Shell script to fix the "unique" identifiers in EPUB files sold by Packt Publishing.

fix-packt-epubs.sh

#!/bin/sh
#
# fixpacktepubs.sh 
#
# Process the EPUB files in the current directory and recreate them with the
# ISBN as the unique value. This script was written because Packt Publishing
# sell EPUBs with non-unique unique identifiers and it assumes that the EPUBs
# being process are structured like Packt's.
#

mkdir -p "new"

find . -type f -maxdepth 1 -iname '*.epub' | while read epub; do
	# Determine the "name" of the EPUB 
	name=$(basename -s ".epub" "${epub}")
	newname="../new/${name}.epub"

	rm -rf "${name}"
	mkdir -p "${name}"
	pushd "${name}" > /dev/null

	# Unpack the contents of the EPUB file.
	unzip -q "../${epub}" > /dev/null

	# Find the ISBN.
	isbn=$(grep -ri 'ISBN' . | sed -Ee 's/.*(ISBN ([0-9-]+)).*/urn:isbn:\2/g' -e 's/-//g')

	# Replace the probably-duplicate value set as the unique identifier with
	# more-likely-to-be-unique ISBN.
	find . -type f -iname '*.opf' -print | while read content; do
		unique=$(grep unique-identifier "${content}" | sed -Ee "s/.*unique-identifier=['\"](.*)[\"'].*/\1/")
		sed -i~ -Ee "s#id=\"${unique}\">.*</#id=\"${unique}\">${isbn}</#g" "${content}"
		rm "${content}~"
	done

	# Reconstruct it.
	zip -X "${newname}" mimetype
	zip -rg "${newname}" META-INF -x \*.DS_Store
	zip -rg "${newname}" OEBPS -x \*.DS_Store

	popd > /dev/null
done