cssStyleList.py (xml.minidom)

#! /usr/bin/env python
import os, glob
from xml.dom import minidom

path = "OEBPS/Text" # your mileage may vary
list = []
new_list = []

def cssList():

    for infile in glob.glob(os.path.join(path, '*html')):
        html = minidom.parse(infile)

        for node in html.getElementsByTagName('p'):
            list.append(node.getAttribute('class'))
            list.append(node.getAttribute('style'))

        for node in html.getElementsByTagName('div'):
            list.append(node.getAttribute('class'))
            list.append(node.getAttribute('style'))

        for node in html.getElementsByTagName('span'):
            list.append(node.getAttribute('class'))
            list.append(node.getAttribute('style'))

        for node in html.getElementsByTagName('i'):
            list.append(node.getAttribute('class'))
            list.append(node.getAttribute('style'))

        for node in html.getElementsByTagName('em'):
            list.append(node.getAttribute('class'))
            list.append(node.getAttribute('style'))

        for node in html.getElementsByTagName('strong'):
            list.append(node.getAttribute('class'))
            list.append(node.getAttribute('style'))

        for node in html.getElementsByTagName('b'):
            list.append(node.getAttribute('class'))
            list.append(node.getAttribute('style'))

        # add tags as required...

        for i in list:
            if i not in new_list:
                if i is not None:
                    if i:
                        new_list.append(i)
    return new_list

if __name__ == "__main__":
    for item in (cssList()):
        print item