Added ability to specify a base URL to handle relative links. (bf930632) · Commits · The Heavy / imgslipstream

imgslipstream.py

+25 −2

Original line number	Diff line number	Diff line
		@@ -32,6 +32,7 @@ if __name__ == '__main__':
		p.add_argument('--files', '-f', action='store_true', help='Make local files instead of embedding.')
		p.add_argument('--maxlen', '-m', action='store', help='Change the max length of URLs allowed (default 100).',
		default=100, type=int)
		p.add_argument('--url', '-u', action='store', help='Provide a relative URL where files can be downloaded from.')
		args = p.parse_args()
		log.info('Arguments: %s', args)

		@@ -43,6 +44,20 @@ if __name__ == '__main__':
		console_handler.setFormatter(logging.Formatter('[%(asctime)s] %(message)s'))
		log.addHandler(console_handler)

		if args.url is not None:
		if args.url[:4] != 'http':
		log.error('Unsupported protocol for URL argument.')
		raise argparse.ArgumentError
		if '://' not in args.url:
		log.error('URL argument is not a valid URL.')
		raise argparse.ArgumentError
		if '.' not in args.url.split('/')[2]:
		log.error('Invalid domain for URL argument.')
		raise argparse.ArgumentError
		if args.url[-1:] != '/':
		log.debug('Adding ending slash to URL argument.')
		args.url = args.url + '/'

		dest = pathlib.Path(args.source).with_suffix('.slipstream' + str(pathlib.Path(args.source).suffix))
		log.info('Destination: %s', dest)
		dest_folder = dest.with_suffix('').with_stem(dest.stem + '_files')
		@@ -114,7 +129,8 @@ if __name__ == '__main__':
		)
		)
		else:
		src_end_quote = document.find('"', src_quote + 6, img_end_tag)
		src_quote = src_quote + 5
		src_end_quote = document.find('"', src_quote + 1, img_end_tag)
		src_quote_type = '"'
		log.debug(
		'using alternate (") src_end_quote = {0}, src_quote_type = {1}'.format(
		@@ -137,6 +153,13 @@ if __name__ == '__main__':
		if url[:4] != 'http':
		if url[:len(dest_folder.name)] == dest_folder.name or url[:11] == 'data:image/':
		log.debug('Quietly ignore already replaced tags.')
		continue
		elif args.url is not None:
		if url[:1] == '/':
		url = args.url[:args.url.find('/', 8)] + url
		else:
		url = args.url + url
		log.debug('Attemping relative url retrieval with url: {0}'.format(url))
		else:
		log.warning('src attribute not http: {0}'.format(document[img_tag - 4:img_end_tag + 40]))
		continue