#!/usr/bin/env python import sys import xml.etree.ElementTree as ET def main(argv): root = ET.parse(argv[1]).getroot() head = root.find('ThML.head') body = root.find('ThML.body') for thing in list(body): parseSomething(thing) def parseSomething(thing): if(len(thing.tag) == 4 and thing.tag[:3] == 'div'): parseDiv(thing) elif (thing.tag == 'p'): parseP(thing) def parseDiv(div): indentLevel = int(div.tag[3:]) title = div.get('title') print('#' * indentLevel + ' ' + title) for child in list(div): parseSomething(child) def parseP(p): text = "".join([x for x in p.itertext()]) print(text.replace("\n", " ")) if __name__ == '__main__': main(sys.argv)