file (at least we hope)

	title="$(grep -o -m 1 "<title>[^<]*</title>" "${input}" | grep -o ">.*<" | sed s/">"/""/ | sed s/"<"/""/)"
	echo -e "${cyan}I found this as a possible title${white} : $u${title}$U." | tee -a /tmp/strapdown2pdf.log

	if [ X"${htm}" = "Xtrue" ]; then
			echo -e "<!DOCTYPE html><html><head><meta charset=\"utf-8\"/><title>${title}</title></head>\n<body>" > "${name}".htm
			python -m markdown -e utf8 -v /tmp/"${name}".md >> "${name}".htm
			echo -e "</body>\n</html>\n" >> "${name}".htm
			echo -e "${cyan}I am done producing ${name}.htm in the current repertory."
			exit 0
	fi

	# Convert md -> tex with lunamark (in /tmp/, as always)
	lunamark -Xhash_enumerators -t latex -o "${name}".tex~ /tmp/"${name}".md || exit 21

	# Because we are proud of this script
	echo -e "%% -*- coding:utf8; mode:latex -*-\n%% LaTeX file automatically generated with [strapdown2pdf](https://bitbucket.org/lbesson/bin/src/master/strapdown2pdf)" > "${name}".tex
	echo -e "%% from ${p}/${input}, the $(date)." >> "${name}".tex

	# Adding two autotex specials comments,
	# One for the title
	echo -e "%autotex% Titre: ${title}" >> "${name}".tex
	# And one for the scale
	echo -e "%autotex% Scale: ${scale}" >> "${name}".tex
	# And one for the police size
	echo -e "%autotex% PoliceSize: ${policesize}" >> "${name}".tex

	# We add the .tex file (it does NOT have any LaTeX headers)
	# Apparently, I have been able to remove all weird espaced math LaTeX code in the input file ...

	# FIXME: here we should also try to improve the LaTeX math code preservation
	cat "${name}".tex~ \
		| sed s/'\\\$'/$/g \
		| sed s/'\\^'/'^'/g \
		| sed s/'\\char92{}'/'\\'/g \
		| sed s/'\\char62{}'/'>'/g \
		| sed s/'\\char60{}'/'<'/g \
		| sed s/'\\char126{}'/'\~'/g \
		| sed s/'\\{'/'{'/g \
		| sed s/'\\}'/'}'/g \
		| sed s/'\\_'/'_'/g \
		| sed s/'\\_{'/'_{'/g \
		| sed s/'\\_'/'_'/g \
		| sed s/'\\emph'/''/g \
		| sed 's/\\&/\&/g' \
		| sed s/'\^{}'/'^'/g \
		| sed s/'\"\([^"]*\)\"'/'« \1 »'/g \
		| sed s/'\([A-Za-z]\)_\([A-Za-z]\)'/'\1\\_\2'/g \
		| sed s_'\(\\href{http[^}]*}\){\([^} ]*\.[^} ]*\)}'_'\1{\\texttt{\2}}'_g \
		| sed s/'\(\\[a-z]*section\){'/'\1*{'/g \
		>> "${name}".tex || exit 16
		# FIXME Add '\([^}]*\)' between on the left and right on {\([^} ]*\.[^} ]*\)} after { and before } ?

	# And a final line to say "Compiled from HTML/MarkDown with StrapDown.js to PDF with ..."
	if [ "$discrete" != "true" ]; then
		echo -e "\n\n\n\n%% Added with strapdown2pdf\n\\\\hspace{\\\\fill}\n\\\\vfill{}\n\n\\\\hspace{\\\\fill}\\\\rule{.6\\\\linewidth}{0.4pt}\\\\hspace{\\\\fill}\n\n\\\\begin{quote}\n\\\\begin{footnotesize}\n (Compiled to \\\\textbf{PDF} from a \\\\texttt{HTML/Markdown} file (powered by \\\\href{http://lbesson.bitbucket.org/md/}{\\\\texttt{StrapDown.js}}) with \\\\textbf{\\\\href{http://lbesson.bitbucket.org/md/strapdown2pdf.html}{strapdown2pdf}}, \\\\texttt{v${version}}.)\n\\\\end{footnotesize}\n\\\\end{quote}\n"  >> "${name}".tex
	fi

	# Compile in batchmode it with autotex, to automatically add headers and packages stuff
	${AUTOTEX} "${name}".tex

	# Compress it (and even gpg sign it !)
	if [ "$sign" = "true" ]; then
		PDFCompress --sign "${name}".pdf
	else
		PDFCompress "${name}".pdf
	fi

	# Clean up local repertory
	mv -vf "${name}".tex* /tmp/
	if [ "$keep" = "true" ]; then
		cp -vf /tmp/"${name}".tex ./ && \
		echo -e "${green} The intermediate LaTeX file ${name}.tex has been kept here, as asked by the option -k or --keep."
	fi

	# Proudly say that we are done
	echo -e "${green}The file ${name}.pdf have been well generated from ${input}, and it should be really beautiful :)" | tee -a /tmp/strapdown2pdf.log
	# read
}

log="/tmp/strapdown2pdf.log"
# And finally treat every arguments.
if [ "$quiet" = "true" ]; then
	echo -e "On "$(date)", strapdown2pdf is running on quiet mode." > "${log}"
	echo -e "On quiet mode, arguments were '$@'." &>> "${log}"
	StrapDown2PDF "${1}" &>> "${log}"
	shift
	for finput in "$@"; do
		echo -e "\n\n---------------------------" &>> "${log}"
		echo -e "Generating the next file..." &>> "${log}"
		StrapDown2PDF "${finput}" &>> "${log}"
	done
else
	StrapDown2PDF "${1}"
	shift
	for finput in "$@"; do
		echo -e "\n\n---------------------------"
		echo -e "${blue}Generating the next file..."
		StrapDown2PDF "${finput}"
	done
fi

## END