gistfile1.py

    # my kingdom for a case statement
    # this should be a DEPTH-FIRST SEARCH AMIRITE
    def walk_tree(self, doc):
        headers = ['h1', 'h2', 'h3', 'h4', 'h5', 'h6']
        lists = ['ol', 'ul']
        paragraphs = ['p']
        specials = ['blockquote']
        passes = ['br', 'root', 'img', 'thead', 'tbody', 'tr', 'td', 'th']
        tables = ['table', ]
        for page_element in self.elements:
            if page_element.tag in headers:
                doc.handle_heading(page_element)
            elif page_element.tag in paragraphs:
                doc.handle_paragraph(page_element)
            elif page_element.tag in tables:
                doc.handle_table(page_element)
            elif page_element.tag in lists:
                doc.handle_lists(page_element)
            elif page_element.tag in specials:
                doc.handle_blockquote(page_element)
            elif page_element.tag in passes:
                pass
            else:
                print "ARGH UNHANDLED ELEMENT: "
                print page_element.tag, page_element.text