Subversion Repositories SmartDukaan

'''

Created on 24-Aug-2011

@author: Varun Gupta

'''

from BeautifulSoup import BeautifulSoup

import urllib

class LetsBuyScraper:

    def __init__(self):

        self.url = None

        self.id = None

    def setUrl(self, url):

        self.url = url

    def scrape(self):

        sock = urllib.urlopen(self.url)

        html = sock.read()

        sock.close()

        self.soup = BeautifulSoup(html)

    def getPhonePrices(self):

        phone_prices = []

        for div in self.soup.findAll('div', {'class': "detailbox"}):

            name = div('h2')[0]('a')[0].string.strip()

            price = div.findAll('span', {'class': "text12_stb"})[0].string.strip()

            print name, price

            phone_prices.append({'name': str(name), 'price': str(price)})

        return phone_prices

    def getNextUrl(self):

        next_url = None

        for anchor in self.soup.findAll('a'):

            try:

                if anchor['title'].strip() == "Next Page":

                    next_url = anchor['href'].strip()

            except KeyError:

                pass

        return next_url

if __name__ == '__main__':

    s = LetsBuyScraper()

    s.setUrl('http://www.letsbuy.com/mobile-phones-mobiles-c-254_88?perpage=192')

    s.scrape()

    print s.getNextUrl()