encode unicode for lxml

[librarian.git] / scripts / book2txt

diff --git a/scripts/book2txt b/scripts/book2txt
index 2963816..f77e512 100755 (executable)
--- a/scripts/book2txt
+++ b/scripts/book2txt
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 #
 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
-# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.  
+# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
 #
 import os
 import optparse
@@ -24,9 +24,9 @@ if __name__ == '__main__':
         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)
@@ -35,10 +35,11 @@ if __name__ == '__main__':
     for input_filename in input_filenames:
         if options.verbose:
             print input_filename
-        
+
         output_filename = os.path.splitext(input_filename)[0] + '.txt'
         try:
-            text.transform(input_filename, output_filename, parse_dublincore=options.parse_dublincore,
+            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' % {
@@ -58,4 +59,4 @@ if __name__ == '__main__':
                 'etype': e.__class__.__name__,
                 'message': e.message,
             }
-            raise e
+            raise