src/librarian/builders/sanitize.py

   1 from lxml import etree
   2 from librarian import OutputFile
   3
   4
   5 class Sanitizer:
   6     identifier = 'sanitize'
   7     file_extension = 'xml2'
   8
   9     def build(self, document, **kwargs):
  10         doc = document.tree.getroot() # TODO: copy
  11         doc.sanitize()
  12         return OutputFile.from_bytes(
  13             etree.tostring(
  14                 doc,
  15                 encoding='utf-8',
  16             )
  17         )
  18