Generate static HTML files for a list of urls and the .htaccess rewrite rules for them.

build-static-cache.py

#!/usr/bin/env python

import os
import urllib2

baseUrl='https://www.thompsontee.com/'
cacheDir='static_cache'
htaccessFile='htaccess'

urls={
    'about-us.html': 'about-us',
    'blog_category_press.html': 'blog/category/press',
    'blog_category_thompson-tee-blog.html': 'blog/category/thompson-tee-blog',
    'frequently-asked-questions.html': 'frequently-asked-questions',
    'index.html': '',
    'made-in-the-usa.html': 'made-in-the-usa',
    'men.html': 'men.html',
    'privacy-policy-cookie-restriction-mode.html': 'privacy-policy-cookie-restriction-mode',
    'returns_asp.html': 'returns.asp',
    'shipping-deliveries.html': 'shipping-deliveries',
    'sweat-101.html': 'sweat-101',
    'technology.html': 'technology',
    'testimonials.html': 'testimonials',
    'women.html': 'women.html'
}

def main():
    print 'Make sure you\'ve disabled static caching on production so fresh html is generated.'
    setup();
    cache();
    htaccess();
    print 'Done.  Remeber to place the files on the server and update the htaccess rules.'

def setup():
    print 'Setting up...'
    if not os.path.isdir(cacheDir):
        print 'Cache directory does not exist, creating...'
        os.makedirs(cacheDir)
    for file in os.listdir(cacheDir):
        path = os.path.join(cacheDir, file)
        try:
            if os.path.isfile(path):
                os.unlink(path)
        except Exception as e:
            print 'Failed to clear the cache directory'
            print e
            exit(1)
    os.unlink(htaccessFile)

def cache():
    print 'Downloading files...'
    for file, url in urls.iteritems():
        response = urllib2.urlopen('%s%s' % (baseUrl, url))
        html = response.read()
        with open('%s/%s' % (cacheDir, file), 'w') as handle:
            handle.write(html.replace('</body>', '<!-- static cache --></body>'))

def htaccess():
    print 'Building .htaccess rules...'
    lines = []
    for file, url in urls.iteritems():
        lines.append('RewriteRule ^/?%s/?$ /static_cache/%s [L,NC]' % (url, file))
    with open(htaccessFile, 'w') as handle:
        handle.write('\n'.join(lines))
        print 'Rules written to %s' % htaccessFile

if __name__ == '__main__':
    main()