tree-distance.py

from zss import Node, simple_distance
import lxml.html

html1 = lxml.html.parse('file1.html')
html2 = lxml.html.parse('file1.html')

def tree_from_el(el):
  node = Node(el.tag)
  for child in el:
    node.addkid(tree_from_el(child))
  return node

tree1 = tree_from_el(html1.getroot())
tree2 = tree_from_el(html2.getroot())
print(simple_distance(tree1, tree2))