def build(self):
doc = document.Document()
ctx = DocBuilder(doc, self.clean_paragraphs)
for bit in self.group.flatten():
typeName = type(bit).__name__
getattr(ctx, "handle_%s" % typeName)(bit)
ctx.flushParagraph()
return doc
def go(self):
soup = BeautifulSoup.BeautifulSoup(
self.source,
convertEntities=BeautifulSoup.BeautifulSoup.HTML_ENTITIES,
fromEncoding=self.encoding,
smartQuotesTo=None)
# Make sure the document content doesn't use multi-lines
soup = self.format(soup)
doc = document.Document()
if self.css_source:
self.css = CSS(self.css_source)
else:
self.css = CSS() # empty css
self.process_into(soup, doc)
return doc
