clean

#!/bin/bash
usage(){
	fmt <<EOF
DESCRIPTION
	Prettify individual source files, or all XHTML and SVG files in a source directory, including canonicalizing XML and minifying SVGs.  Note that this only prettifies the source code; it doesn't perform typography changes.

USAGE
	clean [-v,--verbose] [-s,--single-lines] DIRECTORY,FILE [DIRECTORY,FILE...]
		DIRECTORY is a directory containing one or more XHTML or SVG files.
		FILE is an XHTML or SVG file.
		Multiple directories and files can be included in a single invocation.

		With --single-lines, remove hard line wrapping.
EOF
	exit
}
die(){ printf "Error: ${1}\n" 1>&2; exit 1; }
require(){ command -v $1 > /dev/null 2>&1 || { suggestion=""; if [ ! -z "$2" ]; then suggestion=" $2"; fi; die "$1 is not installed.${suggestion}"; } }
if [ $# -eq 1 ]; then if [ "$1" = "--help" -o "$1" = "-h" ]; then usage; fi fi
#End boilerplate

#Check for dependencies
require "recode" "Try: sudo apt-get install recode"
require "xml2asc" "Try: sudo apt-get install html-xml-utils"

if [ $# -eq 0 ]; then
	usage
fi

verbose="false"
singleLines="false"
dirs=""

while [ $# -gt 0 ]
do
	case "$1" in
		-v|--verbose)
			verbose="true"
		;;
		-s|--single-lines)
			singleLines="true"
		;;
		*)
			dirs=$(printf "%s\n%s" "${dirs}" "$1")
		;;
	esac
	shift
done

#Set xmllint to use tab indentation.
export XMLLINT_INDENT=$(printf "\t")

printf "%s\n" "${dirs}" | while IFS= read -r i;
do
	if [ "${i}" = "" ]; then
		continue
	fi

	srcDir="$(realpath "${i%/}")"

	isDir="false"

	if [ -d "${srcDir}" ]; then
		isDir="true"
	fi

	if [ "${verbose}" = "true" ]; then
		printf "Cleaning %s ..." "${srcDir}"
	fi

	if [ "${isDir}" = "true" ]; then
		if [ "${singleLines}" = "true" ]; then
			if [ -f "${srcDir}"/src/epub/text/colophon.xhtml ]; then
				cp "${srcDir}"/src/epub/text/colophon.xhtml /tmp/colophon.xhtml
			fi

			find "${srcDir}" -iname "*.xhtml" -type f -exec sh -c "perl -i -pe \"s/\n/ /\" "{}"" \;
			find "${srcDir}" -iname "*.xhtml" -type f -exec sh -c "sed --in-place --regexp-extended \"s|\s+| |g\" "{}"" \;

			if [ -f "${srcDir}"/src/epub/text/colophon.xhtml ]; then
				cp /tmp/colophon.xhtml "${srcDir}"/src/epub/text/colophon.xhtml
			fi
		fi

		#Epub3 doesn't allow named entities, so convert them to their unicode equivalents
		#First we use xml2asc to convert *all* special characters to named entities.
		#Then recode converts named entities to unicode.
		find "${srcDir}" -iname "*.xhtml" -type f -exec sh -c "xml2asc < "{}" | recode HTML | sed \"s/&/&amp;/g\" > /tmp/tmp.xhtml; mv /tmp/tmp.xhtml "{}"" \;

		#Clean files with xmllint.
		#The --c14n flag removes the XML declaration, so we re-add it before formatting.  Otherwise we get problems.
		find "${srcDir}" -iname "*.xhtml" -type f -exec sh -c "xmllint --c14n "{}" | (printf '%s\n' '<?xml version=\"1.0\" encoding=\"UTF-8\"?>' && cat) | xmllint --output "{}" --format -" \;
		find "${srcDir}" \( -iname "*.svg" -o -iname "*.opf" -o -iname "*.xml" \) -type f -exec sh -c "xmllint --c14n "{}" | (printf '%s\n' '<?xml version=\"1.0\" encoding=\"UTF-8\"?>' && cat) | xmllint --output "{}" --format -" \;

		#Add trailing newlines to files that don't have one.
		find "${srcDir}" \( -iname "*.xhtml" -o -iname "*.svg" -o -iname "*.css" -o -iname "*.opf" -o -iname "*.xml" \) -type f -exec sed -i -e '$a\' "{}" \;
	else
		if [ "${singleLines}" = "true" ]; then
			perl -i -pe "s/\n/ /" "${srcDir}"
			sed --in-place --regexp-extended "s|\s+| |g" "${srcDir}"
		fi

		xml2asc < "${srcDir}" | recode HTML | sed "s/&/&amp;/g" > /tmp/tmp.xhtml; mv /tmp/tmp.xhtml "${srcDir}"

		xmllint --c14n "${srcDir}" | (printf '%s\n' '<?xml version="1.0" encoding="UTF-8"?>' && cat) | xmllint --output "${srcDir}" --format -

		sed --in-place -e '$a\' "${srcDir}"
	fi

	if [ "${verbose}" = "true" ]; then
		printf "%s\n" " OK"
	fi
done