build-images

#!/usr/bin/env python3

import argparse
import os
import errno
import base64
import fnmatch
import shutil
import subprocess
import regex
import se

def clean_inkscape_svg(filename, clean_path):
	with open(filename, "r+") as file:
		svg = file.read()

		# Time to clean up Inkscape's mess
		svg = regex.sub(r"id=\"[^\"]+?\"", "", svg)
		svg = regex.sub(r"<metadata[^>]*?>.*?</metadata>", "", svg, flags=regex.MULTILINE | regex.DOTALL)
		svg = regex.sub(r"<defs[^>]*?/>", "", svg, flags=regex.MULTILINE | regex.DOTALL)
		svg = regex.sub(r"xmlns:(dc|cc|rdf)=\"[^\"]*?\"", "", svg, flags=regex.MULTILINE | regex.DOTALL)

		# Inkscape includes font and letter-spacing info even though we've removed font information
		svg = regex.sub(r"font-.+?:.+?([;\"])", "\\1", svg)
		svg = regex.sub(r"text-anchor:.+?([;\"])", "\\1", svg)
		svg = regex.sub(r"letter-spacing:.+?([;\"])", "\\1", svg)
		svg = regex.sub(r";+", ";", svg)
		svg = regex.sub(r" style=\";", " style=\"", svg)
		svg = regex.sub(r" style=\"\"", "", svg)

		file.seek(0)
		file.write(svg)
		file.truncate()

	subprocess.run([clean_path, filename])

def remove_metadata(filename, exiftool_path):
	# After lots of research there just isn't a good way to reliably edit exif metadata in pure Python.
	# So unfortunately we're stuck with this dependency on `exiftool`.
	subprocess.run([exiftool_path, "-overwrite_original", "-all=", filename], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)

def main():
	parser = argparse.ArgumentParser(description="Build ebook covers and titlepages for a Standard Ebook source directory, and place the output in DIRECTORY/src/epub/images/.")
	parser.add_argument("-v", "--verbose", action="store_true", help="increase output verbosity")
	parser.add_argument("directories", metavar="DIRECTORY", nargs="+", help="a Standard Ebooks source directory")
	args = parser.parse_args()

	# Check for required utilities
	clean_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), "clean")
	inkscape_path = shutil.which("inkscape")
	exiftool_path = shutil.which("exiftool")

	if inkscape_path is None:
		se.print_error("Couldn't locate Inkscape. Is it installed?")
		exit(1)

	if exiftool_path is None:
		se.print_error("Couldn't locate exiftool. Is it installed?")
		exit(1)

	for directory in args.directories:
		directory = os.path.abspath(directory)

		if args.verbose:
			print("Processing {} ...".format(directory))

		if not os.path.isdir(directory):
			if args.verbose:
				print("\t", end="", flush=True)

			se.print_error("Not a directory: {}".format(directory))
			exit(1)

		source_images_directory = os.path.join(directory, "images")
		source_cover_jpg_filename = os.path.join(source_images_directory, "cover.jpg")
		source_cover_svg_filename = os.path.join(source_images_directory, "cover.svg")
		source_titlepage_svg_filename = os.path.join(source_images_directory, "titlepage.svg")
		dest_images_directory = os.path.join(directory, "src", "epub", "images")
		dest_cover_svg_filename = os.path.join(dest_images_directory, "cover.svg")
		dest_titlepage_svg_filename = os.path.join(dest_images_directory, "titlepage.svg")

		try:
			os.makedirs(dest_images_directory)
		except OSError as exception:
			if exception.errno != errno.EEXIST:
				if args.verbose:
					print("\t", end="", flush=True)

				se.print_error("Couldn't create directory: {}".format(dest_images_directory))
				exit(1)

		# Remove useless metadata from jpg files
		for root, _, filenames in os.walk(source_images_directory):
			for filename in fnmatch.filter(filenames, "cover.source.*"):
				remove_metadata(os.path.join(root, filename), exiftool_path)

		# Build cover.svg
		if os.path.isfile(source_cover_jpg_filename):
			remove_metadata(source_cover_jpg_filename, exiftool_path)

			if os.path.isfile(source_cover_svg_filename):
				if args.verbose:
					print("\tBuilding cover.svg ...", end="", flush=True)

				# base64 encode cover.jpg
				with open(source_cover_jpg_filename, "rb") as file:
					source_cover_jpg_base64 = base64.b64encode(file.read()).decode()

				# Convert text to paths
				# Inkscape adds a ton of crap to the SVG and we clean that crap a little later
				subprocess.run([inkscape_path, source_cover_svg_filename, "--without-gui", "--export-text-to-path", "--export-plain-svg", dest_cover_svg_filename])

				# Embed cover.jpg
				with open(dest_cover_svg_filename, "r+", encoding="utf-8") as file:
					svg = file.read()

					svg = regex.sub(r"xlink:href=\".*?cover\.jpg", "xlink:href=\"data:image/jpeg;base64," + source_cover_jpg_base64, svg, flags=regex.MULTILINE | regex.DOTALL)

					file.seek(0)
					file.write(svg)
					file.truncate()

				clean_inkscape_svg(dest_cover_svg_filename, clean_path)

				if args.verbose:
					print(" OK")
			else:
				if args.verbose:
					print("\t./images/cover.svg not found, skipping ...")
		else:
			if args.verbose:
				print("\t./images/cover.jpg not found, skipping ...")

		# Build titlepage.svg
		if os.path.isfile(source_titlepage_svg_filename):
			if args.verbose:
				print("\tBuilding titlepage.svg ...", end="", flush=True)

			# Convert text to paths
			# inkscape adds a ton of crap to the SVG and we clean that crap a little later
			subprocess.run([inkscape_path, source_titlepage_svg_filename, "--without-gui", "--export-text-to-path", "--export-plain-svg", dest_titlepage_svg_filename])

			clean_inkscape_svg(dest_titlepage_svg_filename, clean_path)

			if args.verbose:
				print(" OK")
		else:
			if args.verbose:
				print("\t./images/titlepage.svg not found, skipping ...")

if __name__ == "__main__":
	main()