farts.py

import xml.sax


class FartHandler(xml.sax.ContentHandler):
    FART_LEVEL = 3  # the schema is bad and you should feel bad...

    def __init__(self):
        self.lvl = 0
        self.infarts = False
        self.curfart = {}
        self.chars = ""
        self.nodename = ""
        self.allfarts = []
        #print "started..."

    def startElement(self, name, attrs):
        #print "start element %s, lvl %d " % (name, self.lvl)
        self.lvl += 1
        if self.lvl == FartHandler.FART_LEVEL and name == "fart":
            #print "Entered a fart... level: %d" % self.lvl
            self.infarts = True

        if not self.infarts:
            return

        self.nodename = name
        #print "new node: " + name

    def characters(self, content):
        self.chars += content

    def endElement(self, name):
        self.lvl -= 1
        if self.infarts:
            if self.lvl < FartHandler.FART_LEVEL:
                if "content" not in self.curfart:
                    self.curfart = {}
                    return
                self.allfarts.append(self.curfart)
                self.curfart = {}
                #print "======Exited fart======="
            else:
                self.curfart[name] = self.chars
                self.chars = ""


fp = open("farts.xml", "r")
data = fp.read()
parser = xml.sax.make_parser()
Handler = FartHandler()
parser.setContentHandler( Handler )
parser.parse("farts.xml")
#print "Got %d farts" % len(Handler.allfarts)
search = "fart"
search = search.upper()
for fart in Handler.allfarts:
    if search in fart["content"].upper() or search in fart["title"].upper():
        print fart["content"]