ngram3.py

#!/usr/bin/python
# coding: utf-8

import urllib
from converter import unaccentedMap
from sgmllib import SGMLParser

down = urllib.urlopen('http://pl.wikipedia.org/wiki/Polska')
strona = unicode(down.read(), 'utf-8')
down.close()

strona = strona.translate(unaccentedMap())
#print strona

class wikiParser(SGMLParser):
    def __init__(self):
    self.f = 0
    def reset(self):
    self.pieces = []
    SGMLParser.reset(self)
    def start_p(self):
    self.f += 1
    def end_p(self):
    self.f -= 1
    def unknown_starttag(self, tag, attr):
    self.f -= 1
    def unknown_endtag(self, tag):
    self.f += 1
    def handle_data(self, text):
    if self.f > 0:
        self.pieces.append(text)
    def output(self):
    return ''.join(self.pieces)

parser = wikiParser()
parser.feed(str(strona))
parser.close()
print parser.output()