Subversion Repositories SmartDukaan

'''

Created on 24-Aug-2011

@author: Varun Gupta

'''

from BeautifulSoup import BeautifulSoup

import urllib

class FlipcartScraper:

    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 getPhones(self):

        phones = []

        for div in self.soup.findAll('div', {'class': 'fk-product-thumb fkp-medium'}):

            try:

                anchor = div.findAll('a', {'class': 'title fk-anchor-link'})[0]

                name = anchor['title'].strip()

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

                in_stock = 0 if div.findAll('b').__len__() > 0 else 1

                for span in div.findAll('span'):

                    try:

                        if span['class'].find('price final-price') > -1:

                            price = span.string.strip()

                    except KeyError:

                        pass

                phones.append({'name': str(name), 'price': str(price), 'product_url': str(product_url), 'in_stock': in_stock})

            except KeyError:

                pass

        return phones

    def getNextUrl(self):

        tab_info = self.soup.findAll('div', {'class': 'unit fk-lres-header-text'})[0]('b')

        current_max = int(tab_info[0].string.split('-')[1])

        total = int(tab_info[1].string)

        if current_max < total:

            return 'http://www.flipkart.com/mobiles/all/%d' % (1 + (current_max / 20))

        else:

            return None

if __name__ == '__main__':

    s = FlipcartScraper()

    s.setUrl('http://www.flipkart.com/mobiles/all/24')

    s.scrape()

    phones = s.getPhones()

    for p in phones: print p