X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/82e604c583fb1efdb6d988ea7e8673f8f7c8f119..2c0f1a1f4d11d1507705de614771e20dab83de17:/scripts/book2txt?ds=sidebyside diff --git a/scripts/book2txt b/scripts/book2txt index d56d6ff..c706a07 100755 --- a/scripts/book2txt +++ b/scripts/book2txt @@ -4,59 +4,31 @@ # This file is part of Librarian, licensed under GNU Affero GPLv3 or later. # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information. # -import os -import optparse - -from librarian import text -from librarian import dcparser, ParseError +from librarian.book2anything import Book2Anything, Option +from librarian.parser import WLDocument + + +class Book2Txt(Book2Anything): + format_name = "TXT" + ext = "txt" + uses_cover = False + uses_provider = False + parser_options = [ + Option('-i', '--ignore-dublin-core', dest='parse_dublincore', + action='store_false', default=True, + help='don\'t try to parse dublin core metadata') + ] + transform_options = [ + Option('-w', '--wrap', action='store', type='int', dest='wrapping', default=0, + help='set line wrap column') + ] + transform_flags = [ + Option('-r', '--raw', dest='raw-text', + action='store_true', default=False, + help='Produce raw text, without any surrounding info.') + ] + transform = WLDocument.as_text if __name__ == '__main__': - # Parse commandline arguments - usage = """Usage: %prog [options] SOURCE [SOURCE...] - Convert SOURCE files to TXT format.""" - - parser = optparse.OptionParser(usage=usage) - - parser.add_option('-v', '--verbose', action='store_true', dest='verbose', default=False, - help='print status messages to stdout') - parser.add_option('-w', '--wrap', action='store', type='int', dest='wrapping', default=0, - help='set line wrap column') - parser.add_option('-i', '--ignore-dublin-core', action='store_false', dest='parse_dublincore', default=True, - help='don\'t try to parse dublin core metadata') - - options, input_filenames = parser.parse_args() - - if len(input_filenames) < 1: - parser.print_help() - exit(1) - - # Do some real work - for input_filename in input_filenames: - if options.verbose: - print input_filename - - output_filename = os.path.splitext(input_filename)[0] + '.txt' - try: - output_file = open(output_filename, 'w') - text.transform(open(input_filename), output_file, parse_dublincore=options.parse_dublincore, - wrapping=str(options.wrapping)) - except ParseError, e: - print '%(file)s:%(name)s:%(message)s' % { - 'file': input_filename, - 'name': e.__class__.__name__, - 'message': e - } - except IOError, e: - print '%(file)s:%(name)s:%(message)s' % { - 'file': input_filename, - 'name': e.__class__.__name__, - 'message': e.strerror, - } - except BaseException, e: - print '%(file)s:%(etype)s:%(message)s' % { - 'file': input_filename, - 'etype': e.__class__.__name__, - 'message': e, - } - raise + Book2Txt.run()