Generate a CSS document based on the content of a HTML file.

html2css

#! /usr/bin/env python
from BeautifulSoup import BeautifulSoup
import os.path
import re

def html2css(FILEsource, OUTPUT):
  fname = os.path.join(os.curdir, FILEsource)
  FILE, OUT = open(fname,'r'), open(OUTPUT,'w')
  source, Tags_regexp, HTML, CSS = FILE.read(), "(?i)<\/?\w+((\s+\w+(\s*=\s*(?:\".*?\"|'.*?'|[^'\">\s]+))?)+\s*|\s*)\/?>", '', ''
  IDlist, CLASSlist, TAGlist = [], [], []

  for match in re.finditer(Tags_regexp, source):
      HTML += str(match.group())

  soup = BeautifulSoup(HTML)

  for ids in soup.findAll(attrs={'id' :  True}):
      IDlist.append(ids['id'])

  for classes in soup.findAll(attrs={'class' : True}):
      CLASSlist.append(classes['class'])

  for tags in soup.body.findAll(True):
      TAGlist.append(tags.name)

  IDlist = list(set(IDlist))
  IDlist.sort()
  CLASSlist = list(set(CLASSlist))
  CLASSlist.sort()
  TAGlist = list(set(TAGlist))
  TAGlist.sort()

  for tgs in TAGlist:
      CSS += tgs + ' { }\n'

  for ids in IDlist:
      CSS += '#' + ids + ' { }\n'

  for cls in CLASSlist:
      CSS += '.' + cls + ' { }\n'

  OUT.write(CSS)
  OUT.close()
  FILE.close()

html2css("in.html","out.css")