identify unused css selectors in a page, proof of concept

unusedselectors.py

#!/usr/bin/env python
"""identify unused css selectors."""
from functools import partial
import re
import sys
import urlparse

import cssutils
from lxml import cssselect
from lxml import html

remove_pseudo = partial(re.compile(r':(hover|link|visited|active)').sub, '')

def process(url):
    tree = html.parse(url)
    sheets = tree.xpath('//link[@rel="stylesheet"]/@href')
    for sheet in sheets:
        cssurl = urlparse.urljoin(url, sheet)
        process_stylesheet(tree, cssurl)

def process_stylesheet(doc, cssurl):
    print cssurl
    ss = cssutils.parseUrl(cssurl)
    rules = (rule for rule in ss if rule.type == rule.STYLE_RULE)
    for rule in rules:
        key = rule.selectorText
        count = match_rule(doc, rule)
        if not count:
            print ' -', key

def match_rule(doc, rule):
    selectors = (selector for selector in rule.selectorList)
    selectors = (remove_pseudo(s.selectorText) for s in selectors)
    selectors = (cssselect.CSSSelector(s) for s in selectors)
    return sum(len(s.evaluate(doc)) for s in selectors)

if __name__ == '__main__':
    process(sys.argv[1])