#2044: handle weird whitespace for covers

[librarian.git] / scripts / book2txt

diff --git a/scripts/book2txt b/scripts/book2txt
index f77e512..9cfdef2 100755 (executable)
--- a/scripts/book2txt
+++ b/scripts/book2txt
@@ -7,8 +7,8 @@
 import os
 import optparse
 
 import os
 import optparse
 

-from librarian import text
-from librarian import dcparser, ParseError
+from librarian import ParseError
+from librarian.parser import WLDocument

 
 
 if __name__ == '__main__':
 
 
 if __name__ == '__main__':


@@ -38,14 +38,15 @@ if __name__ == '__main__':
 
         output_filename = os.path.splitext(input_filename)[0] + '.txt'
         try:
 
         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))
+            doc = WLDocument.from_file(input_filename,
+                parse_dublincore=options.parse_dublincore)
+            html = doc.as_text(wrapping=str(options.wrapping))
+            doc.save_output_file(html, output_path=output_filename)

         except ParseError, e:
             print '%(file)s:%(name)s:%(message)s' % {
                 'file': input_filename,
                 'name': e.__class__.__name__,
         except ParseError, e:
             print '%(file)s:%(name)s:%(message)s' % {
                 'file': input_filename,
                 'name': e.__class__.__name__,

-                'message': e.message
+                'message': e

             }
         except IOError, e:
             print '%(file)s:%(name)s:%(message)s' % {
             }
         except IOError, e:
             print '%(file)s:%(name)s:%(message)s' % {


@@ -57,6 +58,6 @@ if __name__ == '__main__':
             print '%(file)s:%(etype)s:%(message)s' % {
                 'file': input_filename,
                 'etype': e.__class__.__name__,
             print '%(file)s:%(etype)s:%(message)s' % {
                 'file': input_filename,
                 'etype': e.__class__.__name__,

-                'message': e.message,
+                'message': e,

             }
             raise
             }
             raise