parse guitarplayer.ru commerce.guitars for prs

guitarplayer.py

# -*- coding: utf-8 -*-
from grab import Grab
import logging
import pickledb
from urlparse import urlparse, parse_qs

# logging.basicConfig(level=logging.DEBUG)
db = pickledb.load('guitarplayer.db', False)

keywords = ['prs', 'paul', 'reed', 'smith', u'прс']
blacklist = ['gibson', 'lespaul', 'les', 'ibanez']

g = Grab()
for i in range(0, 440, 40):
    g.go("http://forum.guitarplayer.ru/index.php?board=20.%s" % i)
    for i in g.doc.select('//div/span[@id]/a'):
        url = i.attr('href')
        title = i.text()
        topic = parse_qs(urlparse(url).query)['topic'][0]
        if db.get(topic) is None:
            db.set(topic, title)
            if any(word in title.lower() for word in keywords) and not any(word in title.lower() for word in blacklist): 
                print('%s %s' % (url, title))

db.dump()