A little script to pull all assets from 2 URL and compare the found asset URLs as well as their actual content.

compare_assets.py

#! /usr/bin/env python
import re
import sys
import difflib
import requests

from purl import URL


URL_REGEX = re.compile(
    r'(href|src)="(?P<domain>((http|ftp)s?:)?//[^/]*)?(?P<link>.*?\.(css|js)?)"')


def get_asset_dict(content, url):
    asset_dict = {}
    for result in URL_REGEX.findall(content):
        __, domain, __, __, path, __ = result
        if not domain:
            domain = "{}://{}".format(url.scheme(), url.netloc())
        full_url = "{}{}".format(domain, path)
        if full_url.startswith('//'):
            full_url = 'http:' + full_url
        asset_dict[path] = full_url


    return asset_dict


local_url = URL(sys.argv[1])
remote_url = URL(sys.argv[2])

print "Get local content from", local_url
local_content = requests.get(local_url.as_string()).content
print "Get remote content from", remote_url
remote_content = requests.get(remote_url.as_string()).content


local_assets = get_asset_dict(local_content, local_url)
remote_assets = get_asset_dict(remote_content, remote_url)

print "-" * 50
print "Checking assets for changes"
print "-" * 50

difference = list(difflib.context_diff(local_assets.keys(), remote_assets.keys()))
if difference:
    print "\n".join(difference)
else:
    print "No difference in included js/css files"

print '-' * 50

merged_assets = {}
for path in local_assets:

    print '-' * 50
    print "Diffing the file '{}' on both sites:".format(path)

    print "downloading", remote_assets[path]
    remote_asset = requests.get(remote_assets[path]).content.split('\n')

    print "downloading", local_assets[path]
    local_asset = requests.get(local_assets[path]).content.split('\n')

    result = difflib.context_diff(remote_asset, local_asset)
    if not result:
        print "NO DIFF for '{}'".format(path)
    else:
        print "\n".join(result)

    print '-' * 50