X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/ca8319931f449468918067367133ff25f9b19f30..0604bdd5f693da9f1c78f9d9fa2276f0c7b6c17b:/src/librarian/builders/sanitize.py diff --git a/src/librarian/builders/sanitize.py b/src/librarian/builders/sanitize.py new file mode 100644 index 0000000..4d7f7f9 --- /dev/null +++ b/src/librarian/builders/sanitize.py @@ -0,0 +1,18 @@ +from lxml import etree +from librarian import OutputFile + + +class Sanitizer: + identifier = 'sanitize' + file_extension = 'xml2' + + def build(self, document): + doc = document.tree.getroot() # TODO: copy + doc.sanitize() + return OutputFile.from_bytes( + etree.tostring( + doc, + encoding='utf-8', + ) + ) +