Albanian

#!/usr/bin/env python3

import re
import json
from urllib.request import urlopen

start, finish = int(input('Start thread number: ')), int(input('Finish thread number: '))
url = 'http://www.forumishqiptar.com/threads/%d'

def find_posts(text):
    posts = re.findall('<h2 class="title icon">(.*?)</h2>.*?<div id="post_message_.*?">(.*?)</div>',text,flags = re.DOTALL)
    filtered = []
    for title, body in posts:
        title = re.sub('<.*?>','', title)
        title = re.sub('\t','', title)
        title = re.sub('\r','', title)
        body = re.sub('<.*?>','', body)
        body = re.sub('\t','', body)
        body = re.sub('\r','', body)
        filtered.append({
            'title': title,
            'body': body
        })
    return filtered

def get_thread(i):
    print('Download thread %d' % i)
    try:
        data = urlopen(url % i).read().decode('windows-1251')
        return find_posts(data)
    except Exception as e:
        print(e)
        return None


with open('albanian_threads.json', 'wb') as f:
    data = {}
    for i in range(start, finish + 1):
        thread = get_thread(i)
        if thread:
            data[url % i] = thread
    f.write(json.dumps(data).encode('utf-8'))