html_dir   = "%s/organizations" % DATA_HTML_PATH
    html_files = os.listdir(html_dir)

    for file in html_files:
        file    = open(html_dir + "/" + file)
        content = file.read()
        
        #name    = BeautifulSoup(content).find('h1', { 'class': 'oldh1' })
        #name    = name.string.replace(" ", "")

        file.close()

import os

html_dir   = "%s/organizations" % DATA_HTML_PATH
for root, dirs, files in os.walk(html_dir):
    for name in files:
        content = open(os.path.join(root, name)).read()